[git.ipfire.org] IPFire 2.x development tree branch, next, updated. 20b27af8634b88fa400b1779415fc92888fbb1d8
This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "IPFire 2.x development tree".
The branch, next has been updated via 20b27af8634b88fa400b1779415fc92888fbb1d8 (commit) via 2775ab9cc63708ea0370d5220d5ee13a4f94142b (commit) via 8a1a3bf3934c8a92c605135ee13db466ea3dcbf4 (commit) via 41410d197b3dbf75dfd05376ed623857adac8977 (commit) via a7b0967db425d27b349d436bf83e4a18143f6fe3 (commit) via 6b86307cac1ed49a3d88836fc5fa90eef98a33ca (commit) via 63fd135400da8e15a1a0519f377c20bad67a6d0e (commit) via 53a6b00c4fc7e863b8a1b22cc01f90b016184363 (commit) via 2c4b9c5004f0db606612eba3965846d4a2623296 (commit) via e2c723627c587275e274948d026ac41a80a1705a (commit) via c8f8bf328f4dd6567513f942c9c1db1dcf73ef93 (commit) via 50354ffe3a946f314b5bf4f8648fa14d14c14667 (commit) via a93bf696170d32a45ccc76b29f68ca9b4df9ddb3 (commit) via d57c6162cb2d00fd4a4989fa3fe6924db528bce1 (commit) via 4071b2d61bdf3e284395d80a06189d5ae7752c5b (commit) via 0f5350608ed5790101caf94c19bf08a0a0c2118f (commit) via a235f229527896f76ffa9404f6b8fb75f19d522b (commit) via b1109b8af5f0a5e3ab7f0b68211d63ab0594c0ac (commit) from 22a504c69cb578f6e34928f3d862beac635e5224 (commit)
Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below.
- Log ----------------------------------------------------------------- commit 20b27af8634b88fa400b1779415fc92888fbb1d8 Author: Michael Tremer michael.tremer@ipfire.org Date: Fri Jan 22 00:55:46 2016 +0000
Update translations
Signed-off-by: Michael Tremer michael.tremer@ipfire.org
commit 2775ab9cc63708ea0370d5220d5ee13a4f94142b Author: Michael Tremer michael.tremer@ipfire.org Date: Fri Jan 22 00:55:25 2016 +0000
core97: Ship iptables conntrack changes
Signed-off-by: Michael Tremer michael.tremer@ipfire.org
commit 8a1a3bf3934c8a92c605135ee13db466ea3dcbf4 Merge: 41410d1 6b86307 Author: Michael Tremer michael.tremer@ipfire.org Date: Fri Jan 22 00:54:14 2016 +0000
Merge remote-tracking branch 'ms/iptables-conntrack' into next
commit 41410d197b3dbf75dfd05376ed623857adac8977 Merge: 22a504c a7b0967 Author: Michael Tremer michael.tremer@ipfire.org Date: Fri Jan 22 00:49:15 2016 +0000
Merge branch 'hyper-v-fixes' into next
commit a7b0967db425d27b349d436bf83e4a18143f6fe3 Author: Michael Tremer michael.tremer@ipfire.org Date: Tue Dec 15 18:30:56 2015 +0000
linux: Backport Hyper-V network driver
Signed-off-by: Michael Tremer michael.tremer@ipfire.org
commit 6b86307cac1ed49a3d88836fc5fa90eef98a33ca Author: Michael Tremer michael.tremer@ipfire.org Date: Thu Aug 20 23:05:40 2015 +0100
firewall options: Fix translation and spelling of new ALG section
Signed-off-by: Michael Tremer michael.tremer@ipfire.org
commit 63fd135400da8e15a1a0519f377c20bad67a6d0e Author: Alexander Marx alexander.marx@ipfire.org Date: Thu Aug 20 10:56:54 2015 +0200
BUG10844 add new options to firewalloptions for conntrack
simple changes, so far no languagefile changes because of simple protocol names. should be clear in all languages.
Signed-off-by: Michael Tremer michael.tremer@ipfire.org
commit 53a6b00c4fc7e863b8a1b22cc01f90b016184363 Author: Michael Tremer michael.tremer@ipfire.org Date: Wed Aug 12 12:46:07 2015 +0100
firewall: Disable the PPTP and AMANDA conntrack helpers by default
These do not seem to work at the moment.
Signed-off-by: Michael Tremer michael.tremer@ipfire.org
commit 2c4b9c5004f0db606612eba3965846d4a2623296 Author: Michael Tremer michael.tremer@ipfire.org Date: Wed Aug 12 12:44:26 2015 +0100
firewall: Fix amanda helper
This helper requires setting a layer 4 protocol.
Signed-off-by: Michael Tremer michael.tremer@ipfire.org
commit e2c723627c587275e274948d026ac41a80a1705a Author: Michael Tremer michael.tremer@ipfire.org Date: Wed Aug 12 12:18:18 2015 +0100
firewall: Fix H.323 helpers
Signed-off-by: Michael Tremer michael.tremer@ipfire.org
commit c8f8bf328f4dd6567513f942c9c1db1dcf73ef93 Author: Michael Tremer michael.tremer@ipfire.org Date: Tue May 12 13:33:08 2015 +0200
firewall: Add H.323 to the conntrack helpers
commit 50354ffe3a946f314b5bf4f8648fa14d14c14667 Author: Michael Tremer michael.tremer@ipfire.org Date: Tue May 12 13:27:24 2015 +0200
firewall: Add IRC to the conntrack helpers
commit a93bf696170d32a45ccc76b29f68ca9b4df9ddb3 Author: Michael Tremer michael.tremer@ipfire.org Date: Tue May 12 13:25:04 2015 +0200
firewall: Add amanda to the conntrack helpers
commit d57c6162cb2d00fd4a4989fa3fe6924db528bce1 Author: Michael Tremer michael.tremer@ipfire.org Date: Tue May 12 13:16:40 2015 +0200
firewall: Make conntrack helpers configurable
commit 4071b2d61bdf3e284395d80a06189d5ae7752c5b Author: Michael Tremer michael.tremer@ipfire.org Date: Mon May 11 13:04:14 2015 +0200
firewall: iptables will load the conntrack modules automatically
commit 0f5350608ed5790101caf94c19bf08a0a0c2118f Author: Michael Tremer michael.tremer@ipfire.org Date: Mon May 11 13:00:34 2015 +0200
firewall: Accept related ICMP packets again
This rule is required to forward ICMP error messages for aborted TCP connections and the like.
commit a235f229527896f76ffa9404f6b8fb75f19d522b Author: Michael Tremer michael.tremer@ipfire.org Date: Wed Apr 22 18:13:56 2015 +0200
firewall: Remove option to disable the SIP ALG
commit b1109b8af5f0a5e3ab7f0b68211d63ab0594c0ac Author: Michael Tremer michael.tremer@ipfire.org Date: Wed Apr 22 18:10:59 2015 +0200
Enhance the security of the netfilter conntrack helpers
This is suggested here https://home.regit.org/netfilter-en/secure-use-of-helpers/ and deprecated in the kernel (#10665).
-----------------------------------------------------------------------
Summary of changes: config/etc/modprobe.d/nf_conntrack.conf | 2 + config/rootfiles/common/stage2 | 1 + config/rootfiles/core/97/filelists/files | 3 + doc/language_issues.de | 2 - doc/language_issues.en | 2 - doc/language_issues.es | 3 +- doc/language_issues.fr | 3 +- doc/language_issues.it | 3 +- doc/language_issues.nl | 3 +- doc/language_issues.pl | 3 +- doc/language_issues.ru | 3 +- doc/language_issues.tr | 3 +- doc/language_missings | 4 + html/cgi-bin/optionsfw.cgi | 32 +- langs/de/cgi-bin/de.pl | 1 + langs/en/cgi-bin/en.pl | 1 + lfs/configroot | 10 + lfs/linux | 28 +- src/initscripts/init.d/firewall | 85 +- ...-hyperv-Get-rid-of-the-rndis_filter_packe.patch | 129 +++ ...support-for-virtual-Receive-Side-Scaling-.patch | 917 +++++++++++++++++++++ ...ivers-net-hyperv-Cleanup-the-receive-path.patch | 112 +++ ...0002-hyperv-Remove-recv_pkt_list-and-lock.patch | 384 +++++++++ ...-hyperv-Cleanup-the-netvsc-receive-callba.patch | 101 +++ ...rv-Simplify-the-send_completion-variables.patch | 105 +++ ...latest-NetVSP-versions-to-auto-negotiatio.patch | 171 ++++ ...Enable-sendbuf-mechanism-on-the-send-path.patch | 407 +++++++++ ...-for-netvsc-build-without-CONFIG_SYSFS-fl.patch | 42 + ...vers-net-hyperv-Enable-scatter-gather-I-O.patch | 212 +++++ ...-Drivers-net-hyperv-Cleanup-the-send-path.patch | 266 ++++++ .../0006-net-get-rid-of-SET_ETHTOOL_OPS.patch | 44 + ...rs-net-hyperv-Enable-offloads-on-the-host.patch | 196 +++++ ...Add-hash-value-into-RNDIS-Per-packet-info.patch | 93 +++ ...-hyperv-Enable-receive-side-IP-checksum-o.patch | 147 ++++ ...apparent-cut-n-paste-error-in-send-path-t.patch | 32 + ...-hyperv-Enable-send-side-checksum-offload.patch | 140 ++++ ...-Fix-error-return-code-in-netvsc_init_buf.patch | 34 + ...vers-net-hyperv-Enable-large-send-offload.patch | 153 ++++ ...010-hyperv-Fix-a-bug-in-netvsc_start_xmit.patch | 47 ++ ...e-the-receive-buffer-size-for-legacy-hos.patch} | 14 +- .../0011-hyperv-Fix-a-bug-in-netvsc_send.patch | 68 ++ ...-hyperv-Allocate-memory-for-all-possible-.patch | 34 + ...-hyperv-Negotiate-suitable-ndis-version-f.patch | 32 + ...rs-net-hyperv-Address-UDP-checksum-issues.patch | 111 +++ ...5-hyperv-Properly-handle-checksum-offload.patch | 42 + ...support-for-virtual-Receive-Side-Scaling-.patch | 917 +++++++++++++++++++++ ...0017-hyperv-Remove-recv_pkt_list-and-lock.patch | 384 +++++++++ ...rv-Simplify-the-send_completion-variables.patch | 105 +++ ...Enable-sendbuf-mechanism-on-the-send-path.patch | 407 +++++++++ ...-for-netvsc-build-without-CONFIG_SYSFS-fl.patch | 42 + ...Add-hash-value-into-RNDIS-Per-packet-info.patch | 93 +++ ...apparent-cut-n-paste-error-in-send-path-t.patch | 32 + ...-Fix-error-return-code-in-netvsc_init_buf.patch | 34 + .../0024-hyperv-Fix-a-bug-in-netvsc_send.patch | 68 ++ ...hv-vmbus-Support-per-channel-driver-state.patch | 51 ++ 55 files changed, 6312 insertions(+), 46 deletions(-) create mode 100644 config/etc/modprobe.d/nf_conntrack.conf create mode 100644 src/patches/linux/0001-Drivers-net-hyperv-Get-rid-of-the-rndis_filter_packe.patch create mode 100644 src/patches/linux/0001-hyperv-Add-support-for-virtual-Receive-Side-Scaling-.patch create mode 100644 src/patches/linux/0002-Drivers-net-hyperv-Cleanup-the-receive-path.patch create mode 100644 src/patches/linux/0002-hyperv-Remove-recv_pkt_list-and-lock.patch create mode 100644 src/patches/linux/0003-Drivers-net-hyperv-Cleanup-the-netvsc-receive-callba.patch create mode 100644 src/patches/linux/0003-hyperv-Simplify-the-send_completion-variables.patch create mode 100644 src/patches/linux/0004-hyperv-Add-latest-NetVSP-versions-to-auto-negotiatio.patch create mode 100644 src/patches/linux/0004-hyperv-Enable-sendbuf-mechanism-on-the-send-path.patch create mode 100644 src/patches/linux/0005-Add-support-for-netvsc-build-without-CONFIG_SYSFS-fl.patch create mode 100644 src/patches/linux/0005-Drivers-net-hyperv-Enable-scatter-gather-I-O.patch create mode 100644 src/patches/linux/0006-Drivers-net-hyperv-Cleanup-the-send-path.patch create mode 100644 src/patches/linux/0006-net-get-rid-of-SET_ETHTOOL_OPS.patch create mode 100644 src/patches/linux/0007-Drivers-net-hyperv-Enable-offloads-on-the-host.patch create mode 100644 src/patches/linux/0007-hyperv-Add-hash-value-into-RNDIS-Per-packet-info.patch create mode 100644 src/patches/linux/0008-Drivers-net-hyperv-Enable-receive-side-IP-checksum-o.patch create mode 100644 src/patches/linux/0008-hyperv-fix-apparent-cut-n-paste-error-in-send-path-t.patch create mode 100644 src/patches/linux/0009-Drivers-net-hyperv-Enable-send-side-checksum-offload.patch create mode 100644 src/patches/linux/0009-hyperv-Fix-error-return-code-in-netvsc_init_buf.patch create mode 100644 src/patches/linux/0010-Drivers-net-hyperv-Enable-large-send-offload.patch create mode 100644 src/patches/linux/0010-hyperv-Fix-a-bug-in-netvsc_start_xmit.patch rename src/patches/{linux-3.14.x-hyperv-2008-fix.patch => linux/0011-hyperv-Change-the-receive-buffer-size-for-legacy-hos.patch} (81%) create mode 100644 src/patches/linux/0011-hyperv-Fix-a-bug-in-netvsc_send.patch create mode 100644 src/patches/linux/0012-Drivers-net-hyperv-Allocate-memory-for-all-possible-.patch create mode 100644 src/patches/linux/0013-Drivers-net-hyperv-Negotiate-suitable-ndis-version-f.patch create mode 100644 src/patches/linux/0014-Drivers-net-hyperv-Address-UDP-checksum-issues.patch create mode 100644 src/patches/linux/0015-hyperv-Properly-handle-checksum-offload.patch create mode 100644 src/patches/linux/0016-hyperv-Add-support-for-virtual-Receive-Side-Scaling-.patch create mode 100644 src/patches/linux/0017-hyperv-Remove-recv_pkt_list-and-lock.patch create mode 100644 src/patches/linux/0018-hyperv-Simplify-the-send_completion-variables.patch create mode 100644 src/patches/linux/0019-hyperv-Enable-sendbuf-mechanism-on-the-send-path.patch create mode 100644 src/patches/linux/0020-Add-support-for-netvsc-build-without-CONFIG_SYSFS-fl.patch create mode 100644 src/patches/linux/0021-hyperv-Add-hash-value-into-RNDIS-Per-packet-info.patch create mode 100644 src/patches/linux/0022-hyperv-fix-apparent-cut-n-paste-error-in-send-path-t.patch create mode 100644 src/patches/linux/0023-hyperv-Fix-error-return-code-in-netvsc_init_buf.patch create mode 100644 src/patches/linux/0024-hyperv-Fix-a-bug-in-netvsc_send.patch create mode 100644 src/patches/linux/0025-Drivers-hv-vmbus-Support-per-channel-driver-state.patch
Difference in files: diff --git a/config/etc/modprobe.d/nf_conntrack.conf b/config/etc/modprobe.d/nf_conntrack.conf new file mode 100644 index 0000000..d5a1813 --- /dev/null +++ b/config/etc/modprobe.d/nf_conntrack.conf @@ -0,0 +1,2 @@ +# Disable automatic conntrack helper assignment +options nf_conntrack nf_conntrack_helper=0 diff --git a/config/rootfiles/common/stage2 b/config/rootfiles/common/stage2 index 5b763fd..07446b7 100644 --- a/config/rootfiles/common/stage2 +++ b/config/rootfiles/common/stage2 @@ -22,6 +22,7 @@ etc/mime.types etc/modprobe.d etc/modprobe.d/btmrvl_sdio.conf etc/modprobe.d/cfg80211.conf +etc/modprobe.d/nf_conntrack.conf etc/modprobe.d/pcspeaker.conf etc/modules.conf etc/mtab diff --git a/config/rootfiles/core/97/filelists/files b/config/rootfiles/core/97/filelists/files index 40bbbcf..3d9d483 100644 --- a/config/rootfiles/core/97/filelists/files +++ b/config/rootfiles/core/97/filelists/files @@ -1,5 +1,8 @@ etc/system-release etc/issue +etc/rc.d/init.d/firewall +etc/modprobe.d/nf_conntrack.conf +srv/web/ipfire/cgi-bin/optionsfw.cgi srv/web/ipfire/cgi-bin/ovpnmain.cgi srv/web/ipfire/cgi-bin/qos.cgi srv/web/ipfire/cgi-bin/webaccess.cgi diff --git a/doc/language_issues.de b/doc/language_issues.de index 3660356..421c40e 100644 --- a/doc/language_issues.de +++ b/doc/language_issues.de @@ -34,13 +34,11 @@ WARNING: translation string unused: advproxy chgwebpwd new password WARNING: translation string unused: advproxy chgwebpwd new password confirm WARNING: translation string unused: advproxy chgwebpwd old password WARNING: translation string unused: advproxy chgwebpwd username -WARNING: translation string unused: advproxy cre disabled WARNING: translation string unused: advproxy errmsg change fail WARNING: translation string unused: advproxy errmsg change success WARNING: translation string unused: advproxy errmsg invalid user WARNING: translation string unused: advproxy errmsg no password WARNING: translation string unused: advproxy errmsg password incorrect -WARNING: translation string unused: advproxy no cre groups WARNING: translation string unused: advproxy ssadvanced proxy WARNING: translation string unused: advproxy update notification WARNING: translation string unused: again diff --git a/doc/language_issues.en b/doc/language_issues.en index a419afa..0a1756f 100644 --- a/doc/language_issues.en +++ b/doc/language_issues.en @@ -33,13 +33,11 @@ WARNING: translation string unused: advproxy chgwebpwd new password WARNING: translation string unused: advproxy chgwebpwd new password confirm WARNING: translation string unused: advproxy chgwebpwd old password WARNING: translation string unused: advproxy chgwebpwd username -WARNING: translation string unused: advproxy cre disabled WARNING: translation string unused: advproxy errmsg change fail WARNING: translation string unused: advproxy errmsg change success WARNING: translation string unused: advproxy errmsg invalid user WARNING: translation string unused: advproxy errmsg no password WARNING: translation string unused: advproxy errmsg password incorrect -WARNING: translation string unused: advproxy no cre groups WARNING: translation string unused: advproxy ssadvanced proxy WARNING: translation string unused: advproxy update information WARNING: translation string unused: advproxy update notification diff --git a/doc/language_issues.es b/doc/language_issues.es index d375f69..17347f6 100644 --- a/doc/language_issues.es +++ b/doc/language_issues.es @@ -33,13 +33,11 @@ WARNING: translation string unused: advproxy chgwebpwd new password WARNING: translation string unused: advproxy chgwebpwd new password confirm WARNING: translation string unused: advproxy chgwebpwd old password WARNING: translation string unused: advproxy chgwebpwd username -WARNING: translation string unused: advproxy cre disabled WARNING: translation string unused: advproxy errmsg change fail WARNING: translation string unused: advproxy errmsg change success WARNING: translation string unused: advproxy errmsg invalid user WARNING: translation string unused: advproxy errmsg no password WARNING: translation string unused: advproxy errmsg password incorrect -WARNING: translation string unused: advproxy no cre groups WARNING: translation string unused: advproxy ssadvanced proxy WARNING: translation string unused: advproxy update information WARNING: translation string unused: advproxy update notification @@ -640,6 +638,7 @@ WARNING: untranslated string: advproxy group access control WARNING: untranslated string: advproxy group required WARNING: untranslated string: advproxy proxy port transparent WARNING: untranslated string: age second +WARNING: untranslated string: application layer gateways WARNING: untranslated string: atm device WARNING: untranslated string: attention WARNING: untranslated string: bit diff --git a/doc/language_issues.fr b/doc/language_issues.fr index f0f5ec4..a93453f 100644 --- a/doc/language_issues.fr +++ b/doc/language_issues.fr @@ -33,13 +33,11 @@ WARNING: translation string unused: advproxy chgwebpwd new password WARNING: translation string unused: advproxy chgwebpwd new password confirm WARNING: translation string unused: advproxy chgwebpwd old password WARNING: translation string unused: advproxy chgwebpwd username -WARNING: translation string unused: advproxy cre disabled WARNING: translation string unused: advproxy errmsg change fail WARNING: translation string unused: advproxy errmsg change success WARNING: translation string unused: advproxy errmsg invalid user WARNING: translation string unused: advproxy errmsg no password WARNING: translation string unused: advproxy errmsg password incorrect -WARNING: translation string unused: advproxy no cre groups WARNING: translation string unused: advproxy ssadvanced proxy WARNING: translation string unused: advproxy update information WARNING: translation string unused: advproxy update notification @@ -647,6 +645,7 @@ WARNING: untranslated string: advproxy group access control WARNING: untranslated string: advproxy group required WARNING: untranslated string: advproxy proxy port transparent WARNING: untranslated string: age second +WARNING: untranslated string: application layer gateways WARNING: untranslated string: atm device WARNING: untranslated string: attention WARNING: untranslated string: bit diff --git a/doc/language_issues.it b/doc/language_issues.it index 65643e8..776b932 100644 --- a/doc/language_issues.it +++ b/doc/language_issues.it @@ -33,13 +33,11 @@ WARNING: translation string unused: advproxy chgwebpwd new password WARNING: translation string unused: advproxy chgwebpwd new password confirm WARNING: translation string unused: advproxy chgwebpwd old password WARNING: translation string unused: advproxy chgwebpwd username -WARNING: translation string unused: advproxy cre disabled WARNING: translation string unused: advproxy errmsg change fail WARNING: translation string unused: advproxy errmsg change success WARNING: translation string unused: advproxy errmsg invalid user WARNING: translation string unused: advproxy errmsg no password WARNING: translation string unused: advproxy errmsg password incorrect -WARNING: translation string unused: advproxy no cre groups WARNING: translation string unused: advproxy ssadvanced proxy WARNING: translation string unused: advproxy update information WARNING: translation string unused: advproxy update notification @@ -700,6 +698,7 @@ WARNING: untranslated string: advproxy AUTH method ntlm auth WARNING: untranslated string: advproxy basic authentication WARNING: untranslated string: advproxy group access control WARNING: untranslated string: advproxy group required +WARNING: untranslated string: application layer gateways WARNING: untranslated string: bytes WARNING: untranslated string: check all WARNING: untranslated string: dhcp dns enable update diff --git a/doc/language_issues.nl b/doc/language_issues.nl index 3b57bdf..77fa1f5 100644 --- a/doc/language_issues.nl +++ b/doc/language_issues.nl @@ -33,13 +33,11 @@ WARNING: translation string unused: advproxy chgwebpwd new password WARNING: translation string unused: advproxy chgwebpwd new password confirm WARNING: translation string unused: advproxy chgwebpwd old password WARNING: translation string unused: advproxy chgwebpwd username -WARNING: translation string unused: advproxy cre disabled WARNING: translation string unused: advproxy errmsg change fail WARNING: translation string unused: advproxy errmsg change success WARNING: translation string unused: advproxy errmsg invalid user WARNING: translation string unused: advproxy errmsg no password WARNING: translation string unused: advproxy errmsg password incorrect -WARNING: translation string unused: advproxy no cre groups WARNING: translation string unused: advproxy ssadvanced proxy WARNING: translation string unused: advproxy update information WARNING: translation string unused: advproxy update notification @@ -697,6 +695,7 @@ WARNING: untranslated string: advproxy AUTH method ntlm auth WARNING: untranslated string: advproxy basic authentication WARNING: untranslated string: advproxy group access control WARNING: untranslated string: advproxy group required +WARNING: untranslated string: application layer gateways WARNING: untranslated string: atm device WARNING: untranslated string: bytes WARNING: untranslated string: capabilities diff --git a/doc/language_issues.pl b/doc/language_issues.pl index d375f69..17347f6 100644 --- a/doc/language_issues.pl +++ b/doc/language_issues.pl @@ -33,13 +33,11 @@ WARNING: translation string unused: advproxy chgwebpwd new password WARNING: translation string unused: advproxy chgwebpwd new password confirm WARNING: translation string unused: advproxy chgwebpwd old password WARNING: translation string unused: advproxy chgwebpwd username -WARNING: translation string unused: advproxy cre disabled WARNING: translation string unused: advproxy errmsg change fail WARNING: translation string unused: advproxy errmsg change success WARNING: translation string unused: advproxy errmsg invalid user WARNING: translation string unused: advproxy errmsg no password WARNING: translation string unused: advproxy errmsg password incorrect -WARNING: translation string unused: advproxy no cre groups WARNING: translation string unused: advproxy ssadvanced proxy WARNING: translation string unused: advproxy update information WARNING: translation string unused: advproxy update notification @@ -640,6 +638,7 @@ WARNING: untranslated string: advproxy group access control WARNING: untranslated string: advproxy group required WARNING: untranslated string: advproxy proxy port transparent WARNING: untranslated string: age second +WARNING: untranslated string: application layer gateways WARNING: untranslated string: atm device WARNING: untranslated string: attention WARNING: untranslated string: bit diff --git a/doc/language_issues.ru b/doc/language_issues.ru index 05d9e91..6446a74 100644 --- a/doc/language_issues.ru +++ b/doc/language_issues.ru @@ -33,13 +33,11 @@ WARNING: translation string unused: advproxy chgwebpwd new password WARNING: translation string unused: advproxy chgwebpwd new password confirm WARNING: translation string unused: advproxy chgwebpwd old password WARNING: translation string unused: advproxy chgwebpwd username -WARNING: translation string unused: advproxy cre disabled WARNING: translation string unused: advproxy errmsg change fail WARNING: translation string unused: advproxy errmsg change success WARNING: translation string unused: advproxy errmsg invalid user WARNING: translation string unused: advproxy errmsg no password WARNING: translation string unused: advproxy errmsg password incorrect -WARNING: translation string unused: advproxy no cre groups WARNING: translation string unused: advproxy ssadvanced proxy WARNING: translation string unused: advproxy update information WARNING: translation string unused: advproxy update notification @@ -640,6 +638,7 @@ WARNING: untranslated string: advproxy group access control WARNING: untranslated string: advproxy group required WARNING: untranslated string: advproxy proxy port transparent WARNING: untranslated string: age second +WARNING: untranslated string: application layer gateways WARNING: untranslated string: atm device WARNING: untranslated string: attention WARNING: untranslated string: bit diff --git a/doc/language_issues.tr b/doc/language_issues.tr index a419afa..5479859 100644 --- a/doc/language_issues.tr +++ b/doc/language_issues.tr @@ -33,13 +33,11 @@ WARNING: translation string unused: advproxy chgwebpwd new password WARNING: translation string unused: advproxy chgwebpwd new password confirm WARNING: translation string unused: advproxy chgwebpwd old password WARNING: translation string unused: advproxy chgwebpwd username -WARNING: translation string unused: advproxy cre disabled WARNING: translation string unused: advproxy errmsg change fail WARNING: translation string unused: advproxy errmsg change success WARNING: translation string unused: advproxy errmsg invalid user WARNING: translation string unused: advproxy errmsg no password WARNING: translation string unused: advproxy errmsg password incorrect -WARNING: translation string unused: advproxy no cre groups WARNING: translation string unused: advproxy ssadvanced proxy WARNING: translation string unused: advproxy update information WARNING: translation string unused: advproxy update notification @@ -703,6 +701,7 @@ WARNING: translation string unused: xtaccess bad transfert WARNING: translation string unused: year-graph WARNING: translation string unused: yearly firewallhits WARNING: untranslated string: Scan for Songs +WARNING: untranslated string: application layer gateways WARNING: untranslated string: bytes WARNING: untranslated string: fwhost cust geoipgrp WARNING: untranslated string: fwhost err hostip diff --git a/doc/language_missings b/doc/language_missings index c490f2d..32e1e48 100644 --- a/doc/language_missings +++ b/doc/language_missings @@ -26,6 +26,7 @@ < age shour < age sminute < age ssecond +< application layer gateways < atm device < attention < bit @@ -636,6 +637,7 @@ < age shour < age sminute < age ssecond +< application layer gateways < Async logging enabled < atm device < attention @@ -1229,6 +1231,7 @@ < age shour < age sminute < age ssecond +< application layer gateways < atm device < attention < bit @@ -1808,6 +1811,7 @@ < age shour < age sminute < age ssecond +< application layer gateways < atm device < attention < bit diff --git a/html/cgi-bin/optionsfw.cgi b/html/cgi-bin/optionsfw.cgi index 34e0cdc..7a0e8e0 100644 --- a/html/cgi-bin/optionsfw.cgi +++ b/html/cgi-bin/optionsfw.cgi @@ -120,6 +120,21 @@ $checked{'SHOWTABLES'}{$settings{'SHOWTABLES'}} = "checked='checked'"; $checked{'SHOWDROPDOWN'}{'off'} = ''; $checked{'SHOWDROPDOWN'}{'on'} = ''; $checked{'SHOWDROPDOWN'}{$settings{'SHOWDROPDOWN'}} = "checked='checked'"; +$checked{'CONNTRACK_FTP'}{'off'} = ''; +$checked{'CONNTRACK_FTP'}{'on'} = ''; +$checked{'CONNTRACK_FTP'}{$settings{'CONNTRACK_FTP'}} = "checked='checked'"; +$checked{'CONNTRACK_H323'}{'off'} = ''; +$checked{'CONNTRACK_H323'}{'on'} = ''; +$checked{'CONNTRACK_H323'}{$settings{'CONNTRACK_H323'}} = "checked='checked'"; +$checked{'CONNTRACK_IRC'}{'off'} = ''; +$checked{'CONNTRACK_IRC'}{'on'} = ''; +$checked{'CONNTRACK_IRC'}{$settings{'CONNTRACK_IRC'}} = "checked='checked'"; +$checked{'CONNTRACK_SIP'}{'off'} = ''; +$checked{'CONNTRACK_SIP'}{'on'} = ''; +$checked{'CONNTRACK_SIP'}{$settings{'CONNTRACK_SIP'}} = "checked='checked'"; +$checked{'CONNTRACK_TFTP'}{'off'} = ''; +$checked{'CONNTRACK_TFTP'}{'on'} = ''; +$checked{'CONNTRACK_TFTP'}{$settings{'CONNTRACK_TFTP'}} = "checked='checked'"; $selected{'FWPOLICY'}{$settings{'FWPOLICY'}}= 'selected'; $selected{'FWPOLICY1'}{$settings{'FWPOLICY1'}}= 'selected'; $selected{'FWPOLICY2'}{$settings{'FWPOLICY2'}}= 'selected'; @@ -223,7 +238,22 @@ END <input type='radio' name='SHOWTABLES' value='off' $checked{'SHOWTABLES'}{'off'} /> off</td></tr> <tr><td align='left' width='60%'>$Lang::tr{'fw settings dropdown'}</td><td align='left'>on <input type='radio' name='SHOWDROPDOWN' value='on' $checked{'SHOWDROPDOWN'}{'on'} />/ <input type='radio' name='SHOWDROPDOWN' value='off' $checked{'SHOWDROPDOWN'}{'off'} /> off</td></tr> -</table> +</table> +<br /> +<table width='95%' cellspacing='0'> +<tr bgcolor='$color{'color20'}'><td colspan='2' align='left'><b>$Lang::tr{'application layer gateways'}</b></td></tr> +<tr><td align='left' width='60%'>FTP</td><td align='left'>on <input type='radio' name='CONNTRACK_FTP' value='on' $checked{'CONNTRACK_FTP'}{'on'} />/ + <input type='radio' name='CONNTRACK_FTP' value='off' $checked{'CONNTRACK_FTP'}{'off'} /> off</td></tr> +<tr><td align='left' width='60%'>H.323</td><td align='left'>on <input type='radio' name='CONNTRACK_H323' value='on' $checked{'CONNTRACK_H323'}{'on'} />/ + <input type='radio' name='CONNTRACK_H323' value='off' $checked{'CONNTRACK_H323'}{'off'} /> off</td></tr> +<tr><td align='left' width='60%'>IRC</td><td align='left'>on <input type='radio' name='CONNTRACK_IRC' value='on' $checked{'CONNTRACK_IRC'}{'on'} />/ + <input type='radio' name='CONNTRACK_IRC' value='off' $checked{'CONNTRACK_IRC'}{'off'} /> off</td></tr> +<tr><td align='left' width='60%'>SIP</td><td align='left'>on <input type='radio' name='CONNTRACK_SIP' value='on' $checked{'CONNTRACK_SIP'}{'on'} />/ + <input type='radio' name='CONNTRACK_SIP' value='off' $checked{'CONNTRACK_SIP'}{'off'} /> off</td></tr> +<tr><td align='left' width='60%'>TFTP</td><td align='left'>on <input type='radio' name='CONNTRACK_TFTP' value='on' $checked{'CONNTRACK_TFTP'}{'on'} />/ + <input type='radio' name='CONNTRACK_TFTP' value='off' $checked{'CONNTRACK_TFTP'}{'off'} /> off</td></tr> + +</table> <br /> <table width='95%' cellspacing='0'> <tr bgcolor='$color{'color20'}'><td colspan='2' align='left'><b>$Lang::tr{'fw default drop'}</b></td></tr> diff --git a/langs/de/cgi-bin/de.pl b/langs/de/cgi-bin/de.pl index 2bca854..7c330bb 100644 --- a/langs/de/cgi-bin/de.pl +++ b/langs/de/cgi-bin/de.pl @@ -367,6 +367,7 @@ 'alt vpn' => 'VPNs', 'and' => 'Und', 'apcupsd' => 'APC-UPS Status', +'application layer gateways' => 'Application-Layer-Gateways', 'apply' => 'Jetzt anwenden', 'april' => 'April', 'archive not exist' => 'Konfigurationsarchiv existiert nicht', diff --git a/langs/en/cgi-bin/en.pl b/langs/en/cgi-bin/en.pl index 4c52392..23f6310 100644 --- a/langs/en/cgi-bin/en.pl +++ b/langs/en/cgi-bin/en.pl @@ -369,6 +369,7 @@ 'and' => 'And', 'ansi t1.483' => 'TO BE REMOVED', 'apcupsd' => 'APC-UPS status', +'application layer gateways' => 'Application Layer Gateways', 'apply' => 'Apply now', 'april' => 'April', 'archive not exist' => 'Configuration archive does not exist', diff --git a/lfs/configroot b/lfs/configroot index cb74996..f8e9ce4 100644 --- a/lfs/configroot +++ b/lfs/configroot @@ -135,6 +135,16 @@ $(TARGET) : echo "POLICY=MODE2" >> $(CONFIG_ROOT)/firewall/settings echo "POLICY1=MODE2" >> $(CONFIG_ROOT)/firewall/settings
+ # Add conntrack helper default settings + for proto in FTP H323 IRC SIP TFTP; do \ + echo "CONNTRACK_$${proto}=on" >> $(CONFIG_ROOT)/optionsfw/settings; \ + done + + # Do not enable these by default because these are broken + for proto in AMANDA PPTP; do \ + echo "CONNTRACK_$${proto}=off" >> $(CONFIG_ROOT)/optionsfw/settings; \ + done + # set converters executable chmod 755 /usr/sbin/convert-*
diff --git a/lfs/linux b/lfs/linux index 1a0ae10..3fd8d8c 100644 --- a/lfs/linux +++ b/lfs/linux @@ -170,8 +170,32 @@ endif # update the queued trim blacklist from kernel 4.2rc1 cd $(DIR_APP) && patch -Np1 < $(DIR_SRC)/src/patches/linux-3.14.43_new_qtrim_blacklist.patch
- # HyperV 2008 patch - cd $(DIR_APP) && patch -Np1 < $(DIR_SRC)/src/patches/linux-3.14.x-hyperv-2008-fix.patch + # Hyper-V patches + cd $(DIR_APP) && patch -Np1 < $(DIR_SRC)/src/patches/linux/0001-Drivers-net-hyperv-Get-rid-of-the-rndis_filter_packe.patch + cd $(DIR_APP) && patch -Np1 < $(DIR_SRC)/src/patches/linux/0002-Drivers-net-hyperv-Cleanup-the-receive-path.patch + cd $(DIR_APP) && patch -Np1 < $(DIR_SRC)/src/patches/linux/0003-Drivers-net-hyperv-Cleanup-the-netvsc-receive-callba.patch + cd $(DIR_APP) && patch -Np1 < $(DIR_SRC)/src/patches/linux/0004-hyperv-Add-latest-NetVSP-versions-to-auto-negotiatio.patch + cd $(DIR_APP) && patch -Np1 < $(DIR_SRC)/src/patches/linux/0005-Drivers-net-hyperv-Enable-scatter-gather-I-O.patch + cd $(DIR_APP) && patch -Np1 < $(DIR_SRC)/src/patches/linux/0006-Drivers-net-hyperv-Cleanup-the-send-path.patch + cd $(DIR_APP) && patch -Np1 < $(DIR_SRC)/src/patches/linux/0007-Drivers-net-hyperv-Enable-offloads-on-the-host.patch + cd $(DIR_APP) && patch -Np1 < $(DIR_SRC)/src/patches/linux/0008-Drivers-net-hyperv-Enable-receive-side-IP-checksum-o.patch + cd $(DIR_APP) && patch -Np1 < $(DIR_SRC)/src/patches/linux/0009-Drivers-net-hyperv-Enable-send-side-checksum-offload.patch + cd $(DIR_APP) && patch -Np1 < $(DIR_SRC)/src/patches/linux/0010-Drivers-net-hyperv-Enable-large-send-offload.patch + cd $(DIR_APP) && patch -Np1 < $(DIR_SRC)/src/patches/linux/0011-hyperv-Change-the-receive-buffer-size-for-legacy-hos.patch + cd $(DIR_APP) && patch -Np1 < $(DIR_SRC)/src/patches/linux/0012-Drivers-net-hyperv-Allocate-memory-for-all-possible-.patch + cd $(DIR_APP) && patch -Np1 < $(DIR_SRC)/src/patches/linux/0013-Drivers-net-hyperv-Negotiate-suitable-ndis-version-f.patch + cd $(DIR_APP) && patch -Np1 < $(DIR_SRC)/src/patches/linux/0014-Drivers-net-hyperv-Address-UDP-checksum-issues.patch + cd $(DIR_APP) && patch -Np1 < $(DIR_SRC)/src/patches/linux/0015-hyperv-Properly-handle-checksum-offload.patch + cd $(DIR_APP) && patch -Np1 < $(DIR_SRC)/src/patches/linux/0016-hyperv-Add-support-for-virtual-Receive-Side-Scaling-.patch + cd $(DIR_APP) && patch -Np1 < $(DIR_SRC)/src/patches/linux/0017-hyperv-Remove-recv_pkt_list-and-lock.patch + cd $(DIR_APP) && patch -Np1 < $(DIR_SRC)/src/patches/linux/0018-hyperv-Simplify-the-send_completion-variables.patch + cd $(DIR_APP) && patch -Np1 < $(DIR_SRC)/src/patches/linux/0019-hyperv-Enable-sendbuf-mechanism-on-the-send-path.patch + cd $(DIR_APP) && patch -Np1 < $(DIR_SRC)/src/patches/linux/0020-Add-support-for-netvsc-build-without-CONFIG_SYSFS-fl.patch + cd $(DIR_APP) && patch -Np1 < $(DIR_SRC)/src/patches/linux/0021-hyperv-Add-hash-value-into-RNDIS-Per-packet-info.patch + cd $(DIR_APP) && patch -Np1 < $(DIR_SRC)/src/patches/linux/0022-hyperv-fix-apparent-cut-n-paste-error-in-send-path-t.patch + cd $(DIR_APP) && patch -Np1 < $(DIR_SRC)/src/patches/linux/0023-hyperv-Fix-error-return-code-in-netvsc_init_buf.patch + cd $(DIR_APP) && patch -Np1 < $(DIR_SRC)/src/patches/linux/0024-hyperv-Fix-a-bug-in-netvsc_send.patch + cd $(DIR_APP) && patch -Np1 < $(DIR_SRC)/src/patches/linux/0025-Drivers-hv-vmbus-Support-per-channel-driver-state.patch
# fix empty symbol crc's cd $(DIR_APP) && patch -Np1 < $(DIR_SRC)/src/patches/linux-genksyms_fix_typeof_handling.patch diff --git a/src/initscripts/init.d/firewall b/src/initscripts/init.d/firewall index 6622071..cb52670 100644 --- a/src/initscripts/init.d/firewall +++ b/src/initscripts/init.d/firewall @@ -21,9 +21,11 @@ iptables_init() { iptables -F iptables -t nat -F iptables -t mangle -F + iptables -t raw -F iptables -X iptables -t nat -X iptables -t mangle -X + iptables -t raw -X
# Set up policies iptables -P INPUT DROP @@ -84,10 +86,71 @@ iptables_init() { iptables -A INPUT -p tcp -j BADTCP iptables -A FORWARD -p tcp -j BADTCP
- # Connection tracking chain + # Connection tracking chains iptables -N CONNTRACK - iptables -A CONNTRACK -m conntrack --ctstate ESTABLISHED,RELATED -j ACCEPT + iptables -A CONNTRACK -m conntrack --ctstate ESTABLISHED -j ACCEPT iptables -A CONNTRACK -m conntrack --ctstate INVALID -j DROP + iptables -A CONNTRACK -p icmp -m conntrack --ctstate RELATED -j ACCEPT + iptables -t raw -N CONNTRACK + iptables -t raw -A PREROUTING -j CONNTRACK + + # Conntrack helpers (https://home.regit.org/netfilter-en/secure-use-of-helpers/) + + # SIP + if [ "${CONNTRACK_SIP}" = "on" ]; then + iptables -A CONNTRACK -m conntrack --ctstate RELATED \ + -m helper --helper sip -j ACCEPT + for proto in udp tcp; do + iptables -t raw -A CONNTRACK -p "${proto}" --dport 5060 -j CT --helper sip + done + fi + + # H.323 + if [ "${CONNTRACK_H323}" = "on" ]; then + iptables -A CONNTRACK -m conntrack --ctstate RELATED \ + -m helper --helper h323 -j ACCEPT + + # Gatekeeper RAS + iptables -t raw -A CONNTRACK -p udp --dport 1719 -j CT --helper RAS + + # Q.931 + iptables -t raw -A CONNTRACK -p tcp --dport 1720 -j CT --helper Q.931 + fi + + # FTP + if [ "${CONNTRACK_FTP}" = "on" ]; then + iptables -A CONNTRACK -m conntrack --ctstate RELATED \ + -m helper --helper ftp -p tcp --dport 1024: -j ACCEPT + iptables -t raw -A CONNTRACK -p tcp --dport 21 -j CT --helper ftp + fi + + # PPTP + if [ "${CONNTRACK_PPTP}" = "on" ]; then + iptables -A CONNTRACK -m conntrack --ctstate RELATED \ + -m helper --helper pptp -j ACCEPT + iptables -t raw -A CONNTRACK -p udp --dport 1723 -j CT --helper pptp + fi + + # TFTP + if [ "${CONNTRACK_TFTP}" = "on" ]; then + iptables -A CONNTRACK -m conntrack --ctstate RELATED \ + -m helper --helper tftp -j ACCEPT + iptables -t raw -A CONNTRACK -p udp --dport 69 -j CT --helper tftp + fi + + # IRC + if [ "${CONNTRACK_IRC}" = "on" ]; then + iptables -A CONNTRACK -m conntrack --ctstate RELATED \ + -m helper --helper irc -j ACCEPT + iptables -t raw -A CONNTRACK -p tcp --dport 6667 -j CT --helper irc + fi + + # Amanda + if [ "${CONNTRACK_AMANDA}" = "on" ]; then + iptables -A CONNTRACK -m conntrack --ctstate RELATED \ + -m helper --helper amanda -j ACCEPT + iptables -t raw -A CONNTRACK -p tcp -j CT --helper amanda + fi
# Fix for braindead ISP's iptables -A FORWARD -p tcp --tcp-flags SYN,RST SYN -j TCPMSS --clamp-mss-to-pmtu @@ -400,24 +463,6 @@ iptables_red_down() { # See how we were called. case "$1" in start) - boot_mesg "Loading firewall modules into the kernel" - modprobe iptable_nat || failed=1 - for i in $(find /lib/modules/$(uname -r) -name nf_conntrack*); do - modprobe $(basename $i | cut -d. -f1) || failed=1 - done - for i in $(find /lib/modules/$(uname -r) -name nf_nat*); do - modprobe $(basename $i | cut -d. -f1) || failed=1 - done - (exit ${failed}) - evaluate_retval - - if [ -e /var/ipfire/main/disable_nf_sip ]; then - rmmod nf_nat_sip - rmmod nf_conntrack_sip - rmmod nf_nat_h323 - rmmod nf_conntrack_h323 - fi - boot_mesg "Setting up firewall" iptables_init evaluate_retval diff --git a/src/patches/linux-3.14.x-hyperv-2008-fix.patch b/src/patches/linux-3.14.x-hyperv-2008-fix.patch deleted file mode 100644 index e538e08..0000000 --- a/src/patches/linux-3.14.x-hyperv-2008-fix.patch +++ /dev/null @@ -1,50 +0,0 @@ -From 99d3016de4f2a29635f5382b0e9bd0e5f2151487 Mon Sep 17 00:00:00 2001 -From: Haiyang Zhang haiyangz@microsoft.com -Date: Sun, 9 Mar 2014 16:10:59 -0700 -Subject: hyperv: Change the receive buffer size for legacy hosts - -Due to a bug in the Hyper-V host verion 2008R2, we need to use a slightly smaller -receive buffer size, otherwise the buffer will not be accepted by the legacy hosts. - -Signed-off-by: Haiyang Zhang haiyangz@microsoft.com -Signed-off-by: David S. Miller davem@davemloft.net - -diff --git a/drivers/net/hyperv/hyperv_net.h b/drivers/net/hyperv/hyperv_net.h -index 7d06b49..13010b4 100644 ---- a/drivers/net/hyperv/hyperv_net.h -+++ b/drivers/net/hyperv/hyperv_net.h -@@ -513,6 +513,7 @@ struct nvsp_message { - #define NETVSC_MTU 65536 - - #define NETVSC_RECEIVE_BUFFER_SIZE (1024*1024*16) /* 16MB */ -+#define NETVSC_RECEIVE_BUFFER_SIZE_LEGACY (1024*1024*15) /* 15MB */ - - #define NETVSC_RECEIVE_BUFFER_ID 0xcafe - -diff --git a/drivers/net/hyperv/netvsc.c b/drivers/net/hyperv/netvsc.c -index 1a0280d..daddea2 100644 ---- a/drivers/net/hyperv/netvsc.c -+++ b/drivers/net/hyperv/netvsc.c -@@ -365,6 +365,11 @@ static int netvsc_connect_vsp(struct hv_device *device) - goto cleanup; - - /* Post the big receive buffer to NetVSP */ -+ if (net_device->nvsp_version <= NVSP_PROTOCOL_VERSION_2) -+ net_device->recv_buf_size = NETVSC_RECEIVE_BUFFER_SIZE_LEGACY; -+ else -+ net_device->recv_buf_size = NETVSC_RECEIVE_BUFFER_SIZE; -+ - ret = netvsc_init_recv_buf(device); - - cleanup: -@@ -898,7 +903,6 @@ int netvsc_device_add(struct hv_device *device, void *additional_info) - ndev = net_device->ndev; - - /* Initialize the NetVSC channel extension */ -- net_device->recv_buf_size = NETVSC_RECEIVE_BUFFER_SIZE; - spin_lock_init(&net_device->recv_pkt_list_lock); - - INIT_LIST_HEAD(&net_device->recv_pkt_list); --- -cgit v0.10.2 - diff --git a/src/patches/linux/0001-Drivers-net-hyperv-Get-rid-of-the-rndis_filter_packe.patch b/src/patches/linux/0001-Drivers-net-hyperv-Get-rid-of-the-rndis_filter_packe.patch new file mode 100644 index 0000000..2e00392 --- /dev/null +++ b/src/patches/linux/0001-Drivers-net-hyperv-Get-rid-of-the-rndis_filter_packe.patch @@ -0,0 +1,129 @@ +From f3f885fa684ff18fa4d223dc22b782f5e5d32560 Mon Sep 17 00:00:00 2001 +From: KY Srinivasan kys@microsoft.com +Date: Sun, 16 Feb 2014 16:38:43 -0800 +Subject: [PATCH 01/25] Drivers: net: hyperv: Get rid of the + rndis_filter_packet structure + +This structure is redundant; get rid of it make the code little more efficient - +get rid of the unnecessary indirection. + +Signed-off-by: K. Y. Srinivasan kys@microsoft.com +Reviewed-by: Haiyang Zhang haiyangz@microsoft.com +Signed-off-by: David S. Miller davem@davemloft.net +--- + drivers/net/hyperv/hyperv_net.h | 6 ------ + drivers/net/hyperv/netvsc_drv.c | 2 +- + drivers/net/hyperv/rndis_filter.c | 41 +++------------------------------------ + 3 files changed, 4 insertions(+), 45 deletions(-) + +diff --git a/drivers/net/hyperv/hyperv_net.h b/drivers/net/hyperv/hyperv_net.h +index 7b594ce3f21d..7645ba38bde8 100644 +--- a/drivers/net/hyperv/hyperv_net.h ++++ b/drivers/net/hyperv/hyperv_net.h +@@ -846,12 +846,6 @@ struct rndis_message { + }; + + +-struct rndis_filter_packet { +- void *completion_ctx; +- void (*completion)(void *context); +- struct rndis_message msg; +-}; +- + /* Handy macros */ + + /* get the size of an RNDIS message. Pass in the message type, */ +diff --git a/drivers/net/hyperv/netvsc_drv.c b/drivers/net/hyperv/netvsc_drv.c +index 3c1c33ceffba..28020f83ba6f 100644 +--- a/drivers/net/hyperv/netvsc_drv.c ++++ b/drivers/net/hyperv/netvsc_drv.c +@@ -156,7 +156,7 @@ static int netvsc_start_xmit(struct sk_buff *skb, struct net_device *net) + /* Allocate a netvsc packet based on # of frags. */ + packet = kzalloc(sizeof(struct hv_netvsc_packet) + + (num_pages * sizeof(struct hv_page_buffer)) + +- sizeof(struct rndis_filter_packet) + ++ sizeof(struct rndis_message) + + NDIS_VLAN_PPI_SIZE, GFP_ATOMIC); + if (!packet) { + /* out of memory, drop packet */ +diff --git a/drivers/net/hyperv/rndis_filter.c b/drivers/net/hyperv/rndis_filter.c +index b54fd257652b..6a9f6021f09c 100644 +--- a/drivers/net/hyperv/rndis_filter.c ++++ b/drivers/net/hyperv/rndis_filter.c +@@ -58,9 +58,6 @@ struct rndis_request { + u8 request_ext[RNDIS_EXT_LEN]; + }; + +-static void rndis_filter_send_completion(void *ctx); +- +- + static struct rndis_device *get_rndis_device(void) + { + struct rndis_device *device; +@@ -297,7 +294,7 @@ static void rndis_filter_receive_response(struct rndis_device *dev, + "rndis response buffer overflow " + "detected (size %u max %zu)\n", + resp->msg_len, +- sizeof(struct rndis_filter_packet)); ++ sizeof(struct rndis_message)); + + if (resp->ndis_msg_type == + RNDIS_MSG_RESET_C) { +@@ -917,17 +914,14 @@ int rndis_filter_close(struct hv_device *dev) + int rndis_filter_send(struct hv_device *dev, + struct hv_netvsc_packet *pkt) + { +- int ret; +- struct rndis_filter_packet *filter_pkt; + struct rndis_message *rndis_msg; + struct rndis_packet *rndis_pkt; + u32 rndis_msg_size; + bool isvlan = pkt->vlan_tci & VLAN_TAG_PRESENT; + + /* Add the rndis header */ +- filter_pkt = (struct rndis_filter_packet *)pkt->extension; ++ rndis_msg = (struct rndis_message *)pkt->extension; + +- rndis_msg = &filter_pkt->msg; + rndis_msg_size = RNDIS_MESSAGE_SIZE(struct rndis_packet); + if (isvlan) + rndis_msg_size += NDIS_VLAN_PPI_SIZE; +@@ -980,34 +974,5 @@ int rndis_filter_send(struct hv_device *dev, + pkt->page_buf[1].len = rndis_msg_size - pkt->page_buf[0].len; + } + +- /* Save the packet send completion and context */ +- filter_pkt->completion = pkt->completion.send.send_completion; +- filter_pkt->completion_ctx = +- pkt->completion.send.send_completion_ctx; +- +- /* Use ours */ +- pkt->completion.send.send_completion = rndis_filter_send_completion; +- pkt->completion.send.send_completion_ctx = filter_pkt; +- +- ret = netvsc_send(dev, pkt); +- if (ret != 0) { +- /* +- * Reset the completion to originals to allow retries from +- * above +- */ +- pkt->completion.send.send_completion = +- filter_pkt->completion; +- pkt->completion.send.send_completion_ctx = +- filter_pkt->completion_ctx; +- } +- +- return ret; +-} +- +-static void rndis_filter_send_completion(void *ctx) +-{ +- struct rndis_filter_packet *filter_pkt = ctx; +- +- /* Pass it back to the original handler */ +- filter_pkt->completion(filter_pkt->completion_ctx); ++ return netvsc_send(dev, pkt); + } +-- +2.4.3 + diff --git a/src/patches/linux/0001-hyperv-Add-support-for-virtual-Receive-Side-Scaling-.patch b/src/patches/linux/0001-hyperv-Add-support-for-virtual-Receive-Side-Scaling-.patch new file mode 100644 index 0000000..52c4da6 --- /dev/null +++ b/src/patches/linux/0001-hyperv-Add-support-for-virtual-Receive-Side-Scaling-.patch @@ -0,0 +1,917 @@ +From 5b54dac856cb5bd6f33f4159012773e4a33704f7 Mon Sep 17 00:00:00 2001 +From: Haiyang Zhang haiyangz@microsoft.com +Date: Mon, 21 Apr 2014 10:20:28 -0700 +Subject: [PATCH 01/11] hyperv: Add support for virtual Receive Side Scaling + (vRSS) + +This feature allows multiple channels to be used by each virtual NIC. +It is available on Hyper-V host 2012 R2. + +Signed-off-by: Haiyang Zhang haiyangz@microsoft.com +Reviewed-by: K. Y. Srinivasan kys@microsoft.com +Signed-off-by: David S. Miller davem@davemloft.net +--- + drivers/net/hyperv/hyperv_net.h | 110 +++++++++++++++++++++- + drivers/net/hyperv/netvsc.c | 136 +++++++++++++++++++++------ + drivers/net/hyperv/netvsc_drv.c | 103 ++++++++++++++++++++- + drivers/net/hyperv/rndis_filter.c | 189 +++++++++++++++++++++++++++++++++++++- + 4 files changed, 504 insertions(+), 34 deletions(-) + +diff --git a/drivers/net/hyperv/hyperv_net.h b/drivers/net/hyperv/hyperv_net.h +index d18f711d0b0c..57eb3f906d64 100644 +--- a/drivers/net/hyperv/hyperv_net.h ++++ b/drivers/net/hyperv/hyperv_net.h +@@ -28,6 +28,96 @@ + #include <linux/hyperv.h> + #include <linux/rndis.h> + ++/* RSS related */ ++#define OID_GEN_RECEIVE_SCALE_CAPABILITIES 0x00010203 /* query only */ ++#define OID_GEN_RECEIVE_SCALE_PARAMETERS 0x00010204 /* query and set */ ++ ++#define NDIS_OBJECT_TYPE_RSS_CAPABILITIES 0x88 ++#define NDIS_OBJECT_TYPE_RSS_PARAMETERS 0x89 ++ ++#define NDIS_RECEIVE_SCALE_CAPABILITIES_REVISION_2 2 ++#define NDIS_RECEIVE_SCALE_PARAMETERS_REVISION_2 2 ++ ++struct ndis_obj_header { ++ u8 type; ++ u8 rev; ++ u16 size; ++} __packed; ++ ++/* ndis_recv_scale_cap/cap_flag */ ++#define NDIS_RSS_CAPS_MESSAGE_SIGNALED_INTERRUPTS 0x01000000 ++#define NDIS_RSS_CAPS_CLASSIFICATION_AT_ISR 0x02000000 ++#define NDIS_RSS_CAPS_CLASSIFICATION_AT_DPC 0x04000000 ++#define NDIS_RSS_CAPS_USING_MSI_X 0x08000000 ++#define NDIS_RSS_CAPS_RSS_AVAILABLE_ON_PORTS 0x10000000 ++#define NDIS_RSS_CAPS_SUPPORTS_MSI_X 0x20000000 ++#define NDIS_RSS_CAPS_HASH_TYPE_TCP_IPV4 0x00000100 ++#define NDIS_RSS_CAPS_HASH_TYPE_TCP_IPV6 0x00000200 ++#define NDIS_RSS_CAPS_HASH_TYPE_TCP_IPV6_EX 0x00000400 ++ ++struct ndis_recv_scale_cap { /* NDIS_RECEIVE_SCALE_CAPABILITIES */ ++ struct ndis_obj_header hdr; ++ u32 cap_flag; ++ u32 num_int_msg; ++ u32 num_recv_que; ++ u16 num_indirect_tabent; ++} __packed; ++ ++ ++/* ndis_recv_scale_param flags */ ++#define NDIS_RSS_PARAM_FLAG_BASE_CPU_UNCHANGED 0x0001 ++#define NDIS_RSS_PARAM_FLAG_HASH_INFO_UNCHANGED 0x0002 ++#define NDIS_RSS_PARAM_FLAG_ITABLE_UNCHANGED 0x0004 ++#define NDIS_RSS_PARAM_FLAG_HASH_KEY_UNCHANGED 0x0008 ++#define NDIS_RSS_PARAM_FLAG_DISABLE_RSS 0x0010 ++ ++/* Hash info bits */ ++#define NDIS_HASH_FUNC_TOEPLITZ 0x00000001 ++#define NDIS_HASH_IPV4 0x00000100 ++#define NDIS_HASH_TCP_IPV4 0x00000200 ++#define NDIS_HASH_IPV6 0x00000400 ++#define NDIS_HASH_IPV6_EX 0x00000800 ++#define NDIS_HASH_TCP_IPV6 0x00001000 ++#define NDIS_HASH_TCP_IPV6_EX 0x00002000 ++ ++#define NDIS_RSS_INDIRECTION_TABLE_MAX_SIZE_REVISION_2 (128 * 4) ++#define NDIS_RSS_HASH_SECRET_KEY_MAX_SIZE_REVISION_2 40 ++ ++#define ITAB_NUM 128 ++#define HASH_KEYLEN NDIS_RSS_HASH_SECRET_KEY_MAX_SIZE_REVISION_2 ++extern u8 netvsc_hash_key[]; ++ ++struct ndis_recv_scale_param { /* NDIS_RECEIVE_SCALE_PARAMETERS */ ++ struct ndis_obj_header hdr; ++ ++ /* Qualifies the rest of the information */ ++ u16 flag; ++ ++ /* The base CPU number to do receive processing. not used */ ++ u16 base_cpu_number; ++ ++ /* This describes the hash function and type being enabled */ ++ u32 hashinfo; ++ ++ /* The size of indirection table array */ ++ u16 indirect_tabsize; ++ ++ /* The offset of the indirection table from the beginning of this ++ * structure ++ */ ++ u32 indirect_taboffset; ++ ++ /* The size of the hash secret key */ ++ u16 hashkey_size; ++ ++ /* The offset of the secret key from the beginning of this structure */ ++ u32 kashkey_offset; ++ ++ u32 processor_masks_offset; ++ u32 num_processor_masks; ++ u32 processor_masks_entry_size; ++}; ++ + /* Fwd declaration */ + struct hv_netvsc_packet; + struct ndis_tcp_ip_checksum_info; +@@ -39,6 +129,8 @@ struct xferpage_packet { + + /* # of netvsc packets this xfer packet contains */ + u32 count; ++ ++ struct vmbus_channel *channel; + }; + + /* +@@ -54,6 +146,9 @@ struct hv_netvsc_packet { + bool is_data_pkt; + u16 vlan_tci; + ++ u16 q_idx; ++ struct vmbus_channel *channel; ++ + /* + * Valid only for receives when we break a xfer page packet + * into multiple netvsc packets +@@ -120,6 +215,7 @@ void netvsc_linkstatus_callback(struct hv_device *device_obj, + int netvsc_recv_callback(struct hv_device *device_obj, + struct hv_netvsc_packet *packet, + struct ndis_tcp_ip_checksum_info *csum_info); ++void netvsc_channel_cb(void *context); + int rndis_filter_open(struct hv_device *dev); + int rndis_filter_close(struct hv_device *dev); + int rndis_filter_device_add(struct hv_device *dev, +@@ -522,6 +618,8 @@ struct nvsp_message { + + #define NETVSC_PACKET_SIZE 2048 + ++#define VRSS_SEND_TAB_SIZE 16 ++ + /* Per netvsc channel-specific */ + struct netvsc_device { + struct hv_device *dev; +@@ -555,10 +653,20 @@ struct netvsc_device { + + struct net_device *ndev; + ++ struct vmbus_channel *chn_table[NR_CPUS]; ++ u32 send_table[VRSS_SEND_TAB_SIZE]; ++ u32 num_chn; ++ atomic_t queue_sends[NR_CPUS]; ++ + /* Holds rndis device info */ + void *extension; +- /* The recive buffer for this device */ ++ ++ int ring_size; ++ ++ /* The primary channel callback buffer */ + unsigned char cb_buffer[NETVSC_PACKET_SIZE]; ++ /* The sub channel callback buffer */ ++ unsigned char *sub_cb_buf; + }; + + /* NdisInitialize message */ +diff --git a/drivers/net/hyperv/netvsc.c b/drivers/net/hyperv/netvsc.c +index f7629ecefa84..e7e77f12bc38 100644 +--- a/drivers/net/hyperv/netvsc.c ++++ b/drivers/net/hyperv/netvsc.c +@@ -422,6 +422,9 @@ int netvsc_device_remove(struct hv_device *device) + kfree(netvsc_packet); + } + ++ if (net_device->sub_cb_buf) ++ vfree(net_device->sub_cb_buf); ++ + kfree(net_device); + return 0; + } +@@ -461,7 +464,9 @@ static void netvsc_send_completion(struct netvsc_device *net_device, + (nvsp_packet->hdr.msg_type == + NVSP_MSG1_TYPE_SEND_RECV_BUF_COMPLETE) || + (nvsp_packet->hdr.msg_type == +- NVSP_MSG1_TYPE_SEND_SEND_BUF_COMPLETE)) { ++ NVSP_MSG1_TYPE_SEND_SEND_BUF_COMPLETE) || ++ (nvsp_packet->hdr.msg_type == ++ NVSP_MSG5_TYPE_SUBCHANNEL)) { + /* Copy the response back */ + memcpy(&net_device->channel_init_pkt, nvsp_packet, + sizeof(struct nvsp_message)); +@@ -469,28 +474,37 @@ static void netvsc_send_completion(struct netvsc_device *net_device, + } else if (nvsp_packet->hdr.msg_type == + NVSP_MSG1_TYPE_SEND_RNDIS_PKT_COMPLETE) { + int num_outstanding_sends; ++ u16 q_idx = 0; ++ struct vmbus_channel *channel = device->channel; ++ int queue_sends; + + /* Get the send context */ + nvsc_packet = (struct hv_netvsc_packet *)(unsigned long) + packet->trans_id; + + /* Notify the layer above us */ +- if (nvsc_packet) ++ if (nvsc_packet) { ++ q_idx = nvsc_packet->q_idx; ++ channel = nvsc_packet->channel; + nvsc_packet->completion.send.send_completion( + nvsc_packet->completion.send. + send_completion_ctx); ++ } + + num_outstanding_sends = + atomic_dec_return(&net_device->num_outstanding_sends); ++ queue_sends = atomic_dec_return(&net_device-> ++ queue_sends[q_idx]); + + if (net_device->destroy && num_outstanding_sends == 0) + wake_up(&net_device->wait_drain); + +- if (netif_queue_stopped(ndev) && !net_device->start_remove && +- (hv_ringbuf_avail_percent(&device->channel->outbound) +- > RING_AVAIL_PERCENT_HIWATER || +- num_outstanding_sends < 1)) +- netif_wake_queue(ndev); ++ if (netif_tx_queue_stopped(netdev_get_tx_queue(ndev, q_idx)) && ++ !net_device->start_remove && ++ (hv_ringbuf_avail_percent(&channel->outbound) > ++ RING_AVAIL_PERCENT_HIWATER || queue_sends < 1)) ++ netif_tx_wake_queue(netdev_get_tx_queue( ++ ndev, q_idx)); + } else { + netdev_err(ndev, "Unknown send completion packet type- " + "%d received!!\n", nvsp_packet->hdr.msg_type); +@@ -505,6 +519,7 @@ int netvsc_send(struct hv_device *device, + int ret = 0; + struct nvsp_message sendMessage; + struct net_device *ndev; ++ struct vmbus_channel *out_channel = NULL; + u64 req_id; + + net_device = get_outbound_net_device(device); +@@ -531,15 +546,20 @@ int netvsc_send(struct hv_device *device, + else + req_id = 0; + ++ out_channel = net_device->chn_table[packet->q_idx]; ++ if (out_channel == NULL) ++ out_channel = device->channel; ++ packet->channel = out_channel; ++ + if (packet->page_buf_cnt) { +- ret = vmbus_sendpacket_pagebuffer(device->channel, ++ ret = vmbus_sendpacket_pagebuffer(out_channel, + packet->page_buf, + packet->page_buf_cnt, + &sendMessage, + sizeof(struct nvsp_message), + req_id); + } else { +- ret = vmbus_sendpacket(device->channel, &sendMessage, ++ ret = vmbus_sendpacket(out_channel, &sendMessage, + sizeof(struct nvsp_message), + req_id, + VM_PKT_DATA_INBAND, +@@ -548,17 +568,24 @@ int netvsc_send(struct hv_device *device, + + if (ret == 0) { + atomic_inc(&net_device->num_outstanding_sends); +- if (hv_ringbuf_avail_percent(&device->channel->outbound) < ++ atomic_inc(&net_device->queue_sends[packet->q_idx]); ++ ++ if (hv_ringbuf_avail_percent(&out_channel->outbound) < + RING_AVAIL_PERCENT_LOWATER) { +- netif_stop_queue(ndev); ++ netif_tx_stop_queue(netdev_get_tx_queue( ++ ndev, packet->q_idx)); ++ + if (atomic_read(&net_device-> +- num_outstanding_sends) < 1) +- netif_wake_queue(ndev); ++ queue_sends[packet->q_idx]) < 1) ++ netif_tx_wake_queue(netdev_get_tx_queue( ++ ndev, packet->q_idx)); + } + } else if (ret == -EAGAIN) { +- netif_stop_queue(ndev); +- if (atomic_read(&net_device->num_outstanding_sends) < 1) { +- netif_wake_queue(ndev); ++ netif_tx_stop_queue(netdev_get_tx_queue( ++ ndev, packet->q_idx)); ++ if (atomic_read(&net_device->queue_sends[packet->q_idx]) < 1) { ++ netif_tx_wake_queue(netdev_get_tx_queue( ++ ndev, packet->q_idx)); + ret = -ENOSPC; + } + } else { +@@ -570,6 +597,7 @@ int netvsc_send(struct hv_device *device, + } + + static void netvsc_send_recv_completion(struct hv_device *device, ++ struct vmbus_channel *channel, + struct netvsc_device *net_device, + u64 transaction_id, u32 status) + { +@@ -587,7 +615,7 @@ static void netvsc_send_recv_completion(struct hv_device *device, + + retry_send_cmplt: + /* Send the completion */ +- ret = vmbus_sendpacket(device->channel, &recvcompMessage, ++ ret = vmbus_sendpacket(channel, &recvcompMessage, + sizeof(struct nvsp_message), transaction_id, + VM_PKT_COMP, 0); + if (ret == 0) { +@@ -618,6 +646,7 @@ static void netvsc_receive_completion(void *context) + { + struct hv_netvsc_packet *packet = context; + struct hv_device *device = packet->device; ++ struct vmbus_channel *channel; + struct netvsc_device *net_device; + u64 transaction_id = 0; + bool fsend_receive_comp = false; +@@ -649,6 +678,7 @@ static void netvsc_receive_completion(void *context) + */ + if (packet->xfer_page_pkt->count == 0) { + fsend_receive_comp = true; ++ channel = packet->xfer_page_pkt->channel; + transaction_id = packet->completion.recv.recv_completion_tid; + status = packet->xfer_page_pkt->status; + list_add_tail(&packet->xfer_page_pkt->list_ent, +@@ -662,12 +692,13 @@ static void netvsc_receive_completion(void *context) + + /* Send a receive completion for the xfer page packet */ + if (fsend_receive_comp) +- netvsc_send_recv_completion(device, net_device, transaction_id, +- status); ++ netvsc_send_recv_completion(device, channel, net_device, ++ transaction_id, status); + + } + + static void netvsc_receive(struct netvsc_device *net_device, ++ struct vmbus_channel *channel, + struct hv_device *device, + struct vmpacket_descriptor *packet) + { +@@ -748,7 +779,7 @@ static void netvsc_receive(struct netvsc_device *net_device, + spin_unlock_irqrestore(&net_device->recv_pkt_list_lock, + flags); + +- netvsc_send_recv_completion(device, net_device, ++ netvsc_send_recv_completion(device, channel, net_device, + vmxferpage_packet->d.trans_id, + NVSP_STAT_FAIL); + +@@ -759,6 +790,7 @@ static void netvsc_receive(struct netvsc_device *net_device, + xferpage_packet = (struct xferpage_packet *)listHead.next; + list_del(&xferpage_packet->list_ent); + xferpage_packet->status = NVSP_STAT_SUCCESS; ++ xferpage_packet->channel = channel; + + /* This is how much we can satisfy */ + xferpage_packet->count = count - 1; +@@ -800,10 +832,45 @@ static void netvsc_receive(struct netvsc_device *net_device, + + } + +-static void netvsc_channel_cb(void *context) ++ ++static void netvsc_send_table(struct hv_device *hdev, ++ struct vmpacket_descriptor *vmpkt) ++{ ++ struct netvsc_device *nvscdev; ++ struct net_device *ndev; ++ struct nvsp_message *nvmsg; ++ int i; ++ u32 count, *tab; ++ ++ nvscdev = get_outbound_net_device(hdev); ++ if (!nvscdev) ++ return; ++ ndev = nvscdev->ndev; ++ ++ nvmsg = (struct nvsp_message *)((unsigned long)vmpkt + ++ (vmpkt->offset8 << 3)); ++ ++ if (nvmsg->hdr.msg_type != NVSP_MSG5_TYPE_SEND_INDIRECTION_TABLE) ++ return; ++ ++ count = nvmsg->msg.v5_msg.send_table.count; ++ if (count != VRSS_SEND_TAB_SIZE) { ++ netdev_err(ndev, "Received wrong send-table size:%u\n", count); ++ return; ++ } ++ ++ tab = (u32 *)((unsigned long)&nvmsg->msg.v5_msg.send_table + ++ nvmsg->msg.v5_msg.send_table.offset); ++ ++ for (i = 0; i < count; i++) ++ nvscdev->send_table[i] = tab[i]; ++} ++ ++void netvsc_channel_cb(void *context) + { + int ret; +- struct hv_device *device = context; ++ struct vmbus_channel *channel = (struct vmbus_channel *)context; ++ struct hv_device *device; + struct netvsc_device *net_device; + u32 bytes_recvd; + u64 request_id; +@@ -812,14 +879,19 @@ static void netvsc_channel_cb(void *context) + int bufferlen = NETVSC_PACKET_SIZE; + struct net_device *ndev; + ++ if (channel->primary_channel != NULL) ++ device = channel->primary_channel->device_obj; ++ else ++ device = channel->device_obj; ++ + net_device = get_inbound_net_device(device); + if (!net_device) + return; + ndev = net_device->ndev; +- buffer = net_device->cb_buffer; ++ buffer = get_per_channel_state(channel); + + do { +- ret = vmbus_recvpacket_raw(device->channel, buffer, bufferlen, ++ ret = vmbus_recvpacket_raw(channel, buffer, bufferlen, + &bytes_recvd, &request_id); + if (ret == 0) { + if (bytes_recvd > 0) { +@@ -831,8 +903,12 @@ static void netvsc_channel_cb(void *context) + break; + + case VM_PKT_DATA_USING_XFER_PAGES: +- netvsc_receive(net_device, +- device, desc); ++ netvsc_receive(net_device, channel, ++ device, desc); ++ break; ++ ++ case VM_PKT_DATA_INBAND: ++ netvsc_send_table(device, desc); + break; + + default: +@@ -893,6 +969,8 @@ int netvsc_device_add(struct hv_device *device, void *additional_info) + goto cleanup; + } + ++ net_device->ring_size = ring_size; ++ + /* + * Coming into this function, struct net_device * is + * registered as the driver private data. +@@ -917,10 +995,12 @@ int netvsc_device_add(struct hv_device *device, void *additional_info) + } + init_completion(&net_device->channel_init_wait); + ++ set_per_channel_state(device->channel, net_device->cb_buffer); ++ + /* Open the channel */ + ret = vmbus_open(device->channel, ring_size * PAGE_SIZE, + ring_size * PAGE_SIZE, NULL, 0, +- netvsc_channel_cb, device); ++ netvsc_channel_cb, device->channel); + + if (ret != 0) { + netdev_err(ndev, "unable to open channel: %d\n", ret); +@@ -930,6 +1010,8 @@ int netvsc_device_add(struct hv_device *device, void *additional_info) + /* Channel is opened */ + pr_info("hv_netvsc channel opened successfully\n"); + ++ net_device->chn_table[0] = device->channel; ++ + /* Connect with the NetVsp */ + ret = netvsc_connect_vsp(device); + if (ret != 0) { +diff --git a/drivers/net/hyperv/netvsc_drv.c b/drivers/net/hyperv/netvsc_drv.c +index 31e55fba7cad..093cf3fc46b8 100644 +--- a/drivers/net/hyperv/netvsc_drv.c ++++ b/drivers/net/hyperv/netvsc_drv.c +@@ -101,7 +101,7 @@ static int netvsc_open(struct net_device *net) + return ret; + } + +- netif_start_queue(net); ++ netif_tx_start_all_queues(net); + + nvdev = hv_get_drvdata(device_obj); + rdev = nvdev->extension; +@@ -149,6 +149,88 @@ static void *init_ppi_data(struct rndis_message *msg, u32 ppi_size, + return ppi; + } + ++union sub_key { ++ u64 k; ++ struct { ++ u8 pad[3]; ++ u8 kb; ++ u32 ka; ++ }; ++}; ++ ++/* Toeplitz hash function ++ * data: network byte order ++ * return: host byte order ++ */ ++static u32 comp_hash(u8 *key, int klen, u8 *data, int dlen) ++{ ++ union sub_key subk; ++ int k_next = 4; ++ u8 dt; ++ int i, j; ++ u32 ret = 0; ++ ++ subk.k = 0; ++ subk.ka = ntohl(*(u32 *)key); ++ ++ for (i = 0; i < dlen; i++) { ++ subk.kb = key[k_next]; ++ k_next = (k_next + 1) % klen; ++ dt = data[i]; ++ for (j = 0; j < 8; j++) { ++ if (dt & 0x80) ++ ret ^= subk.ka; ++ dt <<= 1; ++ subk.k <<= 1; ++ } ++ } ++ ++ return ret; ++} ++ ++static bool netvsc_set_hash(u32 *hash, struct sk_buff *skb) ++{ ++ struct iphdr *iphdr; ++ int data_len; ++ bool ret = false; ++ ++ if (eth_hdr(skb)->h_proto != htons(ETH_P_IP)) ++ return false; ++ ++ iphdr = ip_hdr(skb); ++ ++ if (iphdr->version == 4) { ++ if (iphdr->protocol == IPPROTO_TCP) ++ data_len = 12; ++ else ++ data_len = 8; ++ *hash = comp_hash(netvsc_hash_key, HASH_KEYLEN, ++ (u8 *)&iphdr->saddr, data_len); ++ ret = true; ++ } ++ ++ return ret; ++} ++ ++static u16 netvsc_select_queue(struct net_device *ndev, struct sk_buff *skb, ++ void *accel_priv, select_queue_fallback_t fallback) ++{ ++ struct net_device_context *net_device_ctx = netdev_priv(ndev); ++ struct hv_device *hdev = net_device_ctx->device_ctx; ++ struct netvsc_device *nvsc_dev = hv_get_drvdata(hdev); ++ u32 hash; ++ u16 q_idx = 0; ++ ++ if (nvsc_dev == NULL || ndev->real_num_tx_queues <= 1) ++ return 0; ++ ++ if (netvsc_set_hash(&hash, skb)) ++ q_idx = nvsc_dev->send_table[hash % VRSS_SEND_TAB_SIZE] % ++ ndev->real_num_tx_queues; ++ ++ return q_idx; ++} ++ + static void netvsc_xmit_completion(void *context) + { + struct hv_netvsc_packet *packet = (struct hv_netvsc_packet *)context; +@@ -333,6 +415,8 @@ static int netvsc_start_xmit(struct sk_buff *skb, struct net_device *net) + + packet->vlan_tci = skb->vlan_tci; + ++ packet->q_idx = skb_get_queue_mapping(skb); ++ + packet->is_data_pkt = true; + packet->total_data_buflen = skb->len; + +@@ -554,6 +638,10 @@ int netvsc_recv_callback(struct hv_device *device_obj, + __vlan_hwaccel_put_tag(skb, htons(ETH_P_8021Q), + packet->vlan_tci); + ++ skb_record_rx_queue(skb, packet->xfer_page_pkt->channel-> ++ offermsg.offer.sub_channel_index % ++ net->real_num_rx_queues); ++ + net->stats.rx_packets++; + net->stats.rx_bytes += packet->total_data_buflen; + +@@ -602,7 +690,7 @@ static int netvsc_change_mtu(struct net_device *ndev, int mtu) + hv_set_drvdata(hdev, ndev); + device_info.ring_size = ring_size; + rndis_filter_device_add(hdev, &device_info); +- netif_wake_queue(ndev); ++ netif_tx_wake_all_queues(ndev); + + return 0; + } +@@ -648,6 +736,7 @@ static const struct net_device_ops device_ops = { + .ndo_change_mtu = netvsc_change_mtu, + .ndo_validate_addr = eth_validate_addr, + .ndo_set_mac_address = netvsc_set_mac_addr, ++ .ndo_select_queue = netvsc_select_queue, + }; + + /* +@@ -694,9 +783,11 @@ static int netvsc_probe(struct hv_device *dev, + struct net_device *net = NULL; + struct net_device_context *net_device_ctx; + struct netvsc_device_info device_info; ++ struct netvsc_device *nvdev; + int ret; + +- net = alloc_etherdev(sizeof(struct net_device_context)); ++ net = alloc_etherdev_mq(sizeof(struct net_device_context), ++ num_online_cpus()); + if (!net) + return -ENOMEM; + +@@ -729,6 +820,12 @@ static int netvsc_probe(struct hv_device *dev, + } + memcpy(net->dev_addr, device_info.mac_adr, ETH_ALEN); + ++ nvdev = hv_get_drvdata(dev); ++ netif_set_real_num_tx_queues(net, nvdev->num_chn); ++ netif_set_real_num_rx_queues(net, nvdev->num_chn); ++ dev_info(&dev->device, "real num tx,rx queues:%u, %u\n", ++ net->real_num_tx_queues, net->real_num_rx_queues); ++ + ret = register_netdev(net); + if (ret != 0) { + pr_err("Unable to register netdev.\n"); +diff --git a/drivers/net/hyperv/rndis_filter.c b/drivers/net/hyperv/rndis_filter.c +index 143a98caf618..d92cfbe43410 100644 +--- a/drivers/net/hyperv/rndis_filter.c ++++ b/drivers/net/hyperv/rndis_filter.c +@@ -31,7 +31,7 @@ + #include "hyperv_net.h" + + +-#define RNDIS_EXT_LEN 100 ++#define RNDIS_EXT_LEN PAGE_SIZE + struct rndis_request { + struct list_head list_ent; + struct completion wait_event; +@@ -94,6 +94,8 @@ static struct rndis_request *get_rndis_request(struct rndis_device *dev, + rndis_msg->ndis_msg_type = msg_type; + rndis_msg->msg_len = msg_len; + ++ request->pkt.q_idx = 0; ++ + /* + * Set the request id. This field is always after the rndis header for + * request/response packet types so we just used the SetRequest as a +@@ -509,6 +511,19 @@ static int rndis_filter_query_device(struct rndis_device *dev, u32 oid, + query->info_buflen = 0; + query->dev_vc_handle = 0; + ++ if (oid == OID_GEN_RECEIVE_SCALE_CAPABILITIES) { ++ struct ndis_recv_scale_cap *cap; ++ ++ request->request_msg.msg_len += ++ sizeof(struct ndis_recv_scale_cap); ++ query->info_buflen = sizeof(struct ndis_recv_scale_cap); ++ cap = (struct ndis_recv_scale_cap *)((unsigned long)query + ++ query->info_buf_offset); ++ cap->hdr.type = NDIS_OBJECT_TYPE_RSS_CAPABILITIES; ++ cap->hdr.rev = NDIS_RECEIVE_SCALE_CAPABILITIES_REVISION_2; ++ cap->hdr.size = sizeof(struct ndis_recv_scale_cap); ++ } ++ + ret = rndis_filter_send_request(dev, request); + if (ret != 0) + goto cleanup; +@@ -695,6 +710,89 @@ cleanup: + return ret; + } + ++u8 netvsc_hash_key[HASH_KEYLEN] = { ++ 0x6d, 0x5a, 0x56, 0xda, 0x25, 0x5b, 0x0e, 0xc2, ++ 0x41, 0x67, 0x25, 0x3d, 0x43, 0xa3, 0x8f, 0xb0, ++ 0xd0, 0xca, 0x2b, 0xcb, 0xae, 0x7b, 0x30, 0xb4, ++ 0x77, 0xcb, 0x2d, 0xa3, 0x80, 0x30, 0xf2, 0x0c, ++ 0x6a, 0x42, 0xb7, 0x3b, 0xbe, 0xac, 0x01, 0xfa ++}; ++ ++int rndis_filter_set_rss_param(struct rndis_device *rdev, int num_queue) ++{ ++ struct net_device *ndev = rdev->net_dev->ndev; ++ struct rndis_request *request; ++ struct rndis_set_request *set; ++ struct rndis_set_complete *set_complete; ++ u32 extlen = sizeof(struct ndis_recv_scale_param) + ++ 4*ITAB_NUM + HASH_KEYLEN; ++ struct ndis_recv_scale_param *rssp; ++ u32 *itab; ++ u8 *keyp; ++ int i, t, ret; ++ ++ request = get_rndis_request( ++ rdev, RNDIS_MSG_SET, ++ RNDIS_MESSAGE_SIZE(struct rndis_set_request) + extlen); ++ if (!request) ++ return -ENOMEM; ++ ++ set = &request->request_msg.msg.set_req; ++ set->oid = OID_GEN_RECEIVE_SCALE_PARAMETERS; ++ set->info_buflen = extlen; ++ set->info_buf_offset = sizeof(struct rndis_set_request); ++ set->dev_vc_handle = 0; ++ ++ rssp = (struct ndis_recv_scale_param *)(set + 1); ++ rssp->hdr.type = NDIS_OBJECT_TYPE_RSS_PARAMETERS; ++ rssp->hdr.rev = NDIS_RECEIVE_SCALE_PARAMETERS_REVISION_2; ++ rssp->hdr.size = sizeof(struct ndis_recv_scale_param); ++ rssp->flag = 0; ++ rssp->hashinfo = NDIS_HASH_FUNC_TOEPLITZ | NDIS_HASH_IPV4 | ++ NDIS_HASH_TCP_IPV4; ++ rssp->indirect_tabsize = 4*ITAB_NUM; ++ rssp->indirect_taboffset = sizeof(struct ndis_recv_scale_param); ++ rssp->hashkey_size = HASH_KEYLEN; ++ rssp->kashkey_offset = rssp->indirect_taboffset + ++ rssp->indirect_tabsize; ++ ++ /* Set indirection table entries */ ++ itab = (u32 *)(rssp + 1); ++ for (i = 0; i < ITAB_NUM; i++) ++ itab[i] = i % num_queue; ++ ++ /* Set hask key values */ ++ keyp = (u8 *)((unsigned long)rssp + rssp->kashkey_offset); ++ for (i = 0; i < HASH_KEYLEN; i++) ++ keyp[i] = netvsc_hash_key[i]; ++ ++ ++ ret = rndis_filter_send_request(rdev, request); ++ if (ret != 0) ++ goto cleanup; ++ ++ t = wait_for_completion_timeout(&request->wait_event, 5*HZ); ++ if (t == 0) { ++ netdev_err(ndev, "timeout before we got a set response...\n"); ++ /* can't put_rndis_request, since we may still receive a ++ * send-completion. ++ */ ++ return -ETIMEDOUT; ++ } else { ++ set_complete = &request->response_msg.msg.set_complete; ++ if (set_complete->status != RNDIS_STATUS_SUCCESS) { ++ netdev_err(ndev, "Fail to set RSS parameters:0x%x\n", ++ set_complete->status); ++ ret = -EINVAL; ++ } ++ } ++ ++cleanup: ++ put_rndis_request(rdev, request); ++ return ret; ++} ++ ++ + static int rndis_filter_query_device_link_status(struct rndis_device *dev) + { + u32 size = sizeof(u32); +@@ -886,6 +984,28 @@ static int rndis_filter_close_device(struct rndis_device *dev) + return ret; + } + ++static void netvsc_sc_open(struct vmbus_channel *new_sc) ++{ ++ struct netvsc_device *nvscdev; ++ u16 chn_index = new_sc->offermsg.offer.sub_channel_index; ++ int ret; ++ ++ nvscdev = hv_get_drvdata(new_sc->primary_channel->device_obj); ++ ++ if (chn_index >= nvscdev->num_chn) ++ return; ++ ++ set_per_channel_state(new_sc, nvscdev->sub_cb_buf + (chn_index - 1) * ++ NETVSC_PACKET_SIZE); ++ ++ ret = vmbus_open(new_sc, nvscdev->ring_size * PAGE_SIZE, ++ nvscdev->ring_size * PAGE_SIZE, NULL, 0, ++ netvsc_channel_cb, new_sc); ++ ++ if (ret == 0) ++ nvscdev->chn_table[chn_index] = new_sc; ++} ++ + int rndis_filter_device_add(struct hv_device *dev, + void *additional_info) + { +@@ -894,6 +1014,10 @@ int rndis_filter_device_add(struct hv_device *dev, + struct rndis_device *rndis_device; + struct netvsc_device_info *device_info = additional_info; + struct ndis_offload_params offloads; ++ struct nvsp_message *init_packet; ++ int t; ++ struct ndis_recv_scale_cap rsscap; ++ u32 rsscap_size = sizeof(struct ndis_recv_scale_cap); + + rndis_device = get_rndis_device(); + if (!rndis_device) +@@ -913,6 +1037,7 @@ int rndis_filter_device_add(struct hv_device *dev, + + /* Initialize the rndis device */ + net_device = hv_get_drvdata(dev); ++ net_device->num_chn = 1; + + net_device->extension = rndis_device; + rndis_device->net_dev = net_device; +@@ -952,7 +1077,6 @@ int rndis_filter_device_add(struct hv_device *dev, + if (ret) + goto err_dev_remv; + +- + rndis_filter_query_device_link_status(rndis_device); + + device_info->link_state = rndis_device->link_state; +@@ -961,7 +1085,66 @@ int rndis_filter_device_add(struct hv_device *dev, + rndis_device->hw_mac_adr, + device_info->link_state ? "down" : "up"); + +- return ret; ++ if (net_device->nvsp_version < NVSP_PROTOCOL_VERSION_5) ++ return 0; ++ ++ /* vRSS setup */ ++ memset(&rsscap, 0, rsscap_size); ++ ret = rndis_filter_query_device(rndis_device, ++ OID_GEN_RECEIVE_SCALE_CAPABILITIES, ++ &rsscap, &rsscap_size); ++ if (ret || rsscap.num_recv_que < 2) ++ goto out; ++ ++ net_device->num_chn = (num_online_cpus() < rsscap.num_recv_que) ? ++ num_online_cpus() : rsscap.num_recv_que; ++ if (net_device->num_chn == 1) ++ goto out; ++ ++ net_device->sub_cb_buf = vzalloc((net_device->num_chn - 1) * ++ NETVSC_PACKET_SIZE); ++ if (!net_device->sub_cb_buf) { ++ net_device->num_chn = 1; ++ dev_info(&dev->device, "No memory for subchannels.\n"); ++ goto out; ++ } ++ ++ vmbus_set_sc_create_callback(dev->channel, netvsc_sc_open); ++ ++ init_packet = &net_device->channel_init_pkt; ++ memset(init_packet, 0, sizeof(struct nvsp_message)); ++ init_packet->hdr.msg_type = NVSP_MSG5_TYPE_SUBCHANNEL; ++ init_packet->msg.v5_msg.subchn_req.op = NVSP_SUBCHANNEL_ALLOCATE; ++ init_packet->msg.v5_msg.subchn_req.num_subchannels = ++ net_device->num_chn - 1; ++ ret = vmbus_sendpacket(dev->channel, init_packet, ++ sizeof(struct nvsp_message), ++ (unsigned long)init_packet, ++ VM_PKT_DATA_INBAND, ++ VMBUS_DATA_PACKET_FLAG_COMPLETION_REQUESTED); ++ if (ret) ++ goto out; ++ t = wait_for_completion_timeout(&net_device->channel_init_wait, 5*HZ); ++ if (t == 0) { ++ ret = -ETIMEDOUT; ++ goto out; ++ } ++ if (init_packet->msg.v5_msg.subchn_comp.status != ++ NVSP_STAT_SUCCESS) { ++ ret = -ENODEV; ++ goto out; ++ } ++ net_device->num_chn = 1 + ++ init_packet->msg.v5_msg.subchn_comp.num_subchannels; ++ ++ vmbus_are_subchannels_present(dev->channel); ++ ++ ret = rndis_filter_set_rss_param(rndis_device, net_device->num_chn); ++ ++out: ++ if (ret) ++ net_device->num_chn = 1; ++ return 0; /* return 0 because primary channel can be used alone */ + + err_dev_remv: + rndis_filter_device_remove(dev); +-- +2.4.3 + diff --git a/src/patches/linux/0002-Drivers-net-hyperv-Cleanup-the-receive-path.patch b/src/patches/linux/0002-Drivers-net-hyperv-Cleanup-the-receive-path.patch new file mode 100644 index 0000000..79485b3 --- /dev/null +++ b/src/patches/linux/0002-Drivers-net-hyperv-Cleanup-the-receive-path.patch @@ -0,0 +1,112 @@ +From 348a5d691d84759dda8cdd3cbf9f071115c1240e Mon Sep 17 00:00:00 2001 +From: KY Srinivasan kys@microsoft.com +Date: Sun, 16 Feb 2014 16:38:44 -0800 +Subject: [PATCH 02/25] Drivers: net: hyperv: Cleanup the receive path + +Make the receive path a little more efficient by parameterizing the +required state rather than re-establishing that state. + +Signed-off-by: K. Y. Srinivasan kys@microsoft.com +Reviewed-by: Haiyang Zhang haiyangz@microsoft.com +Signed-off-by: David S. Miller davem@davemloft.net +--- + drivers/net/hyperv/netvsc.c | 29 +++++++++++++---------------- + 1 file changed, 13 insertions(+), 16 deletions(-) + +diff --git a/drivers/net/hyperv/netvsc.c b/drivers/net/hyperv/netvsc.c +index 03a2c6e17158..7fa2bbade327 100644 +--- a/drivers/net/hyperv/netvsc.c ++++ b/drivers/net/hyperv/netvsc.c +@@ -432,17 +432,14 @@ static inline u32 hv_ringbuf_avail_percent( + return avail_write * 100 / ring_info->ring_datasize; + } + +-static void netvsc_send_completion(struct hv_device *device, ++static void netvsc_send_completion(struct netvsc_device *net_device, ++ struct hv_device *device, + struct vmpacket_descriptor *packet) + { +- struct netvsc_device *net_device; + struct nvsp_message *nvsp_packet; + struct hv_netvsc_packet *nvsc_packet; + struct net_device *ndev; + +- net_device = get_inbound_net_device(device); +- if (!net_device) +- return; + ndev = net_device->ndev; + + nvsp_packet = (struct nvsp_message *)((unsigned long)packet + +@@ -561,13 +558,13 @@ int netvsc_send(struct hv_device *device, + } + + static void netvsc_send_recv_completion(struct hv_device *device, ++ struct netvsc_device *net_device, + u64 transaction_id, u32 status) + { + struct nvsp_message recvcompMessage; + int retries = 0; + int ret; + struct net_device *ndev; +- struct netvsc_device *net_device = hv_get_drvdata(device); + + ndev = net_device->ndev; + +@@ -653,14 +650,15 @@ static void netvsc_receive_completion(void *context) + + /* Send a receive completion for the xfer page packet */ + if (fsend_receive_comp) +- netvsc_send_recv_completion(device, transaction_id, status); ++ netvsc_send_recv_completion(device, net_device, transaction_id, ++ status); + + } + +-static void netvsc_receive(struct hv_device *device, +- struct vmpacket_descriptor *packet) ++static void netvsc_receive(struct netvsc_device *net_device, ++ struct hv_device *device, ++ struct vmpacket_descriptor *packet) + { +- struct netvsc_device *net_device; + struct vmtransfer_page_packet_header *vmxferpage_packet; + struct nvsp_message *nvsp_packet; + struct hv_netvsc_packet *netvsc_packet = NULL; +@@ -673,9 +671,6 @@ static void netvsc_receive(struct hv_device *device, + + LIST_HEAD(listHead); + +- net_device = get_inbound_net_device(device); +- if (!net_device) +- return; + ndev = net_device->ndev; + + /* +@@ -741,7 +736,7 @@ static void netvsc_receive(struct hv_device *device, + spin_unlock_irqrestore(&net_device->recv_pkt_list_lock, + flags); + +- netvsc_send_recv_completion(device, ++ netvsc_send_recv_completion(device, net_device, + vmxferpage_packet->d.trans_id, + NVSP_STAT_FAIL); + +@@ -825,11 +820,13 @@ static void netvsc_channel_cb(void *context) + desc = (struct vmpacket_descriptor *)buffer; + switch (desc->type) { + case VM_PKT_COMP: +- netvsc_send_completion(device, desc); ++ netvsc_send_completion(net_device, ++ device, desc); + break; + + case VM_PKT_DATA_USING_XFER_PAGES: +- netvsc_receive(device, desc); ++ netvsc_receive(net_device, ++ device, desc); + break; + + default: +-- +2.4.3 + diff --git a/src/patches/linux/0002-hyperv-Remove-recv_pkt_list-and-lock.patch b/src/patches/linux/0002-hyperv-Remove-recv_pkt_list-and-lock.patch new file mode 100644 index 0000000..bbe0e52 --- /dev/null +++ b/src/patches/linux/0002-hyperv-Remove-recv_pkt_list-and-lock.patch @@ -0,0 +1,384 @@ +From 4baab26129e0540746744232022110dbe9e011e7 Mon Sep 17 00:00:00 2001 +From: Haiyang Zhang haiyangz@microsoft.com +Date: Mon, 21 Apr 2014 14:54:43 -0700 +Subject: [PATCH 02/11] hyperv: Remove recv_pkt_list and lock + +Removed recv_pkt_list and lock, and updated related code, so that +the locking overhead is reduced especially when multiple channels +are in use. + +The recv_pkt_list isn't actually necessary because the packets are +processed sequentially in each channel. It has been replaced by a +local variable, and the related lock for this list is also removed. +The is_data_pkt field is not used in receive path, so its assignment +is cleaned up. + +Signed-off-by: Haiyang Zhang haiyangz@microsoft.com +Reviewed-by: K. Y. Srinivasan kys@microsoft.com +Signed-off-by: David S. Miller davem@davemloft.net +--- + drivers/net/hyperv/hyperv_net.h | 33 -------- + drivers/net/hyperv/netvsc.c | 174 +++----------------------------------- + drivers/net/hyperv/netvsc_drv.c | 2 +- + drivers/net/hyperv/rndis_filter.c | 2 - + 4 files changed, 13 insertions(+), 198 deletions(-) + +diff --git a/drivers/net/hyperv/hyperv_net.h b/drivers/net/hyperv/hyperv_net.h +index 57eb3f906d64..a1af0f7711e2 100644 +--- a/drivers/net/hyperv/hyperv_net.h ++++ b/drivers/net/hyperv/hyperv_net.h +@@ -119,27 +119,14 @@ struct ndis_recv_scale_param { /* NDIS_RECEIVE_SCALE_PARAMETERS */ + }; + + /* Fwd declaration */ +-struct hv_netvsc_packet; + struct ndis_tcp_ip_checksum_info; + +-/* Represent the xfer page packet which contains 1 or more netvsc packet */ +-struct xferpage_packet { +- struct list_head list_ent; +- u32 status; +- +- /* # of netvsc packets this xfer packet contains */ +- u32 count; +- +- struct vmbus_channel *channel; +-}; +- + /* + * Represent netvsc packet which contains 1 RNDIS and 1 ethernet frame + * within the RNDIS + */ + struct hv_netvsc_packet { + /* Bookkeeping stuff */ +- struct list_head list_ent; + u32 status; + + struct hv_device *device; +@@ -149,19 +136,8 @@ struct hv_netvsc_packet { + u16 q_idx; + struct vmbus_channel *channel; + +- /* +- * Valid only for receives when we break a xfer page packet +- * into multiple netvsc packets +- */ +- struct xferpage_packet *xfer_page_pkt; +- + union { + struct { +- u64 recv_completion_tid; +- void *recv_completion_ctx; +- void (*recv_completion)(void *context); +- } recv; +- struct { + u64 send_completion_tid; + void *send_completion_ctx; + void (*send_completion)(void *context); +@@ -613,9 +589,6 @@ struct nvsp_message { + + #define NETVSC_RECEIVE_BUFFER_ID 0xcafe + +-/* Preallocated receive packets */ +-#define NETVSC_RECEIVE_PACKETLIST_COUNT 256 +- + #define NETVSC_PACKET_SIZE 2048 + + #define VRSS_SEND_TAB_SIZE 16 +@@ -630,12 +603,6 @@ struct netvsc_device { + wait_queue_head_t wait_drain; + bool start_remove; + bool destroy; +- /* +- * List of free preallocated hv_netvsc_packet to represent receive +- * packet +- */ +- struct list_head recv_pkt_list; +- spinlock_t recv_pkt_list_lock; + + /* Receive buffer allocated by us but manages by NetVSP */ + void *recv_buf; +diff --git a/drivers/net/hyperv/netvsc.c b/drivers/net/hyperv/netvsc.c +index e7e77f12bc38..b10334773b32 100644 +--- a/drivers/net/hyperv/netvsc.c ++++ b/drivers/net/hyperv/netvsc.c +@@ -387,7 +387,6 @@ static void netvsc_disconnect_vsp(struct netvsc_device *net_device) + int netvsc_device_remove(struct hv_device *device) + { + struct netvsc_device *net_device; +- struct hv_netvsc_packet *netvsc_packet, *pos; + unsigned long flags; + + net_device = hv_get_drvdata(device); +@@ -416,12 +415,6 @@ int netvsc_device_remove(struct hv_device *device) + vmbus_close(device->channel); + + /* Release all resources */ +- list_for_each_entry_safe(netvsc_packet, pos, +- &net_device->recv_pkt_list, list_ent) { +- list_del(&netvsc_packet->list_ent); +- kfree(netvsc_packet); +- } +- + if (net_device->sub_cb_buf) + vfree(net_device->sub_cb_buf); + +@@ -641,62 +634,6 @@ retry_send_cmplt: + } + } + +-/* Send a receive completion packet to RNDIS device (ie NetVsp) */ +-static void netvsc_receive_completion(void *context) +-{ +- struct hv_netvsc_packet *packet = context; +- struct hv_device *device = packet->device; +- struct vmbus_channel *channel; +- struct netvsc_device *net_device; +- u64 transaction_id = 0; +- bool fsend_receive_comp = false; +- unsigned long flags; +- struct net_device *ndev; +- u32 status = NVSP_STAT_NONE; +- +- /* +- * Even though it seems logical to do a GetOutboundNetDevice() here to +- * send out receive completion, we are using GetInboundNetDevice() +- * since we may have disable outbound traffic already. +- */ +- net_device = get_inbound_net_device(device); +- if (!net_device) +- return; +- ndev = net_device->ndev; +- +- /* Overloading use of the lock. */ +- spin_lock_irqsave(&net_device->recv_pkt_list_lock, flags); +- +- if (packet->status != NVSP_STAT_SUCCESS) +- packet->xfer_page_pkt->status = NVSP_STAT_FAIL; +- +- packet->xfer_page_pkt->count--; +- +- /* +- * Last one in the line that represent 1 xfer page packet. +- * Return the xfer page packet itself to the freelist +- */ +- if (packet->xfer_page_pkt->count == 0) { +- fsend_receive_comp = true; +- channel = packet->xfer_page_pkt->channel; +- transaction_id = packet->completion.recv.recv_completion_tid; +- status = packet->xfer_page_pkt->status; +- list_add_tail(&packet->xfer_page_pkt->list_ent, +- &net_device->recv_pkt_list); +- +- } +- +- /* Put the packet back */ +- list_add_tail(&packet->list_ent, &net_device->recv_pkt_list); +- spin_unlock_irqrestore(&net_device->recv_pkt_list_lock, flags); +- +- /* Send a receive completion for the xfer page packet */ +- if (fsend_receive_comp) +- netvsc_send_recv_completion(device, channel, net_device, +- transaction_id, status); +- +-} +- + static void netvsc_receive(struct netvsc_device *net_device, + struct vmbus_channel *channel, + struct hv_device *device, +@@ -704,16 +641,13 @@ static void netvsc_receive(struct netvsc_device *net_device, + { + struct vmtransfer_page_packet_header *vmxferpage_packet; + struct nvsp_message *nvsp_packet; +- struct hv_netvsc_packet *netvsc_packet = NULL; +- /* struct netvsc_driver *netvscDriver; */ +- struct xferpage_packet *xferpage_packet = NULL; ++ struct hv_netvsc_packet nv_pkt; ++ struct hv_netvsc_packet *netvsc_packet = &nv_pkt; ++ u32 status = NVSP_STAT_SUCCESS; + int i; + int count = 0; +- unsigned long flags; + struct net_device *ndev; + +- LIST_HEAD(listHead); +- + ndev = net_device->ndev; + + /* +@@ -746,78 +680,14 @@ static void netvsc_receive(struct netvsc_device *net_device, + return; + } + +- /* +- * Grab free packets (range count + 1) to represent this xfer +- * page packet. +1 to represent the xfer page packet itself. +- * We grab it here so that we know exactly how many we can +- * fulfil +- */ +- spin_lock_irqsave(&net_device->recv_pkt_list_lock, flags); +- while (!list_empty(&net_device->recv_pkt_list)) { +- list_move_tail(net_device->recv_pkt_list.next, &listHead); +- if (++count == vmxferpage_packet->range_cnt + 1) +- break; +- } +- spin_unlock_irqrestore(&net_device->recv_pkt_list_lock, flags); +- +- /* +- * We need at least 2 netvsc pkts (1 to represent the xfer +- * page and at least 1 for the range) i.e. we can handled +- * some of the xfer page packet ranges... +- */ +- if (count < 2) { +- netdev_err(ndev, "Got only %d netvsc pkt...needed " +- "%d pkts. Dropping this xfer page packet completely!\n", +- count, vmxferpage_packet->range_cnt + 1); +- +- /* Return it to the freelist */ +- spin_lock_irqsave(&net_device->recv_pkt_list_lock, flags); +- for (i = count; i != 0; i--) { +- list_move_tail(listHead.next, +- &net_device->recv_pkt_list); +- } +- spin_unlock_irqrestore(&net_device->recv_pkt_list_lock, +- flags); +- +- netvsc_send_recv_completion(device, channel, net_device, +- vmxferpage_packet->d.trans_id, +- NVSP_STAT_FAIL); +- +- return; +- } +- +- /* Remove the 1st packet to represent the xfer page packet itself */ +- xferpage_packet = (struct xferpage_packet *)listHead.next; +- list_del(&xferpage_packet->list_ent); +- xferpage_packet->status = NVSP_STAT_SUCCESS; +- xferpage_packet->channel = channel; +- +- /* This is how much we can satisfy */ +- xferpage_packet->count = count - 1; +- +- if (xferpage_packet->count != vmxferpage_packet->range_cnt) { +- netdev_err(ndev, "Needed %d netvsc pkts to satisfy " +- "this xfer page...got %d\n", +- vmxferpage_packet->range_cnt, xferpage_packet->count); +- } ++ count = vmxferpage_packet->range_cnt; ++ netvsc_packet->device = device; ++ netvsc_packet->channel = channel; + + /* Each range represents 1 RNDIS pkt that contains 1 ethernet frame */ +- for (i = 0; i < (count - 1); i++) { +- netvsc_packet = (struct hv_netvsc_packet *)listHead.next; +- list_del(&netvsc_packet->list_ent); +- ++ for (i = 0; i < count; i++) { + /* Initialize the netvsc packet */ + netvsc_packet->status = NVSP_STAT_SUCCESS; +- netvsc_packet->xfer_page_pkt = xferpage_packet; +- netvsc_packet->completion.recv.recv_completion = +- netvsc_receive_completion; +- netvsc_packet->completion.recv.recv_completion_ctx = +- netvsc_packet; +- netvsc_packet->device = device; +- /* Save this so that we can send it back */ +- netvsc_packet->completion.recv.recv_completion_tid = +- vmxferpage_packet->d.trans_id; +- + netvsc_packet->data = (void *)((unsigned long)net_device-> + recv_buf + vmxferpage_packet->ranges[i].byte_offset); + netvsc_packet->total_data_buflen = +@@ -826,10 +696,12 @@ static void netvsc_receive(struct netvsc_device *net_device, + /* Pass it to the upper layer */ + rndis_filter_receive(device, netvsc_packet); + +- netvsc_receive_completion(netvsc_packet-> +- completion.recv.recv_completion_ctx); ++ if (netvsc_packet->status != NVSP_STAT_SUCCESS) ++ status = NVSP_STAT_FAIL; + } + ++ netvsc_send_recv_completion(device, channel, net_device, ++ vmxferpage_packet->d.trans_id, status); + } + + +@@ -956,11 +828,9 @@ void netvsc_channel_cb(void *context) + int netvsc_device_add(struct hv_device *device, void *additional_info) + { + int ret = 0; +- int i; + int ring_size = + ((struct netvsc_device_info *)additional_info)->ring_size; + struct netvsc_device *net_device; +- struct hv_netvsc_packet *packet, *pos; + struct net_device *ndev; + + net_device = alloc_net_device(device); +@@ -981,18 +851,6 @@ int netvsc_device_add(struct hv_device *device, void *additional_info) + ndev = net_device->ndev; + + /* Initialize the NetVSC channel extension */ +- spin_lock_init(&net_device->recv_pkt_list_lock); +- +- INIT_LIST_HEAD(&net_device->recv_pkt_list); +- +- for (i = 0; i < NETVSC_RECEIVE_PACKETLIST_COUNT; i++) { +- packet = kzalloc(sizeof(struct hv_netvsc_packet), GFP_KERNEL); +- if (!packet) +- break; +- +- list_add_tail(&packet->list_ent, +- &net_device->recv_pkt_list); +- } + init_completion(&net_device->channel_init_wait); + + set_per_channel_state(device->channel, net_device->cb_buffer); +@@ -1028,16 +886,8 @@ close: + + cleanup: + +- if (net_device) { +- list_for_each_entry_safe(packet, pos, +- &net_device->recv_pkt_list, +- list_ent) { +- list_del(&packet->list_ent); +- kfree(packet); +- } +- ++ if (net_device) + kfree(net_device); +- } + + return ret; + } +diff --git a/drivers/net/hyperv/netvsc_drv.c b/drivers/net/hyperv/netvsc_drv.c +index 093cf3fc46b8..8f6d53a2ed95 100644 +--- a/drivers/net/hyperv/netvsc_drv.c ++++ b/drivers/net/hyperv/netvsc_drv.c +@@ -638,7 +638,7 @@ int netvsc_recv_callback(struct hv_device *device_obj, + __vlan_hwaccel_put_tag(skb, htons(ETH_P_8021Q), + packet->vlan_tci); + +- skb_record_rx_queue(skb, packet->xfer_page_pkt->channel-> ++ skb_record_rx_queue(skb, packet->channel-> + offermsg.offer.sub_channel_index % + net->real_num_rx_queues); + +diff --git a/drivers/net/hyperv/rndis_filter.c b/drivers/net/hyperv/rndis_filter.c +index d92cfbe43410..48f5a0fbd674 100644 +--- a/drivers/net/hyperv/rndis_filter.c ++++ b/drivers/net/hyperv/rndis_filter.c +@@ -401,8 +401,6 @@ static void rndis_filter_receive_data(struct rndis_device *dev, + pkt->total_data_buflen = rndis_pkt->data_len; + pkt->data = (void *)((unsigned long)pkt->data + data_offset); + +- pkt->is_data_pkt = true; +- + vlan = rndis_get_ppi(rndis_pkt, IEEE_8021Q_INFO); + if (vlan) { + pkt->vlan_tci = VLAN_TAG_PRESENT | vlan->vlanid | +-- +2.4.3 + diff --git a/src/patches/linux/0003-Drivers-net-hyperv-Cleanup-the-netvsc-receive-callba.patch b/src/patches/linux/0003-Drivers-net-hyperv-Cleanup-the-netvsc-receive-callba.patch new file mode 100644 index 0000000..e0b82ea --- /dev/null +++ b/src/patches/linux/0003-Drivers-net-hyperv-Cleanup-the-netvsc-receive-callba.patch @@ -0,0 +1,101 @@ +From c9f2db35ac4f789930522d9d36200cb71b442bed Mon Sep 17 00:00:00 2001 +From: KY Srinivasan kys@microsoft.com +Date: Sun, 16 Feb 2014 16:38:45 -0800 +Subject: [PATCH 03/25] Drivers: net: hyperv: Cleanup the netvsc receive + callback functio + +Get rid of the buffer allocation in the receive path for normal packets. + +Signed-off-by: K. Y. Srinivasan kys@microsoft.com +Signed-off-by: David S. Miller davem@davemloft.net +--- + drivers/net/hyperv/hyperv_net.h | 2 ++ + drivers/net/hyperv/netvsc.c | 33 ++++++++++----------------------- + 2 files changed, 12 insertions(+), 23 deletions(-) + +diff --git a/drivers/net/hyperv/hyperv_net.h b/drivers/net/hyperv/hyperv_net.h +index 7645ba38bde8..01a16ea77a5a 100644 +--- a/drivers/net/hyperv/hyperv_net.h ++++ b/drivers/net/hyperv/hyperv_net.h +@@ -506,6 +506,8 @@ struct netvsc_device { + + /* Holds rndis device info */ + void *extension; ++ /* The recive buffer for this device */ ++ unsigned char cb_buffer[NETVSC_PACKET_SIZE]; + }; + + /* NdisInitialize message */ +diff --git a/drivers/net/hyperv/netvsc.c b/drivers/net/hyperv/netvsc.c +index 7fa2bbade327..9a0e9c6f1414 100644 +--- a/drivers/net/hyperv/netvsc.c ++++ b/drivers/net/hyperv/netvsc.c +@@ -795,22 +795,16 @@ static void netvsc_channel_cb(void *context) + struct netvsc_device *net_device; + u32 bytes_recvd; + u64 request_id; +- unsigned char *packet; + struct vmpacket_descriptor *desc; + unsigned char *buffer; + int bufferlen = NETVSC_PACKET_SIZE; + struct net_device *ndev; + +- packet = kzalloc(NETVSC_PACKET_SIZE * sizeof(unsigned char), +- GFP_ATOMIC); +- if (!packet) +- return; +- buffer = packet; +- + net_device = get_inbound_net_device(device); + if (!net_device) +- goto out; ++ return; + ndev = net_device->ndev; ++ buffer = net_device->cb_buffer; + + do { + ret = vmbus_recvpacket_raw(device->channel, buffer, bufferlen, +@@ -838,23 +832,16 @@ static void netvsc_channel_cb(void *context) + break; + } + +- /* reset */ +- if (bufferlen > NETVSC_PACKET_SIZE) { +- kfree(buffer); +- buffer = packet; +- bufferlen = NETVSC_PACKET_SIZE; +- } + } else { +- /* reset */ +- if (bufferlen > NETVSC_PACKET_SIZE) { +- kfree(buffer); +- buffer = packet; +- bufferlen = NETVSC_PACKET_SIZE; +- } +- ++ /* ++ * We are done for this pass. ++ */ + break; + } ++ + } else if (ret == -ENOBUFS) { ++ if (bufferlen > NETVSC_PACKET_SIZE) ++ kfree(buffer); + /* Handle large packet */ + buffer = kmalloc(bytes_recvd, GFP_ATOMIC); + if (buffer == NULL) { +@@ -869,8 +856,8 @@ static void netvsc_channel_cb(void *context) + } + } while (1); + +-out: +- kfree(buffer); ++ if (bufferlen > NETVSC_PACKET_SIZE) ++ kfree(buffer); + return; + } + +-- +2.4.3 + diff --git a/src/patches/linux/0003-hyperv-Simplify-the-send_completion-variables.patch b/src/patches/linux/0003-hyperv-Simplify-the-send_completion-variables.patch new file mode 100644 index 0000000..dcab56f --- /dev/null +++ b/src/patches/linux/0003-hyperv-Simplify-the-send_completion-variables.patch @@ -0,0 +1,105 @@ +From 893f66277799cd46bdf97429cc5d16a815a51273 Mon Sep 17 00:00:00 2001 +From: Haiyang Zhang haiyangz@microsoft.com +Date: Mon, 21 Apr 2014 14:54:44 -0700 +Subject: [PATCH 03/11] hyperv: Simplify the send_completion variables + +The union contains only one member now, so we use the variables in it directly. + +Signed-off-by: Haiyang Zhang haiyangz@microsoft.com +Reviewed-by: K. Y. Srinivasan kys@microsoft.com +Signed-off-by: David S. Miller davem@davemloft.net +--- + drivers/net/hyperv/hyperv_net.h | 10 +++------- + drivers/net/hyperv/netvsc.c | 7 +++---- + drivers/net/hyperv/netvsc_drv.c | 8 ++++---- + drivers/net/hyperv/rndis_filter.c | 2 +- + 4 files changed, 11 insertions(+), 16 deletions(-) + +diff --git a/drivers/net/hyperv/hyperv_net.h b/drivers/net/hyperv/hyperv_net.h +index a1af0f7711e2..d1f7826aa75f 100644 +--- a/drivers/net/hyperv/hyperv_net.h ++++ b/drivers/net/hyperv/hyperv_net.h +@@ -136,13 +136,9 @@ struct hv_netvsc_packet { + u16 q_idx; + struct vmbus_channel *channel; + +- union { +- struct { +- u64 send_completion_tid; +- void *send_completion_ctx; +- void (*send_completion)(void *context); +- } send; +- } completion; ++ u64 send_completion_tid; ++ void *send_completion_ctx; ++ void (*send_completion)(void *context); + + /* This points to the memory after page_buf */ + struct rndis_message *rndis_msg; +diff --git a/drivers/net/hyperv/netvsc.c b/drivers/net/hyperv/netvsc.c +index b10334773b32..bbee44635035 100644 +--- a/drivers/net/hyperv/netvsc.c ++++ b/drivers/net/hyperv/netvsc.c +@@ -479,9 +479,8 @@ static void netvsc_send_completion(struct netvsc_device *net_device, + if (nvsc_packet) { + q_idx = nvsc_packet->q_idx; + channel = nvsc_packet->channel; +- nvsc_packet->completion.send.send_completion( +- nvsc_packet->completion.send. +- send_completion_ctx); ++ nvsc_packet->send_completion(nvsc_packet-> ++ send_completion_ctx); + } + + num_outstanding_sends = +@@ -534,7 +533,7 @@ int netvsc_send(struct hv_device *device, + 0xFFFFFFFF; + sendMessage.msg.v1_msg.send_rndis_pkt.send_buf_section_size = 0; + +- if (packet->completion.send.send_completion) ++ if (packet->send_completion) + req_id = (ulong)packet; + else + req_id = 0; +diff --git a/drivers/net/hyperv/netvsc_drv.c b/drivers/net/hyperv/netvsc_drv.c +index 8f6d53a2ed95..c76b66515e92 100644 +--- a/drivers/net/hyperv/netvsc_drv.c ++++ b/drivers/net/hyperv/netvsc_drv.c +@@ -235,7 +235,7 @@ static void netvsc_xmit_completion(void *context) + { + struct hv_netvsc_packet *packet = (struct hv_netvsc_packet *)context; + struct sk_buff *skb = (struct sk_buff *) +- (unsigned long)packet->completion.send.send_completion_tid; ++ (unsigned long)packet->send_completion_tid; + + kfree(packet); + +@@ -425,9 +425,9 @@ static int netvsc_start_xmit(struct sk_buff *skb, struct net_device *net) + (num_data_pgs * sizeof(struct hv_page_buffer))); + + /* Set the completion routine */ +- packet->completion.send.send_completion = netvsc_xmit_completion; +- packet->completion.send.send_completion_ctx = packet; +- packet->completion.send.send_completion_tid = (unsigned long)skb; ++ packet->send_completion = netvsc_xmit_completion; ++ packet->send_completion_ctx = packet; ++ packet->send_completion_tid = (unsigned long)skb; + + isvlan = packet->vlan_tci & VLAN_TAG_PRESENT; + +diff --git a/drivers/net/hyperv/rndis_filter.c b/drivers/net/hyperv/rndis_filter.c +index 48f5a0fbd674..99c527adae5b 100644 +--- a/drivers/net/hyperv/rndis_filter.c ++++ b/drivers/net/hyperv/rndis_filter.c +@@ -236,7 +236,7 @@ static int rndis_filter_send_request(struct rndis_device *dev, + packet->page_buf[0].len; + } + +- packet->completion.send.send_completion = NULL; ++ packet->send_completion = NULL; + + ret = netvsc_send(dev->net_dev->dev, packet); + return ret; +-- +2.4.3 + diff --git a/src/patches/linux/0004-hyperv-Add-latest-NetVSP-versions-to-auto-negotiatio.patch b/src/patches/linux/0004-hyperv-Add-latest-NetVSP-versions-to-auto-negotiatio.patch new file mode 100644 index 0000000..f9cd827 --- /dev/null +++ b/src/patches/linux/0004-hyperv-Add-latest-NetVSP-versions-to-auto-negotiatio.patch @@ -0,0 +1,171 @@ +From 3c2a271d9681cc017947c5e027acc64707c30dee Mon Sep 17 00:00:00 2001 +From: Haiyang Zhang haiyangz@microsoft.com +Date: Wed, 19 Feb 2014 15:49:45 -0800 +Subject: [PATCH 04/25] hyperv: Add latest NetVSP versions to auto negotiation + +It auto negotiates the highest NetVSP version supported by both guest and host. + +Signed-off-by: Haiyang Zhang haiyangz@microsoft.com +Reviewed-by: K. Y. Srinivasan kys@microsoft.com +Signed-off-by: David S. Miller davem@davemloft.net +--- + drivers/net/hyperv/hyperv_net.h | 53 +++++++++++++++++++++++++++++++++++++++++ + drivers/net/hyperv/netvsc.c | 25 ++++++++++++------- + drivers/net/hyperv/netvsc_drv.c | 2 +- + 3 files changed, 70 insertions(+), 10 deletions(-) + +diff --git a/drivers/net/hyperv/hyperv_net.h b/drivers/net/hyperv/hyperv_net.h +index 01a16ea77a5a..39fc230f5c20 100644 +--- a/drivers/net/hyperv/hyperv_net.h ++++ b/drivers/net/hyperv/hyperv_net.h +@@ -139,6 +139,8 @@ int rndis_filter_set_device_mac(struct hv_device *hdev, char *mac); + + #define NVSP_PROTOCOL_VERSION_1 2 + #define NVSP_PROTOCOL_VERSION_2 0x30002 ++#define NVSP_PROTOCOL_VERSION_4 0x40000 ++#define NVSP_PROTOCOL_VERSION_5 0x50000 + + enum { + NVSP_MSG_TYPE_NONE = 0, +@@ -193,6 +195,23 @@ enum { + + NVSP_MSG2_TYPE_ALLOC_CHIMNEY_HANDLE, + NVSP_MSG2_TYPE_ALLOC_CHIMNEY_HANDLE_COMP, ++ ++ NVSP_MSG2_MAX = NVSP_MSG2_TYPE_ALLOC_CHIMNEY_HANDLE_COMP, ++ ++ /* Version 4 messages */ ++ NVSP_MSG4_TYPE_SEND_VF_ASSOCIATION, ++ NVSP_MSG4_TYPE_SWITCH_DATA_PATH, ++ NVSP_MSG4_TYPE_UPLINK_CONNECT_STATE_DEPRECATED, ++ ++ NVSP_MSG4_MAX = NVSP_MSG4_TYPE_UPLINK_CONNECT_STATE_DEPRECATED, ++ ++ /* Version 5 messages */ ++ NVSP_MSG5_TYPE_OID_QUERY_EX, ++ NVSP_MSG5_TYPE_OID_QUERY_EX_COMP, ++ NVSP_MSG5_TYPE_SUBCHANNEL, ++ NVSP_MSG5_TYPE_SEND_INDIRECTION_TABLE, ++ ++ NVSP_MSG5_MAX = NVSP_MSG5_TYPE_SEND_INDIRECTION_TABLE, + }; + + enum { +@@ -447,10 +466,44 @@ union nvsp_2_message_uber { + struct nvsp_2_free_rxbuf free_rxbuf; + } __packed; + ++enum nvsp_subchannel_operation { ++ NVSP_SUBCHANNEL_NONE = 0, ++ NVSP_SUBCHANNEL_ALLOCATE, ++ NVSP_SUBCHANNEL_MAX ++}; ++ ++struct nvsp_5_subchannel_request { ++ u32 op; ++ u32 num_subchannels; ++} __packed; ++ ++struct nvsp_5_subchannel_complete { ++ u32 status; ++ u32 num_subchannels; /* Actual number of subchannels allocated */ ++} __packed; ++ ++struct nvsp_5_send_indirect_table { ++ /* The number of entries in the send indirection table */ ++ u32 count; ++ ++ /* The offset of the send indireciton table from top of this struct. ++ * The send indirection table tells which channel to put the send ++ * traffic on. Each entry is a channel number. ++ */ ++ u32 offset; ++} __packed; ++ ++union nvsp_5_message_uber { ++ struct nvsp_5_subchannel_request subchn_req; ++ struct nvsp_5_subchannel_complete subchn_comp; ++ struct nvsp_5_send_indirect_table send_table; ++} __packed; ++ + union nvsp_all_messages { + union nvsp_message_init_uber init_msg; + union nvsp_1_message_uber v1_msg; + union nvsp_2_message_uber v2_msg; ++ union nvsp_5_message_uber v5_msg; + } __packed; + + /* ALL Messages */ +diff --git a/drivers/net/hyperv/netvsc.c b/drivers/net/hyperv/netvsc.c +index 9a0e9c6f1414..1a0280dcba7e 100644 +--- a/drivers/net/hyperv/netvsc.c ++++ b/drivers/net/hyperv/netvsc.c +@@ -290,7 +290,7 @@ static int negotiate_nvsp_ver(struct hv_device *device, + NVSP_STAT_SUCCESS) + return -EINVAL; + +- if (nvsp_ver != NVSP_PROTOCOL_VERSION_2) ++ if (nvsp_ver == NVSP_PROTOCOL_VERSION_1) + return 0; + + /* NVSPv2 only: Send NDIS config */ +@@ -314,6 +314,9 @@ static int netvsc_connect_vsp(struct hv_device *device) + struct nvsp_message *init_packet; + int ndis_version; + struct net_device *ndev; ++ u32 ver_list[] = { NVSP_PROTOCOL_VERSION_1, NVSP_PROTOCOL_VERSION_2, ++ NVSP_PROTOCOL_VERSION_4, NVSP_PROTOCOL_VERSION_5 }; ++ int i, num_ver = 4; /* number of different NVSP versions */ + + net_device = get_outbound_net_device(device); + if (!net_device) +@@ -323,13 +326,14 @@ static int netvsc_connect_vsp(struct hv_device *device) + init_packet = &net_device->channel_init_pkt; + + /* Negotiate the latest NVSP protocol supported */ +- if (negotiate_nvsp_ver(device, net_device, init_packet, +- NVSP_PROTOCOL_VERSION_2) == 0) { +- net_device->nvsp_version = NVSP_PROTOCOL_VERSION_2; +- } else if (negotiate_nvsp_ver(device, net_device, init_packet, +- NVSP_PROTOCOL_VERSION_1) == 0) { +- net_device->nvsp_version = NVSP_PROTOCOL_VERSION_1; +- } else { ++ for (i = num_ver - 1; i >= 0; i--) ++ if (negotiate_nvsp_ver(device, net_device, init_packet, ++ ver_list[i]) == 0) { ++ net_device->nvsp_version = ver_list[i]; ++ break; ++ } ++ ++ if (i < 0) { + ret = -EPROTO; + goto cleanup; + } +@@ -339,7 +343,10 @@ static int netvsc_connect_vsp(struct hv_device *device) + /* Send the ndis version */ + memset(init_packet, 0, sizeof(struct nvsp_message)); + +- ndis_version = 0x00050001; ++ if (net_device->nvsp_version <= NVSP_PROTOCOL_VERSION_4) ++ ndis_version = 0x00050001; ++ else ++ ndis_version = 0x0006001e; + + init_packet->hdr.msg_type = NVSP_MSG1_TYPE_SEND_NDIS_VER; + init_packet->msg.v1_msg. +diff --git a/drivers/net/hyperv/netvsc_drv.c b/drivers/net/hyperv/netvsc_drv.c +index 28020f83ba6f..8e3a0b00099b 100644 +--- a/drivers/net/hyperv/netvsc_drv.c ++++ b/drivers/net/hyperv/netvsc_drv.c +@@ -328,7 +328,7 @@ static int netvsc_change_mtu(struct net_device *ndev, int mtu) + if (nvdev == NULL || nvdev->destroy) + return -ENODEV; + +- if (nvdev->nvsp_version == NVSP_PROTOCOL_VERSION_2) ++ if (nvdev->nvsp_version >= NVSP_PROTOCOL_VERSION_2) + limit = NETVSC_MTU; + + if (mtu < 68 || mtu > limit) +-- +2.4.3 + diff --git a/src/patches/linux/0004-hyperv-Enable-sendbuf-mechanism-on-the-send-path.patch b/src/patches/linux/0004-hyperv-Enable-sendbuf-mechanism-on-the-send-path.patch new file mode 100644 index 0000000..bc09122 --- /dev/null +++ b/src/patches/linux/0004-hyperv-Enable-sendbuf-mechanism-on-the-send-path.patch @@ -0,0 +1,407 @@ +From c25aaf814a63f9d9c4e45416f13d70ef0aa0be2e Mon Sep 17 00:00:00 2001 +From: KY Srinivasan kys@microsoft.com +Date: Wed, 30 Apr 2014 10:14:31 -0700 +Subject: [PATCH 04/11] hyperv: Enable sendbuf mechanism on the send path + +We send packets using a copy-free mechanism (this is the Guest to Host transport +via VMBUS). While this is obviously optimal for large packets, +it may not be optimal for small packets. Hyper-V host supports +a second mechanism for sending packets that is "copy based". We implement that +mechanism in this patch. + +In this version of the patch I have addressed a comment from David Miller. + +With this patch (and all of the other offload and VRSS patches), we are now able +to almost saturate a 10G interface between Linux VMs on Hyper-V +on different hosts - close to 9 Gbps as measured via iperf. + +Signed-off-by: K. Y. Srinivasan kys@microsoft.com +Reviewed-by: Haiyang Zhang haiyangz@microsoft.com +Signed-off-by: David S. Miller davem@davemloft.net +--- + drivers/net/hyperv/hyperv_net.h | 14 +++ + drivers/net/hyperv/netvsc.c | 226 ++++++++++++++++++++++++++++++++++++++-- + drivers/net/hyperv/netvsc_drv.c | 3 +- + 3 files changed, 234 insertions(+), 9 deletions(-) + +diff --git a/drivers/net/hyperv/hyperv_net.h b/drivers/net/hyperv/hyperv_net.h +index d1f7826aa75f..4b7df5a5c966 100644 +--- a/drivers/net/hyperv/hyperv_net.h ++++ b/drivers/net/hyperv/hyperv_net.h +@@ -140,6 +140,8 @@ struct hv_netvsc_packet { + void *send_completion_ctx; + void (*send_completion)(void *context); + ++ u32 send_buf_index; ++ + /* This points to the memory after page_buf */ + struct rndis_message *rndis_msg; + +@@ -582,6 +584,9 @@ struct nvsp_message { + + #define NETVSC_RECEIVE_BUFFER_SIZE (1024*1024*16) /* 16MB */ + #define NETVSC_RECEIVE_BUFFER_SIZE_LEGACY (1024*1024*15) /* 15MB */ ++#define NETVSC_SEND_BUFFER_SIZE (1024 * 1024) /* 1MB */ ++#define NETVSC_INVALID_INDEX -1 ++ + + #define NETVSC_RECEIVE_BUFFER_ID 0xcafe + +@@ -607,6 +612,15 @@ struct netvsc_device { + u32 recv_section_cnt; + struct nvsp_1_receive_buffer_section *recv_section; + ++ /* Send buffer allocated by us */ ++ void *send_buf; ++ u32 send_buf_size; ++ u32 send_buf_gpadl_handle; ++ u32 send_section_cnt; ++ u32 send_section_size; ++ unsigned long *send_section_map; ++ int map_words; ++ + /* Used for NetVSP initialization protocol */ + struct completion channel_init_wait; + struct nvsp_message channel_init_pkt; +diff --git a/drivers/net/hyperv/netvsc.c b/drivers/net/hyperv/netvsc.c +index bbee44635035..c041f63a6d30 100644 +--- a/drivers/net/hyperv/netvsc.c ++++ b/drivers/net/hyperv/netvsc.c +@@ -28,6 +28,7 @@ + #include <linux/slab.h> + #include <linux/netdevice.h> + #include <linux/if_ether.h> ++#include <asm/sync_bitops.h> + + #include "hyperv_net.h" + +@@ -80,7 +81,7 @@ get_in_err: + } + + +-static int netvsc_destroy_recv_buf(struct netvsc_device *net_device) ++static int netvsc_destroy_buf(struct netvsc_device *net_device) + { + struct nvsp_message *revoke_packet; + int ret = 0; +@@ -146,10 +147,62 @@ static int netvsc_destroy_recv_buf(struct netvsc_device *net_device) + net_device->recv_section = NULL; + } + ++ /* Deal with the send buffer we may have setup. ++ * If we got a send section size, it means we received a ++ * SendsendBufferComplete msg (ie sent ++ * NvspMessage1TypeSendReceiveBuffer msg) therefore, we need ++ * to send a revoke msg here ++ */ ++ if (net_device->send_section_size) { ++ /* Send the revoke receive buffer */ ++ revoke_packet = &net_device->revoke_packet; ++ memset(revoke_packet, 0, sizeof(struct nvsp_message)); ++ ++ revoke_packet->hdr.msg_type = ++ NVSP_MSG1_TYPE_REVOKE_SEND_BUF; ++ revoke_packet->msg.v1_msg.revoke_recv_buf.id = 0; ++ ++ ret = vmbus_sendpacket(net_device->dev->channel, ++ revoke_packet, ++ sizeof(struct nvsp_message), ++ (unsigned long)revoke_packet, ++ VM_PKT_DATA_INBAND, 0); ++ /* If we failed here, we might as well return and ++ * have a leak rather than continue and a bugchk ++ */ ++ if (ret != 0) { ++ netdev_err(ndev, "unable to send " ++ "revoke send buffer to netvsp\n"); ++ return ret; ++ } ++ } ++ /* Teardown the gpadl on the vsp end */ ++ if (net_device->send_buf_gpadl_handle) { ++ ret = vmbus_teardown_gpadl(net_device->dev->channel, ++ net_device->send_buf_gpadl_handle); ++ ++ /* If we failed here, we might as well return and have a leak ++ * rather than continue and a bugchk ++ */ ++ if (ret != 0) { ++ netdev_err(ndev, ++ "unable to teardown send buffer's gpadl\n"); ++ return ret; ++ } ++ net_device->recv_buf_gpadl_handle = 0; ++ } ++ if (net_device->send_buf) { ++ /* Free up the receive buffer */ ++ free_pages((unsigned long)net_device->send_buf, ++ get_order(net_device->send_buf_size)); ++ net_device->send_buf = NULL; ++ } ++ kfree(net_device->send_section_map); ++ + return ret; + } + +-static int netvsc_init_recv_buf(struct hv_device *device) ++static int netvsc_init_buf(struct hv_device *device) + { + int ret = 0; + int t; +@@ -248,10 +301,90 @@ static int netvsc_init_recv_buf(struct hv_device *device) + goto cleanup; + } + ++ /* Now setup the send buffer. ++ */ ++ net_device->send_buf = ++ (void *)__get_free_pages(GFP_KERNEL|__GFP_ZERO, ++ get_order(net_device->send_buf_size)); ++ if (!net_device->send_buf) { ++ netdev_err(ndev, "unable to allocate send " ++ "buffer of size %d\n", net_device->send_buf_size); ++ ret = -ENOMEM; ++ goto cleanup; ++ } ++ ++ /* Establish the gpadl handle for this buffer on this ++ * channel. Note: This call uses the vmbus connection rather ++ * than the channel to establish the gpadl handle. ++ */ ++ ret = vmbus_establish_gpadl(device->channel, net_device->send_buf, ++ net_device->send_buf_size, ++ &net_device->send_buf_gpadl_handle); ++ if (ret != 0) { ++ netdev_err(ndev, ++ "unable to establish send buffer's gpadl\n"); ++ goto cleanup; ++ } ++ ++ /* Notify the NetVsp of the gpadl handle */ ++ init_packet = &net_device->channel_init_pkt; ++ memset(init_packet, 0, sizeof(struct nvsp_message)); ++ init_packet->hdr.msg_type = NVSP_MSG1_TYPE_SEND_SEND_BUF; ++ init_packet->msg.v1_msg.send_recv_buf.gpadl_handle = ++ net_device->send_buf_gpadl_handle; ++ init_packet->msg.v1_msg.send_recv_buf.id = 0; ++ ++ /* Send the gpadl notification request */ ++ ret = vmbus_sendpacket(device->channel, init_packet, ++ sizeof(struct nvsp_message), ++ (unsigned long)init_packet, ++ VM_PKT_DATA_INBAND, ++ VMBUS_DATA_PACKET_FLAG_COMPLETION_REQUESTED); ++ if (ret != 0) { ++ netdev_err(ndev, ++ "unable to send send buffer's gpadl to netvsp\n"); ++ goto cleanup; ++ } ++ ++ t = wait_for_completion_timeout(&net_device->channel_init_wait, 5*HZ); ++ BUG_ON(t == 0); ++ ++ /* Check the response */ ++ if (init_packet->msg.v1_msg. ++ send_send_buf_complete.status != NVSP_STAT_SUCCESS) { ++ netdev_err(ndev, "Unable to complete send buffer " ++ "initialization with NetVsp - status %d\n", ++ init_packet->msg.v1_msg. ++ send_recv_buf_complete.status); ++ ret = -EINVAL; ++ goto cleanup; ++ } ++ ++ /* Parse the response */ ++ net_device->send_section_size = init_packet->msg. ++ v1_msg.send_send_buf_complete.section_size; ++ ++ /* Section count is simply the size divided by the section size. ++ */ ++ net_device->send_section_cnt = ++ net_device->send_buf_size/net_device->send_section_size; ++ ++ dev_info(&device->device, "Send section size: %d, Section count:%d\n", ++ net_device->send_section_size, net_device->send_section_cnt); ++ ++ /* Setup state for managing the send buffer. */ ++ net_device->map_words = DIV_ROUND_UP(net_device->send_section_cnt, ++ BITS_PER_LONG); ++ ++ net_device->send_section_map = ++ kzalloc(net_device->map_words * sizeof(ulong), GFP_KERNEL); ++ if (net_device->send_section_map == NULL) ++ goto cleanup; ++ + goto exit; + + cleanup: +- netvsc_destroy_recv_buf(net_device); ++ netvsc_destroy_buf(net_device); + + exit: + return ret; +@@ -369,8 +502,9 @@ static int netvsc_connect_vsp(struct hv_device *device) + net_device->recv_buf_size = NETVSC_RECEIVE_BUFFER_SIZE_LEGACY; + else + net_device->recv_buf_size = NETVSC_RECEIVE_BUFFER_SIZE; ++ net_device->send_buf_size = NETVSC_SEND_BUFFER_SIZE; + +- ret = netvsc_init_recv_buf(device); ++ ret = netvsc_init_buf(device); + + cleanup: + return ret; +@@ -378,7 +512,7 @@ cleanup: + + static void netvsc_disconnect_vsp(struct netvsc_device *net_device) + { +- netvsc_destroy_recv_buf(net_device); ++ netvsc_destroy_buf(net_device); + } + + /* +@@ -440,6 +574,12 @@ static inline u32 hv_ringbuf_avail_percent( + return avail_write * 100 / ring_info->ring_datasize; + } + ++static inline void netvsc_free_send_slot(struct netvsc_device *net_device, ++ u32 index) ++{ ++ sync_change_bit(index, net_device->send_section_map); ++} ++ + static void netvsc_send_completion(struct netvsc_device *net_device, + struct hv_device *device, + struct vmpacket_descriptor *packet) +@@ -447,6 +587,7 @@ static void netvsc_send_completion(struct netvsc_device *net_device, + struct nvsp_message *nvsp_packet; + struct hv_netvsc_packet *nvsc_packet; + struct net_device *ndev; ++ u32 send_index; + + ndev = net_device->ndev; + +@@ -477,6 +618,9 @@ static void netvsc_send_completion(struct netvsc_device *net_device, + + /* Notify the layer above us */ + if (nvsc_packet) { ++ send_index = nvsc_packet->send_buf_index; ++ if (send_index != NETVSC_INVALID_INDEX) ++ netvsc_free_send_slot(net_device, send_index); + q_idx = nvsc_packet->q_idx; + channel = nvsc_packet->channel; + nvsc_packet->send_completion(nvsc_packet-> +@@ -504,6 +648,52 @@ static void netvsc_send_completion(struct netvsc_device *net_device, + + } + ++static u32 netvsc_get_next_send_section(struct netvsc_device *net_device) ++{ ++ unsigned long index; ++ u32 max_words = net_device->map_words; ++ unsigned long *map_addr = (unsigned long *)net_device->send_section_map; ++ u32 section_cnt = net_device->send_section_cnt; ++ int ret_val = NETVSC_INVALID_INDEX; ++ int i; ++ int prev_val; ++ ++ for (i = 0; i < max_words; i++) { ++ if (!~(map_addr[i])) ++ continue; ++ index = ffz(map_addr[i]); ++ prev_val = sync_test_and_set_bit(index, &map_addr[i]); ++ if (prev_val) ++ continue; ++ if ((index + (i * BITS_PER_LONG)) >= section_cnt) ++ break; ++ ret_val = (index + (i * BITS_PER_LONG)); ++ break; ++ } ++ return ret_val; ++} ++ ++u32 netvsc_copy_to_send_buf(struct netvsc_device *net_device, ++ unsigned int section_index, ++ struct hv_netvsc_packet *packet) ++{ ++ char *start = net_device->send_buf; ++ char *dest = (start + (section_index * net_device->send_section_size)); ++ int i; ++ u32 msg_size = 0; ++ ++ for (i = 0; i < packet->page_buf_cnt; i++) { ++ char *src = phys_to_virt(packet->page_buf[i].pfn << PAGE_SHIFT); ++ u32 offset = packet->page_buf[i].offset; ++ u32 len = packet->page_buf[i].len; ++ ++ memcpy(dest, (src + offset), len); ++ msg_size += len; ++ dest += len; ++ } ++ return msg_size; ++} ++ + int netvsc_send(struct hv_device *device, + struct hv_netvsc_packet *packet) + { +@@ -513,6 +703,10 @@ int netvsc_send(struct hv_device *device, + struct net_device *ndev; + struct vmbus_channel *out_channel = NULL; + u64 req_id; ++ unsigned int section_index = NETVSC_INVALID_INDEX; ++ u32 msg_size = 0; ++ struct sk_buff *skb; ++ + + net_device = get_outbound_net_device(device); + if (!net_device) +@@ -528,10 +722,26 @@ int netvsc_send(struct hv_device *device, + sendMessage.msg.v1_msg.send_rndis_pkt.channel_type = 1; + } + +- /* Not using send buffer section */ ++ /* Attempt to send via sendbuf */ ++ if (packet->total_data_buflen < net_device->send_section_size) { ++ section_index = netvsc_get_next_send_section(net_device); ++ if (section_index != NETVSC_INVALID_INDEX) { ++ msg_size = netvsc_copy_to_send_buf(net_device, ++ section_index, ++ packet); ++ skb = (struct sk_buff *) ++ (unsigned long)packet->send_completion_tid; ++ if (skb) ++ dev_kfree_skb_any(skb); ++ packet->page_buf_cnt = 0; ++ } ++ } ++ packet->send_buf_index = section_index; ++ ++ + sendMessage.msg.v1_msg.send_rndis_pkt.send_buf_section_index = +- 0xFFFFFFFF; +- sendMessage.msg.v1_msg.send_rndis_pkt.send_buf_section_size = 0; ++ section_index; ++ sendMessage.msg.v1_msg.send_rndis_pkt.send_buf_section_size = msg_size; + + if (packet->send_completion) + req_id = (ulong)packet; +diff --git a/drivers/net/hyperv/netvsc_drv.c b/drivers/net/hyperv/netvsc_drv.c +index c76b66515e92..939e3af60ec4 100644 +--- a/drivers/net/hyperv/netvsc_drv.c ++++ b/drivers/net/hyperv/netvsc_drv.c +@@ -236,10 +236,11 @@ static void netvsc_xmit_completion(void *context) + struct hv_netvsc_packet *packet = (struct hv_netvsc_packet *)context; + struct sk_buff *skb = (struct sk_buff *) + (unsigned long)packet->send_completion_tid; ++ u32 index = packet->send_buf_index; + + kfree(packet); + +- if (skb) ++ if (skb && (index == NETVSC_INVALID_INDEX)) + dev_kfree_skb_any(skb); + } + +-- +2.4.3 + diff --git a/src/patches/linux/0005-Add-support-for-netvsc-build-without-CONFIG_SYSFS-fl.patch b/src/patches/linux/0005-Add-support-for-netvsc-build-without-CONFIG_SYSFS-fl.patch new file mode 100644 index 0000000..4d00ad2 --- /dev/null +++ b/src/patches/linux/0005-Add-support-for-netvsc-build-without-CONFIG_SYSFS-fl.patch @@ -0,0 +1,42 @@ +From e565e803d437b36c4fb4ced5e346827981183284 Mon Sep 17 00:00:00 2001 +From: Haiyang Zhang haiyangz@microsoft.com +Date: Thu, 8 May 2014 15:14:10 -0700 +Subject: [PATCH 05/11] Add support for netvsc build without CONFIG_SYSFS flag + +This change ensures the driver can be built successfully without the +CONFIG_SYSFS flag. +MS-TFS: 182270 + +Signed-off-by: Haiyang Zhang haiyangz@microsoft.com +Reviewed-by: K. Y. Srinivasan kys@microsoft.com +Signed-off-by: David S. Miller davem@davemloft.net +--- + drivers/net/hyperv/netvsc_drv.c | 5 +---- + 1 file changed, 1 insertion(+), 4 deletions(-) + +diff --git a/drivers/net/hyperv/netvsc_drv.c b/drivers/net/hyperv/netvsc_drv.c +index 939e3af60ec4..083d084396d3 100644 +--- a/drivers/net/hyperv/netvsc_drv.c ++++ b/drivers/net/hyperv/netvsc_drv.c +@@ -640,8 +640,7 @@ int netvsc_recv_callback(struct hv_device *device_obj, + packet->vlan_tci); + + skb_record_rx_queue(skb, packet->channel-> +- offermsg.offer.sub_channel_index % +- net->real_num_rx_queues); ++ offermsg.offer.sub_channel_index); + + net->stats.rx_packets++; + net->stats.rx_bytes += packet->total_data_buflen; +@@ -824,8 +823,6 @@ static int netvsc_probe(struct hv_device *dev, + nvdev = hv_get_drvdata(dev); + netif_set_real_num_tx_queues(net, nvdev->num_chn); + netif_set_real_num_rx_queues(net, nvdev->num_chn); +- dev_info(&dev->device, "real num tx,rx queues:%u, %u\n", +- net->real_num_tx_queues, net->real_num_rx_queues); + + ret = register_netdev(net); + if (ret != 0) { +-- +2.4.3 + diff --git a/src/patches/linux/0005-Drivers-net-hyperv-Enable-scatter-gather-I-O.patch b/src/patches/linux/0005-Drivers-net-hyperv-Enable-scatter-gather-I-O.patch new file mode 100644 index 0000000..27fb2d9 --- /dev/null +++ b/src/patches/linux/0005-Drivers-net-hyperv-Enable-scatter-gather-I-O.patch @@ -0,0 +1,212 @@ +From 4c06034001e20ff9f6e2a1a3dfa155bf3f31440c Mon Sep 17 00:00:00 2001 +From: KY Srinivasan kys@microsoft.com +Date: Sat, 8 Mar 2014 19:23:13 -0800 +Subject: [PATCH 05/25] Drivers: net: hyperv: Enable scatter gather I/O + +Cleanup the code and enable scatter gather I/O. + +Signed-off-by: K. Y. Srinivasan kys@microsoft.com +Reviewed-by: Haiyang Zhang haiyangz@microsoft.com +Signed-off-by: David S. Miller davem@davemloft.net +--- + drivers/net/hyperv/netvsc_drv.c | 153 ++++++++++++++++++++++++++++++---------- + 1 file changed, 114 insertions(+), 39 deletions(-) + +diff --git a/drivers/net/hyperv/netvsc_drv.c b/drivers/net/hyperv/netvsc_drv.c +index 8e3a0b00099b..72961741be54 100644 +--- a/drivers/net/hyperv/netvsc_drv.c ++++ b/drivers/net/hyperv/netvsc_drv.c +@@ -140,22 +140,124 @@ static void netvsc_xmit_completion(void *context) + dev_kfree_skb_any(skb); + } + ++static u32 fill_pg_buf(struct page *page, u32 offset, u32 len, ++ struct hv_page_buffer *pb) ++{ ++ int j = 0; ++ ++ /* Deal with compund pages by ignoring unused part ++ * of the page. ++ */ ++ page += (offset >> PAGE_SHIFT); ++ offset &= ~PAGE_MASK; ++ ++ while (len > 0) { ++ unsigned long bytes; ++ ++ bytes = PAGE_SIZE - offset; ++ if (bytes > len) ++ bytes = len; ++ pb[j].pfn = page_to_pfn(page); ++ pb[j].offset = offset; ++ pb[j].len = bytes; ++ ++ offset += bytes; ++ len -= bytes; ++ ++ if (offset == PAGE_SIZE && len) { ++ page++; ++ offset = 0; ++ j++; ++ } ++ } ++ ++ return j + 1; ++} ++ ++static void init_page_array(void *hdr, u32 len, struct sk_buff *skb, ++ struct hv_page_buffer *pb) ++{ ++ u32 slots_used = 0; ++ char *data = skb->data; ++ int frags = skb_shinfo(skb)->nr_frags; ++ int i; ++ ++ /* The packet is laid out thus: ++ * 1. hdr ++ * 2. skb linear data ++ * 3. skb fragment data ++ */ ++ if (hdr != NULL) ++ slots_used += fill_pg_buf(virt_to_page(hdr), ++ offset_in_page(hdr), ++ len, &pb[slots_used]); ++ ++ slots_used += fill_pg_buf(virt_to_page(data), ++ offset_in_page(data), ++ skb_headlen(skb), &pb[slots_used]); ++ ++ for (i = 0; i < frags; i++) { ++ skb_frag_t *frag = skb_shinfo(skb)->frags + i; ++ ++ slots_used += fill_pg_buf(skb_frag_page(frag), ++ frag->page_offset, ++ skb_frag_size(frag), &pb[slots_used]); ++ } ++} ++ ++static int count_skb_frag_slots(struct sk_buff *skb) ++{ ++ int i, frags = skb_shinfo(skb)->nr_frags; ++ int pages = 0; ++ ++ for (i = 0; i < frags; i++) { ++ skb_frag_t *frag = skb_shinfo(skb)->frags + i; ++ unsigned long size = skb_frag_size(frag); ++ unsigned long offset = frag->page_offset; ++ ++ /* Skip unused frames from start of page */ ++ offset &= ~PAGE_MASK; ++ pages += PFN_UP(offset + size); ++ } ++ return pages; ++} ++ ++static int netvsc_get_slots(struct sk_buff *skb) ++{ ++ char *data = skb->data; ++ unsigned int offset = offset_in_page(data); ++ unsigned int len = skb_headlen(skb); ++ int slots; ++ int frag_slots; ++ ++ slots = DIV_ROUND_UP(offset + len, PAGE_SIZE); ++ frag_slots = count_skb_frag_slots(skb); ++ return slots + frag_slots; ++} ++ + static int netvsc_start_xmit(struct sk_buff *skb, struct net_device *net) + { + struct net_device_context *net_device_ctx = netdev_priv(net); + struct hv_netvsc_packet *packet; + int ret; +- unsigned int i, num_pages, npg_data; ++ unsigned int num_data_pages; + u32 skb_length = skb->len; + +- /* Add multipages for skb->data and additional 2 for RNDIS */ +- npg_data = (((unsigned long)skb->data + skb_headlen(skb) - 1) +- >> PAGE_SHIFT) - ((unsigned long)skb->data >> PAGE_SHIFT) + 1; +- num_pages = skb_shinfo(skb)->nr_frags + npg_data + 2; ++ /* We will atmost need two pages to describe the rndis ++ * header. We can only transmit MAX_PAGE_BUFFER_COUNT number ++ * of pages in a single packet. ++ */ ++ num_data_pages = netvsc_get_slots(skb) + 2; ++ if (num_data_pages > MAX_PAGE_BUFFER_COUNT) { ++ netdev_err(net, "Packet too big: %u\n", skb->len); ++ dev_kfree_skb(skb); ++ net->stats.tx_dropped++; ++ return NETDEV_TX_OK; ++ } + + /* Allocate a netvsc packet based on # of frags. */ + packet = kzalloc(sizeof(struct hv_netvsc_packet) + +- (num_pages * sizeof(struct hv_page_buffer)) + ++ (num_data_pages * sizeof(struct hv_page_buffer)) + + sizeof(struct rndis_message) + + NDIS_VLAN_PPI_SIZE, GFP_ATOMIC); + if (!packet) { +@@ -170,44 +272,17 @@ static int netvsc_start_xmit(struct sk_buff *skb, struct net_device *net) + packet->vlan_tci = skb->vlan_tci; + + packet->extension = (void *)(unsigned long)packet + +- sizeof(struct hv_netvsc_packet) + +- (num_pages * sizeof(struct hv_page_buffer)); ++ sizeof(struct hv_netvsc_packet) + ++ (num_data_pages * sizeof(struct hv_page_buffer)); + + /* If the rndis msg goes beyond 1 page, we will add 1 later */ +- packet->page_buf_cnt = num_pages - 1; ++ packet->page_buf_cnt = num_data_pages - 1; + + /* Initialize it from the skb */ + packet->total_data_buflen = skb->len; + + /* Start filling in the page buffers starting after RNDIS buffer. */ +- packet->page_buf[1].pfn = virt_to_phys(skb->data) >> PAGE_SHIFT; +- packet->page_buf[1].offset +- = (unsigned long)skb->data & (PAGE_SIZE - 1); +- if (npg_data == 1) +- packet->page_buf[1].len = skb_headlen(skb); +- else +- packet->page_buf[1].len = PAGE_SIZE +- - packet->page_buf[1].offset; +- +- for (i = 2; i <= npg_data; i++) { +- packet->page_buf[i].pfn = virt_to_phys(skb->data +- + PAGE_SIZE * (i-1)) >> PAGE_SHIFT; +- packet->page_buf[i].offset = 0; +- packet->page_buf[i].len = PAGE_SIZE; +- } +- if (npg_data > 1) +- packet->page_buf[npg_data].len = (((unsigned long)skb->data +- + skb_headlen(skb) - 1) & (PAGE_SIZE - 1)) + 1; +- +- /* Additional fragments are after SKB data */ +- for (i = 0; i < skb_shinfo(skb)->nr_frags; i++) { +- const skb_frag_t *f = &skb_shinfo(skb)->frags[i]; +- +- packet->page_buf[i+npg_data+1].pfn = +- page_to_pfn(skb_frag_page(f)); +- packet->page_buf[i+npg_data+1].offset = f->page_offset; +- packet->page_buf[i+npg_data+1].len = skb_frag_size(f); +- } ++ init_page_array(NULL, 0, skb, &packet->page_buf[1]); + + /* Set the completion routine */ + packet->completion.send.send_completion = netvsc_xmit_completion; +@@ -454,8 +529,8 @@ static int netvsc_probe(struct hv_device *dev, + net->netdev_ops = &device_ops; + + /* TODO: Add GSO and Checksum offload */ +- net->hw_features = 0; +- net->features = NETIF_F_HW_VLAN_CTAG_TX; ++ net->hw_features = NETIF_F_SG; ++ net->features = NETIF_F_HW_VLAN_CTAG_TX | NETIF_F_SG; + + SET_ETHTOOL_OPS(net, ðtool_ops); + SET_NETDEV_DEV(net, &dev->device); +-- +2.4.3 + diff --git a/src/patches/linux/0006-Drivers-net-hyperv-Cleanup-the-send-path.patch b/src/patches/linux/0006-Drivers-net-hyperv-Cleanup-the-send-path.patch new file mode 100644 index 0000000..20fcb3f --- /dev/null +++ b/src/patches/linux/0006-Drivers-net-hyperv-Cleanup-the-send-path.patch @@ -0,0 +1,266 @@ +From d972eb71fb95660fe74616901b55b0d7a336daed Mon Sep 17 00:00:00 2001 +From: KY Srinivasan kys@microsoft.com +Date: Sat, 8 Mar 2014 19:23:14 -0800 +Subject: [PATCH 06/25] Drivers: net: hyperv: Cleanup the send path + +In preparation for enabling offloads, cleanup the send path. + +Signed-off-by: K. Y. Srinivasan kys@microsoft.com +Reviewed-by: Haiyang Zhang haiyangz@microsoft.com +Signed-off-by: David S. Miller davem@davemloft.net +--- + drivers/net/hyperv/hyperv_net.h | 7 +--- + drivers/net/hyperv/netvsc_drv.c | 88 +++++++++++++++++++++++++++++++-------- + drivers/net/hyperv/rndis_filter.c | 66 ----------------------------- + 3 files changed, 71 insertions(+), 90 deletions(-) + +diff --git a/drivers/net/hyperv/hyperv_net.h b/drivers/net/hyperv/hyperv_net.h +index 39fc230f5c20..694bf7cada90 100644 +--- a/drivers/net/hyperv/hyperv_net.h ++++ b/drivers/net/hyperv/hyperv_net.h +@@ -73,7 +73,7 @@ struct hv_netvsc_packet { + } completion; + + /* This points to the memory after page_buf */ +- void *extension; ++ struct rndis_message *rndis_msg; + + u32 total_data_buflen; + /* Points to the send/receive buffer where the ethernet frame is */ +@@ -126,11 +126,6 @@ void rndis_filter_device_remove(struct hv_device *dev); + int rndis_filter_receive(struct hv_device *dev, + struct hv_netvsc_packet *pkt); + +- +- +-int rndis_filter_send(struct hv_device *dev, +- struct hv_netvsc_packet *pkt); +- + int rndis_filter_set_packet_filter(struct rndis_device *dev, u32 new_filter); + int rndis_filter_set_device_mac(struct hv_device *hdev, char *mac); + +diff --git a/drivers/net/hyperv/netvsc_drv.c b/drivers/net/hyperv/netvsc_drv.c +index 72961741be54..87293a15e470 100644 +--- a/drivers/net/hyperv/netvsc_drv.c ++++ b/drivers/net/hyperv/netvsc_drv.c +@@ -128,6 +128,27 @@ static int netvsc_close(struct net_device *net) + return ret; + } + ++static void *init_ppi_data(struct rndis_message *msg, u32 ppi_size, ++ int pkt_type) ++{ ++ struct rndis_packet *rndis_pkt; ++ struct rndis_per_packet_info *ppi; ++ ++ rndis_pkt = &msg->msg.pkt; ++ rndis_pkt->data_offset += ppi_size; ++ ++ ppi = (struct rndis_per_packet_info *)((void *)rndis_pkt + ++ rndis_pkt->per_pkt_info_offset + rndis_pkt->per_pkt_info_len); ++ ++ ppi->size = ppi_size; ++ ppi->type = pkt_type; ++ ppi->ppi_offset = sizeof(struct rndis_per_packet_info); ++ ++ rndis_pkt->per_pkt_info_len += ppi_size; ++ ++ return ppi; ++} ++ + static void netvsc_xmit_completion(void *context) + { + struct hv_netvsc_packet *packet = (struct hv_netvsc_packet *)context; +@@ -174,8 +195,8 @@ static u32 fill_pg_buf(struct page *page, u32 offset, u32 len, + return j + 1; + } + +-static void init_page_array(void *hdr, u32 len, struct sk_buff *skb, +- struct hv_page_buffer *pb) ++static u32 init_page_array(void *hdr, u32 len, struct sk_buff *skb, ++ struct hv_page_buffer *pb) + { + u32 slots_used = 0; + char *data = skb->data; +@@ -203,6 +224,7 @@ static void init_page_array(void *hdr, u32 len, struct sk_buff *skb, + frag->page_offset, + skb_frag_size(frag), &pb[slots_used]); + } ++ return slots_used; + } + + static int count_skb_frag_slots(struct sk_buff *skb) +@@ -240,15 +262,20 @@ static int netvsc_start_xmit(struct sk_buff *skb, struct net_device *net) + struct net_device_context *net_device_ctx = netdev_priv(net); + struct hv_netvsc_packet *packet; + int ret; +- unsigned int num_data_pages; + u32 skb_length = skb->len; ++ unsigned int num_data_pgs; ++ struct rndis_message *rndis_msg; ++ struct rndis_packet *rndis_pkt; ++ u32 rndis_msg_size; ++ bool isvlan; ++ struct rndis_per_packet_info *ppi; + + /* We will atmost need two pages to describe the rndis + * header. We can only transmit MAX_PAGE_BUFFER_COUNT number + * of pages in a single packet. + */ +- num_data_pages = netvsc_get_slots(skb) + 2; +- if (num_data_pages > MAX_PAGE_BUFFER_COUNT) { ++ num_data_pgs = netvsc_get_slots(skb) + 2; ++ if (num_data_pgs > MAX_PAGE_BUFFER_COUNT) { + netdev_err(net, "Packet too big: %u\n", skb->len); + dev_kfree_skb(skb); + net->stats.tx_dropped++; +@@ -257,7 +284,7 @@ static int netvsc_start_xmit(struct sk_buff *skb, struct net_device *net) + + /* Allocate a netvsc packet based on # of frags. */ + packet = kzalloc(sizeof(struct hv_netvsc_packet) + +- (num_data_pages * sizeof(struct hv_page_buffer)) + ++ (num_data_pgs * sizeof(struct hv_page_buffer)) + + sizeof(struct rndis_message) + + NDIS_VLAN_PPI_SIZE, GFP_ATOMIC); + if (!packet) { +@@ -271,26 +298,51 @@ static int netvsc_start_xmit(struct sk_buff *skb, struct net_device *net) + + packet->vlan_tci = skb->vlan_tci; + +- packet->extension = (void *)(unsigned long)packet + +- sizeof(struct hv_netvsc_packet) + +- (num_data_pages * sizeof(struct hv_page_buffer)); +- +- /* If the rndis msg goes beyond 1 page, we will add 1 later */ +- packet->page_buf_cnt = num_data_pages - 1; +- +- /* Initialize it from the skb */ ++ packet->is_data_pkt = true; + packet->total_data_buflen = skb->len; + +- /* Start filling in the page buffers starting after RNDIS buffer. */ +- init_page_array(NULL, 0, skb, &packet->page_buf[1]); ++ packet->rndis_msg = (struct rndis_message *)((unsigned long)packet + ++ sizeof(struct hv_netvsc_packet) + ++ (num_data_pgs * sizeof(struct hv_page_buffer))); + + /* Set the completion routine */ + packet->completion.send.send_completion = netvsc_xmit_completion; + packet->completion.send.send_completion_ctx = packet; + packet->completion.send.send_completion_tid = (unsigned long)skb; + +- ret = rndis_filter_send(net_device_ctx->device_ctx, +- packet); ++ isvlan = packet->vlan_tci & VLAN_TAG_PRESENT; ++ ++ /* Add the rndis header */ ++ rndis_msg = packet->rndis_msg; ++ rndis_msg->ndis_msg_type = RNDIS_MSG_PACKET; ++ rndis_msg->msg_len = packet->total_data_buflen; ++ rndis_pkt = &rndis_msg->msg.pkt; ++ rndis_pkt->data_offset = sizeof(struct rndis_packet); ++ rndis_pkt->data_len = packet->total_data_buflen; ++ rndis_pkt->per_pkt_info_offset = sizeof(struct rndis_packet); ++ ++ rndis_msg_size = RNDIS_MESSAGE_SIZE(struct rndis_packet); ++ ++ if (isvlan) { ++ struct ndis_pkt_8021q_info *vlan; ++ ++ rndis_msg_size += NDIS_VLAN_PPI_SIZE; ++ ppi = init_ppi_data(rndis_msg, NDIS_VLAN_PPI_SIZE, ++ IEEE_8021Q_INFO); ++ vlan = (struct ndis_pkt_8021q_info *)((void *)ppi + ++ ppi->ppi_offset); ++ vlan->vlanid = packet->vlan_tci & VLAN_VID_MASK; ++ vlan->pri = (packet->vlan_tci & VLAN_PRIO_MASK) >> ++ VLAN_PRIO_SHIFT; ++ } ++ ++ /* Start filling in the page buffers with the rndis hdr */ ++ rndis_msg->msg_len += rndis_msg_size; ++ packet->page_buf_cnt = init_page_array(rndis_msg, rndis_msg_size, ++ skb, &packet->page_buf[0]); ++ ++ ret = netvsc_send(net_device_ctx->device_ctx, packet); ++ + if (ret == 0) { + net->stats.tx_bytes += skb_length; + net->stats.tx_packets++; +diff --git a/drivers/net/hyperv/rndis_filter.c b/drivers/net/hyperv/rndis_filter.c +index 6a9f6021f09c..dcbf144ea7da 100644 +--- a/drivers/net/hyperv/rndis_filter.c ++++ b/drivers/net/hyperv/rndis_filter.c +@@ -910,69 +910,3 @@ int rndis_filter_close(struct hv_device *dev) + + return rndis_filter_close_device(nvdev->extension); + } +- +-int rndis_filter_send(struct hv_device *dev, +- struct hv_netvsc_packet *pkt) +-{ +- struct rndis_message *rndis_msg; +- struct rndis_packet *rndis_pkt; +- u32 rndis_msg_size; +- bool isvlan = pkt->vlan_tci & VLAN_TAG_PRESENT; +- +- /* Add the rndis header */ +- rndis_msg = (struct rndis_message *)pkt->extension; +- +- rndis_msg_size = RNDIS_MESSAGE_SIZE(struct rndis_packet); +- if (isvlan) +- rndis_msg_size += NDIS_VLAN_PPI_SIZE; +- +- rndis_msg->ndis_msg_type = RNDIS_MSG_PACKET; +- rndis_msg->msg_len = pkt->total_data_buflen + +- rndis_msg_size; +- +- rndis_pkt = &rndis_msg->msg.pkt; +- rndis_pkt->data_offset = sizeof(struct rndis_packet); +- if (isvlan) +- rndis_pkt->data_offset += NDIS_VLAN_PPI_SIZE; +- rndis_pkt->data_len = pkt->total_data_buflen; +- +- if (isvlan) { +- struct rndis_per_packet_info *ppi; +- struct ndis_pkt_8021q_info *vlan; +- +- rndis_pkt->per_pkt_info_offset = sizeof(struct rndis_packet); +- rndis_pkt->per_pkt_info_len = NDIS_VLAN_PPI_SIZE; +- +- ppi = (struct rndis_per_packet_info *)((ulong)rndis_pkt + +- rndis_pkt->per_pkt_info_offset); +- ppi->size = NDIS_VLAN_PPI_SIZE; +- ppi->type = IEEE_8021Q_INFO; +- ppi->ppi_offset = sizeof(struct rndis_per_packet_info); +- +- vlan = (struct ndis_pkt_8021q_info *)((ulong)ppi + +- ppi->ppi_offset); +- vlan->vlanid = pkt->vlan_tci & VLAN_VID_MASK; +- vlan->pri = (pkt->vlan_tci & VLAN_PRIO_MASK) >> VLAN_PRIO_SHIFT; +- } +- +- pkt->is_data_pkt = true; +- pkt->page_buf[0].pfn = virt_to_phys(rndis_msg) >> PAGE_SHIFT; +- pkt->page_buf[0].offset = +- (unsigned long)rndis_msg & (PAGE_SIZE-1); +- pkt->page_buf[0].len = rndis_msg_size; +- +- /* Add one page_buf if the rndis msg goes beyond page boundary */ +- if (pkt->page_buf[0].offset + rndis_msg_size > PAGE_SIZE) { +- int i; +- for (i = pkt->page_buf_cnt; i > 1; i--) +- pkt->page_buf[i] = pkt->page_buf[i-1]; +- pkt->page_buf_cnt++; +- pkt->page_buf[0].len = PAGE_SIZE - pkt->page_buf[0].offset; +- pkt->page_buf[1].pfn = virt_to_phys((void *)((ulong) +- rndis_msg + pkt->page_buf[0].len)) >> PAGE_SHIFT; +- pkt->page_buf[1].offset = 0; +- pkt->page_buf[1].len = rndis_msg_size - pkt->page_buf[0].len; +- } +- +- return netvsc_send(dev, pkt); +-} +-- +2.4.3 + diff --git a/src/patches/linux/0006-net-get-rid-of-SET_ETHTOOL_OPS.patch b/src/patches/linux/0006-net-get-rid-of-SET_ETHTOOL_OPS.patch new file mode 100644 index 0000000..e4a7a46 --- /dev/null +++ b/src/patches/linux/0006-net-get-rid-of-SET_ETHTOOL_OPS.patch @@ -0,0 +1,44 @@ +From 7ad24ea4bf620a32631d7b3069c3e30c078b0c3e Mon Sep 17 00:00:00 2001 +From: Wilfried Klaebe w-lkml@lebenslange-mailadresse.de +Date: Sun, 11 May 2014 00:12:32 +0000 +Subject: [PATCH 06/11] net: get rid of SET_ETHTOOL_OPS + +net: get rid of SET_ETHTOOL_OPS + +Dave Miller mentioned he'd like to see SET_ETHTOOL_OPS gone. +This does that. + +Mostly done via coccinelle script: +@@ +struct ethtool_ops *ops; +struct net_device *dev; +@@ +- SET_ETHTOOL_OPS(dev, ops); ++ dev->ethtool_ops = ops; + +Compile tested only, but I'd seriously wonder if this broke anything. + +Suggested-by: Dave Miller davem@davemloft.net +Signed-off-by: Wilfried Klaebe w-lkml@lebenslange-mailadresse.de +Acked-by: Felipe Balbi balbi@ti.com +Signed-off-by: David S. Miller davem@davemloft.net +--- + drivers/net/hyperv/netvsc_drv.c | 2 +- + 1 file changed, 1 insertion(+), 1 deletion(-) + +diff --git a/drivers/net/hyperv/netvsc_drv.c b/drivers/net/hyperv/netvsc_drv.c +index 1de3ef5dd5d2..2e967a7bdb33 100644 +--- a/drivers/net/hyperv/netvsc_drv.c ++++ b/drivers/net/hyperv/netvsc_drv.c +@@ -810,7 +810,7 @@ static int netvsc_probe(struct hv_device *dev, + net->features = NETIF_F_HW_VLAN_CTAG_TX | NETIF_F_SG | NETIF_F_RXCSUM | + NETIF_F_IP_CSUM | NETIF_F_TSO; + +- SET_ETHTOOL_OPS(net, ðtool_ops); ++ net->ethtool_ops = ðtool_ops; + SET_NETDEV_DEV(net, &dev->device); + + /* Notify the netvsc driver of the new device */ +-- +2.4.3 + diff --git a/src/patches/linux/0007-Drivers-net-hyperv-Enable-offloads-on-the-host.patch b/src/patches/linux/0007-Drivers-net-hyperv-Enable-offloads-on-the-host.patch new file mode 100644 index 0000000..a109f04 --- /dev/null +++ b/src/patches/linux/0007-Drivers-net-hyperv-Enable-offloads-on-the-host.patch @@ -0,0 +1,196 @@ +From 6b15b5f37e976a5f3840c7ea59560e10c6251250 Mon Sep 17 00:00:00 2001 +From: KY Srinivasan kys@microsoft.com +Date: Sat, 8 Mar 2014 19:23:15 -0800 +Subject: [PATCH 07/25] Drivers: net: hyperv: Enable offloads on the host + +Prior to enabling guest side offloads, enable the offloads on the host. + +Signed-off-by: K. Y. Srinivasan kys@microsoft.com +Reviewed-by: Haiyang Zhang haiyangz@microsoft.com +Signed-off-by: David S. Miller davem@davemloft.net +--- + drivers/net/hyperv/hyperv_net.h | 55 +++++++++++++++++++++++++++ + drivers/net/hyperv/rndis_filter.c | 80 +++++++++++++++++++++++++++++++++++++++ + 2 files changed, 135 insertions(+) + +diff --git a/drivers/net/hyperv/hyperv_net.h b/drivers/net/hyperv/hyperv_net.h +index 694bf7cada90..8bc4e766589b 100644 +--- a/drivers/net/hyperv/hyperv_net.h ++++ b/drivers/net/hyperv/hyperv_net.h +@@ -721,6 +721,61 @@ struct ndis_pkt_8021q_info { + }; + }; + ++struct ndis_oject_header { ++ u8 type; ++ u8 revision; ++ u16 size; ++}; ++ ++#define NDIS_OBJECT_TYPE_DEFAULT 0x80 ++#define NDIS_OFFLOAD_PARAMETERS_REVISION_3 3 ++#define NDIS_OFFLOAD_PARAMETERS_NO_CHANGE 0 ++#define NDIS_OFFLOAD_PARAMETERS_LSOV2_DISABLED 1 ++#define NDIS_OFFLOAD_PARAMETERS_LSOV2_ENABLED 2 ++#define NDIS_OFFLOAD_PARAMETERS_LSOV1_ENABLED 2 ++#define NDIS_OFFLOAD_PARAMETERS_RSC_DISABLED 1 ++#define NDIS_OFFLOAD_PARAMETERS_RSC_ENABLED 2 ++#define NDIS_OFFLOAD_PARAMETERS_TX_RX_DISABLED 1 ++#define NDIS_OFFLOAD_PARAMETERS_TX_ENABLED_RX_DISABLED 2 ++#define NDIS_OFFLOAD_PARAMETERS_RX_ENABLED_TX_DISABLED 3 ++#define NDIS_OFFLOAD_PARAMETERS_TX_RX_ENABLED 4 ++ ++/* ++ * New offload OIDs for NDIS 6 ++ */ ++#define OID_TCP_OFFLOAD_CURRENT_CONFIG 0xFC01020B /* query only */ ++#define OID_TCP_OFFLOAD_PARAMETERS 0xFC01020C /* set only */ ++#define OID_TCP_OFFLOAD_HARDWARE_CAPABILITIES 0xFC01020D/* query only */ ++#define OID_TCP_CONNECTION_OFFLOAD_CURRENT_CONFIG 0xFC01020E /* query only */ ++#define OID_TCP_CONNECTION_OFFLOAD_HARDWARE_CAPABILITIES 0xFC01020F /* query */ ++#define OID_OFFLOAD_ENCAPSULATION 0x0101010A /* set/query */ ++ ++struct ndis_offload_params { ++ struct ndis_oject_header header; ++ u8 ip_v4_csum; ++ u8 tcp_ip_v4_csum; ++ u8 udp_ip_v4_csum; ++ u8 tcp_ip_v6_csum; ++ u8 udp_ip_v6_csum; ++ u8 lso_v1; ++ u8 ip_sec_v1; ++ u8 lso_v2_ipv4; ++ u8 lso_v2_ipv6; ++ u8 tcp_connection_ip_v4; ++ u8 tcp_connection_ip_v6; ++ u32 flags; ++ u8 ip_sec_v2; ++ u8 ip_sec_v2_ip_v4; ++ struct { ++ u8 rsc_ip_v4; ++ u8 rsc_ip_v6; ++ }; ++ struct { ++ u8 encapsulated_packet_task_offload; ++ u8 encapsulation_types; ++ }; ++}; ++ + #define NDIS_VLAN_PPI_SIZE (sizeof(struct rndis_per_packet_info) + \ + sizeof(struct ndis_pkt_8021q_info)) + +diff --git a/drivers/net/hyperv/rndis_filter.c b/drivers/net/hyperv/rndis_filter.c +index dcbf144ea7da..9b02f21097a7 100644 +--- a/drivers/net/hyperv/rndis_filter.c ++++ b/drivers/net/hyperv/rndis_filter.c +@@ -627,6 +627,61 @@ cleanup: + return ret; + } + ++int rndis_filter_set_offload_params(struct hv_device *hdev, ++ struct ndis_offload_params *req_offloads) ++{ ++ struct netvsc_device *nvdev = hv_get_drvdata(hdev); ++ struct rndis_device *rdev = nvdev->extension; ++ struct net_device *ndev = nvdev->ndev; ++ struct rndis_request *request; ++ struct rndis_set_request *set; ++ struct ndis_offload_params *offload_params; ++ struct rndis_set_complete *set_complete; ++ u32 extlen = sizeof(struct ndis_offload_params); ++ int ret, t; ++ ++ request = get_rndis_request(rdev, RNDIS_MSG_SET, ++ RNDIS_MESSAGE_SIZE(struct rndis_set_request) + extlen); ++ if (!request) ++ return -ENOMEM; ++ ++ set = &request->request_msg.msg.set_req; ++ set->oid = OID_TCP_OFFLOAD_PARAMETERS; ++ set->info_buflen = extlen; ++ set->info_buf_offset = sizeof(struct rndis_set_request); ++ set->dev_vc_handle = 0; ++ ++ offload_params = (struct ndis_offload_params *)((ulong)set + ++ set->info_buf_offset); ++ *offload_params = *req_offloads; ++ offload_params->header.type = NDIS_OBJECT_TYPE_DEFAULT; ++ offload_params->header.revision = NDIS_OFFLOAD_PARAMETERS_REVISION_3; ++ offload_params->header.size = extlen; ++ ++ ret = rndis_filter_send_request(rdev, request); ++ if (ret != 0) ++ goto cleanup; ++ ++ t = wait_for_completion_timeout(&request->wait_event, 5*HZ); ++ if (t == 0) { ++ netdev_err(ndev, "timeout before we got aOFFLOAD set response...\n"); ++ /* can't put_rndis_request, since we may still receive a ++ * send-completion. ++ */ ++ return -EBUSY; ++ } else { ++ set_complete = &request->response_msg.msg.set_complete; ++ if (set_complete->status != RNDIS_STATUS_SUCCESS) { ++ netdev_err(ndev, "Fail to set MAC on host side:0x%x\n", ++ set_complete->status); ++ ret = -EINVAL; ++ } ++ } ++ ++cleanup: ++ put_rndis_request(rdev, request); ++ return ret; ++} + + static int rndis_filter_query_device_link_status(struct rndis_device *dev) + { +@@ -826,6 +881,7 @@ int rndis_filter_device_add(struct hv_device *dev, + struct netvsc_device *net_device; + struct rndis_device *rndis_device; + struct netvsc_device_info *device_info = additional_info; ++ struct ndis_offload_params offloads; + + rndis_device = get_rndis_device(); + if (!rndis_device) +@@ -865,6 +921,26 @@ int rndis_filter_device_add(struct hv_device *dev, + + memcpy(device_info->mac_adr, rndis_device->hw_mac_adr, ETH_ALEN); + ++ /* Turn on the offloads; the host supports all of the relevant ++ * offloads. ++ */ ++ memset(&offloads, 0, sizeof(struct ndis_offload_params)); ++ /* A value of zero means "no change"; now turn on what we ++ * want. ++ */ ++ offloads.ip_v4_csum = NDIS_OFFLOAD_PARAMETERS_TX_RX_ENABLED; ++ offloads.tcp_ip_v4_csum = NDIS_OFFLOAD_PARAMETERS_TX_RX_ENABLED; ++ offloads.udp_ip_v4_csum = NDIS_OFFLOAD_PARAMETERS_TX_RX_ENABLED; ++ offloads.tcp_ip_v6_csum = NDIS_OFFLOAD_PARAMETERS_TX_RX_ENABLED; ++ offloads.udp_ip_v6_csum = NDIS_OFFLOAD_PARAMETERS_TX_RX_ENABLED; ++ offloads.lso_v2_ipv4 = NDIS_OFFLOAD_PARAMETERS_LSOV2_ENABLED; ++ ++ ++ ret = rndis_filter_set_offload_params(dev, &offloads); ++ if (ret) ++ goto err_dev_remv; ++ ++ + rndis_filter_query_device_link_status(rndis_device); + + device_info->link_state = rndis_device->link_state; +@@ -874,6 +950,10 @@ int rndis_filter_device_add(struct hv_device *dev, + device_info->link_state ? "down" : "up"); + + return ret; ++ ++err_dev_remv: ++ rndis_filter_device_remove(dev); ++ return ret; + } + + void rndis_filter_device_remove(struct hv_device *dev) +-- +2.4.3 + diff --git a/src/patches/linux/0007-hyperv-Add-hash-value-into-RNDIS-Per-packet-info.patch b/src/patches/linux/0007-hyperv-Add-hash-value-into-RNDIS-Per-packet-info.patch new file mode 100644 index 0000000..0db5f72 --- /dev/null +++ b/src/patches/linux/0007-hyperv-Add-hash-value-into-RNDIS-Per-packet-info.patch @@ -0,0 +1,93 @@ +From 307f099520b66504cf6c5638f3f404c48b9fb45b Mon Sep 17 00:00:00 2001 +From: Haiyang Zhang haiyangz@microsoft.com +Date: Wed, 21 May 2014 12:55:39 -0700 +Subject: [PATCH 07/11] hyperv: Add hash value into RNDIS Per-packet info + +It passes the hash value as the RNDIS Per-packet info to the Hyper-V host, +so that the send completion notices can be spread across multiple channels. +MS-TFS: 140273 + +Signed-off-by: Haiyang Zhang haiyangz@microsoft.com +Signed-off-by: David S. Miller davem@davemloft.net +--- + drivers/net/hyperv/hyperv_net.h | 4 ++++ + drivers/net/hyperv/netvsc_drv.c | 18 ++++++++++++++---- + 2 files changed, 18 insertions(+), 4 deletions(-) + +diff --git a/drivers/net/hyperv/hyperv_net.h b/drivers/net/hyperv/hyperv_net.h +index 4b7df5a5c966..6cc37c15e0bf 100644 +--- a/drivers/net/hyperv/hyperv_net.h ++++ b/drivers/net/hyperv/hyperv_net.h +@@ -791,6 +791,7 @@ enum ndis_per_pkt_info_type { + IEEE_8021Q_INFO, + ORIGINAL_PKTINFO, + PACKET_CANCEL_ID, ++ NBL_HASH_VALUE = PACKET_CANCEL_ID, + ORIGINAL_NET_BUFLIST, + CACHED_NET_BUFLIST, + SHORT_PKT_PADINFO, +@@ -937,6 +938,9 @@ struct ndis_tcp_lso_info { + #define NDIS_LSO_PPI_SIZE (sizeof(struct rndis_per_packet_info) + \ + sizeof(struct ndis_tcp_lso_info)) + ++#define NDIS_HASH_PPI_SIZE (sizeof(struct rndis_per_packet_info) + \ ++ sizeof(u32)) ++ + /* Format of Information buffer passed in a SetRequest for the OID */ + /* OID_GEN_RNDIS_CONFIG_PARAMETER. */ + struct rndis_config_parameter_info { +diff --git a/drivers/net/hyperv/netvsc_drv.c b/drivers/net/hyperv/netvsc_drv.c +index 2e967a7bdb33..4fd71b75e666 100644 +--- a/drivers/net/hyperv/netvsc_drv.c ++++ b/drivers/net/hyperv/netvsc_drv.c +@@ -224,9 +224,11 @@ static u16 netvsc_select_queue(struct net_device *ndev, struct sk_buff *skb, + if (nvsc_dev == NULL || ndev->real_num_tx_queues <= 1) + return 0; + +- if (netvsc_set_hash(&hash, skb)) ++ if (netvsc_set_hash(&hash, skb)) { + q_idx = nvsc_dev->send_table[hash % VRSS_SEND_TAB_SIZE] % + ndev->real_num_tx_queues; ++ skb_set_hash(skb, hash, PKT_HASH_TYPE_L3); ++ } + + return q_idx; + } +@@ -384,6 +386,7 @@ static int netvsc_start_xmit(struct sk_buff *skb, struct net_device *net) + struct ndis_tcp_lso_info *lso_info; + int hdr_offset; + u32 net_trans_info; ++ u32 hash; + + + /* We will atmost need two pages to describe the rndis +@@ -402,9 +405,8 @@ static int netvsc_start_xmit(struct sk_buff *skb, struct net_device *net) + packet = kzalloc(sizeof(struct hv_netvsc_packet) + + (num_data_pgs * sizeof(struct hv_page_buffer)) + + sizeof(struct rndis_message) + +- NDIS_VLAN_PPI_SIZE + +- NDIS_CSUM_PPI_SIZE + +- NDIS_LSO_PPI_SIZE, GFP_ATOMIC); ++ NDIS_VLAN_PPI_SIZE + NDIS_CSUM_PPI_SIZE + ++ NDIS_LSO_PPI_SIZE + NDIS_HASH_PPI_SIZE, GFP_ATOMIC); + if (!packet) { + /* out of memory, drop packet */ + netdev_err(net, "unable to allocate hv_netvsc_packet\n"); +@@ -443,6 +445,14 @@ static int netvsc_start_xmit(struct sk_buff *skb, struct net_device *net) + + rndis_msg_size = RNDIS_MESSAGE_SIZE(struct rndis_packet); + ++ hash = skb_get_hash_raw(skb); ++ if (hash != 0 && net->real_num_tx_queues > 1) { ++ rndis_msg_size += NDIS_HASH_PPI_SIZE; ++ ppi = init_ppi_data(rndis_msg, NDIS_HASH_PPI_SIZE, ++ NBL_HASH_VALUE); ++ *(u32 *)((void *)ppi + ppi->ppi_offset) = hash; ++ } ++ + if (isvlan) { + struct ndis_pkt_8021q_info *vlan; + +-- +2.4.3 + diff --git a/src/patches/linux/0008-Drivers-net-hyperv-Enable-receive-side-IP-checksum-o.patch b/src/patches/linux/0008-Drivers-net-hyperv-Enable-receive-side-IP-checksum-o.patch new file mode 100644 index 0000000..e77e9f2 --- /dev/null +++ b/src/patches/linux/0008-Drivers-net-hyperv-Enable-receive-side-IP-checksum-o.patch @@ -0,0 +1,147 @@ +From 59e0c70c618668522a9431686f7e3a69ef396dff Mon Sep 17 00:00:00 2001 +From: KY Srinivasan kys@microsoft.com +Date: Sat, 8 Mar 2014 19:23:16 -0800 +Subject: [PATCH 08/25] Drivers: net: hyperv: Enable receive side IP checksum + offload + +Enable receive side checksum offload. + +Signed-off-by: K. Y. Srinivasan kys@microsoft.com +Reviewed-by: Haiyang Zhang haiyangz@microsoft.com +Signed-off-by: David S. Miller davem@davemloft.net +--- + drivers/net/hyperv/hyperv_net.h | 33 ++++++++++++++++++++++++++++++++- + drivers/net/hyperv/netvsc_drv.c | 19 +++++++++++++++---- + drivers/net/hyperv/rndis_filter.c | 4 +++- + 3 files changed, 50 insertions(+), 6 deletions(-) + +diff --git a/drivers/net/hyperv/hyperv_net.h b/drivers/net/hyperv/hyperv_net.h +index 8bc4e766589b..faeb74623fbd 100644 +--- a/drivers/net/hyperv/hyperv_net.h ++++ b/drivers/net/hyperv/hyperv_net.h +@@ -30,6 +30,7 @@ + + /* Fwd declaration */ + struct hv_netvsc_packet; ++struct ndis_tcp_ip_checksum_info; + + /* Represent the xfer page packet which contains 1 or more netvsc packet */ + struct xferpage_packet { +@@ -117,7 +118,8 @@ int netvsc_send(struct hv_device *device, + void netvsc_linkstatus_callback(struct hv_device *device_obj, + unsigned int status); + int netvsc_recv_callback(struct hv_device *device_obj, +- struct hv_netvsc_packet *packet); ++ struct hv_netvsc_packet *packet, ++ struct ndis_tcp_ip_checksum_info *csum_info); + int rndis_filter_open(struct hv_device *dev); + int rndis_filter_close(struct hv_device *dev); + int rndis_filter_device_add(struct hv_device *dev, +@@ -776,9 +778,38 @@ struct ndis_offload_params { + }; + }; + ++struct ndis_tcp_ip_checksum_info { ++ union { ++ struct { ++ u32 is_ipv4:1; ++ u32 is_ipv6:1; ++ u32 tcp_checksum:1; ++ u32 udp_checksum:1; ++ u32 ip_header_checksum:1; ++ u32 reserved:11; ++ u32 tcp_header_offset:10; ++ } transmit; ++ struct { ++ u32 tcp_checksum_failed:1; ++ u32 udp_checksum_failed:1; ++ u32 ip_checksum_failed:1; ++ u32 tcp_checksum_succeeded:1; ++ u32 udp_checksum_succeeded:1; ++ u32 ip_checksum_succeeded:1; ++ u32 loopback:1; ++ u32 tcp_checksum_value_invalid:1; ++ u32 ip_checksum_value_invalid:1; ++ } receive; ++ u32 value; ++ }; ++}; ++ + #define NDIS_VLAN_PPI_SIZE (sizeof(struct rndis_per_packet_info) + \ + sizeof(struct ndis_pkt_8021q_info)) + ++#define NDIS_CSUM_PPI_SIZE (sizeof(struct rndis_per_packet_info) + \ ++ sizeof(struct ndis_tcp_ip_checksum_info)) ++ + /* Format of Information buffer passed in a SetRequest for the OID */ + /* OID_GEN_RNDIS_CONFIG_PARAMETER. */ + struct rndis_config_parameter_info { +diff --git a/drivers/net/hyperv/netvsc_drv.c b/drivers/net/hyperv/netvsc_drv.c +index 87293a15e470..7438360b7a02 100644 +--- a/drivers/net/hyperv/netvsc_drv.c ++++ b/drivers/net/hyperv/netvsc_drv.c +@@ -392,7 +392,8 @@ void netvsc_linkstatus_callback(struct hv_device *device_obj, + * "wire" on the specified device. + */ + int netvsc_recv_callback(struct hv_device *device_obj, +- struct hv_netvsc_packet *packet) ++ struct hv_netvsc_packet *packet, ++ struct ndis_tcp_ip_checksum_info *csum_info) + { + struct net_device *net; + struct sk_buff *skb; +@@ -419,7 +420,17 @@ int netvsc_recv_callback(struct hv_device *device_obj, + packet->total_data_buflen); + + skb->protocol = eth_type_trans(skb, net); +- skb->ip_summed = CHECKSUM_NONE; ++ if (csum_info) { ++ /* We only look at the IP checksum here. ++ * Should we be dropping the packet if checksum ++ * failed? How do we deal with other checksums - TCP/UDP? ++ */ ++ if (csum_info->receive.ip_checksum_succeeded) ++ skb->ip_summed = CHECKSUM_UNNECESSARY; ++ else ++ skb->ip_summed = CHECKSUM_NONE; ++ } ++ + if (packet->vlan_tci & VLAN_TAG_PRESENT) + __vlan_hwaccel_put_tag(skb, htons(ETH_P_8021Q), + packet->vlan_tci); +@@ -581,8 +592,8 @@ static int netvsc_probe(struct hv_device *dev, + net->netdev_ops = &device_ops; + + /* TODO: Add GSO and Checksum offload */ +- net->hw_features = NETIF_F_SG; +- net->features = NETIF_F_HW_VLAN_CTAG_TX | NETIF_F_SG; ++ net->hw_features = NETIF_F_RXCSUM | NETIF_F_SG; ++ net->features = NETIF_F_HW_VLAN_CTAG_TX | NETIF_F_SG | NETIF_F_RXCSUM; + + SET_ETHTOOL_OPS(net, ðtool_ops); + SET_NETDEV_DEV(net, &dev->device); +diff --git a/drivers/net/hyperv/rndis_filter.c b/drivers/net/hyperv/rndis_filter.c +index 9b02f21097a7..4a37e3db9e32 100644 +--- a/drivers/net/hyperv/rndis_filter.c ++++ b/drivers/net/hyperv/rndis_filter.c +@@ -370,6 +370,7 @@ static void rndis_filter_receive_data(struct rndis_device *dev, + struct rndis_packet *rndis_pkt; + u32 data_offset; + struct ndis_pkt_8021q_info *vlan; ++ struct ndis_tcp_ip_checksum_info *csum_info; + + rndis_pkt = &msg->msg.pkt; + +@@ -408,7 +409,8 @@ static void rndis_filter_receive_data(struct rndis_device *dev, + pkt->vlan_tci = 0; + } + +- netvsc_recv_callback(dev->net_dev->dev, pkt); ++ csum_info = rndis_get_ppi(rndis_pkt, TCPIP_CHKSUM_PKTINFO); ++ netvsc_recv_callback(dev->net_dev->dev, pkt, csum_info); + } + + int rndis_filter_receive(struct hv_device *dev, +-- +2.4.3 + diff --git a/src/patches/linux/0008-hyperv-fix-apparent-cut-n-paste-error-in-send-path-t.patch b/src/patches/linux/0008-hyperv-fix-apparent-cut-n-paste-error-in-send-path-t.patch new file mode 100644 index 0000000..b774f81 --- /dev/null +++ b/src/patches/linux/0008-hyperv-fix-apparent-cut-n-paste-error-in-send-path-t.patch @@ -0,0 +1,32 @@ +From 2f18423d7ebf8044a9edaca840be5ae199fb32f6 Mon Sep 17 00:00:00 2001 +From: Dave Jones davej@redhat.com +Date: Mon, 16 Jun 2014 16:59:02 -0400 +Subject: [PATCH 08/11] hyperv: fix apparent cut-n-paste error in send path + teardown + +c25aaf814a63: "hyperv: Enable sendbuf mechanism on the send path" added +some teardown code that looks like it was copied from the recieve path +above, but missed a variable name replacement. + +Signed-off-by: Dave Jones davej@redhat.com +Signed-off-by: David S. Miller davem@davemloft.net +--- + drivers/net/hyperv/netvsc.c | 2 +- + 1 file changed, 1 insertion(+), 1 deletion(-) + +diff --git a/drivers/net/hyperv/netvsc.c b/drivers/net/hyperv/netvsc.c +index c041f63a6d30..4ed38eaecea8 100644 +--- a/drivers/net/hyperv/netvsc.c ++++ b/drivers/net/hyperv/netvsc.c +@@ -189,7 +189,7 @@ static int netvsc_destroy_buf(struct netvsc_device *net_device) + "unable to teardown send buffer's gpadl\n"); + return ret; + } +- net_device->recv_buf_gpadl_handle = 0; ++ net_device->send_buf_gpadl_handle = 0; + } + if (net_device->send_buf) { + /* Free up the receive buffer */ +-- +2.4.3 + diff --git a/src/patches/linux/0009-Drivers-net-hyperv-Enable-send-side-checksum-offload.patch b/src/patches/linux/0009-Drivers-net-hyperv-Enable-send-side-checksum-offload.patch new file mode 100644 index 0000000..db42108 --- /dev/null +++ b/src/patches/linux/0009-Drivers-net-hyperv-Enable-send-side-checksum-offload.patch @@ -0,0 +1,140 @@ +From d2a0be7f4263eb669af84240c5424a72cce4cdb4 Mon Sep 17 00:00:00 2001 +From: KY Srinivasan kys@microsoft.com +Date: Sat, 8 Mar 2014 19:23:17 -0800 +Subject: [PATCH 09/25] Drivers: net: hyperv: Enable send side checksum offload + +Enable send side checksum offload. + +Signed-off-by: K. Y. Srinivasan kys@microsoft.com +Reviewed-by: Haiyang Zhang haiyangz@microsoft.com +Signed-off-by: David S. Miller davem@davemloft.net +--- + drivers/net/hyperv/hyperv_net.h | 10 ++++++ + drivers/net/hyperv/netvsc_drv.c | 69 +++++++++++++++++++++++++++++++++++++++-- + 2 files changed, 77 insertions(+), 2 deletions(-) + +diff --git a/drivers/net/hyperv/hyperv_net.h b/drivers/net/hyperv/hyperv_net.h +index faeb74623fbd..4cf238234321 100644 +--- a/drivers/net/hyperv/hyperv_net.h ++++ b/drivers/net/hyperv/hyperv_net.h +@@ -1035,6 +1035,16 @@ struct rndis_message { + #define NDIS_PACKET_TYPE_FUNCTIONAL 0x00000400 + #define NDIS_PACKET_TYPE_MAC_FRAME 0x00000800 + ++#define INFO_IPV4 2 ++#define INFO_IPV6 4 ++#define INFO_TCP 2 ++#define INFO_UDP 4 ++ ++#define TRANSPORT_INFO_NOT_IP 0 ++#define TRANSPORT_INFO_IPV4_TCP ((INFO_IPV4 << 16) | INFO_TCP) ++#define TRANSPORT_INFO_IPV4_UDP ((INFO_IPV4 << 16) | INFO_UDP) ++#define TRANSPORT_INFO_IPV6_TCP ((INFO_IPV6 << 16) | INFO_TCP) ++#define TRANSPORT_INFO_IPV6_UDP ((INFO_IPV6 << 16) | INFO_UDP) + + + #endif /* _HYPERV_NET_H */ +diff --git a/drivers/net/hyperv/netvsc_drv.c b/drivers/net/hyperv/netvsc_drv.c +index 7438360b7a02..2fab69ac61ef 100644 +--- a/drivers/net/hyperv/netvsc_drv.c ++++ b/drivers/net/hyperv/netvsc_drv.c +@@ -257,6 +257,35 @@ static int netvsc_get_slots(struct sk_buff *skb) + return slots + frag_slots; + } + ++static u32 get_net_transport_info(struct sk_buff *skb, u32 *trans_off) ++{ ++ u32 ret_val = TRANSPORT_INFO_NOT_IP; ++ ++ if ((eth_hdr(skb)->h_proto != htons(ETH_P_IP)) && ++ (eth_hdr(skb)->h_proto != htons(ETH_P_IPV6))) { ++ goto not_ip; ++ } ++ ++ *trans_off = skb_transport_offset(skb); ++ ++ if ((eth_hdr(skb)->h_proto == htons(ETH_P_IP))) { ++ struct iphdr *iphdr = ip_hdr(skb); ++ ++ if (iphdr->protocol == IPPROTO_TCP) ++ ret_val = TRANSPORT_INFO_IPV4_TCP; ++ else if (iphdr->protocol == IPPROTO_UDP) ++ ret_val = TRANSPORT_INFO_IPV4_UDP; ++ } else { ++ if (ipv6_hdr(skb)->nexthdr == IPPROTO_TCP) ++ ret_val = TRANSPORT_INFO_IPV6_TCP; ++ else if (ipv6_hdr(skb)->nexthdr == IPPROTO_UDP) ++ ret_val = TRANSPORT_INFO_IPV6_UDP; ++ } ++ ++not_ip: ++ return ret_val; ++} ++ + static int netvsc_start_xmit(struct sk_buff *skb, struct net_device *net) + { + struct net_device_context *net_device_ctx = netdev_priv(net); +@@ -269,6 +298,10 @@ static int netvsc_start_xmit(struct sk_buff *skb, struct net_device *net) + u32 rndis_msg_size; + bool isvlan; + struct rndis_per_packet_info *ppi; ++ struct ndis_tcp_ip_checksum_info *csum_info; ++ int hdr_offset; ++ u32 net_trans_info; ++ + + /* We will atmost need two pages to describe the rndis + * header. We can only transmit MAX_PAGE_BUFFER_COUNT number +@@ -336,6 +369,37 @@ static int netvsc_start_xmit(struct sk_buff *skb, struct net_device *net) + VLAN_PRIO_SHIFT; + } + ++ net_trans_info = get_net_transport_info(skb, &hdr_offset); ++ if (net_trans_info == TRANSPORT_INFO_NOT_IP) ++ goto do_send; ++ ++ /* ++ * Setup the sendside checksum offload only if this is not a ++ * GSO packet. ++ */ ++ if (skb_is_gso(skb)) ++ goto do_send; ++ ++ rndis_msg_size += NDIS_CSUM_PPI_SIZE; ++ ppi = init_ppi_data(rndis_msg, NDIS_CSUM_PPI_SIZE, ++ TCPIP_CHKSUM_PKTINFO); ++ ++ csum_info = (struct ndis_tcp_ip_checksum_info *)((void *)ppi + ++ ppi->ppi_offset); ++ ++ if (net_trans_info & (INFO_IPV4 << 16)) ++ csum_info->transmit.is_ipv4 = 1; ++ else ++ csum_info->transmit.is_ipv6 = 1; ++ ++ if (net_trans_info & INFO_TCP) { ++ csum_info->transmit.tcp_checksum = 1; ++ csum_info->transmit.tcp_header_offset = hdr_offset; ++ } else if (net_trans_info & INFO_UDP) { ++ csum_info->transmit.udp_checksum = 1; ++ } ++ ++do_send: + /* Start filling in the page buffers with the rndis hdr */ + rndis_msg->msg_len += rndis_msg_size; + packet->page_buf_cnt = init_page_array(rndis_msg, rndis_msg_size, +@@ -592,8 +656,9 @@ static int netvsc_probe(struct hv_device *dev, + net->netdev_ops = &device_ops; + + /* TODO: Add GSO and Checksum offload */ +- net->hw_features = NETIF_F_RXCSUM | NETIF_F_SG; +- net->features = NETIF_F_HW_VLAN_CTAG_TX | NETIF_F_SG | NETIF_F_RXCSUM; ++ net->hw_features = NETIF_F_RXCSUM | NETIF_F_SG | NETIF_F_IP_CSUM; ++ net->features = NETIF_F_HW_VLAN_CTAG_TX | NETIF_F_SG | NETIF_F_RXCSUM | ++ NETIF_F_IP_CSUM; + + SET_ETHTOOL_OPS(net, ðtool_ops); + SET_NETDEV_DEV(net, &dev->device); +-- +2.4.3 + diff --git a/src/patches/linux/0009-hyperv-Fix-error-return-code-in-netvsc_init_buf.patch b/src/patches/linux/0009-hyperv-Fix-error-return-code-in-netvsc_init_buf.patch new file mode 100644 index 0000000..fcae531 --- /dev/null +++ b/src/patches/linux/0009-hyperv-Fix-error-return-code-in-netvsc_init_buf.patch @@ -0,0 +1,34 @@ +From dd1d3f8f9920926aa426589e542eed6bf58b7354 Mon Sep 17 00:00:00 2001 +From: Wei Yongjun yongjun_wei@trendmicro.com.cn +Date: Wed, 23 Jul 2014 09:00:35 +0800 +Subject: [PATCH 09/11] hyperv: Fix error return code in netvsc_init_buf() + +Fix to return -ENOMEM from the kalloc error handling +case instead of 0. + +Signed-off-by: Wei Yongjun yongjun_wei@trendmicro.com.cn +Reviewed-by: Haiyang Zhang haiyangz@microsoft.com +Signed-off-by: David S. Miller davem@davemloft.net +--- + drivers/net/hyperv/netvsc.c | 4 +++- + 1 file changed, 3 insertions(+), 1 deletion(-) + +diff --git a/drivers/net/hyperv/netvsc.c b/drivers/net/hyperv/netvsc.c +index 4ed38eaecea8..d97d5f39a04e 100644 +--- a/drivers/net/hyperv/netvsc.c ++++ b/drivers/net/hyperv/netvsc.c +@@ -378,8 +378,10 @@ static int netvsc_init_buf(struct hv_device *device) + + net_device->send_section_map = + kzalloc(net_device->map_words * sizeof(ulong), GFP_KERNEL); +- if (net_device->send_section_map == NULL) ++ if (net_device->send_section_map == NULL) { ++ ret = -ENOMEM; + goto cleanup; ++ } + + goto exit; + +-- +2.4.3 + diff --git a/src/patches/linux/0010-Drivers-net-hyperv-Enable-large-send-offload.patch b/src/patches/linux/0010-Drivers-net-hyperv-Enable-large-send-offload.patch new file mode 100644 index 0000000..36be2a8 --- /dev/null +++ b/src/patches/linux/0010-Drivers-net-hyperv-Enable-large-send-offload.patch @@ -0,0 +1,153 @@ +From a4ec4f58017b456281ee17c35fb82dfe4eab2193 Mon Sep 17 00:00:00 2001 +From: KY Srinivasan kys@microsoft.com +Date: Sat, 8 Mar 2014 19:23:18 -0800 +Subject: [PATCH 10/25] Drivers: net: hyperv: Enable large send offload + +Enable segmentation offload. + +Signed-off-by: K. Y. Srinivasan kys@microsoft.com +Reviewed-by: Haiyang Zhang haiyangz@microsoft.com +Signed-off-by: David S. Miller davem@davemloft.net +--- + drivers/net/hyperv/hyperv_net.h | 40 ++++++++++++++++++++++++++++++++++++++++ + drivers/net/hyperv/netvsc_drv.c | 38 ++++++++++++++++++++++++++++++++++---- + 2 files changed, 74 insertions(+), 4 deletions(-) + +diff --git a/drivers/net/hyperv/hyperv_net.h b/drivers/net/hyperv/hyperv_net.h +index 4cf238234321..7d06b4959383 100644 +--- a/drivers/net/hyperv/hyperv_net.h ++++ b/drivers/net/hyperv/hyperv_net.h +@@ -742,6 +742,10 @@ struct ndis_oject_header { + #define NDIS_OFFLOAD_PARAMETERS_RX_ENABLED_TX_DISABLED 3 + #define NDIS_OFFLOAD_PARAMETERS_TX_RX_ENABLED 4 + ++#define NDIS_TCP_LARGE_SEND_OFFLOAD_V2_TYPE 1 ++#define NDIS_TCP_LARGE_SEND_OFFLOAD_IPV4 0 ++#define NDIS_TCP_LARGE_SEND_OFFLOAD_IPV6 1 ++ + /* + * New offload OIDs for NDIS 6 + */ +@@ -804,12 +808,48 @@ struct ndis_tcp_ip_checksum_info { + }; + }; + ++struct ndis_tcp_lso_info { ++ union { ++ struct { ++ u32 unused:30; ++ u32 type:1; ++ u32 reserved2:1; ++ } transmit; ++ struct { ++ u32 mss:20; ++ u32 tcp_header_offset:10; ++ u32 type:1; ++ u32 reserved2:1; ++ } lso_v1_transmit; ++ struct { ++ u32 tcp_payload:30; ++ u32 type:1; ++ u32 reserved2:1; ++ } lso_v1_transmit_complete; ++ struct { ++ u32 mss:20; ++ u32 tcp_header_offset:10; ++ u32 type:1; ++ u32 ip_version:1; ++ } lso_v2_transmit; ++ struct { ++ u32 reserved:30; ++ u32 type:1; ++ u32 reserved2:1; ++ } lso_v2_transmit_complete; ++ u32 value; ++ }; ++}; ++ + #define NDIS_VLAN_PPI_SIZE (sizeof(struct rndis_per_packet_info) + \ + sizeof(struct ndis_pkt_8021q_info)) + + #define NDIS_CSUM_PPI_SIZE (sizeof(struct rndis_per_packet_info) + \ + sizeof(struct ndis_tcp_ip_checksum_info)) + ++#define NDIS_LSO_PPI_SIZE (sizeof(struct rndis_per_packet_info) + \ ++ sizeof(struct ndis_tcp_lso_info)) ++ + /* Format of Information buffer passed in a SetRequest for the OID */ + /* OID_GEN_RNDIS_CONFIG_PARAMETER. */ + struct rndis_config_parameter_info { +diff --git a/drivers/net/hyperv/netvsc_drv.c b/drivers/net/hyperv/netvsc_drv.c +index 2fab69ac61ef..5baa1fa7e692 100644 +--- a/drivers/net/hyperv/netvsc_drv.c ++++ b/drivers/net/hyperv/netvsc_drv.c +@@ -299,6 +299,7 @@ static int netvsc_start_xmit(struct sk_buff *skb, struct net_device *net) + bool isvlan; + struct rndis_per_packet_info *ppi; + struct ndis_tcp_ip_checksum_info *csum_info; ++ struct ndis_tcp_lso_info *lso_info; + int hdr_offset; + u32 net_trans_info; + +@@ -378,7 +379,7 @@ static int netvsc_start_xmit(struct sk_buff *skb, struct net_device *net) + * GSO packet. + */ + if (skb_is_gso(skb)) +- goto do_send; ++ goto do_lso; + + rndis_msg_size += NDIS_CSUM_PPI_SIZE; + ppi = init_ppi_data(rndis_msg, NDIS_CSUM_PPI_SIZE, +@@ -398,6 +399,35 @@ static int netvsc_start_xmit(struct sk_buff *skb, struct net_device *net) + } else if (net_trans_info & INFO_UDP) { + csum_info->transmit.udp_checksum = 1; + } ++ goto do_send; ++ ++do_lso: ++ rndis_msg_size += NDIS_LSO_PPI_SIZE; ++ ppi = init_ppi_data(rndis_msg, NDIS_LSO_PPI_SIZE, ++ TCP_LARGESEND_PKTINFO); ++ ++ lso_info = (struct ndis_tcp_lso_info *)((void *)ppi + ++ ppi->ppi_offset); ++ ++ lso_info->lso_v2_transmit.type = NDIS_TCP_LARGE_SEND_OFFLOAD_V2_TYPE; ++ if (net_trans_info & (INFO_IPV4 << 16)) { ++ lso_info->lso_v2_transmit.ip_version = ++ NDIS_TCP_LARGE_SEND_OFFLOAD_IPV4; ++ ip_hdr(skb)->tot_len = 0; ++ ip_hdr(skb)->check = 0; ++ tcp_hdr(skb)->check = ++ ~csum_tcpudp_magic(ip_hdr(skb)->saddr, ++ ip_hdr(skb)->daddr, 0, IPPROTO_TCP, 0); ++ } else { ++ lso_info->lso_v2_transmit.ip_version = ++ NDIS_TCP_LARGE_SEND_OFFLOAD_IPV6; ++ ipv6_hdr(skb)->payload_len = 0; ++ tcp_hdr(skb)->check = ++ ~csum_ipv6_magic(&ipv6_hdr(skb)->saddr, ++ &ipv6_hdr(skb)->daddr, 0, IPPROTO_TCP, 0); ++ } ++ lso_info->lso_v2_transmit.tcp_header_offset = hdr_offset; ++ lso_info->lso_v2_transmit.mss = skb_shinfo(skb)->gso_size; + + do_send: + /* Start filling in the page buffers with the rndis hdr */ +@@ -655,10 +685,10 @@ static int netvsc_probe(struct hv_device *dev, + + net->netdev_ops = &device_ops; + +- /* TODO: Add GSO and Checksum offload */ +- net->hw_features = NETIF_F_RXCSUM | NETIF_F_SG | NETIF_F_IP_CSUM; ++ net->hw_features = NETIF_F_RXCSUM | NETIF_F_SG | NETIF_F_IP_CSUM | ++ NETIF_F_TSO; + net->features = NETIF_F_HW_VLAN_CTAG_TX | NETIF_F_SG | NETIF_F_RXCSUM | +- NETIF_F_IP_CSUM; ++ NETIF_F_IP_CSUM | NETIF_F_TSO; + + SET_ETHTOOL_OPS(net, ðtool_ops); + SET_NETDEV_DEV(net, &dev->device); +-- +2.4.3 + diff --git a/src/patches/linux/0010-hyperv-Fix-a-bug-in-netvsc_start_xmit.patch b/src/patches/linux/0010-hyperv-Fix-a-bug-in-netvsc_start_xmit.patch new file mode 100644 index 0000000..5e08805 --- /dev/null +++ b/src/patches/linux/0010-hyperv-Fix-a-bug-in-netvsc_start_xmit.patch @@ -0,0 +1,47 @@ +From b3e774263908a834c1c0d5abf3a7658280e42fc7 Mon Sep 17 00:00:00 2001 +From: KY Srinivasan kys@microsoft.com +Date: Sun, 28 Sep 2014 22:16:43 -0700 +Subject: [PATCH 10/11] hyperv: Fix a bug in netvsc_start_xmit() + +[ Upstream commit dedb845ded56ded1c62f5398a94ffa8615d4592d ] + +After the packet is successfully sent, we should not touch the skb +as it may have been freed. This patch is based on the work done by +Long Li longli@microsoft.com. + +In this version of the patch I have fixed issues pointed out by David. +David, please queue this up for stable. + +Signed-off-by: K. Y. Srinivasan kys@microsoft.com +Tested-by: Long Li longli@microsoft.com +Tested-by: Sitsofe Wheeler sitsofe@yahoo.com +Signed-off-by: David S. Miller davem@davemloft.net +Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org +--- + drivers/net/hyperv/netvsc_drv.c | 3 ++- + 1 file changed, 2 insertions(+), 1 deletion(-) + +diff --git a/drivers/net/hyperv/netvsc_drv.c b/drivers/net/hyperv/netvsc_drv.c +index 4fd71b75e666..f15297201777 100644 +--- a/drivers/net/hyperv/netvsc_drv.c ++++ b/drivers/net/hyperv/netvsc_drv.c +@@ -387,6 +387,7 @@ static int netvsc_start_xmit(struct sk_buff *skb, struct net_device *net) + int hdr_offset; + u32 net_trans_info; + u32 hash; ++ u32 skb_length = skb->len; + + + /* We will atmost need two pages to describe the rndis +@@ -562,7 +563,7 @@ do_send: + + drop: + if (ret == 0) { +- net->stats.tx_bytes += skb->len; ++ net->stats.tx_bytes += skb_length; + net->stats.tx_packets++; + } else { + kfree(packet); +-- +2.4.3 + diff --git a/src/patches/linux/0011-hyperv-Change-the-receive-buffer-size-for-legacy-hos.patch b/src/patches/linux/0011-hyperv-Change-the-receive-buffer-size-for-legacy-hos.patch new file mode 100644 index 0000000..e3ee8cb --- /dev/null +++ b/src/patches/linux/0011-hyperv-Change-the-receive-buffer-size-for-legacy-hos.patch @@ -0,0 +1,54 @@ +From 56b3f72d3fed7f9b8d17dcf5d81455fa1b4327d7 Mon Sep 17 00:00:00 2001 +From: Haiyang Zhang haiyangz@microsoft.com +Date: Sun, 9 Mar 2014 16:10:59 -0700 +Subject: [PATCH 11/25] hyperv: Change the receive buffer size for legacy hosts + +Due to a bug in the Hyper-V host verion 2008R2, we need to use a slightly smaller +receive buffer size, otherwise the buffer will not be accepted by the legacy hosts. + +Signed-off-by: Haiyang Zhang haiyangz@microsoft.com +Signed-off-by: David S. Miller davem@davemloft.net +--- + drivers/net/hyperv/hyperv_net.h | 1 + + drivers/net/hyperv/netvsc.c | 6 +++++- + 2 files changed, 6 insertions(+), 1 deletion(-) + +diff --git a/drivers/net/hyperv/hyperv_net.h b/drivers/net/hyperv/hyperv_net.h +index 7d06b4959383..13010b4dae5b 100644 +--- a/drivers/net/hyperv/hyperv_net.h ++++ b/drivers/net/hyperv/hyperv_net.h +@@ -513,6 +513,7 @@ struct nvsp_message { + #define NETVSC_MTU 65536 + + #define NETVSC_RECEIVE_BUFFER_SIZE (1024*1024*16) /* 16MB */ ++#define NETVSC_RECEIVE_BUFFER_SIZE_LEGACY (1024*1024*15) /* 15MB */ + + #define NETVSC_RECEIVE_BUFFER_ID 0xcafe + +diff --git a/drivers/net/hyperv/netvsc.c b/drivers/net/hyperv/netvsc.c +index 1a0280dcba7e..daddea2654ce 100644 +--- a/drivers/net/hyperv/netvsc.c ++++ b/drivers/net/hyperv/netvsc.c +@@ -365,6 +365,11 @@ static int netvsc_connect_vsp(struct hv_device *device) + goto cleanup; + + /* Post the big receive buffer to NetVSP */ ++ if (net_device->nvsp_version <= NVSP_PROTOCOL_VERSION_2) ++ net_device->recv_buf_size = NETVSC_RECEIVE_BUFFER_SIZE_LEGACY; ++ else ++ net_device->recv_buf_size = NETVSC_RECEIVE_BUFFER_SIZE; ++ + ret = netvsc_init_recv_buf(device); + + cleanup: +@@ -898,7 +903,6 @@ int netvsc_device_add(struct hv_device *device, void *additional_info) + ndev = net_device->ndev; + + /* Initialize the NetVSC channel extension */ +- net_device->recv_buf_size = NETVSC_RECEIVE_BUFFER_SIZE; + spin_lock_init(&net_device->recv_pkt_list_lock); + + INIT_LIST_HEAD(&net_device->recv_pkt_list); +-- +2.4.3 + diff --git a/src/patches/linux/0011-hyperv-Fix-a-bug-in-netvsc_send.patch b/src/patches/linux/0011-hyperv-Fix-a-bug-in-netvsc_send.patch new file mode 100644 index 0000000..00f9c5d --- /dev/null +++ b/src/patches/linux/0011-hyperv-Fix-a-bug-in-netvsc_send.patch @@ -0,0 +1,68 @@ +From 26875bba869bd91a1d8fef9229a56a1e6d9fef2b Mon Sep 17 00:00:00 2001 +From: KY Srinivasan kys@microsoft.com +Date: Sun, 5 Oct 2014 10:42:51 -0700 +Subject: [PATCH 11/11] hyperv: Fix a bug in netvsc_send() + +[ Upstream commit 3a67c9ccad926a168d8b7891537a452018368a5b ] + +After the packet is successfully sent, we should not touch the packet +as it may have been freed. This patch is based on the work done by +Long Li longli@microsoft.com. + +David, please queue this up for stable. + +Signed-off-by: K. Y. Srinivasan kys@microsoft.com +Reported-by: Sitsofe Wheeler sitsofe@yahoo.com +Signed-off-by: David S. Miller davem@davemloft.net +Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org +--- + drivers/net/hyperv/netvsc.c | 15 ++++++++------- + 1 file changed, 8 insertions(+), 7 deletions(-) + +diff --git a/drivers/net/hyperv/netvsc.c b/drivers/net/hyperv/netvsc.c +index d97d5f39a04e..7edf976ecfa0 100644 +--- a/drivers/net/hyperv/netvsc.c ++++ b/drivers/net/hyperv/netvsc.c +@@ -708,6 +708,7 @@ int netvsc_send(struct hv_device *device, + unsigned int section_index = NETVSC_INVALID_INDEX; + u32 msg_size = 0; + struct sk_buff *skb; ++ u16 q_idx = packet->q_idx; + + + net_device = get_outbound_net_device(device); +@@ -772,24 +773,24 @@ int netvsc_send(struct hv_device *device, + + if (ret == 0) { + atomic_inc(&net_device->num_outstanding_sends); +- atomic_inc(&net_device->queue_sends[packet->q_idx]); ++ atomic_inc(&net_device->queue_sends[q_idx]); + + if (hv_ringbuf_avail_percent(&out_channel->outbound) < + RING_AVAIL_PERCENT_LOWATER) { + netif_tx_stop_queue(netdev_get_tx_queue( +- ndev, packet->q_idx)); ++ ndev, q_idx)); + + if (atomic_read(&net_device-> +- queue_sends[packet->q_idx]) < 1) ++ queue_sends[q_idx]) < 1) + netif_tx_wake_queue(netdev_get_tx_queue( +- ndev, packet->q_idx)); ++ ndev, q_idx)); + } + } else if (ret == -EAGAIN) { + netif_tx_stop_queue(netdev_get_tx_queue( +- ndev, packet->q_idx)); +- if (atomic_read(&net_device->queue_sends[packet->q_idx]) < 1) { ++ ndev, q_idx)); ++ if (atomic_read(&net_device->queue_sends[q_idx]) < 1) { + netif_tx_wake_queue(netdev_get_tx_queue( +- ndev, packet->q_idx)); ++ ndev, q_idx)); + ret = -ENOSPC; + } + } else { +-- +2.4.3 + diff --git a/src/patches/linux/0012-Drivers-net-hyperv-Allocate-memory-for-all-possible-.patch b/src/patches/linux/0012-Drivers-net-hyperv-Allocate-memory-for-all-possible-.patch new file mode 100644 index 0000000..c626bb0 --- /dev/null +++ b/src/patches/linux/0012-Drivers-net-hyperv-Allocate-memory-for-all-possible-.patch @@ -0,0 +1,34 @@ +From e0f6906eb7e5b395370da9499189d13b59020382 Mon Sep 17 00:00:00 2001 +From: KY Srinivasan kys@microsoft.com +Date: Wed, 9 Apr 2014 15:00:45 -0700 +Subject: [PATCH 12/25] Drivers: net: hyperv: Allocate memory for all possible + per-pecket information + +An outgoing packet can potentially need per-packet information for +all the offloads and VLAN tagging. Fix this issue. + +Signed-off-by: K. Y. Srinivasan kys@microsoft.com +Reviewed-by: Haiyang Zhang haiyangz@microsoft.com +Signed-off-by: David S. Miller davem@davemloft.net +--- + drivers/net/hyperv/netvsc_drv.c | 4 +++- + 1 file changed, 3 insertions(+), 1 deletion(-) + +diff --git a/drivers/net/hyperv/netvsc_drv.c b/drivers/net/hyperv/netvsc_drv.c +index 5baa1fa7e692..c76c85176644 100644 +--- a/drivers/net/hyperv/netvsc_drv.c ++++ b/drivers/net/hyperv/netvsc_drv.c +@@ -320,7 +320,9 @@ static int netvsc_start_xmit(struct sk_buff *skb, struct net_device *net) + packet = kzalloc(sizeof(struct hv_netvsc_packet) + + (num_data_pgs * sizeof(struct hv_page_buffer)) + + sizeof(struct rndis_message) + +- NDIS_VLAN_PPI_SIZE, GFP_ATOMIC); ++ NDIS_VLAN_PPI_SIZE + ++ NDIS_CSUM_PPI_SIZE + ++ NDIS_LSO_PPI_SIZE, GFP_ATOMIC); + if (!packet) { + /* out of memory, drop packet */ + netdev_err(net, "unable to allocate hv_netvsc_packet\n"); +-- +2.4.3 + diff --git a/src/patches/linux/0013-Drivers-net-hyperv-Negotiate-suitable-ndis-version-f.patch b/src/patches/linux/0013-Drivers-net-hyperv-Negotiate-suitable-ndis-version-f.patch new file mode 100644 index 0000000..7044c52 --- /dev/null +++ b/src/patches/linux/0013-Drivers-net-hyperv-Negotiate-suitable-ndis-version-f.patch @@ -0,0 +1,32 @@ +From 8a7882bada78e7d7355aafc0ca3c5696d25eb443 Mon Sep 17 00:00:00 2001 +From: KY Srinivasan kys@microsoft.com +Date: Wed, 9 Apr 2014 15:00:46 -0700 +Subject: [PATCH 13/25] Drivers: net: hyperv: Negotiate suitable ndis version + for offload support + +Ws2008R2 supports ndis_version 6.1 and 6.1 is the minimal version required +for various offloads. Negotiate ndis_version 6.1 when on ws2008r2. + +Signed-off-by: K. Y. Srinivasan kys@microsoft.com +Reviewed-by: Haiyang Zhang haiyangz@microsoft.com +Signed-off-by: David S. Miller davem@davemloft.net +--- + drivers/net/hyperv/netvsc.c | 2 +- + 1 file changed, 1 insertion(+), 1 deletion(-) + +diff --git a/drivers/net/hyperv/netvsc.c b/drivers/net/hyperv/netvsc.c +index daddea2654ce..f7629ecefa84 100644 +--- a/drivers/net/hyperv/netvsc.c ++++ b/drivers/net/hyperv/netvsc.c +@@ -344,7 +344,7 @@ static int netvsc_connect_vsp(struct hv_device *device) + memset(init_packet, 0, sizeof(struct nvsp_message)); + + if (net_device->nvsp_version <= NVSP_PROTOCOL_VERSION_4) +- ndis_version = 0x00050001; ++ ndis_version = 0x00060001; + else + ndis_version = 0x0006001e; + +-- +2.4.3 + diff --git a/src/patches/linux/0014-Drivers-net-hyperv-Address-UDP-checksum-issues.patch b/src/patches/linux/0014-Drivers-net-hyperv-Address-UDP-checksum-issues.patch new file mode 100644 index 0000000..eaf343e --- /dev/null +++ b/src/patches/linux/0014-Drivers-net-hyperv-Address-UDP-checksum-issues.patch @@ -0,0 +1,111 @@ +From b822ee7a17efd1bc8c7584da5d0a2c042e9ca5b6 Mon Sep 17 00:00:00 2001 +From: KY Srinivasan kys@microsoft.com +Date: Wed, 9 Apr 2014 15:00:47 -0700 +Subject: [PATCH 14/25] Drivers: net: hyperv: Address UDP checksum issues + +ws2008r2 does not support UDP checksum offload. Thus, we cannnot turn on +UDP offload in the host. Also, on ws2012 and ws2012 r2, there appear to be +an issue with UDP checksum offload. +Fix this issue by computing the UDP checksum in the Hyper-V driver. + +Based on Dave Miller's comments, in this version, I have COWed the skb +before modifying the UDP header (the checksum field). + +Signed-off-by: K. Y. Srinivasan kys@microsoft.com +Reviewed-by: Haiyang Zhang haiyangz@microsoft.com +Signed-off-by: David S. Miller davem@davemloft.net +--- + drivers/net/hyperv/hyperv_net.h | 1 + + drivers/net/hyperv/netvsc_drv.c | 26 +++++++++++++++++++++++++- + drivers/net/hyperv/rndis_filter.c | 12 +++++++++++- + 3 files changed, 37 insertions(+), 2 deletions(-) + +diff --git a/drivers/net/hyperv/hyperv_net.h b/drivers/net/hyperv/hyperv_net.h +index 13010b4dae5b..d18f711d0b0c 100644 +--- a/drivers/net/hyperv/hyperv_net.h ++++ b/drivers/net/hyperv/hyperv_net.h +@@ -747,6 +747,7 @@ struct ndis_oject_header { + #define NDIS_TCP_LARGE_SEND_OFFLOAD_IPV4 0 + #define NDIS_TCP_LARGE_SEND_OFFLOAD_IPV6 1 + ++#define VERSION_4_OFFLOAD_SIZE 22 + /* + * New offload OIDs for NDIS 6 + */ +diff --git a/drivers/net/hyperv/netvsc_drv.c b/drivers/net/hyperv/netvsc_drv.c +index c76c85176644..0d898876689e 100644 +--- a/drivers/net/hyperv/netvsc_drv.c ++++ b/drivers/net/hyperv/netvsc_drv.c +@@ -399,7 +399,30 @@ static int netvsc_start_xmit(struct sk_buff *skb, struct net_device *net) + csum_info->transmit.tcp_checksum = 1; + csum_info->transmit.tcp_header_offset = hdr_offset; + } else if (net_trans_info & INFO_UDP) { +- csum_info->transmit.udp_checksum = 1; ++ /* UDP checksum offload is not supported on ws2008r2. ++ * Furthermore, on ws2012 and ws2012r2, there are some ++ * issues with udp checksum offload from Linux guests. ++ * (these are host issues). ++ * For now compute the checksum here. ++ */ ++ struct udphdr *uh; ++ u16 udp_len; ++ ++ ret = skb_cow_head(skb, 0); ++ if (ret) ++ goto drop; ++ ++ uh = udp_hdr(skb); ++ udp_len = ntohs(uh->len); ++ uh->check = 0; ++ uh->check = csum_tcpudp_magic(ip_hdr(skb)->saddr, ++ ip_hdr(skb)->daddr, ++ udp_len, IPPROTO_UDP, ++ csum_partial(uh, udp_len, 0)); ++ if (uh->check == 0) ++ uh->check = CSUM_MANGLED_0; ++ ++ csum_info->transmit.udp_checksum = 0; + } + goto do_send; + +@@ -439,6 +462,7 @@ do_send: + + ret = netvsc_send(net_device_ctx->device_ctx, packet); + ++drop: + if (ret == 0) { + net->stats.tx_bytes += skb_length; + net->stats.tx_packets++; +diff --git a/drivers/net/hyperv/rndis_filter.c b/drivers/net/hyperv/rndis_filter.c +index 4a37e3db9e32..143a98caf618 100644 +--- a/drivers/net/hyperv/rndis_filter.c ++++ b/drivers/net/hyperv/rndis_filter.c +@@ -641,6 +641,16 @@ int rndis_filter_set_offload_params(struct hv_device *hdev, + struct rndis_set_complete *set_complete; + u32 extlen = sizeof(struct ndis_offload_params); + int ret, t; ++ u32 vsp_version = nvdev->nvsp_version; ++ ++ if (vsp_version <= NVSP_PROTOCOL_VERSION_4) { ++ extlen = VERSION_4_OFFLOAD_SIZE; ++ /* On NVSP_PROTOCOL_VERSION_4 and below, we do not support ++ * UDP checksum offload. ++ */ ++ req_offloads->udp_ip_v4_csum = 0; ++ req_offloads->udp_ip_v6_csum = 0; ++ } + + request = get_rndis_request(rdev, RNDIS_MSG_SET, + RNDIS_MESSAGE_SIZE(struct rndis_set_request) + extlen); +@@ -674,7 +684,7 @@ int rndis_filter_set_offload_params(struct hv_device *hdev, + } else { + set_complete = &request->response_msg.msg.set_complete; + if (set_complete->status != RNDIS_STATUS_SUCCESS) { +- netdev_err(ndev, "Fail to set MAC on host side:0x%x\n", ++ netdev_err(ndev, "Fail to set offload on host side:0x%x\n", + set_complete->status); + ret = -EINVAL; + } +-- +2.4.3 + diff --git a/src/patches/linux/0015-hyperv-Properly-handle-checksum-offload.patch b/src/patches/linux/0015-hyperv-Properly-handle-checksum-offload.patch new file mode 100644 index 0000000..0760833 --- /dev/null +++ b/src/patches/linux/0015-hyperv-Properly-handle-checksum-offload.patch @@ -0,0 +1,42 @@ +From 540360773bad2b81f14c38ca92b62797fa32cbc6 Mon Sep 17 00:00:00 2001 +From: KY Srinivasan kys@microsoft.com +Date: Wed, 30 Apr 2014 11:58:25 -0700 +Subject: [PATCH 15/25] hyperv: Properly handle checksum offload + +Do checksum offload only if the client of the driver wants checksum to be +offloaded. + +In V1 version of this patch, I addressed comments from +Stephen Hemminger stephen@networkplumber.org and +Eric Dumazet eric.dumazet@gmail.com. + +In this version of the patch I have addressed comments from +David Miller. + +This patch fixes a bug that is exposed in gateway scenarios. + +Signed-off-by: K. Y. Srinivasan kys@microsoft.com +Reviewed-by: Haiyang Zhang haiyangz@microsoft.com +Signed-off-by: David S. Miller davem@davemloft.net +--- + drivers/net/hyperv/netvsc_drv.c | 4 ++++ + 1 file changed, 4 insertions(+) + +diff --git a/drivers/net/hyperv/netvsc_drv.c b/drivers/net/hyperv/netvsc_drv.c +index 0d898876689e..ce6d870dd7ae 100644 +--- a/drivers/net/hyperv/netvsc_drv.c ++++ b/drivers/net/hyperv/netvsc_drv.c +@@ -383,6 +383,10 @@ static int netvsc_start_xmit(struct sk_buff *skb, struct net_device *net) + if (skb_is_gso(skb)) + goto do_lso; + ++ if ((skb->ip_summed == CHECKSUM_NONE) || ++ (skb->ip_summed == CHECKSUM_UNNECESSARY)) ++ goto do_send; ++ + rndis_msg_size += NDIS_CSUM_PPI_SIZE; + ppi = init_ppi_data(rndis_msg, NDIS_CSUM_PPI_SIZE, + TCPIP_CHKSUM_PKTINFO); +-- +2.4.3 + diff --git a/src/patches/linux/0016-hyperv-Add-support-for-virtual-Receive-Side-Scaling-.patch b/src/patches/linux/0016-hyperv-Add-support-for-virtual-Receive-Side-Scaling-.patch new file mode 100644 index 0000000..d01a428 --- /dev/null +++ b/src/patches/linux/0016-hyperv-Add-support-for-virtual-Receive-Side-Scaling-.patch @@ -0,0 +1,917 @@ +From 44559a96c6864eb6e95db0ae896c621b82e605f3 Mon Sep 17 00:00:00 2001 +From: Haiyang Zhang haiyangz@microsoft.com +Date: Mon, 21 Apr 2014 10:20:28 -0700 +Subject: [PATCH 16/25] hyperv: Add support for virtual Receive Side Scaling + (vRSS) + +This feature allows multiple channels to be used by each virtual NIC. +It is available on Hyper-V host 2012 R2. + +Signed-off-by: Haiyang Zhang haiyangz@microsoft.com +Reviewed-by: K. Y. Srinivasan kys@microsoft.com +Signed-off-by: David S. Miller davem@davemloft.net +--- + drivers/net/hyperv/hyperv_net.h | 110 +++++++++++++++++++++- + drivers/net/hyperv/netvsc.c | 136 +++++++++++++++++++++------ + drivers/net/hyperv/netvsc_drv.c | 103 ++++++++++++++++++++- + drivers/net/hyperv/rndis_filter.c | 189 +++++++++++++++++++++++++++++++++++++- + 4 files changed, 504 insertions(+), 34 deletions(-) + +diff --git a/drivers/net/hyperv/hyperv_net.h b/drivers/net/hyperv/hyperv_net.h +index d18f711d0b0c..57eb3f906d64 100644 +--- a/drivers/net/hyperv/hyperv_net.h ++++ b/drivers/net/hyperv/hyperv_net.h +@@ -28,6 +28,96 @@ + #include <linux/hyperv.h> + #include <linux/rndis.h> + ++/* RSS related */ ++#define OID_GEN_RECEIVE_SCALE_CAPABILITIES 0x00010203 /* query only */ ++#define OID_GEN_RECEIVE_SCALE_PARAMETERS 0x00010204 /* query and set */ ++ ++#define NDIS_OBJECT_TYPE_RSS_CAPABILITIES 0x88 ++#define NDIS_OBJECT_TYPE_RSS_PARAMETERS 0x89 ++ ++#define NDIS_RECEIVE_SCALE_CAPABILITIES_REVISION_2 2 ++#define NDIS_RECEIVE_SCALE_PARAMETERS_REVISION_2 2 ++ ++struct ndis_obj_header { ++ u8 type; ++ u8 rev; ++ u16 size; ++} __packed; ++ ++/* ndis_recv_scale_cap/cap_flag */ ++#define NDIS_RSS_CAPS_MESSAGE_SIGNALED_INTERRUPTS 0x01000000 ++#define NDIS_RSS_CAPS_CLASSIFICATION_AT_ISR 0x02000000 ++#define NDIS_RSS_CAPS_CLASSIFICATION_AT_DPC 0x04000000 ++#define NDIS_RSS_CAPS_USING_MSI_X 0x08000000 ++#define NDIS_RSS_CAPS_RSS_AVAILABLE_ON_PORTS 0x10000000 ++#define NDIS_RSS_CAPS_SUPPORTS_MSI_X 0x20000000 ++#define NDIS_RSS_CAPS_HASH_TYPE_TCP_IPV4 0x00000100 ++#define NDIS_RSS_CAPS_HASH_TYPE_TCP_IPV6 0x00000200 ++#define NDIS_RSS_CAPS_HASH_TYPE_TCP_IPV6_EX 0x00000400 ++ ++struct ndis_recv_scale_cap { /* NDIS_RECEIVE_SCALE_CAPABILITIES */ ++ struct ndis_obj_header hdr; ++ u32 cap_flag; ++ u32 num_int_msg; ++ u32 num_recv_que; ++ u16 num_indirect_tabent; ++} __packed; ++ ++ ++/* ndis_recv_scale_param flags */ ++#define NDIS_RSS_PARAM_FLAG_BASE_CPU_UNCHANGED 0x0001 ++#define NDIS_RSS_PARAM_FLAG_HASH_INFO_UNCHANGED 0x0002 ++#define NDIS_RSS_PARAM_FLAG_ITABLE_UNCHANGED 0x0004 ++#define NDIS_RSS_PARAM_FLAG_HASH_KEY_UNCHANGED 0x0008 ++#define NDIS_RSS_PARAM_FLAG_DISABLE_RSS 0x0010 ++ ++/* Hash info bits */ ++#define NDIS_HASH_FUNC_TOEPLITZ 0x00000001 ++#define NDIS_HASH_IPV4 0x00000100 ++#define NDIS_HASH_TCP_IPV4 0x00000200 ++#define NDIS_HASH_IPV6 0x00000400 ++#define NDIS_HASH_IPV6_EX 0x00000800 ++#define NDIS_HASH_TCP_IPV6 0x00001000 ++#define NDIS_HASH_TCP_IPV6_EX 0x00002000 ++ ++#define NDIS_RSS_INDIRECTION_TABLE_MAX_SIZE_REVISION_2 (128 * 4) ++#define NDIS_RSS_HASH_SECRET_KEY_MAX_SIZE_REVISION_2 40 ++ ++#define ITAB_NUM 128 ++#define HASH_KEYLEN NDIS_RSS_HASH_SECRET_KEY_MAX_SIZE_REVISION_2 ++extern u8 netvsc_hash_key[]; ++ ++struct ndis_recv_scale_param { /* NDIS_RECEIVE_SCALE_PARAMETERS */ ++ struct ndis_obj_header hdr; ++ ++ /* Qualifies the rest of the information */ ++ u16 flag; ++ ++ /* The base CPU number to do receive processing. not used */ ++ u16 base_cpu_number; ++ ++ /* This describes the hash function and type being enabled */ ++ u32 hashinfo; ++ ++ /* The size of indirection table array */ ++ u16 indirect_tabsize; ++ ++ /* The offset of the indirection table from the beginning of this ++ * structure ++ */ ++ u32 indirect_taboffset; ++ ++ /* The size of the hash secret key */ ++ u16 hashkey_size; ++ ++ /* The offset of the secret key from the beginning of this structure */ ++ u32 kashkey_offset; ++ ++ u32 processor_masks_offset; ++ u32 num_processor_masks; ++ u32 processor_masks_entry_size; ++}; ++ + /* Fwd declaration */ + struct hv_netvsc_packet; + struct ndis_tcp_ip_checksum_info; +@@ -39,6 +129,8 @@ struct xferpage_packet { + + /* # of netvsc packets this xfer packet contains */ + u32 count; ++ ++ struct vmbus_channel *channel; + }; + + /* +@@ -54,6 +146,9 @@ struct hv_netvsc_packet { + bool is_data_pkt; + u16 vlan_tci; + ++ u16 q_idx; ++ struct vmbus_channel *channel; ++ + /* + * Valid only for receives when we break a xfer page packet + * into multiple netvsc packets +@@ -120,6 +215,7 @@ void netvsc_linkstatus_callback(struct hv_device *device_obj, + int netvsc_recv_callback(struct hv_device *device_obj, + struct hv_netvsc_packet *packet, + struct ndis_tcp_ip_checksum_info *csum_info); ++void netvsc_channel_cb(void *context); + int rndis_filter_open(struct hv_device *dev); + int rndis_filter_close(struct hv_device *dev); + int rndis_filter_device_add(struct hv_device *dev, +@@ -522,6 +618,8 @@ struct nvsp_message { + + #define NETVSC_PACKET_SIZE 2048 + ++#define VRSS_SEND_TAB_SIZE 16 ++ + /* Per netvsc channel-specific */ + struct netvsc_device { + struct hv_device *dev; +@@ -555,10 +653,20 @@ struct netvsc_device { + + struct net_device *ndev; + ++ struct vmbus_channel *chn_table[NR_CPUS]; ++ u32 send_table[VRSS_SEND_TAB_SIZE]; ++ u32 num_chn; ++ atomic_t queue_sends[NR_CPUS]; ++ + /* Holds rndis device info */ + void *extension; +- /* The recive buffer for this device */ ++ ++ int ring_size; ++ ++ /* The primary channel callback buffer */ + unsigned char cb_buffer[NETVSC_PACKET_SIZE]; ++ /* The sub channel callback buffer */ ++ unsigned char *sub_cb_buf; + }; + + /* NdisInitialize message */ +diff --git a/drivers/net/hyperv/netvsc.c b/drivers/net/hyperv/netvsc.c +index f7629ecefa84..e7e77f12bc38 100644 +--- a/drivers/net/hyperv/netvsc.c ++++ b/drivers/net/hyperv/netvsc.c +@@ -422,6 +422,9 @@ int netvsc_device_remove(struct hv_device *device) + kfree(netvsc_packet); + } + ++ if (net_device->sub_cb_buf) ++ vfree(net_device->sub_cb_buf); ++ + kfree(net_device); + return 0; + } +@@ -461,7 +464,9 @@ static void netvsc_send_completion(struct netvsc_device *net_device, + (nvsp_packet->hdr.msg_type == + NVSP_MSG1_TYPE_SEND_RECV_BUF_COMPLETE) || + (nvsp_packet->hdr.msg_type == +- NVSP_MSG1_TYPE_SEND_SEND_BUF_COMPLETE)) { ++ NVSP_MSG1_TYPE_SEND_SEND_BUF_COMPLETE) || ++ (nvsp_packet->hdr.msg_type == ++ NVSP_MSG5_TYPE_SUBCHANNEL)) { + /* Copy the response back */ + memcpy(&net_device->channel_init_pkt, nvsp_packet, + sizeof(struct nvsp_message)); +@@ -469,28 +474,37 @@ static void netvsc_send_completion(struct netvsc_device *net_device, + } else if (nvsp_packet->hdr.msg_type == + NVSP_MSG1_TYPE_SEND_RNDIS_PKT_COMPLETE) { + int num_outstanding_sends; ++ u16 q_idx = 0; ++ struct vmbus_channel *channel = device->channel; ++ int queue_sends; + + /* Get the send context */ + nvsc_packet = (struct hv_netvsc_packet *)(unsigned long) + packet->trans_id; + + /* Notify the layer above us */ +- if (nvsc_packet) ++ if (nvsc_packet) { ++ q_idx = nvsc_packet->q_idx; ++ channel = nvsc_packet->channel; + nvsc_packet->completion.send.send_completion( + nvsc_packet->completion.send. + send_completion_ctx); ++ } + + num_outstanding_sends = + atomic_dec_return(&net_device->num_outstanding_sends); ++ queue_sends = atomic_dec_return(&net_device-> ++ queue_sends[q_idx]); + + if (net_device->destroy && num_outstanding_sends == 0) + wake_up(&net_device->wait_drain); + +- if (netif_queue_stopped(ndev) && !net_device->start_remove && +- (hv_ringbuf_avail_percent(&device->channel->outbound) +- > RING_AVAIL_PERCENT_HIWATER || +- num_outstanding_sends < 1)) +- netif_wake_queue(ndev); ++ if (netif_tx_queue_stopped(netdev_get_tx_queue(ndev, q_idx)) && ++ !net_device->start_remove && ++ (hv_ringbuf_avail_percent(&channel->outbound) > ++ RING_AVAIL_PERCENT_HIWATER || queue_sends < 1)) ++ netif_tx_wake_queue(netdev_get_tx_queue( ++ ndev, q_idx)); + } else { + netdev_err(ndev, "Unknown send completion packet type- " + "%d received!!\n", nvsp_packet->hdr.msg_type); +@@ -505,6 +519,7 @@ int netvsc_send(struct hv_device *device, + int ret = 0; + struct nvsp_message sendMessage; + struct net_device *ndev; ++ struct vmbus_channel *out_channel = NULL; + u64 req_id; + + net_device = get_outbound_net_device(device); +@@ -531,15 +546,20 @@ int netvsc_send(struct hv_device *device, + else + req_id = 0; + ++ out_channel = net_device->chn_table[packet->q_idx]; ++ if (out_channel == NULL) ++ out_channel = device->channel; ++ packet->channel = out_channel; ++ + if (packet->page_buf_cnt) { +- ret = vmbus_sendpacket_pagebuffer(device->channel, ++ ret = vmbus_sendpacket_pagebuffer(out_channel, + packet->page_buf, + packet->page_buf_cnt, + &sendMessage, + sizeof(struct nvsp_message), + req_id); + } else { +- ret = vmbus_sendpacket(device->channel, &sendMessage, ++ ret = vmbus_sendpacket(out_channel, &sendMessage, + sizeof(struct nvsp_message), + req_id, + VM_PKT_DATA_INBAND, +@@ -548,17 +568,24 @@ int netvsc_send(struct hv_device *device, + + if (ret == 0) { + atomic_inc(&net_device->num_outstanding_sends); +- if (hv_ringbuf_avail_percent(&device->channel->outbound) < ++ atomic_inc(&net_device->queue_sends[packet->q_idx]); ++ ++ if (hv_ringbuf_avail_percent(&out_channel->outbound) < + RING_AVAIL_PERCENT_LOWATER) { +- netif_stop_queue(ndev); ++ netif_tx_stop_queue(netdev_get_tx_queue( ++ ndev, packet->q_idx)); ++ + if (atomic_read(&net_device-> +- num_outstanding_sends) < 1) +- netif_wake_queue(ndev); ++ queue_sends[packet->q_idx]) < 1) ++ netif_tx_wake_queue(netdev_get_tx_queue( ++ ndev, packet->q_idx)); + } + } else if (ret == -EAGAIN) { +- netif_stop_queue(ndev); +- if (atomic_read(&net_device->num_outstanding_sends) < 1) { +- netif_wake_queue(ndev); ++ netif_tx_stop_queue(netdev_get_tx_queue( ++ ndev, packet->q_idx)); ++ if (atomic_read(&net_device->queue_sends[packet->q_idx]) < 1) { ++ netif_tx_wake_queue(netdev_get_tx_queue( ++ ndev, packet->q_idx)); + ret = -ENOSPC; + } + } else { +@@ -570,6 +597,7 @@ int netvsc_send(struct hv_device *device, + } + + static void netvsc_send_recv_completion(struct hv_device *device, ++ struct vmbus_channel *channel, + struct netvsc_device *net_device, + u64 transaction_id, u32 status) + { +@@ -587,7 +615,7 @@ static void netvsc_send_recv_completion(struct hv_device *device, + + retry_send_cmplt: + /* Send the completion */ +- ret = vmbus_sendpacket(device->channel, &recvcompMessage, ++ ret = vmbus_sendpacket(channel, &recvcompMessage, + sizeof(struct nvsp_message), transaction_id, + VM_PKT_COMP, 0); + if (ret == 0) { +@@ -618,6 +646,7 @@ static void netvsc_receive_completion(void *context) + { + struct hv_netvsc_packet *packet = context; + struct hv_device *device = packet->device; ++ struct vmbus_channel *channel; + struct netvsc_device *net_device; + u64 transaction_id = 0; + bool fsend_receive_comp = false; +@@ -649,6 +678,7 @@ static void netvsc_receive_completion(void *context) + */ + if (packet->xfer_page_pkt->count == 0) { + fsend_receive_comp = true; ++ channel = packet->xfer_page_pkt->channel; + transaction_id = packet->completion.recv.recv_completion_tid; + status = packet->xfer_page_pkt->status; + list_add_tail(&packet->xfer_page_pkt->list_ent, +@@ -662,12 +692,13 @@ static void netvsc_receive_completion(void *context) + + /* Send a receive completion for the xfer page packet */ + if (fsend_receive_comp) +- netvsc_send_recv_completion(device, net_device, transaction_id, +- status); ++ netvsc_send_recv_completion(device, channel, net_device, ++ transaction_id, status); + + } + + static void netvsc_receive(struct netvsc_device *net_device, ++ struct vmbus_channel *channel, + struct hv_device *device, + struct vmpacket_descriptor *packet) + { +@@ -748,7 +779,7 @@ static void netvsc_receive(struct netvsc_device *net_device, + spin_unlock_irqrestore(&net_device->recv_pkt_list_lock, + flags); + +- netvsc_send_recv_completion(device, net_device, ++ netvsc_send_recv_completion(device, channel, net_device, + vmxferpage_packet->d.trans_id, + NVSP_STAT_FAIL); + +@@ -759,6 +790,7 @@ static void netvsc_receive(struct netvsc_device *net_device, + xferpage_packet = (struct xferpage_packet *)listHead.next; + list_del(&xferpage_packet->list_ent); + xferpage_packet->status = NVSP_STAT_SUCCESS; ++ xferpage_packet->channel = channel; + + /* This is how much we can satisfy */ + xferpage_packet->count = count - 1; +@@ -800,10 +832,45 @@ static void netvsc_receive(struct netvsc_device *net_device, + + } + +-static void netvsc_channel_cb(void *context) ++ ++static void netvsc_send_table(struct hv_device *hdev, ++ struct vmpacket_descriptor *vmpkt) ++{ ++ struct netvsc_device *nvscdev; ++ struct net_device *ndev; ++ struct nvsp_message *nvmsg; ++ int i; ++ u32 count, *tab; ++ ++ nvscdev = get_outbound_net_device(hdev); ++ if (!nvscdev) ++ return; ++ ndev = nvscdev->ndev; ++ ++ nvmsg = (struct nvsp_message *)((unsigned long)vmpkt + ++ (vmpkt->offset8 << 3)); ++ ++ if (nvmsg->hdr.msg_type != NVSP_MSG5_TYPE_SEND_INDIRECTION_TABLE) ++ return; ++ ++ count = nvmsg->msg.v5_msg.send_table.count; ++ if (count != VRSS_SEND_TAB_SIZE) { ++ netdev_err(ndev, "Received wrong send-table size:%u\n", count); ++ return; ++ } ++ ++ tab = (u32 *)((unsigned long)&nvmsg->msg.v5_msg.send_table + ++ nvmsg->msg.v5_msg.send_table.offset); ++ ++ for (i = 0; i < count; i++) ++ nvscdev->send_table[i] = tab[i]; ++} ++ ++void netvsc_channel_cb(void *context) + { + int ret; +- struct hv_device *device = context; ++ struct vmbus_channel *channel = (struct vmbus_channel *)context; ++ struct hv_device *device; + struct netvsc_device *net_device; + u32 bytes_recvd; + u64 request_id; +@@ -812,14 +879,19 @@ static void netvsc_channel_cb(void *context) + int bufferlen = NETVSC_PACKET_SIZE; + struct net_device *ndev; + ++ if (channel->primary_channel != NULL) ++ device = channel->primary_channel->device_obj; ++ else ++ device = channel->device_obj; ++ + net_device = get_inbound_net_device(device); + if (!net_device) + return; + ndev = net_device->ndev; +- buffer = net_device->cb_buffer; ++ buffer = get_per_channel_state(channel); + + do { +- ret = vmbus_recvpacket_raw(device->channel, buffer, bufferlen, ++ ret = vmbus_recvpacket_raw(channel, buffer, bufferlen, + &bytes_recvd, &request_id); + if (ret == 0) { + if (bytes_recvd > 0) { +@@ -831,8 +903,12 @@ static void netvsc_channel_cb(void *context) + break; + + case VM_PKT_DATA_USING_XFER_PAGES: +- netvsc_receive(net_device, +- device, desc); ++ netvsc_receive(net_device, channel, ++ device, desc); ++ break; ++ ++ case VM_PKT_DATA_INBAND: ++ netvsc_send_table(device, desc); + break; + + default: +@@ -893,6 +969,8 @@ int netvsc_device_add(struct hv_device *device, void *additional_info) + goto cleanup; + } + ++ net_device->ring_size = ring_size; ++ + /* + * Coming into this function, struct net_device * is + * registered as the driver private data. +@@ -917,10 +995,12 @@ int netvsc_device_add(struct hv_device *device, void *additional_info) + } + init_completion(&net_device->channel_init_wait); + ++ set_per_channel_state(device->channel, net_device->cb_buffer); ++ + /* Open the channel */ + ret = vmbus_open(device->channel, ring_size * PAGE_SIZE, + ring_size * PAGE_SIZE, NULL, 0, +- netvsc_channel_cb, device); ++ netvsc_channel_cb, device->channel); + + if (ret != 0) { + netdev_err(ndev, "unable to open channel: %d\n", ret); +@@ -930,6 +1010,8 @@ int netvsc_device_add(struct hv_device *device, void *additional_info) + /* Channel is opened */ + pr_info("hv_netvsc channel opened successfully\n"); + ++ net_device->chn_table[0] = device->channel; ++ + /* Connect with the NetVsp */ + ret = netvsc_connect_vsp(device); + if (ret != 0) { +diff --git a/drivers/net/hyperv/netvsc_drv.c b/drivers/net/hyperv/netvsc_drv.c +index ce6d870dd7ae..e486dbd33f61 100644 +--- a/drivers/net/hyperv/netvsc_drv.c ++++ b/drivers/net/hyperv/netvsc_drv.c +@@ -101,7 +101,7 @@ static int netvsc_open(struct net_device *net) + return ret; + } + +- netif_start_queue(net); ++ netif_tx_start_all_queues(net); + + nvdev = hv_get_drvdata(device_obj); + rdev = nvdev->extension; +@@ -149,6 +149,88 @@ static void *init_ppi_data(struct rndis_message *msg, u32 ppi_size, + return ppi; + } + ++union sub_key { ++ u64 k; ++ struct { ++ u8 pad[3]; ++ u8 kb; ++ u32 ka; ++ }; ++}; ++ ++/* Toeplitz hash function ++ * data: network byte order ++ * return: host byte order ++ */ ++static u32 comp_hash(u8 *key, int klen, u8 *data, int dlen) ++{ ++ union sub_key subk; ++ int k_next = 4; ++ u8 dt; ++ int i, j; ++ u32 ret = 0; ++ ++ subk.k = 0; ++ subk.ka = ntohl(*(u32 *)key); ++ ++ for (i = 0; i < dlen; i++) { ++ subk.kb = key[k_next]; ++ k_next = (k_next + 1) % klen; ++ dt = data[i]; ++ for (j = 0; j < 8; j++) { ++ if (dt & 0x80) ++ ret ^= subk.ka; ++ dt <<= 1; ++ subk.k <<= 1; ++ } ++ } ++ ++ return ret; ++} ++ ++static bool netvsc_set_hash(u32 *hash, struct sk_buff *skb) ++{ ++ struct iphdr *iphdr; ++ int data_len; ++ bool ret = false; ++ ++ if (eth_hdr(skb)->h_proto != htons(ETH_P_IP)) ++ return false; ++ ++ iphdr = ip_hdr(skb); ++ ++ if (iphdr->version == 4) { ++ if (iphdr->protocol == IPPROTO_TCP) ++ data_len = 12; ++ else ++ data_len = 8; ++ *hash = comp_hash(netvsc_hash_key, HASH_KEYLEN, ++ (u8 *)&iphdr->saddr, data_len); ++ ret = true; ++ } ++ ++ return ret; ++} ++ ++static u16 netvsc_select_queue(struct net_device *ndev, struct sk_buff *skb, ++ void *accel_priv, select_queue_fallback_t fallback) ++{ ++ struct net_device_context *net_device_ctx = netdev_priv(ndev); ++ struct hv_device *hdev = net_device_ctx->device_ctx; ++ struct netvsc_device *nvsc_dev = hv_get_drvdata(hdev); ++ u32 hash; ++ u16 q_idx = 0; ++ ++ if (nvsc_dev == NULL || ndev->real_num_tx_queues <= 1) ++ return 0; ++ ++ if (netvsc_set_hash(&hash, skb)) ++ q_idx = nvsc_dev->send_table[hash % VRSS_SEND_TAB_SIZE] % ++ ndev->real_num_tx_queues; ++ ++ return q_idx; ++} ++ + static void netvsc_xmit_completion(void *context) + { + struct hv_netvsc_packet *packet = (struct hv_netvsc_packet *)context; +@@ -334,6 +416,8 @@ static int netvsc_start_xmit(struct sk_buff *skb, struct net_device *net) + + packet->vlan_tci = skb->vlan_tci; + ++ packet->q_idx = skb_get_queue_mapping(skb); ++ + packet->is_data_pkt = true; + packet->total_data_buflen = skb->len; + +@@ -559,6 +643,10 @@ int netvsc_recv_callback(struct hv_device *device_obj, + __vlan_hwaccel_put_tag(skb, htons(ETH_P_8021Q), + packet->vlan_tci); + ++ skb_record_rx_queue(skb, packet->xfer_page_pkt->channel-> ++ offermsg.offer.sub_channel_index % ++ net->real_num_rx_queues); ++ + net->stats.rx_packets++; + net->stats.rx_bytes += packet->total_data_buflen; + +@@ -607,7 +695,7 @@ static int netvsc_change_mtu(struct net_device *ndev, int mtu) + hv_set_drvdata(hdev, ndev); + device_info.ring_size = ring_size; + rndis_filter_device_add(hdev, &device_info); +- netif_wake_queue(ndev); ++ netif_tx_wake_all_queues(ndev); + + return 0; + } +@@ -653,6 +741,7 @@ static const struct net_device_ops device_ops = { + .ndo_change_mtu = netvsc_change_mtu, + .ndo_validate_addr = eth_validate_addr, + .ndo_set_mac_address = netvsc_set_mac_addr, ++ .ndo_select_queue = netvsc_select_queue, + }; + + /* +@@ -699,9 +788,11 @@ static int netvsc_probe(struct hv_device *dev, + struct net_device *net = NULL; + struct net_device_context *net_device_ctx; + struct netvsc_device_info device_info; ++ struct netvsc_device *nvdev; + int ret; + +- net = alloc_etherdev(sizeof(struct net_device_context)); ++ net = alloc_etherdev_mq(sizeof(struct net_device_context), ++ num_online_cpus()); + if (!net) + return -ENOMEM; + +@@ -734,6 +825,12 @@ static int netvsc_probe(struct hv_device *dev, + } + memcpy(net->dev_addr, device_info.mac_adr, ETH_ALEN); + ++ nvdev = hv_get_drvdata(dev); ++ netif_set_real_num_tx_queues(net, nvdev->num_chn); ++ netif_set_real_num_rx_queues(net, nvdev->num_chn); ++ dev_info(&dev->device, "real num tx,rx queues:%u, %u\n", ++ net->real_num_tx_queues, net->real_num_rx_queues); ++ + ret = register_netdev(net); + if (ret != 0) { + pr_err("Unable to register netdev.\n"); +diff --git a/drivers/net/hyperv/rndis_filter.c b/drivers/net/hyperv/rndis_filter.c +index 143a98caf618..d92cfbe43410 100644 +--- a/drivers/net/hyperv/rndis_filter.c ++++ b/drivers/net/hyperv/rndis_filter.c +@@ -31,7 +31,7 @@ + #include "hyperv_net.h" + + +-#define RNDIS_EXT_LEN 100 ++#define RNDIS_EXT_LEN PAGE_SIZE + struct rndis_request { + struct list_head list_ent; + struct completion wait_event; +@@ -94,6 +94,8 @@ static struct rndis_request *get_rndis_request(struct rndis_device *dev, + rndis_msg->ndis_msg_type = msg_type; + rndis_msg->msg_len = msg_len; + ++ request->pkt.q_idx = 0; ++ + /* + * Set the request id. This field is always after the rndis header for + * request/response packet types so we just used the SetRequest as a +@@ -509,6 +511,19 @@ static int rndis_filter_query_device(struct rndis_device *dev, u32 oid, + query->info_buflen = 0; + query->dev_vc_handle = 0; + ++ if (oid == OID_GEN_RECEIVE_SCALE_CAPABILITIES) { ++ struct ndis_recv_scale_cap *cap; ++ ++ request->request_msg.msg_len += ++ sizeof(struct ndis_recv_scale_cap); ++ query->info_buflen = sizeof(struct ndis_recv_scale_cap); ++ cap = (struct ndis_recv_scale_cap *)((unsigned long)query + ++ query->info_buf_offset); ++ cap->hdr.type = NDIS_OBJECT_TYPE_RSS_CAPABILITIES; ++ cap->hdr.rev = NDIS_RECEIVE_SCALE_CAPABILITIES_REVISION_2; ++ cap->hdr.size = sizeof(struct ndis_recv_scale_cap); ++ } ++ + ret = rndis_filter_send_request(dev, request); + if (ret != 0) + goto cleanup; +@@ -695,6 +710,89 @@ cleanup: + return ret; + } + ++u8 netvsc_hash_key[HASH_KEYLEN] = { ++ 0x6d, 0x5a, 0x56, 0xda, 0x25, 0x5b, 0x0e, 0xc2, ++ 0x41, 0x67, 0x25, 0x3d, 0x43, 0xa3, 0x8f, 0xb0, ++ 0xd0, 0xca, 0x2b, 0xcb, 0xae, 0x7b, 0x30, 0xb4, ++ 0x77, 0xcb, 0x2d, 0xa3, 0x80, 0x30, 0xf2, 0x0c, ++ 0x6a, 0x42, 0xb7, 0x3b, 0xbe, 0xac, 0x01, 0xfa ++}; ++ ++int rndis_filter_set_rss_param(struct rndis_device *rdev, int num_queue) ++{ ++ struct net_device *ndev = rdev->net_dev->ndev; ++ struct rndis_request *request; ++ struct rndis_set_request *set; ++ struct rndis_set_complete *set_complete; ++ u32 extlen = sizeof(struct ndis_recv_scale_param) + ++ 4*ITAB_NUM + HASH_KEYLEN; ++ struct ndis_recv_scale_param *rssp; ++ u32 *itab; ++ u8 *keyp; ++ int i, t, ret; ++ ++ request = get_rndis_request( ++ rdev, RNDIS_MSG_SET, ++ RNDIS_MESSAGE_SIZE(struct rndis_set_request) + extlen); ++ if (!request) ++ return -ENOMEM; ++ ++ set = &request->request_msg.msg.set_req; ++ set->oid = OID_GEN_RECEIVE_SCALE_PARAMETERS; ++ set->info_buflen = extlen; ++ set->info_buf_offset = sizeof(struct rndis_set_request); ++ set->dev_vc_handle = 0; ++ ++ rssp = (struct ndis_recv_scale_param *)(set + 1); ++ rssp->hdr.type = NDIS_OBJECT_TYPE_RSS_PARAMETERS; ++ rssp->hdr.rev = NDIS_RECEIVE_SCALE_PARAMETERS_REVISION_2; ++ rssp->hdr.size = sizeof(struct ndis_recv_scale_param); ++ rssp->flag = 0; ++ rssp->hashinfo = NDIS_HASH_FUNC_TOEPLITZ | NDIS_HASH_IPV4 | ++ NDIS_HASH_TCP_IPV4; ++ rssp->indirect_tabsize = 4*ITAB_NUM; ++ rssp->indirect_taboffset = sizeof(struct ndis_recv_scale_param); ++ rssp->hashkey_size = HASH_KEYLEN; ++ rssp->kashkey_offset = rssp->indirect_taboffset + ++ rssp->indirect_tabsize; ++ ++ /* Set indirection table entries */ ++ itab = (u32 *)(rssp + 1); ++ for (i = 0; i < ITAB_NUM; i++) ++ itab[i] = i % num_queue; ++ ++ /* Set hask key values */ ++ keyp = (u8 *)((unsigned long)rssp + rssp->kashkey_offset); ++ for (i = 0; i < HASH_KEYLEN; i++) ++ keyp[i] = netvsc_hash_key[i]; ++ ++ ++ ret = rndis_filter_send_request(rdev, request); ++ if (ret != 0) ++ goto cleanup; ++ ++ t = wait_for_completion_timeout(&request->wait_event, 5*HZ); ++ if (t == 0) { ++ netdev_err(ndev, "timeout before we got a set response...\n"); ++ /* can't put_rndis_request, since we may still receive a ++ * send-completion. ++ */ ++ return -ETIMEDOUT; ++ } else { ++ set_complete = &request->response_msg.msg.set_complete; ++ if (set_complete->status != RNDIS_STATUS_SUCCESS) { ++ netdev_err(ndev, "Fail to set RSS parameters:0x%x\n", ++ set_complete->status); ++ ret = -EINVAL; ++ } ++ } ++ ++cleanup: ++ put_rndis_request(rdev, request); ++ return ret; ++} ++ ++ + static int rndis_filter_query_device_link_status(struct rndis_device *dev) + { + u32 size = sizeof(u32); +@@ -886,6 +984,28 @@ static int rndis_filter_close_device(struct rndis_device *dev) + return ret; + } + ++static void netvsc_sc_open(struct vmbus_channel *new_sc) ++{ ++ struct netvsc_device *nvscdev; ++ u16 chn_index = new_sc->offermsg.offer.sub_channel_index; ++ int ret; ++ ++ nvscdev = hv_get_drvdata(new_sc->primary_channel->device_obj); ++ ++ if (chn_index >= nvscdev->num_chn) ++ return; ++ ++ set_per_channel_state(new_sc, nvscdev->sub_cb_buf + (chn_index - 1) * ++ NETVSC_PACKET_SIZE); ++ ++ ret = vmbus_open(new_sc, nvscdev->ring_size * PAGE_SIZE, ++ nvscdev->ring_size * PAGE_SIZE, NULL, 0, ++ netvsc_channel_cb, new_sc); ++ ++ if (ret == 0) ++ nvscdev->chn_table[chn_index] = new_sc; ++} ++ + int rndis_filter_device_add(struct hv_device *dev, + void *additional_info) + { +@@ -894,6 +1014,10 @@ int rndis_filter_device_add(struct hv_device *dev, + struct rndis_device *rndis_device; + struct netvsc_device_info *device_info = additional_info; + struct ndis_offload_params offloads; ++ struct nvsp_message *init_packet; ++ int t; ++ struct ndis_recv_scale_cap rsscap; ++ u32 rsscap_size = sizeof(struct ndis_recv_scale_cap); + + rndis_device = get_rndis_device(); + if (!rndis_device) +@@ -913,6 +1037,7 @@ int rndis_filter_device_add(struct hv_device *dev, + + /* Initialize the rndis device */ + net_device = hv_get_drvdata(dev); ++ net_device->num_chn = 1; + + net_device->extension = rndis_device; + rndis_device->net_dev = net_device; +@@ -952,7 +1077,6 @@ int rndis_filter_device_add(struct hv_device *dev, + if (ret) + goto err_dev_remv; + +- + rndis_filter_query_device_link_status(rndis_device); + + device_info->link_state = rndis_device->link_state; +@@ -961,7 +1085,66 @@ int rndis_filter_device_add(struct hv_device *dev, + rndis_device->hw_mac_adr, + device_info->link_state ? "down" : "up"); + +- return ret; ++ if (net_device->nvsp_version < NVSP_PROTOCOL_VERSION_5) ++ return 0; ++ ++ /* vRSS setup */ ++ memset(&rsscap, 0, rsscap_size); ++ ret = rndis_filter_query_device(rndis_device, ++ OID_GEN_RECEIVE_SCALE_CAPABILITIES, ++ &rsscap, &rsscap_size); ++ if (ret || rsscap.num_recv_que < 2) ++ goto out; ++ ++ net_device->num_chn = (num_online_cpus() < rsscap.num_recv_que) ? ++ num_online_cpus() : rsscap.num_recv_que; ++ if (net_device->num_chn == 1) ++ goto out; ++ ++ net_device->sub_cb_buf = vzalloc((net_device->num_chn - 1) * ++ NETVSC_PACKET_SIZE); ++ if (!net_device->sub_cb_buf) { ++ net_device->num_chn = 1; ++ dev_info(&dev->device, "No memory for subchannels.\n"); ++ goto out; ++ } ++ ++ vmbus_set_sc_create_callback(dev->channel, netvsc_sc_open); ++ ++ init_packet = &net_device->channel_init_pkt; ++ memset(init_packet, 0, sizeof(struct nvsp_message)); ++ init_packet->hdr.msg_type = NVSP_MSG5_TYPE_SUBCHANNEL; ++ init_packet->msg.v5_msg.subchn_req.op = NVSP_SUBCHANNEL_ALLOCATE; ++ init_packet->msg.v5_msg.subchn_req.num_subchannels = ++ net_device->num_chn - 1; ++ ret = vmbus_sendpacket(dev->channel, init_packet, ++ sizeof(struct nvsp_message), ++ (unsigned long)init_packet, ++ VM_PKT_DATA_INBAND, ++ VMBUS_DATA_PACKET_FLAG_COMPLETION_REQUESTED); ++ if (ret) ++ goto out; ++ t = wait_for_completion_timeout(&net_device->channel_init_wait, 5*HZ); ++ if (t == 0) { ++ ret = -ETIMEDOUT; ++ goto out; ++ } ++ if (init_packet->msg.v5_msg.subchn_comp.status != ++ NVSP_STAT_SUCCESS) { ++ ret = -ENODEV; ++ goto out; ++ } ++ net_device->num_chn = 1 + ++ init_packet->msg.v5_msg.subchn_comp.num_subchannels; ++ ++ vmbus_are_subchannels_present(dev->channel); ++ ++ ret = rndis_filter_set_rss_param(rndis_device, net_device->num_chn); ++ ++out: ++ if (ret) ++ net_device->num_chn = 1; ++ return 0; /* return 0 because primary channel can be used alone */ + + err_dev_remv: + rndis_filter_device_remove(dev); +-- +2.4.3 + diff --git a/src/patches/linux/0017-hyperv-Remove-recv_pkt_list-and-lock.patch b/src/patches/linux/0017-hyperv-Remove-recv_pkt_list-and-lock.patch new file mode 100644 index 0000000..31465a1 --- /dev/null +++ b/src/patches/linux/0017-hyperv-Remove-recv_pkt_list-and-lock.patch @@ -0,0 +1,384 @@ +From d6bf5567c1438b4f3b1bcff1a1525ddb1754df19 Mon Sep 17 00:00:00 2001 +From: Haiyang Zhang haiyangz@microsoft.com +Date: Mon, 21 Apr 2014 14:54:43 -0700 +Subject: [PATCH 17/25] hyperv: Remove recv_pkt_list and lock + +Removed recv_pkt_list and lock, and updated related code, so that +the locking overhead is reduced especially when multiple channels +are in use. + +The recv_pkt_list isn't actually necessary because the packets are +processed sequentially in each channel. It has been replaced by a +local variable, and the related lock for this list is also removed. +The is_data_pkt field is not used in receive path, so its assignment +is cleaned up. + +Signed-off-by: Haiyang Zhang haiyangz@microsoft.com +Reviewed-by: K. Y. Srinivasan kys@microsoft.com +Signed-off-by: David S. Miller davem@davemloft.net +--- + drivers/net/hyperv/hyperv_net.h | 33 -------- + drivers/net/hyperv/netvsc.c | 174 +++----------------------------------- + drivers/net/hyperv/netvsc_drv.c | 2 +- + drivers/net/hyperv/rndis_filter.c | 2 - + 4 files changed, 13 insertions(+), 198 deletions(-) + +diff --git a/drivers/net/hyperv/hyperv_net.h b/drivers/net/hyperv/hyperv_net.h +index 57eb3f906d64..a1af0f7711e2 100644 +--- a/drivers/net/hyperv/hyperv_net.h ++++ b/drivers/net/hyperv/hyperv_net.h +@@ -119,27 +119,14 @@ struct ndis_recv_scale_param { /* NDIS_RECEIVE_SCALE_PARAMETERS */ + }; + + /* Fwd declaration */ +-struct hv_netvsc_packet; + struct ndis_tcp_ip_checksum_info; + +-/* Represent the xfer page packet which contains 1 or more netvsc packet */ +-struct xferpage_packet { +- struct list_head list_ent; +- u32 status; +- +- /* # of netvsc packets this xfer packet contains */ +- u32 count; +- +- struct vmbus_channel *channel; +-}; +- + /* + * Represent netvsc packet which contains 1 RNDIS and 1 ethernet frame + * within the RNDIS + */ + struct hv_netvsc_packet { + /* Bookkeeping stuff */ +- struct list_head list_ent; + u32 status; + + struct hv_device *device; +@@ -149,19 +136,8 @@ struct hv_netvsc_packet { + u16 q_idx; + struct vmbus_channel *channel; + +- /* +- * Valid only for receives when we break a xfer page packet +- * into multiple netvsc packets +- */ +- struct xferpage_packet *xfer_page_pkt; +- + union { + struct { +- u64 recv_completion_tid; +- void *recv_completion_ctx; +- void (*recv_completion)(void *context); +- } recv; +- struct { + u64 send_completion_tid; + void *send_completion_ctx; + void (*send_completion)(void *context); +@@ -613,9 +589,6 @@ struct nvsp_message { + + #define NETVSC_RECEIVE_BUFFER_ID 0xcafe + +-/* Preallocated receive packets */ +-#define NETVSC_RECEIVE_PACKETLIST_COUNT 256 +- + #define NETVSC_PACKET_SIZE 2048 + + #define VRSS_SEND_TAB_SIZE 16 +@@ -630,12 +603,6 @@ struct netvsc_device { + wait_queue_head_t wait_drain; + bool start_remove; + bool destroy; +- /* +- * List of free preallocated hv_netvsc_packet to represent receive +- * packet +- */ +- struct list_head recv_pkt_list; +- spinlock_t recv_pkt_list_lock; + + /* Receive buffer allocated by us but manages by NetVSP */ + void *recv_buf; +diff --git a/drivers/net/hyperv/netvsc.c b/drivers/net/hyperv/netvsc.c +index e7e77f12bc38..b10334773b32 100644 +--- a/drivers/net/hyperv/netvsc.c ++++ b/drivers/net/hyperv/netvsc.c +@@ -387,7 +387,6 @@ static void netvsc_disconnect_vsp(struct netvsc_device *net_device) + int netvsc_device_remove(struct hv_device *device) + { + struct netvsc_device *net_device; +- struct hv_netvsc_packet *netvsc_packet, *pos; + unsigned long flags; + + net_device = hv_get_drvdata(device); +@@ -416,12 +415,6 @@ int netvsc_device_remove(struct hv_device *device) + vmbus_close(device->channel); + + /* Release all resources */ +- list_for_each_entry_safe(netvsc_packet, pos, +- &net_device->recv_pkt_list, list_ent) { +- list_del(&netvsc_packet->list_ent); +- kfree(netvsc_packet); +- } +- + if (net_device->sub_cb_buf) + vfree(net_device->sub_cb_buf); + +@@ -641,62 +634,6 @@ retry_send_cmplt: + } + } + +-/* Send a receive completion packet to RNDIS device (ie NetVsp) */ +-static void netvsc_receive_completion(void *context) +-{ +- struct hv_netvsc_packet *packet = context; +- struct hv_device *device = packet->device; +- struct vmbus_channel *channel; +- struct netvsc_device *net_device; +- u64 transaction_id = 0; +- bool fsend_receive_comp = false; +- unsigned long flags; +- struct net_device *ndev; +- u32 status = NVSP_STAT_NONE; +- +- /* +- * Even though it seems logical to do a GetOutboundNetDevice() here to +- * send out receive completion, we are using GetInboundNetDevice() +- * since we may have disable outbound traffic already. +- */ +- net_device = get_inbound_net_device(device); +- if (!net_device) +- return; +- ndev = net_device->ndev; +- +- /* Overloading use of the lock. */ +- spin_lock_irqsave(&net_device->recv_pkt_list_lock, flags); +- +- if (packet->status != NVSP_STAT_SUCCESS) +- packet->xfer_page_pkt->status = NVSP_STAT_FAIL; +- +- packet->xfer_page_pkt->count--; +- +- /* +- * Last one in the line that represent 1 xfer page packet. +- * Return the xfer page packet itself to the freelist +- */ +- if (packet->xfer_page_pkt->count == 0) { +- fsend_receive_comp = true; +- channel = packet->xfer_page_pkt->channel; +- transaction_id = packet->completion.recv.recv_completion_tid; +- status = packet->xfer_page_pkt->status; +- list_add_tail(&packet->xfer_page_pkt->list_ent, +- &net_device->recv_pkt_list); +- +- } +- +- /* Put the packet back */ +- list_add_tail(&packet->list_ent, &net_device->recv_pkt_list); +- spin_unlock_irqrestore(&net_device->recv_pkt_list_lock, flags); +- +- /* Send a receive completion for the xfer page packet */ +- if (fsend_receive_comp) +- netvsc_send_recv_completion(device, channel, net_device, +- transaction_id, status); +- +-} +- + static void netvsc_receive(struct netvsc_device *net_device, + struct vmbus_channel *channel, + struct hv_device *device, +@@ -704,16 +641,13 @@ static void netvsc_receive(struct netvsc_device *net_device, + { + struct vmtransfer_page_packet_header *vmxferpage_packet; + struct nvsp_message *nvsp_packet; +- struct hv_netvsc_packet *netvsc_packet = NULL; +- /* struct netvsc_driver *netvscDriver; */ +- struct xferpage_packet *xferpage_packet = NULL; ++ struct hv_netvsc_packet nv_pkt; ++ struct hv_netvsc_packet *netvsc_packet = &nv_pkt; ++ u32 status = NVSP_STAT_SUCCESS; + int i; + int count = 0; +- unsigned long flags; + struct net_device *ndev; + +- LIST_HEAD(listHead); +- + ndev = net_device->ndev; + + /* +@@ -746,78 +680,14 @@ static void netvsc_receive(struct netvsc_device *net_device, + return; + } + +- /* +- * Grab free packets (range count + 1) to represent this xfer +- * page packet. +1 to represent the xfer page packet itself. +- * We grab it here so that we know exactly how many we can +- * fulfil +- */ +- spin_lock_irqsave(&net_device->recv_pkt_list_lock, flags); +- while (!list_empty(&net_device->recv_pkt_list)) { +- list_move_tail(net_device->recv_pkt_list.next, &listHead); +- if (++count == vmxferpage_packet->range_cnt + 1) +- break; +- } +- spin_unlock_irqrestore(&net_device->recv_pkt_list_lock, flags); +- +- /* +- * We need at least 2 netvsc pkts (1 to represent the xfer +- * page and at least 1 for the range) i.e. we can handled +- * some of the xfer page packet ranges... +- */ +- if (count < 2) { +- netdev_err(ndev, "Got only %d netvsc pkt...needed " +- "%d pkts. Dropping this xfer page packet completely!\n", +- count, vmxferpage_packet->range_cnt + 1); +- +- /* Return it to the freelist */ +- spin_lock_irqsave(&net_device->recv_pkt_list_lock, flags); +- for (i = count; i != 0; i--) { +- list_move_tail(listHead.next, +- &net_device->recv_pkt_list); +- } +- spin_unlock_irqrestore(&net_device->recv_pkt_list_lock, +- flags); +- +- netvsc_send_recv_completion(device, channel, net_device, +- vmxferpage_packet->d.trans_id, +- NVSP_STAT_FAIL); +- +- return; +- } +- +- /* Remove the 1st packet to represent the xfer page packet itself */ +- xferpage_packet = (struct xferpage_packet *)listHead.next; +- list_del(&xferpage_packet->list_ent); +- xferpage_packet->status = NVSP_STAT_SUCCESS; +- xferpage_packet->channel = channel; +- +- /* This is how much we can satisfy */ +- xferpage_packet->count = count - 1; +- +- if (xferpage_packet->count != vmxferpage_packet->range_cnt) { +- netdev_err(ndev, "Needed %d netvsc pkts to satisfy " +- "this xfer page...got %d\n", +- vmxferpage_packet->range_cnt, xferpage_packet->count); +- } ++ count = vmxferpage_packet->range_cnt; ++ netvsc_packet->device = device; ++ netvsc_packet->channel = channel; + + /* Each range represents 1 RNDIS pkt that contains 1 ethernet frame */ +- for (i = 0; i < (count - 1); i++) { +- netvsc_packet = (struct hv_netvsc_packet *)listHead.next; +- list_del(&netvsc_packet->list_ent); +- ++ for (i = 0; i < count; i++) { + /* Initialize the netvsc packet */ + netvsc_packet->status = NVSP_STAT_SUCCESS; +- netvsc_packet->xfer_page_pkt = xferpage_packet; +- netvsc_packet->completion.recv.recv_completion = +- netvsc_receive_completion; +- netvsc_packet->completion.recv.recv_completion_ctx = +- netvsc_packet; +- netvsc_packet->device = device; +- /* Save this so that we can send it back */ +- netvsc_packet->completion.recv.recv_completion_tid = +- vmxferpage_packet->d.trans_id; +- + netvsc_packet->data = (void *)((unsigned long)net_device-> + recv_buf + vmxferpage_packet->ranges[i].byte_offset); + netvsc_packet->total_data_buflen = +@@ -826,10 +696,12 @@ static void netvsc_receive(struct netvsc_device *net_device, + /* Pass it to the upper layer */ + rndis_filter_receive(device, netvsc_packet); + +- netvsc_receive_completion(netvsc_packet-> +- completion.recv.recv_completion_ctx); ++ if (netvsc_packet->status != NVSP_STAT_SUCCESS) ++ status = NVSP_STAT_FAIL; + } + ++ netvsc_send_recv_completion(device, channel, net_device, ++ vmxferpage_packet->d.trans_id, status); + } + + +@@ -956,11 +828,9 @@ void netvsc_channel_cb(void *context) + int netvsc_device_add(struct hv_device *device, void *additional_info) + { + int ret = 0; +- int i; + int ring_size = + ((struct netvsc_device_info *)additional_info)->ring_size; + struct netvsc_device *net_device; +- struct hv_netvsc_packet *packet, *pos; + struct net_device *ndev; + + net_device = alloc_net_device(device); +@@ -981,18 +851,6 @@ int netvsc_device_add(struct hv_device *device, void *additional_info) + ndev = net_device->ndev; + + /* Initialize the NetVSC channel extension */ +- spin_lock_init(&net_device->recv_pkt_list_lock); +- +- INIT_LIST_HEAD(&net_device->recv_pkt_list); +- +- for (i = 0; i < NETVSC_RECEIVE_PACKETLIST_COUNT; i++) { +- packet = kzalloc(sizeof(struct hv_netvsc_packet), GFP_KERNEL); +- if (!packet) +- break; +- +- list_add_tail(&packet->list_ent, +- &net_device->recv_pkt_list); +- } + init_completion(&net_device->channel_init_wait); + + set_per_channel_state(device->channel, net_device->cb_buffer); +@@ -1028,16 +886,8 @@ close: + + cleanup: + +- if (net_device) { +- list_for_each_entry_safe(packet, pos, +- &net_device->recv_pkt_list, +- list_ent) { +- list_del(&packet->list_ent); +- kfree(packet); +- } +- ++ if (net_device) + kfree(net_device); +- } + + return ret; + } +diff --git a/drivers/net/hyperv/netvsc_drv.c b/drivers/net/hyperv/netvsc_drv.c +index e486dbd33f61..6cc4db064fec 100644 +--- a/drivers/net/hyperv/netvsc_drv.c ++++ b/drivers/net/hyperv/netvsc_drv.c +@@ -643,7 +643,7 @@ int netvsc_recv_callback(struct hv_device *device_obj, + __vlan_hwaccel_put_tag(skb, htons(ETH_P_8021Q), + packet->vlan_tci); + +- skb_record_rx_queue(skb, packet->xfer_page_pkt->channel-> ++ skb_record_rx_queue(skb, packet->channel-> + offermsg.offer.sub_channel_index % + net->real_num_rx_queues); + +diff --git a/drivers/net/hyperv/rndis_filter.c b/drivers/net/hyperv/rndis_filter.c +index d92cfbe43410..48f5a0fbd674 100644 +--- a/drivers/net/hyperv/rndis_filter.c ++++ b/drivers/net/hyperv/rndis_filter.c +@@ -401,8 +401,6 @@ static void rndis_filter_receive_data(struct rndis_device *dev, + pkt->total_data_buflen = rndis_pkt->data_len; + pkt->data = (void *)((unsigned long)pkt->data + data_offset); + +- pkt->is_data_pkt = true; +- + vlan = rndis_get_ppi(rndis_pkt, IEEE_8021Q_INFO); + if (vlan) { + pkt->vlan_tci = VLAN_TAG_PRESENT | vlan->vlanid | +-- +2.4.3 + diff --git a/src/patches/linux/0018-hyperv-Simplify-the-send_completion-variables.patch b/src/patches/linux/0018-hyperv-Simplify-the-send_completion-variables.patch new file mode 100644 index 0000000..82b742b --- /dev/null +++ b/src/patches/linux/0018-hyperv-Simplify-the-send_completion-variables.patch @@ -0,0 +1,105 @@ +From d6eeeb452c1e6e7cf14f4a581a8f2ea2c50ec17a Mon Sep 17 00:00:00 2001 +From: Haiyang Zhang haiyangz@microsoft.com +Date: Mon, 21 Apr 2014 14:54:44 -0700 +Subject: [PATCH 18/25] hyperv: Simplify the send_completion variables + +The union contains only one member now, so we use the variables in it directly. + +Signed-off-by: Haiyang Zhang haiyangz@microsoft.com +Reviewed-by: K. Y. Srinivasan kys@microsoft.com +Signed-off-by: David S. Miller davem@davemloft.net +--- + drivers/net/hyperv/hyperv_net.h | 10 +++------- + drivers/net/hyperv/netvsc.c | 7 +++---- + drivers/net/hyperv/netvsc_drv.c | 8 ++++---- + drivers/net/hyperv/rndis_filter.c | 2 +- + 4 files changed, 11 insertions(+), 16 deletions(-) + +diff --git a/drivers/net/hyperv/hyperv_net.h b/drivers/net/hyperv/hyperv_net.h +index a1af0f7711e2..d1f7826aa75f 100644 +--- a/drivers/net/hyperv/hyperv_net.h ++++ b/drivers/net/hyperv/hyperv_net.h +@@ -136,13 +136,9 @@ struct hv_netvsc_packet { + u16 q_idx; + struct vmbus_channel *channel; + +- union { +- struct { +- u64 send_completion_tid; +- void *send_completion_ctx; +- void (*send_completion)(void *context); +- } send; +- } completion; ++ u64 send_completion_tid; ++ void *send_completion_ctx; ++ void (*send_completion)(void *context); + + /* This points to the memory after page_buf */ + struct rndis_message *rndis_msg; +diff --git a/drivers/net/hyperv/netvsc.c b/drivers/net/hyperv/netvsc.c +index b10334773b32..bbee44635035 100644 +--- a/drivers/net/hyperv/netvsc.c ++++ b/drivers/net/hyperv/netvsc.c +@@ -479,9 +479,8 @@ static void netvsc_send_completion(struct netvsc_device *net_device, + if (nvsc_packet) { + q_idx = nvsc_packet->q_idx; + channel = nvsc_packet->channel; +- nvsc_packet->completion.send.send_completion( +- nvsc_packet->completion.send. +- send_completion_ctx); ++ nvsc_packet->send_completion(nvsc_packet-> ++ send_completion_ctx); + } + + num_outstanding_sends = +@@ -534,7 +533,7 @@ int netvsc_send(struct hv_device *device, + 0xFFFFFFFF; + sendMessage.msg.v1_msg.send_rndis_pkt.send_buf_section_size = 0; + +- if (packet->completion.send.send_completion) ++ if (packet->send_completion) + req_id = (ulong)packet; + else + req_id = 0; +diff --git a/drivers/net/hyperv/netvsc_drv.c b/drivers/net/hyperv/netvsc_drv.c +index 6cc4db064fec..f3d3525ed42d 100644 +--- a/drivers/net/hyperv/netvsc_drv.c ++++ b/drivers/net/hyperv/netvsc_drv.c +@@ -235,7 +235,7 @@ static void netvsc_xmit_completion(void *context) + { + struct hv_netvsc_packet *packet = (struct hv_netvsc_packet *)context; + struct sk_buff *skb = (struct sk_buff *) +- (unsigned long)packet->completion.send.send_completion_tid; ++ (unsigned long)packet->send_completion_tid; + + kfree(packet); + +@@ -426,9 +426,9 @@ static int netvsc_start_xmit(struct sk_buff *skb, struct net_device *net) + (num_data_pgs * sizeof(struct hv_page_buffer))); + + /* Set the completion routine */ +- packet->completion.send.send_completion = netvsc_xmit_completion; +- packet->completion.send.send_completion_ctx = packet; +- packet->completion.send.send_completion_tid = (unsigned long)skb; ++ packet->send_completion = netvsc_xmit_completion; ++ packet->send_completion_ctx = packet; ++ packet->send_completion_tid = (unsigned long)skb; + + isvlan = packet->vlan_tci & VLAN_TAG_PRESENT; + +diff --git a/drivers/net/hyperv/rndis_filter.c b/drivers/net/hyperv/rndis_filter.c +index 48f5a0fbd674..99c527adae5b 100644 +--- a/drivers/net/hyperv/rndis_filter.c ++++ b/drivers/net/hyperv/rndis_filter.c +@@ -236,7 +236,7 @@ static int rndis_filter_send_request(struct rndis_device *dev, + packet->page_buf[0].len; + } + +- packet->completion.send.send_completion = NULL; ++ packet->send_completion = NULL; + + ret = netvsc_send(dev->net_dev->dev, packet); + return ret; +-- +2.4.3 + diff --git a/src/patches/linux/0019-hyperv-Enable-sendbuf-mechanism-on-the-send-path.patch b/src/patches/linux/0019-hyperv-Enable-sendbuf-mechanism-on-the-send-path.patch new file mode 100644 index 0000000..3940865 --- /dev/null +++ b/src/patches/linux/0019-hyperv-Enable-sendbuf-mechanism-on-the-send-path.patch @@ -0,0 +1,407 @@ +From 4685e50349d5dc5fe485c898ca3ce539e93a0118 Mon Sep 17 00:00:00 2001 +From: KY Srinivasan kys@microsoft.com +Date: Wed, 30 Apr 2014 10:14:31 -0700 +Subject: [PATCH 19/25] hyperv: Enable sendbuf mechanism on the send path + +We send packets using a copy-free mechanism (this is the Guest to Host transport +via VMBUS). While this is obviously optimal for large packets, +it may not be optimal for small packets. Hyper-V host supports +a second mechanism for sending packets that is "copy based". We implement that +mechanism in this patch. + +In this version of the patch I have addressed a comment from David Miller. + +With this patch (and all of the other offload and VRSS patches), we are now able +to almost saturate a 10G interface between Linux VMs on Hyper-V +on different hosts - close to 9 Gbps as measured via iperf. + +Signed-off-by: K. Y. Srinivasan kys@microsoft.com +Reviewed-by: Haiyang Zhang haiyangz@microsoft.com +Signed-off-by: David S. Miller davem@davemloft.net +--- + drivers/net/hyperv/hyperv_net.h | 14 +++ + drivers/net/hyperv/netvsc.c | 226 ++++++++++++++++++++++++++++++++++++++-- + drivers/net/hyperv/netvsc_drv.c | 3 +- + 3 files changed, 234 insertions(+), 9 deletions(-) + +diff --git a/drivers/net/hyperv/hyperv_net.h b/drivers/net/hyperv/hyperv_net.h +index d1f7826aa75f..4b7df5a5c966 100644 +--- a/drivers/net/hyperv/hyperv_net.h ++++ b/drivers/net/hyperv/hyperv_net.h +@@ -140,6 +140,8 @@ struct hv_netvsc_packet { + void *send_completion_ctx; + void (*send_completion)(void *context); + ++ u32 send_buf_index; ++ + /* This points to the memory after page_buf */ + struct rndis_message *rndis_msg; + +@@ -582,6 +584,9 @@ struct nvsp_message { + + #define NETVSC_RECEIVE_BUFFER_SIZE (1024*1024*16) /* 16MB */ + #define NETVSC_RECEIVE_BUFFER_SIZE_LEGACY (1024*1024*15) /* 15MB */ ++#define NETVSC_SEND_BUFFER_SIZE (1024 * 1024) /* 1MB */ ++#define NETVSC_INVALID_INDEX -1 ++ + + #define NETVSC_RECEIVE_BUFFER_ID 0xcafe + +@@ -607,6 +612,15 @@ struct netvsc_device { + u32 recv_section_cnt; + struct nvsp_1_receive_buffer_section *recv_section; + ++ /* Send buffer allocated by us */ ++ void *send_buf; ++ u32 send_buf_size; ++ u32 send_buf_gpadl_handle; ++ u32 send_section_cnt; ++ u32 send_section_size; ++ unsigned long *send_section_map; ++ int map_words; ++ + /* Used for NetVSP initialization protocol */ + struct completion channel_init_wait; + struct nvsp_message channel_init_pkt; +diff --git a/drivers/net/hyperv/netvsc.c b/drivers/net/hyperv/netvsc.c +index bbee44635035..c041f63a6d30 100644 +--- a/drivers/net/hyperv/netvsc.c ++++ b/drivers/net/hyperv/netvsc.c +@@ -28,6 +28,7 @@ + #include <linux/slab.h> + #include <linux/netdevice.h> + #include <linux/if_ether.h> ++#include <asm/sync_bitops.h> + + #include "hyperv_net.h" + +@@ -80,7 +81,7 @@ get_in_err: + } + + +-static int netvsc_destroy_recv_buf(struct netvsc_device *net_device) ++static int netvsc_destroy_buf(struct netvsc_device *net_device) + { + struct nvsp_message *revoke_packet; + int ret = 0; +@@ -146,10 +147,62 @@ static int netvsc_destroy_recv_buf(struct netvsc_device *net_device) + net_device->recv_section = NULL; + } + ++ /* Deal with the send buffer we may have setup. ++ * If we got a send section size, it means we received a ++ * SendsendBufferComplete msg (ie sent ++ * NvspMessage1TypeSendReceiveBuffer msg) therefore, we need ++ * to send a revoke msg here ++ */ ++ if (net_device->send_section_size) { ++ /* Send the revoke receive buffer */ ++ revoke_packet = &net_device->revoke_packet; ++ memset(revoke_packet, 0, sizeof(struct nvsp_message)); ++ ++ revoke_packet->hdr.msg_type = ++ NVSP_MSG1_TYPE_REVOKE_SEND_BUF; ++ revoke_packet->msg.v1_msg.revoke_recv_buf.id = 0; ++ ++ ret = vmbus_sendpacket(net_device->dev->channel, ++ revoke_packet, ++ sizeof(struct nvsp_message), ++ (unsigned long)revoke_packet, ++ VM_PKT_DATA_INBAND, 0); ++ /* If we failed here, we might as well return and ++ * have a leak rather than continue and a bugchk ++ */ ++ if (ret != 0) { ++ netdev_err(ndev, "unable to send " ++ "revoke send buffer to netvsp\n"); ++ return ret; ++ } ++ } ++ /* Teardown the gpadl on the vsp end */ ++ if (net_device->send_buf_gpadl_handle) { ++ ret = vmbus_teardown_gpadl(net_device->dev->channel, ++ net_device->send_buf_gpadl_handle); ++ ++ /* If we failed here, we might as well return and have a leak ++ * rather than continue and a bugchk ++ */ ++ if (ret != 0) { ++ netdev_err(ndev, ++ "unable to teardown send buffer's gpadl\n"); ++ return ret; ++ } ++ net_device->recv_buf_gpadl_handle = 0; ++ } ++ if (net_device->send_buf) { ++ /* Free up the receive buffer */ ++ free_pages((unsigned long)net_device->send_buf, ++ get_order(net_device->send_buf_size)); ++ net_device->send_buf = NULL; ++ } ++ kfree(net_device->send_section_map); ++ + return ret; + } + +-static int netvsc_init_recv_buf(struct hv_device *device) ++static int netvsc_init_buf(struct hv_device *device) + { + int ret = 0; + int t; +@@ -248,10 +301,90 @@ static int netvsc_init_recv_buf(struct hv_device *device) + goto cleanup; + } + ++ /* Now setup the send buffer. ++ */ ++ net_device->send_buf = ++ (void *)__get_free_pages(GFP_KERNEL|__GFP_ZERO, ++ get_order(net_device->send_buf_size)); ++ if (!net_device->send_buf) { ++ netdev_err(ndev, "unable to allocate send " ++ "buffer of size %d\n", net_device->send_buf_size); ++ ret = -ENOMEM; ++ goto cleanup; ++ } ++ ++ /* Establish the gpadl handle for this buffer on this ++ * channel. Note: This call uses the vmbus connection rather ++ * than the channel to establish the gpadl handle. ++ */ ++ ret = vmbus_establish_gpadl(device->channel, net_device->send_buf, ++ net_device->send_buf_size, ++ &net_device->send_buf_gpadl_handle); ++ if (ret != 0) { ++ netdev_err(ndev, ++ "unable to establish send buffer's gpadl\n"); ++ goto cleanup; ++ } ++ ++ /* Notify the NetVsp of the gpadl handle */ ++ init_packet = &net_device->channel_init_pkt; ++ memset(init_packet, 0, sizeof(struct nvsp_message)); ++ init_packet->hdr.msg_type = NVSP_MSG1_TYPE_SEND_SEND_BUF; ++ init_packet->msg.v1_msg.send_recv_buf.gpadl_handle = ++ net_device->send_buf_gpadl_handle; ++ init_packet->msg.v1_msg.send_recv_buf.id = 0; ++ ++ /* Send the gpadl notification request */ ++ ret = vmbus_sendpacket(device->channel, init_packet, ++ sizeof(struct nvsp_message), ++ (unsigned long)init_packet, ++ VM_PKT_DATA_INBAND, ++ VMBUS_DATA_PACKET_FLAG_COMPLETION_REQUESTED); ++ if (ret != 0) { ++ netdev_err(ndev, ++ "unable to send send buffer's gpadl to netvsp\n"); ++ goto cleanup; ++ } ++ ++ t = wait_for_completion_timeout(&net_device->channel_init_wait, 5*HZ); ++ BUG_ON(t == 0); ++ ++ /* Check the response */ ++ if (init_packet->msg.v1_msg. ++ send_send_buf_complete.status != NVSP_STAT_SUCCESS) { ++ netdev_err(ndev, "Unable to complete send buffer " ++ "initialization with NetVsp - status %d\n", ++ init_packet->msg.v1_msg. ++ send_recv_buf_complete.status); ++ ret = -EINVAL; ++ goto cleanup; ++ } ++ ++ /* Parse the response */ ++ net_device->send_section_size = init_packet->msg. ++ v1_msg.send_send_buf_complete.section_size; ++ ++ /* Section count is simply the size divided by the section size. ++ */ ++ net_device->send_section_cnt = ++ net_device->send_buf_size/net_device->send_section_size; ++ ++ dev_info(&device->device, "Send section size: %d, Section count:%d\n", ++ net_device->send_section_size, net_device->send_section_cnt); ++ ++ /* Setup state for managing the send buffer. */ ++ net_device->map_words = DIV_ROUND_UP(net_device->send_section_cnt, ++ BITS_PER_LONG); ++ ++ net_device->send_section_map = ++ kzalloc(net_device->map_words * sizeof(ulong), GFP_KERNEL); ++ if (net_device->send_section_map == NULL) ++ goto cleanup; ++ + goto exit; + + cleanup: +- netvsc_destroy_recv_buf(net_device); ++ netvsc_destroy_buf(net_device); + + exit: + return ret; +@@ -369,8 +502,9 @@ static int netvsc_connect_vsp(struct hv_device *device) + net_device->recv_buf_size = NETVSC_RECEIVE_BUFFER_SIZE_LEGACY; + else + net_device->recv_buf_size = NETVSC_RECEIVE_BUFFER_SIZE; ++ net_device->send_buf_size = NETVSC_SEND_BUFFER_SIZE; + +- ret = netvsc_init_recv_buf(device); ++ ret = netvsc_init_buf(device); + + cleanup: + return ret; +@@ -378,7 +512,7 @@ cleanup: + + static void netvsc_disconnect_vsp(struct netvsc_device *net_device) + { +- netvsc_destroy_recv_buf(net_device); ++ netvsc_destroy_buf(net_device); + } + + /* +@@ -440,6 +574,12 @@ static inline u32 hv_ringbuf_avail_percent( + return avail_write * 100 / ring_info->ring_datasize; + } + ++static inline void netvsc_free_send_slot(struct netvsc_device *net_device, ++ u32 index) ++{ ++ sync_change_bit(index, net_device->send_section_map); ++} ++ + static void netvsc_send_completion(struct netvsc_device *net_device, + struct hv_device *device, + struct vmpacket_descriptor *packet) +@@ -447,6 +587,7 @@ static void netvsc_send_completion(struct netvsc_device *net_device, + struct nvsp_message *nvsp_packet; + struct hv_netvsc_packet *nvsc_packet; + struct net_device *ndev; ++ u32 send_index; + + ndev = net_device->ndev; + +@@ -477,6 +618,9 @@ static void netvsc_send_completion(struct netvsc_device *net_device, + + /* Notify the layer above us */ + if (nvsc_packet) { ++ send_index = nvsc_packet->send_buf_index; ++ if (send_index != NETVSC_INVALID_INDEX) ++ netvsc_free_send_slot(net_device, send_index); + q_idx = nvsc_packet->q_idx; + channel = nvsc_packet->channel; + nvsc_packet->send_completion(nvsc_packet-> +@@ -504,6 +648,52 @@ static void netvsc_send_completion(struct netvsc_device *net_device, + + } + ++static u32 netvsc_get_next_send_section(struct netvsc_device *net_device) ++{ ++ unsigned long index; ++ u32 max_words = net_device->map_words; ++ unsigned long *map_addr = (unsigned long *)net_device->send_section_map; ++ u32 section_cnt = net_device->send_section_cnt; ++ int ret_val = NETVSC_INVALID_INDEX; ++ int i; ++ int prev_val; ++ ++ for (i = 0; i < max_words; i++) { ++ if (!~(map_addr[i])) ++ continue; ++ index = ffz(map_addr[i]); ++ prev_val = sync_test_and_set_bit(index, &map_addr[i]); ++ if (prev_val) ++ continue; ++ if ((index + (i * BITS_PER_LONG)) >= section_cnt) ++ break; ++ ret_val = (index + (i * BITS_PER_LONG)); ++ break; ++ } ++ return ret_val; ++} ++ ++u32 netvsc_copy_to_send_buf(struct netvsc_device *net_device, ++ unsigned int section_index, ++ struct hv_netvsc_packet *packet) ++{ ++ char *start = net_device->send_buf; ++ char *dest = (start + (section_index * net_device->send_section_size)); ++ int i; ++ u32 msg_size = 0; ++ ++ for (i = 0; i < packet->page_buf_cnt; i++) { ++ char *src = phys_to_virt(packet->page_buf[i].pfn << PAGE_SHIFT); ++ u32 offset = packet->page_buf[i].offset; ++ u32 len = packet->page_buf[i].len; ++ ++ memcpy(dest, (src + offset), len); ++ msg_size += len; ++ dest += len; ++ } ++ return msg_size; ++} ++ + int netvsc_send(struct hv_device *device, + struct hv_netvsc_packet *packet) + { +@@ -513,6 +703,10 @@ int netvsc_send(struct hv_device *device, + struct net_device *ndev; + struct vmbus_channel *out_channel = NULL; + u64 req_id; ++ unsigned int section_index = NETVSC_INVALID_INDEX; ++ u32 msg_size = 0; ++ struct sk_buff *skb; ++ + + net_device = get_outbound_net_device(device); + if (!net_device) +@@ -528,10 +722,26 @@ int netvsc_send(struct hv_device *device, + sendMessage.msg.v1_msg.send_rndis_pkt.channel_type = 1; + } + +- /* Not using send buffer section */ ++ /* Attempt to send via sendbuf */ ++ if (packet->total_data_buflen < net_device->send_section_size) { ++ section_index = netvsc_get_next_send_section(net_device); ++ if (section_index != NETVSC_INVALID_INDEX) { ++ msg_size = netvsc_copy_to_send_buf(net_device, ++ section_index, ++ packet); ++ skb = (struct sk_buff *) ++ (unsigned long)packet->send_completion_tid; ++ if (skb) ++ dev_kfree_skb_any(skb); ++ packet->page_buf_cnt = 0; ++ } ++ } ++ packet->send_buf_index = section_index; ++ ++ + sendMessage.msg.v1_msg.send_rndis_pkt.send_buf_section_index = +- 0xFFFFFFFF; +- sendMessage.msg.v1_msg.send_rndis_pkt.send_buf_section_size = 0; ++ section_index; ++ sendMessage.msg.v1_msg.send_rndis_pkt.send_buf_section_size = msg_size; + + if (packet->send_completion) + req_id = (ulong)packet; +diff --git a/drivers/net/hyperv/netvsc_drv.c b/drivers/net/hyperv/netvsc_drv.c +index f3d3525ed42d..9a19aa5672e6 100644 +--- a/drivers/net/hyperv/netvsc_drv.c ++++ b/drivers/net/hyperv/netvsc_drv.c +@@ -236,10 +236,11 @@ static void netvsc_xmit_completion(void *context) + struct hv_netvsc_packet *packet = (struct hv_netvsc_packet *)context; + struct sk_buff *skb = (struct sk_buff *) + (unsigned long)packet->send_completion_tid; ++ u32 index = packet->send_buf_index; + + kfree(packet); + +- if (skb) ++ if (skb && (index == NETVSC_INVALID_INDEX)) + dev_kfree_skb_any(skb); + } + +-- +2.4.3 + diff --git a/src/patches/linux/0020-Add-support-for-netvsc-build-without-CONFIG_SYSFS-fl.patch b/src/patches/linux/0020-Add-support-for-netvsc-build-without-CONFIG_SYSFS-fl.patch new file mode 100644 index 0000000..b610b54 --- /dev/null +++ b/src/patches/linux/0020-Add-support-for-netvsc-build-without-CONFIG_SYSFS-fl.patch @@ -0,0 +1,42 @@ +From 3f0b77385356301cf4718a94f76a4068588ecb8e Mon Sep 17 00:00:00 2001 +From: Haiyang Zhang haiyangz@microsoft.com +Date: Thu, 8 May 2014 15:14:10 -0700 +Subject: [PATCH 20/25] Add support for netvsc build without CONFIG_SYSFS flag + +This change ensures the driver can be built successfully without the +CONFIG_SYSFS flag. +MS-TFS: 182270 + +Signed-off-by: Haiyang Zhang haiyangz@microsoft.com +Reviewed-by: K. Y. Srinivasan kys@microsoft.com +Signed-off-by: David S. Miller davem@davemloft.net +--- + drivers/net/hyperv/netvsc_drv.c | 5 +---- + 1 file changed, 1 insertion(+), 4 deletions(-) + +diff --git a/drivers/net/hyperv/netvsc_drv.c b/drivers/net/hyperv/netvsc_drv.c +index 9a19aa5672e6..346f1aeb9c24 100644 +--- a/drivers/net/hyperv/netvsc_drv.c ++++ b/drivers/net/hyperv/netvsc_drv.c +@@ -645,8 +645,7 @@ int netvsc_recv_callback(struct hv_device *device_obj, + packet->vlan_tci); + + skb_record_rx_queue(skb, packet->channel-> +- offermsg.offer.sub_channel_index % +- net->real_num_rx_queues); ++ offermsg.offer.sub_channel_index); + + net->stats.rx_packets++; + net->stats.rx_bytes += packet->total_data_buflen; +@@ -829,8 +828,6 @@ static int netvsc_probe(struct hv_device *dev, + nvdev = hv_get_drvdata(dev); + netif_set_real_num_tx_queues(net, nvdev->num_chn); + netif_set_real_num_rx_queues(net, nvdev->num_chn); +- dev_info(&dev->device, "real num tx,rx queues:%u, %u\n", +- net->real_num_tx_queues, net->real_num_rx_queues); + + ret = register_netdev(net); + if (ret != 0) { +-- +2.4.3 + diff --git a/src/patches/linux/0021-hyperv-Add-hash-value-into-RNDIS-Per-packet-info.patch b/src/patches/linux/0021-hyperv-Add-hash-value-into-RNDIS-Per-packet-info.patch new file mode 100644 index 0000000..c522d28 --- /dev/null +++ b/src/patches/linux/0021-hyperv-Add-hash-value-into-RNDIS-Per-packet-info.patch @@ -0,0 +1,93 @@ +From 1267d9b235e7612f8cdfa842a0433bb82e75f7fb Mon Sep 17 00:00:00 2001 +From: Haiyang Zhang haiyangz@microsoft.com +Date: Wed, 21 May 2014 12:55:39 -0700 +Subject: [PATCH 21/25] hyperv: Add hash value into RNDIS Per-packet info + +It passes the hash value as the RNDIS Per-packet info to the Hyper-V host, +so that the send completion notices can be spread across multiple channels. +MS-TFS: 140273 + +Signed-off-by: Haiyang Zhang haiyangz@microsoft.com +Signed-off-by: David S. Miller davem@davemloft.net +--- + drivers/net/hyperv/hyperv_net.h | 4 ++++ + drivers/net/hyperv/netvsc_drv.c | 18 ++++++++++++++---- + 2 files changed, 18 insertions(+), 4 deletions(-) + +diff --git a/drivers/net/hyperv/hyperv_net.h b/drivers/net/hyperv/hyperv_net.h +index 4b7df5a5c966..6cc37c15e0bf 100644 +--- a/drivers/net/hyperv/hyperv_net.h ++++ b/drivers/net/hyperv/hyperv_net.h +@@ -791,6 +791,7 @@ enum ndis_per_pkt_info_type { + IEEE_8021Q_INFO, + ORIGINAL_PKTINFO, + PACKET_CANCEL_ID, ++ NBL_HASH_VALUE = PACKET_CANCEL_ID, + ORIGINAL_NET_BUFLIST, + CACHED_NET_BUFLIST, + SHORT_PKT_PADINFO, +@@ -937,6 +938,9 @@ struct ndis_tcp_lso_info { + #define NDIS_LSO_PPI_SIZE (sizeof(struct rndis_per_packet_info) + \ + sizeof(struct ndis_tcp_lso_info)) + ++#define NDIS_HASH_PPI_SIZE (sizeof(struct rndis_per_packet_info) + \ ++ sizeof(u32)) ++ + /* Format of Information buffer passed in a SetRequest for the OID */ + /* OID_GEN_RNDIS_CONFIG_PARAMETER. */ + struct rndis_config_parameter_info { +diff --git a/drivers/net/hyperv/netvsc_drv.c b/drivers/net/hyperv/netvsc_drv.c +index 346f1aeb9c24..bd3b3acd04dc 100644 +--- a/drivers/net/hyperv/netvsc_drv.c ++++ b/drivers/net/hyperv/netvsc_drv.c +@@ -224,9 +224,11 @@ static u16 netvsc_select_queue(struct net_device *ndev, struct sk_buff *skb, + if (nvsc_dev == NULL || ndev->real_num_tx_queues <= 1) + return 0; + +- if (netvsc_set_hash(&hash, skb)) ++ if (netvsc_set_hash(&hash, skb)) { + q_idx = nvsc_dev->send_table[hash % VRSS_SEND_TAB_SIZE] % + ndev->real_num_tx_queues; ++ skb_set_hash(skb, hash, PKT_HASH_TYPE_L3); ++ } + + return q_idx; + } +@@ -385,6 +387,7 @@ static int netvsc_start_xmit(struct sk_buff *skb, struct net_device *net) + struct ndis_tcp_lso_info *lso_info; + int hdr_offset; + u32 net_trans_info; ++ u32 hash; + + + /* We will atmost need two pages to describe the rndis +@@ -403,9 +406,8 @@ static int netvsc_start_xmit(struct sk_buff *skb, struct net_device *net) + packet = kzalloc(sizeof(struct hv_netvsc_packet) + + (num_data_pgs * sizeof(struct hv_page_buffer)) + + sizeof(struct rndis_message) + +- NDIS_VLAN_PPI_SIZE + +- NDIS_CSUM_PPI_SIZE + +- NDIS_LSO_PPI_SIZE, GFP_ATOMIC); ++ NDIS_VLAN_PPI_SIZE + NDIS_CSUM_PPI_SIZE + ++ NDIS_LSO_PPI_SIZE + NDIS_HASH_PPI_SIZE, GFP_ATOMIC); + if (!packet) { + /* out of memory, drop packet */ + netdev_err(net, "unable to allocate hv_netvsc_packet\n"); +@@ -444,6 +446,14 @@ static int netvsc_start_xmit(struct sk_buff *skb, struct net_device *net) + + rndis_msg_size = RNDIS_MESSAGE_SIZE(struct rndis_packet); + ++ hash = skb_get_hash_raw(skb); ++ if (hash != 0 && net->real_num_tx_queues > 1) { ++ rndis_msg_size += NDIS_HASH_PPI_SIZE; ++ ppi = init_ppi_data(rndis_msg, NDIS_HASH_PPI_SIZE, ++ NBL_HASH_VALUE); ++ *(u32 *)((void *)ppi + ppi->ppi_offset) = hash; ++ } ++ + if (isvlan) { + struct ndis_pkt_8021q_info *vlan; + +-- +2.4.3 + diff --git a/src/patches/linux/0022-hyperv-fix-apparent-cut-n-paste-error-in-send-path-t.patch b/src/patches/linux/0022-hyperv-fix-apparent-cut-n-paste-error-in-send-path-t.patch new file mode 100644 index 0000000..d849ebb --- /dev/null +++ b/src/patches/linux/0022-hyperv-fix-apparent-cut-n-paste-error-in-send-path-t.patch @@ -0,0 +1,32 @@ +From ee99150c7f34737e4382e7199d2ffe3dfbb54a5c Mon Sep 17 00:00:00 2001 +From: Dave Jones davej@redhat.com +Date: Mon, 16 Jun 2014 16:59:02 -0400 +Subject: [PATCH 22/25] hyperv: fix apparent cut-n-paste error in send path + teardown + +c25aaf814a63: "hyperv: Enable sendbuf mechanism on the send path" added +some teardown code that looks like it was copied from the recieve path +above, but missed a variable name replacement. + +Signed-off-by: Dave Jones davej@redhat.com +Signed-off-by: David S. Miller davem@davemloft.net +--- + drivers/net/hyperv/netvsc.c | 2 +- + 1 file changed, 1 insertion(+), 1 deletion(-) + +diff --git a/drivers/net/hyperv/netvsc.c b/drivers/net/hyperv/netvsc.c +index c041f63a6d30..4ed38eaecea8 100644 +--- a/drivers/net/hyperv/netvsc.c ++++ b/drivers/net/hyperv/netvsc.c +@@ -189,7 +189,7 @@ static int netvsc_destroy_buf(struct netvsc_device *net_device) + "unable to teardown send buffer's gpadl\n"); + return ret; + } +- net_device->recv_buf_gpadl_handle = 0; ++ net_device->send_buf_gpadl_handle = 0; + } + if (net_device->send_buf) { + /* Free up the receive buffer */ +-- +2.4.3 + diff --git a/src/patches/linux/0023-hyperv-Fix-error-return-code-in-netvsc_init_buf.patch b/src/patches/linux/0023-hyperv-Fix-error-return-code-in-netvsc_init_buf.patch new file mode 100644 index 0000000..ab0960d --- /dev/null +++ b/src/patches/linux/0023-hyperv-Fix-error-return-code-in-netvsc_init_buf.patch @@ -0,0 +1,34 @@ +From f1009dfec7c439a958bd9ed8893dd6aa692c61f5 Mon Sep 17 00:00:00 2001 +From: Wei Yongjun yongjun_wei@trendmicro.com.cn +Date: Wed, 23 Jul 2014 09:00:35 +0800 +Subject: [PATCH 23/25] hyperv: Fix error return code in netvsc_init_buf() + +Fix to return -ENOMEM from the kalloc error handling +case instead of 0. + +Signed-off-by: Wei Yongjun yongjun_wei@trendmicro.com.cn +Reviewed-by: Haiyang Zhang haiyangz@microsoft.com +Signed-off-by: David S. Miller davem@davemloft.net +--- + drivers/net/hyperv/netvsc.c | 4 +++- + 1 file changed, 3 insertions(+), 1 deletion(-) + +diff --git a/drivers/net/hyperv/netvsc.c b/drivers/net/hyperv/netvsc.c +index 4ed38eaecea8..d97d5f39a04e 100644 +--- a/drivers/net/hyperv/netvsc.c ++++ b/drivers/net/hyperv/netvsc.c +@@ -378,8 +378,10 @@ static int netvsc_init_buf(struct hv_device *device) + + net_device->send_section_map = + kzalloc(net_device->map_words * sizeof(ulong), GFP_KERNEL); +- if (net_device->send_section_map == NULL) ++ if (net_device->send_section_map == NULL) { ++ ret = -ENOMEM; + goto cleanup; ++ } + + goto exit; + +-- +2.4.3 + diff --git a/src/patches/linux/0024-hyperv-Fix-a-bug-in-netvsc_send.patch b/src/patches/linux/0024-hyperv-Fix-a-bug-in-netvsc_send.patch new file mode 100644 index 0000000..8495fa2 --- /dev/null +++ b/src/patches/linux/0024-hyperv-Fix-a-bug-in-netvsc_send.patch @@ -0,0 +1,68 @@ +From 38bca2d5bd6fdaa0b8e1e415f79d89322c6825a8 Mon Sep 17 00:00:00 2001 +From: KY Srinivasan kys@microsoft.com +Date: Sun, 5 Oct 2014 10:42:51 -0700 +Subject: [PATCH 24/25] hyperv: Fix a bug in netvsc_send() + +[ Upstream commit 3a67c9ccad926a168d8b7891537a452018368a5b ] + +After the packet is successfully sent, we should not touch the packet +as it may have been freed. This patch is based on the work done by +Long Li longli@microsoft.com. + +David, please queue this up for stable. + +Signed-off-by: K. Y. Srinivasan kys@microsoft.com +Reported-by: Sitsofe Wheeler sitsofe@yahoo.com +Signed-off-by: David S. Miller davem@davemloft.net +Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org +--- + drivers/net/hyperv/netvsc.c | 15 ++++++++------- + 1 file changed, 8 insertions(+), 7 deletions(-) + +diff --git a/drivers/net/hyperv/netvsc.c b/drivers/net/hyperv/netvsc.c +index d97d5f39a04e..7edf976ecfa0 100644 +--- a/drivers/net/hyperv/netvsc.c ++++ b/drivers/net/hyperv/netvsc.c +@@ -708,6 +708,7 @@ int netvsc_send(struct hv_device *device, + unsigned int section_index = NETVSC_INVALID_INDEX; + u32 msg_size = 0; + struct sk_buff *skb; ++ u16 q_idx = packet->q_idx; + + + net_device = get_outbound_net_device(device); +@@ -772,24 +773,24 @@ int netvsc_send(struct hv_device *device, + + if (ret == 0) { + atomic_inc(&net_device->num_outstanding_sends); +- atomic_inc(&net_device->queue_sends[packet->q_idx]); ++ atomic_inc(&net_device->queue_sends[q_idx]); + + if (hv_ringbuf_avail_percent(&out_channel->outbound) < + RING_AVAIL_PERCENT_LOWATER) { + netif_tx_stop_queue(netdev_get_tx_queue( +- ndev, packet->q_idx)); ++ ndev, q_idx)); + + if (atomic_read(&net_device-> +- queue_sends[packet->q_idx]) < 1) ++ queue_sends[q_idx]) < 1) + netif_tx_wake_queue(netdev_get_tx_queue( +- ndev, packet->q_idx)); ++ ndev, q_idx)); + } + } else if (ret == -EAGAIN) { + netif_tx_stop_queue(netdev_get_tx_queue( +- ndev, packet->q_idx)); +- if (atomic_read(&net_device->queue_sends[packet->q_idx]) < 1) { ++ ndev, q_idx)); ++ if (atomic_read(&net_device->queue_sends[q_idx]) < 1) { + netif_tx_wake_queue(netdev_get_tx_queue( +- ndev, packet->q_idx)); ++ ndev, q_idx)); + ret = -ENOSPC; + } + } else { +-- +2.4.3 + diff --git a/src/patches/linux/0025-Drivers-hv-vmbus-Support-per-channel-driver-state.patch b/src/patches/linux/0025-Drivers-hv-vmbus-Support-per-channel-driver-state.patch new file mode 100644 index 0000000..47037be --- /dev/null +++ b/src/patches/linux/0025-Drivers-hv-vmbus-Support-per-channel-driver-state.patch @@ -0,0 +1,51 @@ +From 4b71288758aa1d510402b84ca93b1ef566575d22 Mon Sep 17 00:00:00 2001 +From: "K. Y. Srinivasan" kys@microsoft.com +Date: Mon, 3 Feb 2014 12:42:45 -0800 +Subject: [PATCH 25/25] Drivers: hv: vmbus: Support per-channel driver state + +As we implement Virtual Receive Side Scaling on the networking side +(the VRSS patches are currently under review), it will be useful to have +per-channel state that vmbus drivers can manage. Add support for +managing per-channel state. + +Signed-off-by: K. Y. Srinivasan kys@microsoft.com +Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org +--- + include/linux/hyperv.h | 14 ++++++++++++++ + 1 file changed, 14 insertions(+) + +diff --git a/include/linux/hyperv.h b/include/linux/hyperv.h +index 6088058a3e00..732dc7e37e96 100644 +--- a/include/linux/hyperv.h ++++ b/include/linux/hyperv.h +@@ -1045,6 +1045,10 @@ struct vmbus_channel { + * This will be NULL for the primary channel. + */ + struct vmbus_channel *primary_channel; ++ /* ++ * Support per-channel state for use by vmbus drivers. ++ */ ++ void *per_channel_state; + }; + + static inline void set_channel_read_state(struct vmbus_channel *c, bool state) +@@ -1052,6 +1056,16 @@ static inline void set_channel_read_state(struct vmbus_channel *c, bool state) + c->batched_reading = state; + } + ++static inline void set_per_channel_state(struct vmbus_channel *c, void *s) ++{ ++ c->per_channel_state = s; ++} ++ ++static inline void *get_per_channel_state(struct vmbus_channel *c) ++{ ++ return c->per_channel_state; ++} ++ + void vmbus_onmessage(void *context); + + int vmbus_request_offers(void); +-- +2.4.3 +
hooks/post-receive -- IPFire 2.x development tree
-
git＠ipfire.org