public inbox for development@lists.ipfire.org
 help / color / mirror / Atom feed
From: Adolf Belka <adolf.belka@ipfire.org>
To: development@lists.ipfire.org
Cc: Adolf Belka <adolf.belka@ipfire.org>
Subject: [PATCH] libffi: Update to version 3.8.1
Date: Tue, 17 Nov 2026 22:25:01 +0100	[thread overview]
Message-ID: <20261117212501.3424309-2-adolf.belka@ipfire.org> (raw)
In-Reply-To: <20261117212501.3424309-1-adolf.belka@ipfire.org>

- Update from version 3.8.0 to 3.8.1
- Update of rootfile
- 6 GHSA security fixes. Each is relevant to a specific platform.
- Changelog
3.8.1
    AArch64 — ffi_call placed the pointer to a by-value composite larger
     than 16 bytes at the wrong stack offset, and homogeneous-aggregate
     (HFA/HVA) detection did not match AAPCS64, so a callee or closure could use
     the wrong pointer. (GHSA-h52h-3759-xqm4, GHSA-vg77-j273-647r)
    macOS on Apple Silicon (Darwin arm64) — heap use-after-free in the
     trampoline-table closure allocator once a process used more than one
     trampoline table (over 1024 live closures). (GHSA-23hx-9rxv-9qfp)
    Linux and Cygwin (static trampolines) — the /proc/self/maps path was
     truncated at the first space and the reopened file was never verified, so a
     file at a truncated or replaced path could be mapped as trampoline code.
     (GHSA-fq46-69xf-x5gf)
    s390x — closures mishandled 128-bit integer arguments, passing an
     uninitialized pointer to the callback; they are now passed by reference like
     complex types. (GHSA-wf8m-qr3x-cch4)
    x86-64 (System V) — vector types whose calling convention is undefined
     (sub-8-byte vectors and <1 x double>, where GCC and Clang disagree) were
     misclassified; libffi now rejects them with FFI_BAD_TYPEDEF.
     (GHSA-5pc3-hr6f-q3mm)
    Raw API with static trampolines (notably i386) — raw-API closures passed
     the wrong pointer as the handler's data (and the i386 native path corrupted
     the closure), causing a wild indirect call. (GHSA-w6j9-w53f-2857)
Bug fixes and ABI conformance
    Broader ABI-conformance work and test coverage for complex-number calls,
     closures, and small aggregates across ARM, x86, PowerPC, MIPS, RISC-V,
     LoongArch, and SPARC.
    AArch64 / ARM big-endian — fixed closures receiving garbage sub-word
     integer and float arguments, and sub-word and float return values
     (#1012, closes #1011, #675).
    x86-64 — return a single-long double struct in %st0 like a bare
     long double (#1058); copy the upper half of 16-byte vector arguments in
     reusable call plans.
    powerpc64 — fixed _Complex long double when the ABI's long double is
     64 bits (e.g. -mlong-double-64 or musl).
    RISC-V — ignore the caller's return buffer for void calls.
    MIPS — emit CFI in the N32/N64 assembly so unwinding works through
     libffi frames (#1063).
    wasm — correct heap accesses at or above 2 GiB (#1062), make closures
     work with stack switching, and declare the stack-helper dependencies in
     EM_JS_DEPS for newer Emscripten (#1019).
    Closures / trampolines (all targets) — fail allocation gracefully when
     the page-size lookup fails (#1015) and bound the executable-mmap retry loop
     (#1055).
Build and packaging
    Keep the bundled dlmalloc symbols static so they cannot clash with an
     application's allocator, and emit subsections_via_symbols on Apple
     platforms for linker dead-stripping (#1017).
    Honor an explicit --enable-docs, respect every --libdir spelling, and
     keep CCASFLAGS in sync with CFLAGS (#1060).

Signed-off-by: Adolf Belka <adolf.belka@ipfire.org>
---
 config/rootfiles/common/libffi | 2 +-
 lfs/libffi                     | 4 ++--
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/config/rootfiles/common/libffi b/config/rootfiles/common/libffi
index 41cc47b9e..6d6bb7b42 100644
--- a/config/rootfiles/common/libffi
+++ b/config/rootfiles/common/libffi
@@ -4,7 +4,7 @@
 #usr/lib/libffi.la
 #usr/lib/libffi.so
 usr/lib/libffi.so.8
-usr/lib/libffi.so.8.5.0
+usr/lib/libffi.so.8.5.1
 #usr/lib/pkgconfig/libffi.pc
 #usr/share/info/libffi.info
 #usr/share/man/man3/ffi.3
diff --git a/lfs/libffi b/lfs/libffi
index df8f5c3ac..1205650e3 100644
--- a/lfs/libffi
+++ b/lfs/libffi
@@ -24,7 +24,7 @@
 
 include Config
 
-VER        = 3.8.0
+VER        = 3.8.1
 
 THISAPP    = libffi-$(VER)
 DL_FILE    = $(THISAPP).tar.gz
@@ -40,7 +40,7 @@ objects = $(DL_FILE)
 
 $(DL_FILE) = $(DL_FROM)/$(DL_FILE)
 
-$(DL_FILE)_BLAKE2 = 589f87f79ea637dbddd00db3e320090f17855da5b1669a42b154f1c61bcafd1e8a964738d25b424c95ba5e148565df4b7379800ba42f40149c74d749f4f4b096
+$(DL_FILE)_BLAKE2 = 67a84bb411d4cdf6eb0d307c2b9e733f731f58780aa8f7d817b63605d0f8ac5e87224fffa212d6f560e38bed98a5ccc19f15fe54b74274da4f8908f55d6d4377
 
 install : $(TARGET)
 
-- 
2.56.1



      reply	other threads:[~2026-11-17 21:25 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-11-17 21:25 [PATCH] core207: Ship libffi Adolf Belka
2026-11-17 21:25 ` Adolf Belka [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20261117212501.3424309-2-adolf.belka@ipfire.org \
    --to=adolf.belka@ipfire.org \
    --cc=development@lists.ipfire.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox