rsync: Fix multiple CVEs
v2 cover rsync: Fix multiple CVEs
From: Ashishkumar Parmar <[email protected]>
Backport the upstream rsync security fixes for CVE-2026-29518, CVE-2026-43617, CVE-2026-43618, CVE-2026-43619, CVE-2026-43620, and CVE-2026-45232 to Wrynose.
Changes in v2:
- Updated the CVE-2026-43619 backport to restore the upstream commit description and explanatory comments for the openat2/Android fallback handling.
- Clarified the Wrynose-specific Backport Changes metadata.
- No functional changes were made to the other five CVE patches.
Ashishkumar Parmar (6):
rsync: Fix CVE-2026-29518
rsync: Fix CVE-2026-43619
rsync: Fix CVE-2026-43618
rsync: Fix CVE-2026-43620
rsync: Fix CVE-2026-43617
rsync: Fix CVE-2026-45232
Signature · 1 line
2.35.6
PATCH v2 1/6 rsync: Fix CVE-2026-29518
From: Ashishkumar Parmar <[email protected]>
This patch backports the upstream rsync v3.4.3 security fix train for CVE-2026-29518. The direct CVE fixes are [3] and [4]:
- [3] enables secure_relative_open() for daemon modules running with "use chroot = no", closing the receiver-side basis-file TOCTOU.
- [4] routes the sender read path through secure_relative_open() from the trusted module root, closing the matching sender-side TOCTOU.
This patch also carries [1] and [2] from the same upstream v3.4.3 security/update train. These are supporting secure_relative_open() changes, not standalone direct CVE fixes: [1] uses Linux openat2(RESOLVE_BENEATH) to preserve the same confinement while allowing legitimate in-tree symlinks, fixing upstream issue #715; [2] adds the equivalent FreeBSD/macOS O_RESOLVE_BENEATH path. Debian's rsync 3.4.1+ds1-5+deb13u3 security update also carries these commits before the direct CVE-2026-29518 commits.
The upstream fixed release is referenced in [5], and the public CVE record is referenced in [6]. Individual backported commit links are also recorded in the embedded patch headers.
[1] https://github.com/RsyncProject/rsync/commit/4fa7156ccdb2ad34b034d18fe2fd6cd79adef8a1
[2] https://github.com/RsyncProject/rsync/commit/7f60ec001a0be63b770707ec8b829524c3809a43
[3] https://github.com/RsyncProject/rsync/commit/f1c24ab03bc85cb2638a569faa60216582fb6c5d
[4] https://github.com/RsyncProject/rsync/commit/859d44fa4f1420775e4ba050337ef32092f2894c
[5] https://github.com/RsyncProject/rsync/releases/tag/v3.4.3
[6] https://www.cve.org/CVERecord?id=CVE-2026-29518
--- .../rsync/files/CVE-2026-29518_p1.patch | 392 ++++++++++++++++++ .../rsync/files/CVE-2026-29518_p2.patch | 98 +++++ .../rsync/files/CVE-2026-29518_p3.patch | 328 +++++++++++++++ .../rsync/files/CVE-2026-29518_p4.patch | 71 ++++ meta/recipes-devtools/rsync/rsync_3.4.1.bb | 4 + 5 files changed, 893 insertions(+) create mode 100644 meta/recipes-devtools/rsync/files/CVE-2026-29518_p1.patch create mode 100644 meta/recipes-devtools/rsync/files/CVE-2026-29518_p2.patch create mode 100644 meta/recipes-devtools/rsync/files/CVE-2026-29518_p3.patch create mode 100644 meta/recipes-devtools/rsync/files/CVE-2026-29518_p4.patch
meta/recipes-devtools/rsync/files/CVE-2026-29518_p1.patch new file +392 -0
@@ -0,0 +1,392 @@ +From e7b575413b962ac042079a55530577b20c7eee44 Mon Sep 17 00:00:00 2001 +From: Andrew Tridgell <[email protected]> +Date: Thu, 30 Apr 2026 08:39:22 +1000 +Subject: [PATCH] syscall: use openat2(RESOLVE_BENEATH) on Linux for + secure_relative_open + +The CVE fix in commit c35e283 made secure_relative_open() walk every +component of relpath with O_NOFOLLOW. That blocks every symlink in the +path, which is stricter than the threat model required: legitimate +directory symlinks within the destination tree (e.g. when using -K / +--copy-dirlinks) are also rejected, breaking delta transfers with +"failed verification -- update discarded". See issue #715. + +On Linux 5.6+, openat2(RESOLVE_BENEATH | RESOLVE_NO_MAGICLINKS) gives +us exactly what we want: the kernel rejects any resolution that would +escape the starting directory (via "..", absolute paths, or symlinks +pointing outside dirfd) while still following symlinks that resolve +within it. /proc magic-links are blocked too. + +Use openat2 first; fall back to the existing per-component O_NOFOLLOW +walk on ENOSYS (kernel < 5.6). The lexical "../" checks at the head +of the function are kept as defense in depth. The Linux gate is +plain #ifdef __linux__: the runtime ENOSYS fallback covers the only +case that actually matters (header present + old kernel), and any +Linux build environment without linux/openat2.h will fail with a +clear "no such file" error rather than silently disabling the +protection. + +Verified manually that openat2(RESOLVE_BENEATH) blocks all four +escape patterns (absolute symlink, ../ symlink, lexical .., absolute +path) while allowing direct and within-tree symlinks. The new +testsuite/symlink-dirlink-basis.test (taken from PR #864 by Samuel +Henrique) exercises the issue #715 regression and passes; full +make check passes 47/47. + +Test: testsuite/symlink-dirlink-basis.test (8 scenarios) +Fixes: https://github.com/RsyncProject/rsync/issues/715 + +Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]> + +CVE: CVE-2026-29518 +Upstream-Status: Backport [https://github.com/RsyncProject/rsync/commit/4fa7156ccdb2ad34b034d18fe2fd6cd79adef8a1] + +(cherry picked from commit 4fa7156ccdb2ad34b034d18fe2fd6cd79adef8a1) +Signed-off-by: Ashishkumar Parmar <[email protected]> +--- + syscall.c | 62 ++++++- + testsuite/symlink-dirlink-basis.test | 247 +++++++++++++++++++++++++++ + 2 files changed, 304 insertions(+), 5 deletions(-) + create mode 100755 testsuite/symlink-dirlink-basis.test + +diff --git a/syscall.c b/syscall.c +index 34a9bba0..d66c9fdd 100644 +--- a/syscall.c ++++ b/syscall.c +@@ -33,6 +33,11 @@ + #include <sys/syscall.h> + #endif + ++#ifdef __linux__ ++#include <sys/syscall.h> ++#include <linux/openat2.h> ++#endif ++ + #include "ifuncs.h" + + extern int dry_run; +@@ -715,12 +720,49 @@ int do_open_nofollow(const char *pathname, int flags) + /* + open a file relative to a base directory. The basedir can be NULL, + in which case the current working directory is used. The relpath +- must be a relative path, and the relpath must not contain any +- elements in the path which follow symlinks (ie. like O_NOFOLLOW, but +- applies to all path components, not just the last component) +- +- The relpath must also not contain any ../ elements in the path ++ must be a relative path. The kernel must guarantee that resolution ++ cannot escape basedir (or the cwd, when basedir is NULL): no ".." ++ jumps above the start, no symlinks pointing outside, no absolute ++ paths, no /proc magic-link tricks. ++ ++ Symlinks *within* basedir are followed normally — earlier rsync ++ versions rejected every symlink with O_NOFOLLOW on each component, ++ which broke legitimate directory symlinks on the receiver side ++ (https://github.com/RsyncProject/rsync/issues/715). The escape ++ prevention is handled by the kernel via openat2(RESOLVE_BENEATH) ++ on Linux 5.6+; older systems fall back to the per-component ++ O_NOFOLLOW walk below. ++ ++ The relpath must also not contain any ../ elements in the path. + */ ++ ++#ifdef __linux__ ++static int secure_relative_open_linux(const char *basedir, const char *relpath, int flags, mode_t mode) ++{ ++ struct open_how how; ++ int dirfd, retfd; ++ ++ memset(&how, 0, sizeof how); ++ how.flags = flags; ++ how.mode = mode; ++ how.resolve = RESOLVE_BENEATH | RESOLVE_NO_MAGICLINKS; ++ ++ if (basedir == NULL) { ++ dirfd = AT_FDCWD; ++ } else { ++ dirfd = openat(AT_FDCWD, basedir, O_RDONLY | O_DIRECTORY); ++ if (dirfd == -1) ++ return -1; ++ } ++ ++ retfd = syscall(SYS_openat2, dirfd, relpath, &how, sizeof how); ++ ++ if (dirfd != AT_FDCWD) ++ close(dirfd); ++ return retfd; ++} ++#endif ++ + int secure_relative_open(const char *basedir, const char *relpath, int flags, mode_t mode) + { + if (!relpath || relpath[0] == '/') { +@@ -734,6 +776,16 @@ int secure_relative_open(const char *basedir, const char *relpath, int flags, mo + return -1; + } + ++#ifdef __linux__ ++ { ++ int fd = secure_relative_open_linux(basedir, relpath, flags, mode); ++ /* ENOSYS = kernel < 5.6 doesn't have the syscall even though ++ * glibc/kernel-headers do; fall through to the portable path. */ ++ if (fd != -1 || errno != ENOSYS) ++ return fd; ++ } ++#endif ++ + #if !defined(O_NOFOLLOW) || !defined(O_DIRECTORY) || !defined(AT_FDCWD) + // really old system, all we can do is live with the risks + if (!basedir) { +diff --git a/testsuite/symlink-dirlink-basis.test b/testsuite/symlink-dirlink-basis.test +new file mode 100755 +index 00000000..9065dd81 +--- /dev/null ++++ b/testsuite/symlink-dirlink-basis.test +@@ -0,0 +1,247 @@ ++#!/bin/sh ++ ++# Test that updating a file through a directory symlink works when using ++# -K (--copy-dirlinks). This is a regression test for: ++# https://github.com/RsyncProject/rsync/issues/715 ++# ++# The CVE fix in commit c35e283 introduced secure_relative_open() which ++# uses O_NOFOLLOW on all path components, breaking legitimate directory ++# symlinks on the receiver side. The fix splits the path into basedir ++# (dirname, symlinks followed) and basename (O_NOFOLLOW) so that ++# directory symlinks are traversed while the final file component is ++# still protected. ++# ++# The regression only manifests when delta matching is triggered (i.e., ++# the sender finds matching blocks in the old file). Small files with ++# completely different content are transferred in full and don't trigger ++# the bug. We use a large file with a small modification to ensure ++# delta transfer is used. ++# ++# In addition to the original regression, this test covers edge cases ++# in the fix itself: ++# - --backup with directory symlinks (finish_transfer pointer identity) ++# - --partial-dir with protocol < 29 (fnamecmp != partialptr guard) ++# - --inplace with directory symlinks (updating_basis_or_equiv check) ++# - Files without a dirname (top-level files, no split needed) ++ ++. "$suitedir/rsync.fns" ++ ++RSYNC_RSH="$scratchdir/src/support/lsh.sh" ++export RSYNC_RSH ++ ++# $HOME is set to $scratchdir by rsync.fns ++# localhost: destination will cd to $HOME (i.e., $scratchdir) ++ ++# Helper: create a large file suitable for delta transfers. ++# ~32KB is large enough for rsync's block matching to find matches. ++make_testfile() { ++ dd if=/dev/urandom of="$1" bs=1024 count=32 2>/dev/null \ ++ || test_fail "failed to create test file $1" ++} ++ ++# Set up source tree ++srcbase="$tmpdir/src" ++ ++###################################################################### ++# Test 1: Basic directory symlink update (the original issue #715) ++###################################################################### ++ ++mkdir -p "$HOME/real-dir" ++ln -s real-dir "$HOME/dir" ++ ++mkdir -p "$srcbase/dir" ++make_testfile "$srcbase/dir/file" ++ ++# First transfer (initial): should create the file through the symlink ++(cd "$srcbase" && $RSYNC -KRlptv --rsync-path="$RSYNC" dir/file localhost:) \ ++ || test_fail "test 1: initial transfer failed" ++ ++if [ ! -f "$HOME/real-dir/file" ]; then ++ test_fail "test 1: initial transfer did not create file through symlink" ++fi ++ ++diff "$srcbase/dir/file" "$HOME/real-dir/file" >/dev/null \ ++ || test_fail "test 1: initial transfer content mismatch" ++ ++# Small modification to trigger delta transfer ++echo "appended update" >> "$srcbase/dir/file" ++sleep 1 ++touch "$srcbase/dir/file" ++ ++# Second transfer (update): was failing with "failed verification" ++(cd "$srcbase" && $RSYNC -KRlptv --rsync-path="$RSYNC" dir/file localhost:) \ ++ || test_fail "test 1: update through directory symlink failed" ++ ++diff "$srcbase/dir/file" "$HOME/real-dir/file" >/dev/null \ ++ || test_fail "test 1: update transfer content mismatch" ++ ++###################################################################### ++# Test 2: Compression (-z) as in the original reproducer ++###################################################################### ++ ++echo "another line" >> "$srcbase/dir/file" ++sleep 1 ++touch "$srcbase/dir/file" ++ ++(cd "$srcbase" && $RSYNC -KRlptzv --rsync-path="$RSYNC" dir/file localhost:) \ ++ || test_fail "test 2: compressed update through directory symlink failed" ++ ++diff "$srcbase/dir/file" "$HOME/real-dir/file" >/dev/null \ ++ || test_fail "test 2: compressed update content mismatch" ++ ++###################################################################### ++# Test 3: Nested directory symlinks (nested/sub/data.txt where ++# "nested" is a symlink to "nested_real") ++###################################################################### ++ ++mkdir -p "$HOME/nested_real/sub" ++ln -s nested_real "$HOME/nested" ++ ++mkdir -p "$srcbase/nested/sub" ++make_testfile "$srcbase/nested/sub/data.txt" ++ ++(cd "$srcbase" && $RSYNC -KRlptv --rsync-path="$RSYNC" nested/sub/data.txt localhost:) \ ++ || test_fail "test 3: initial nested transfer failed" ++ ++echo "appended nested" >> "$srcbase/nested/sub/data.txt" ++sleep 1 ++touch "$srcbase/nested/sub/data.txt" ++ ++(cd "$srcbase" && $RSYNC -KRlptv --rsync-path="$RSYNC" nested/sub/data.txt localhost:) \ ++ || test_fail "test 3: update through nested directory symlink failed" ++ ++diff "$srcbase/nested/sub/data.txt" "$HOME/nested_real/sub/data.txt" >/dev/null \ ++ || test_fail "test 3: nested update content mismatch" ++ ++###################################################################### ++# Test 4: --backup with directory symlinks ++# ++# Exercises the finish_transfer() "fnamecmp == fname" pointer ++# comparison that determines whether to update fnamecmp to the ++# backup name. If broken, --backup would reference a renamed file ++# for xattr handling. ++###################################################################### ++ ++# Reset destination ++rm -f "$HOME/real-dir/file" "$HOME/real-dir/file~" ++ ++make_testfile "$srcbase/dir/file" ++ ++(cd "$srcbase" && $RSYNC -KRlptv --rsync-path="$RSYNC" dir/file localhost:) \ ++ || test_fail "test 4: initial transfer for backup test failed" ++ ++echo "backup update" >> "$srcbase/dir/file" ++sleep 1 ++touch "$srcbase/dir/file" ++ ++(cd "$srcbase" && $RSYNC -KRlptv --backup --rsync-path="$RSYNC" dir/file localhost:) \ ++ || test_fail "test 4: update with --backup through directory symlink failed" ++ ++diff "$srcbase/dir/file" "$HOME/real-dir/file" >/dev/null \ ++ || test_fail "test 4: backup update content mismatch" ++ ++if [ ! -f "$HOME/real-dir/file~" ]; then ++ test_fail "test 4: backup file was not created" ++fi ++ ++###################################################################### ++# Test 5: --inplace with directory symlinks ++# ++# Exercises the updating_basis_or_equiv check which uses ++# "fnamecmp == fname". With --inplace, rsync writes directly to ++# the destination file instead of a temp file. ++###################################################################### ++ ++rm -f "$HOME/real-dir/file" "$HOME/real-dir/file~" ++ ++make_testfile "$srcbase/dir/file" ++ ++(cd "$srcbase" && $RSYNC -KRlptv --inplace --rsync-path="$RSYNC" dir/file localhost:) \ ++ || test_fail "test 5: initial inplace transfer failed" ++ ++echo "inplace update" >> "$srcbase/dir/file" ++sleep 1 ++touch "$srcbase/dir/file" ++ ++(cd "$srcbase" && $RSYNC -KRlptv --inplace --rsync-path="$RSYNC" dir/file localhost:) \ ++ || test_fail "test 5: inplace update through directory symlink failed" ++ ++diff "$srcbase/dir/file" "$HOME/real-dir/file" >/dev/null \ ++ || test_fail "test 5: inplace update content mismatch" ++ ++###################################################################### ++# Test 6: Top-level file (no dirname, no split needed) ++# ++# Ensures the dirname/basename split is not attempted for files ++# at the top level (file->dirname is NULL). ++###################################################################### ++ ++make_testfile "$srcbase/topfile" ++mkdir -p "$HOME" ++ ++(cd "$srcbase" && $RSYNC -Rlptv --rsync-path="$RSYNC" topfile localhost:) \ ++ || test_fail "test 6: initial top-level transfer failed" ++ ++echo "toplevel update" >> "$srcbase/topfile" ++sleep 1 ++touch "$srcbase/topfile" ++ ++(cd "$srcbase" && $RSYNC -Rlptv --rsync-path="$RSYNC" topfile localhost:) \ ++ || test_fail "test 6: top-level update failed" ++ ++diff "$srcbase/topfile" "$HOME/topfile" >/dev/null \ ++ || test_fail "test 6: top-level update content mismatch" ++ ++###################################################################### ++# Test 7: --partial-dir with protocol < 29 ++# ++# For protocol < 29, fnamecmp_type stays FNAMECMP_FNAME even when ++# fnamecmp is set to partialptr. The dirname/basename split must ++# NOT trigger in this case (guarded by "fnamecmp == fname"). ++###################################################################### ++ ++rm -f "$HOME/real-dir/file" ++make_testfile "$srcbase/dir/file" ++ ++(cd "$srcbase" && $RSYNC -KRlptv --protocol=28 --partial-dir=.rsync-partial \ ++ --rsync-path="$RSYNC" dir/file localhost:) \ ++ || test_fail "test 7: initial proto28 partial-dir transfer failed" ++ ++echo "partial-dir update" >> "$srcbase/dir/file" ++sleep 1 ++touch "$srcbase/dir/file" ++ ++(cd "$srcbase" && $RSYNC -KRlptv --protocol=28 --partial-dir=.rsync-partial \ ++ --rsync-path="$RSYNC" dir/file localhost:) \ ++ || test_fail "test 7: proto28 partial-dir update through dirlink failed" ++ ++diff "$srcbase/dir/file" "$HOME/real-dir/file" >/dev/null \ ++ || test_fail "test 7: proto28 partial-dir update content mismatch" ++ ++###################################################################### ++# Test 8: Protocol < 29 basic directory symlink update ++# ++# Exercises the protocol < 29 code path and its fallback logic ++# (clearing basedir on retry). ++###################################################################### ++ ++rm -f "$HOME/real-dir/file" ++make_testfile "$srcbase/dir/file" ++ ++(cd "$srcbase" && $RSYNC -KRlptv --protocol=28 \ ++ --rsync-path="$RSYNC" dir/file localhost:) \ ++ || test_fail "test 8: initial proto28 transfer failed" ++ ++echo "proto28 update" >> "$srcbase/dir/file" ++sleep 1 ++touch "$srcbase/dir/file" ++ ++(cd "$srcbase" && $RSYNC -KRlptv --protocol=28 \ ++ --rsync-path="$RSYNC" dir/file localhost:) \ ++ || test_fail "test 8: proto28 update through directory symlink failed" ++ ++diff "$srcbase/dir/file" "$HOME/real-dir/file" >/dev/null \ ++ || test_fail "test 8: proto28 update content mismatch" ++ ++# The script would have aborted on error, so getting here means we've won. ++exit 0
meta/recipes-devtools/rsync/files/CVE-2026-29518_p2.patch new file +98 -0
98 lines not on this page · raw message
meta/recipes-devtools/rsync/files/CVE-2026-29518_p3.patch new file +328 -0
328 lines not on this page · raw message
meta/recipes-devtools/rsync/files/CVE-2026-29518_p4.patch new file +71 -0
71 lines not on this page · raw message
meta/recipes-devtools/rsync/rsync_3.4.1.bb +4 -0
@@ -16,6 +16,10 @@ SRC_URI = "https://download.samba.org/pub/${BPN}/src/${BP}.tar.gz \ file://determism.patch \ file://0001-Add-missing-prototypes-to-function-declarations.patch \ file://CVE-2025-10158.patch \ + file://CVE-2026-29518_p1.patch \ + file://CVE-2026-29518_p2.patch \ + file://CVE-2026-29518_p3.patch \ + file://CVE-2026-29518_p4.patch \ " SRC_URI[sha256sum] = "2924bcb3a1ed8b551fc101f740b9f0fe0a202b115027647cf69850d65fd88c52"
Signature · 1 line
2.35.6
Re: [wrynose][PATCH v2 1/6] rsync: Fix CVE-2026-29518
Hi,
Please ignore this patch series. I will send corrected patches again.
Regards,
Ashish.
________________________________
From: Ashishkumar Parmar X (asparmar - E INFOCHIPS PRIVATE LIMITED at Cisco) <[email protected]>
Sent: 17 July 2026 10:50
To: [email protected] <[email protected]>
Cc: Ashishkumar Parmar -X (asparmar - E INFOCHIPS PRIVATE LIMITED at Cisco) <[email protected]>
Subject: [wrynose][PATCH v2 1/6] rsync: Fix CVE-2026-29518
From: Ashishkumar Parmar <[email protected]>
This patch backports the upstream rsync v3.4.3 security fix train for CVE-2026-29518. The direct CVE fixes are [3] and [4]:
- [3] enables secure_relative_open() for daemon modules running with "use chroot = no", closing the receiver-side basis-file TOCTOU.
- [4] routes the sender read path through secure_relative_open() from the trusted module root, closing the matching sender-side TOCTOU.
This patch also carries [1] and [2] from the same upstream v3.4.3 security/update train. These are supporting secure_relative_open() changes, not standalone direct CVE fixes: [1] uses Linux openat2(RESOLVE_BENEATH) to preserve the same confinement while allowing legitimate in-tree symlinks, fixing upstream issue #715; [2] adds the equivalent FreeBSD/macOS O_RESOLVE_BENEATH path. Debian's rsync 3.4.1+ds1-5+deb13u3 security update also carries these commits before the direct CVE-2026-29518 commits.
The upstream fixed release is referenced in [5], and the public CVE record is referenced in [6]. Individual backported commit links are also recorded in the embedded patch headers.
[1] https://github.com/RsyncProject/rsync/commit/4fa7156ccdb2ad34b034d18fe2fd6cd79adef8a1
[2] https://github.com/RsyncProject/rsync/commit/7f60ec001a0be63b770707ec8b829524c3809a43
[3] https://github.com/RsyncProject/rsync/commit/f1c24ab03bc85cb2638a569faa60216582fb6c5d
[4] https://github.com/RsyncProject/rsync/commit/859d44fa4f1420775e4ba050337ef32092f2894c
[5] https://github.com/RsyncProject/rsync/releases/tag/v3.4.3
[6] https://www.cve.org/CVERecord?id=CVE-2026-29518
--- .../rsync/files/CVE-2026-29518_p1.patch | 392 ++++++++++++++++++ .../rsync/files/CVE-2026-29518_p2.patch | 98 +++++ .../rsync/files/CVE-2026-29518_p3.patch | 328 +++++++++++++++ .../rsync/files/CVE-2026-29518_p4.patch | 71 ++++ meta/recipes-devtools/rsync/rsync_3.4.1.bb | 4 + 5 files changed, 893 insertions(+) create mode 100644 meta/recipes-devtools/rsync/files/CVE-2026-29518_p1.patch create mode 100644 meta/recipes-devtools/rsync/files/CVE-2026-29518_p2.patch create mode 100644 meta/recipes-devtools/rsync/files/CVE-2026-29518_p3.patch create mode 100644 meta/recipes-devtools/rsync/files/CVE-2026-29518_p4.patch
meta/recipes-devtools/rsync/files/CVE-2026-29518_p1.patch new file +392 -0
@@ -0,0 +1,392 @@ +From e7b575413b962ac042079a55530577b20c7eee44 Mon Sep 17 00:00:00 2001 +From: Andrew Tridgell <[email protected]> +Date: Thu, 30 Apr 2026 08:39:22 +1000 +Subject: [PATCH] syscall: use openat2(RESOLVE_BENEATH) on Linux for + secure_relative_open + +The CVE fix in commit c35e283 made secure_relative_open() walk every +component of relpath with O_NOFOLLOW. That blocks every symlink in the +path, which is stricter than the threat model required: legitimate +directory symlinks within the destination tree (e.g. when using -K / +--copy-dirlinks) are also rejected, breaking delta transfers with +"failed verification -- update discarded". See issue #715. + +On Linux 5.6+, openat2(RESOLVE_BENEATH | RESOLVE_NO_MAGICLINKS) gives +us exactly what we want: the kernel rejects any resolution that would +escape the starting directory (via "..", absolute paths, or symlinks +pointing outside dirfd) while still following symlinks that resolve +within it. /proc magic-links are blocked too. + +Use openat2 first; fall back to the existing per-component O_NOFOLLOW +walk on ENOSYS (kernel < 5.6). The lexical "../" checks at the head +of the function are kept as defense in depth. The Linux gate is +plain #ifdef __linux__: the runtime ENOSYS fallback covers the only +case that actually matters (header present + old kernel), and any +Linux build environment without linux/openat2.h will fail with a +clear "no such file" error rather than silently disabling the +protection. + +Verified manually that openat2(RESOLVE_BENEATH) blocks all four +escape patterns (absolute symlink, ../ symlink, lexical .., absolute +path) while allowing direct and within-tree symlinks. The new +testsuite/symlink-dirlink-basis.test (taken from PR #864 by Samuel +Henrique) exercises the issue #715 regression and passes; full +make check passes 47/47. + +Test: testsuite/symlink-dirlink-basis.test (8 scenarios) +Fixes: https://github.com/RsyncProject/rsync/issues/715 + +Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]> + +CVE: CVE-2026-29518 +Upstream-Status: Backport [https://github.com/RsyncProject/rsync/commit/4fa7156ccdb2ad34b034d18fe2fd6cd79adef8a1] + +(cherry picked from commit 4fa7156ccdb2ad34b034d18fe2fd6cd79adef8a1) +Signed-off-by: Ashishkumar Parmar <[email protected]> +--- + syscall.c | 62 ++++++- + testsuite/symlink-dirlink-basis.test | 247 +++++++++++++++++++++++++++ + 2 files changed, 304 insertions(+), 5 deletions(-) + create mode 100755 testsuite/symlink-dirlink-basis.test + +diff --git a/syscall.c b/syscall.c +index 34a9bba0..d66c9fdd 100644 +--- a/syscall.c ++++ b/syscall.c +@@ -33,6 +33,11 @@ + #include <sys/syscall.h> + #endif + ++#ifdef __linux__ ++#include <sys/syscall.h> ++#include <linux/openat2.h> ++#endif ++ + #include "ifuncs.h" + + extern int dry_run; +@@ -715,12 +720,49 @@ int do_open_nofollow(const char *pathname, int flags) + /* + open a file relative to a base directory. The basedir can be NULL, + in which case the current working directory is used. The relpath +- must be a relative path, and the relpath must not contain any +- elements in the path which follow symlinks (ie. like O_NOFOLLOW, but +- applies to all path components, not just the last component) +- +- The relpath must also not contain any ../ elements in the path ++ must be a relative path. The kernel must guarantee that resolution ++ cannot escape basedir (or the cwd, when basedir is NULL): no ".." ++ jumps above the start, no symlinks pointing outside, no absolute ++ paths, no /proc magic-link tricks. ++ ++ Symlinks *within* basedir are followed normally — earlier rsync ++ versions rejected every symlink with O_NOFOLLOW on each component, ++ which broke legitimate directory symlinks on the receiver side ++ (https://github.com/RsyncProject/rsync/issues/715). The escape ++ prevention is handled by the kernel via openat2(RESOLVE_BENEATH) ++ on Linux 5.6+; older systems fall back to the per-component ++ O_NOFOLLOW walk below. ++ ++ The relpath must also not contain any ../ elements in the path. + */ ++ ++#ifdef __linux__ ++static int secure_relative_open_linux(const char *basedir, const char *relpath, int flags, mode_t mode) ++{ ++ struct open_how how; ++ int dirfd, retfd; ++ ++ memset(&how, 0, sizeof how); ++ how.flags = flags; ++ how.mode = mode; ++ how.resolve = RESOLVE_BENEATH | RESOLVE_NO_MAGICLINKS; ++ ++ if (basedir == NULL) { ++ dirfd = AT_FDCWD; ++ } else { ++ dirfd = openat(AT_FDCWD, basedir, O_RDONLY | O_DIRECTORY); ++ if (dirfd == -1) ++ return -1; ++ } ++ ++ retfd = syscall(SYS_openat2, dirfd, relpath, &how, sizeof how); ++ ++ if (dirfd != AT_FDCWD) ++ close(dirfd); ++ return retfd; ++} ++#endif ++ + int secure_relative_open(const char *basedir, const char *relpath, int flags, mode_t mode) + { + if (!relpath || relpath[0] == '/') { +@@ -734,6 +776,16 @@ int secure_relative_open(const char *basedir, const char *relpath, int flags, mo + return -1; + } + ++#ifdef __linux__ ++ { ++ int fd = secure_relative_open_linux(basedir, relpath, flags, mode); ++ /* ENOSYS = kernel < 5.6 doesn't have the syscall even though ++ * glibc/kernel-headers do; fall through to the portable path. */ ++ if (fd != -1 || errno != ENOSYS) ++ return fd; ++ } ++#endif ++ + #if !defined(O_NOFOLLOW) || !defined(O_DIRECTORY) || !defined(AT_FDCWD) + // really old system, all we can do is live with the risks + if (!basedir) { +diff --git a/testsuite/symlink-dirlink-basis.test b/testsuite/symlink-dirlink-basis.test +new file mode 100755 +index 00000000..9065dd81 +--- /dev/null ++++ b/testsuite/symlink-dirlink-basis.test +@@ -0,0 +1,247 @@ ++#!/bin/sh ++ ++# Test that updating a file through a directory symlink works when using ++# -K (--copy-dirlinks). This is a regression test for: ++# https://github.com/RsyncProject/rsync/issues/715 ++# ++# The CVE fix in commit c35e283 introduced secure_relative_open() which ++# uses O_NOFOLLOW on all path components, breaking legitimate directory ++# symlinks on the receiver side. The fix splits the path into basedir ++# (dirname, symlinks followed) and basename (O_NOFOLLOW) so that ++# directory symlinks are traversed while the final file component is ++# still protected. ++# ++# The regression only manifests when delta matching is triggered (i.e., ++# the sender finds matching blocks in the old file). Small files with ++# completely different content are transferred in full and don't trigger ++# the bug. We use a large file with a small modification to ensure ++# delta transfer is used. ++# ++# In addition to the original regression, this test covers edge cases ++# in the fix itself: ++# - --backup with directory symlinks (finish_transfer pointer identity) ++# - --partial-dir with protocol < 29 (fnamecmp != partialptr guard) ++# - --inplace with directory symlinks (updating_basis_or_equiv check) ++# - Files without a dirname (top-level files, no split needed) ++ ++. "$suitedir/rsync.fns" ++ ++RSYNC_RSH="$scratchdir/src/support/lsh.sh" ++export RSYNC_RSH ++ ++# $HOME is set to $scratchdir by rsync.fns ++# localhost: destination will cd to $HOME (i.e., $scratchdir) ++ ++# Helper: create a large file suitable for delta transfers. ++# ~32KB is large enough for rsync's block matching to find matches. ++make_testfile() { ++ dd if=/dev/urandom of="$1" bs=1024 count=32 2>/dev/null \ ++ || test_fail "failed to create test file $1" ++} ++ ++# Set up source tree ++srcbase="$tmpdir/src" ++ ++###################################################################### ++# Test 1: Basic directory symlink update (the original issue #715) ++###################################################################### ++ ++mkdir -p "$HOME/real-dir" ++ln -s real-dir "$HOME/dir" ++ ++mkdir -p "$srcbase/dir" ++make_testfile "$srcbase/dir/file" ++ ++# First transfer (initial): should create the file through the symlink ++(cd "$srcbase" && $RSYNC -KRlptv --rsync-path="$RSYNC" dir/file localhost:) \ ++ || test_fail "test 1: initial transfer failed" ++ ++if [ ! -f "$HOME/real-dir/file" ]; then ++ test_fail "test 1: initial transfer did not create file through symlink" ++fi ++ ++diff "$srcbase/dir/file" "$HOME/real-dir/file" >/dev/null \ ++ || test_fail "test 1: initial transfer content mismatch" ++ ++# Small modification to trigger delta transfer ++echo "appended update" >> "$srcbase/dir/file" ++sleep 1 ++touch "$srcbase/dir/file" ++ ++# Second transfer (update): was failing with "failed verification" ++(cd "$srcbase" && $RSYNC -KRlptv --rsync-path="$RSYNC" dir/file localhost:) \ ++ || test_fail "test 1: update through directory symlink failed" ++ ++diff "$srcbase/dir/file" "$HOME/real-dir/file" >/dev/null \ ++ || test_fail "test 1: update transfer content mismatch" ++ ++###################################################################### ++# Test 2: Compression (-z) as in the original reproducer ++###################################################################### ++ ++echo "another line" >> "$srcbase/dir/file" ++sleep 1 ++touch "$srcbase/dir/file" ++ ++(cd "$srcbase" && $RSYNC -KRlptzv --rsync-path="$RSYNC" dir/file localhost:) \ ++ || test_fail "test 2: compressed update through directory symlink failed" ++ ++diff "$srcbase/dir/file" "$HOME/real-dir/file" >/dev/null \ ++ || test_fail "test 2: compressed update content mismatch" ++ ++###################################################################### ++# Test 3: Nested directory symlinks (nested/sub/data.txt where ++# "nested" is a symlink to "nested_real") ++###################################################################### ++ ++mkdir -p "$HOME/nested_real/sub" ++ln -s nested_real "$HOME/nested" ++ ++mkdir -p "$srcbase/nested/sub" ++make_testfile "$srcbase/nested/sub/data.txt" ++ ++(cd "$srcbase" && $RSYNC -KRlptv --rsync-path="$RSYNC" nested/sub/data.txt localhost:) \ ++ || test_fail "test 3: initial nested transfer failed" ++ ++echo "appended nested" >> "$srcbase/nested/sub/data.txt" ++sleep 1 ++touch "$srcbase/nested/sub/data.txt" ++ ++(cd "$srcbase" && $RSYNC -KRlptv --rsync-path="$RSYNC" nested/sub/data.txt localhost:) \ ++ || test_fail "test 3: update through nested directory symlink failed" ++ ++diff "$srcbase/nested/sub/data.txt" "$HOME/nested_real/sub/data.txt" >/dev/null \ ++ || test_fail "test 3: nested update content mismatch" ++ ++###################################################################### ++# Test 4: --backup with directory symlinks ++# ++# Exercises the finish_transfer() "fnamecmp == fname" pointer ++# comparison that determines whether to update fnamecmp to the ++# backup name. If broken, --backup would reference a renamed file ++# for xattr handling. ++###################################################################### ++ ++# Reset destination ++rm -f "$HOME/real-dir/file" "$HOME/real-dir/file~" ++ ++make_testfile "$srcbase/dir/file" ++ ++(cd "$srcbase" && $RSYNC -KRlptv --rsync-path="$RSYNC" dir/file localhost:) \ ++ || test_fail "test 4: initial transfer for backup test failed" ++ ++echo "backup update" >> "$srcbase/dir/file" ++sleep 1 ++touch "$srcbase/dir/file" ++ ++(cd "$srcbase" && $RSYNC -KRlptv --backup --rsync-path="$RSYNC" dir/file localhost:) \ ++ || test_fail "test 4: update with --backup through directory symlink failed" ++ ++diff "$srcbase/dir/file" "$HOME/real-dir/file" >/dev/null \ ++ || test_fail "test 4: backup update content mismatch" ++ ++if [ ! -f "$HOME/real-dir/file~" ]; then ++ test_fail "test 4: backup file was not created" ++fi ++ ++###################################################################### ++# Test 5: --inplace with directory symlinks ++# ++# Exercises the updating_basis_or_equiv check which uses ++# "fnamecmp == fname". With --inplace, rsync writes directly to ++# the destination file instead of a temp file. ++###################################################################### ++ ++rm -f "$HOME/real-dir/file" "$HOME/real-dir/file~" ++ ++make_testfile "$srcbase/dir/file" ++ ++(cd "$srcbase" && $RSYNC -KRlptv --inplace --rsync-path="$RSYNC" dir/file localhost:) \ ++ || test_fail "test 5: initial inplace transfer failed" ++ ++echo "inplace update" >> "$srcbase/dir/file" ++sleep 1 ++touch "$srcbase/dir/file" ++ ++(cd "$srcbase" && $RSYNC -KRlptv --inplace --rsync-path="$RSYNC" dir/file localhost:) \ ++ || test_fail "test 5: inplace update through directory symlink failed" ++ ++diff "$srcbase/dir/file" "$HOME/real-dir/file" >/dev/null \ ++ || test_fail "test 5: inplace update content mismatch" ++ ++###################################################################### ++# Test 6: Top-level file (no dirname, no split needed) ++# ++# Ensures the dirname/basename split is not attempted for files ++# at the top level (file->dirname is NULL). ++###################################################################### ++ ++make_testfile "$srcbase/topfile" ++mkdir -p "$HOME" ++ ++(cd "$srcbase" && $RSYNC -Rlptv --rsync-path="$RSYNC" topfile localhost:) \ ++ || test_fail "test 6: initial top-level transfer failed" ++ ++echo "toplevel update" >> "$srcbase/topfile" ++sleep 1 ++touch "$srcbase/topfile" ++ ++(cd "$srcbase" && $RSYNC -Rlptv --rsync-path="$RSYNC" topfile localhost:) \ ++ || test_fail "test 6: top-level update failed" ++ ++diff "$srcbase/topfile" "$HOME/topfile" >/dev/null \ ++ || test_fail "test 6: top-level update content mismatch" ++ ++###################################################################### ++# Test 7: --partial-dir with protocol < 29 ++# ++# For protocol < 29, fnamecmp_type stays FNAMECMP_FNAME even when ++# fnamecmp is set to partialptr. The dirname/basename split must ++# NOT trigger in this case (guarded by "fnamecmp == fname"). ++###################################################################### ++ ++rm -f "$HOME/real-dir/file" ++make_testfile "$srcbase/dir/file" ++ ++(cd "$srcbase" && $RSYNC -KRlptv --protocol=28 --partial-dir=.rsync-partial \ ++ --rsync-path="$RSYNC" dir/file localhost:) \ ++ || test_fail "test 7: initial proto28 partial-dir transfer failed" ++ ++echo "partial-dir update" >> "$srcbase/dir/file" ++sleep 1 ++touch "$srcbase/dir/file" ++ ++(cd "$srcbase" && $RSYNC -KRlptv --protocol=28 --partial-dir=.rsync-partial \ ++ --rsync-path="$RSYNC" dir/file localhost:) \ ++ || test_fail "test 7: proto28 partial-dir update through dirlink failed" ++ ++diff "$srcbase/dir/file" "$HOME/real-dir/file" >/dev/null \ ++ || test_fail "test 7: proto28 partial-dir update content mismatch" ++ ++###################################################################### ++# Test 8: Protocol < 29 basic directory symlink update ++# ++# Exercises the protocol < 29 code path and its fallback logic ++# (clearing basedir on retry). ++###################################################################### ++ ++rm -f "$HOME/real-dir/file" ++make_testfile "$srcbase/dir/file" ++ ++(cd "$srcbase" && $RSYNC -KRlptv --protocol=28 \ ++ --rsync-path="$RSYNC" dir/file localhost:) \ ++ || test_fail "test 8: initial proto28 transfer failed" ++ ++echo "proto28 update" >> "$srcbase/dir/file" ++sleep 1 ++touch "$srcbase/dir/file" ++ ++(cd "$srcbase" && $RSYNC -KRlptv --protocol=28 \ ++ --rsync-path="$RSYNC" dir/file localhost:) \ ++ || test_fail "test 8: proto28 update through directory symlink failed" ++ ++diff "$srcbase/dir/file" "$HOME/real-dir/file" >/dev/null \ ++ || test_fail "test 8: proto28 update content mismatch" ++ ++# The script would have aborted on error, so getting here means we've won. ++exit 0
meta/recipes-devtools/rsync/files/CVE-2026-29518_p2.patch new file +98 -0
98 lines not on this page · raw message
meta/recipes-devtools/rsync/files/CVE-2026-29518_p3.patch new file +328 -0
328 lines not on this page · raw message
meta/recipes-devtools/rsync/files/CVE-2026-29518_p4.patch new file +71 -0
71 lines not on this page · raw message
meta/recipes-devtools/rsync/rsync_3.4.1.bb +4 -0
@@ -16,6 +16,10 @@ SRC_URI = "https://download.samba.org/pub/${BPN}/src/${BP}.tar.gz \ file://determism.patch \ file://0001-Add-missing-prototypes-to-function-declarations.patch \ file://CVE-2025-10158.patch \ + file://CVE-2026-29518_p1.patch \ + file://CVE-2026-29518_p2.patch \ + file://CVE-2026-29518_p3.patch \ + file://CVE-2026-29518_p4.patch \ " SRC_URI[sha256sum] = "2924bcb3a1ed8b551fc101f740b9f0fe0a202b115027647cf69850d65fd88c52"
--
2.35.6
PATCH v2 2/6 rsync: Fix CVE-2026-43619
In reply to Ashishkumar Parmar X (asparmar - E INFOCHIPS PRIVATE LIMITED at Cisco)From: Ashishkumar Parmar <[email protected]>
This patch backports the upstream rsync v3.4.3 symlink-race fix chain for CVE-2026-43619. The upstream NEWS entry describes this CVE as symlink races on path-based system calls in "use chroot = no" daemon mode, generalizing CVE-2026-29518 beyond the receiver's open() path.
The backported commits form the following linear upstream chain:
- [1] closes the receiver-side do_chmod() path.
- [2] closes the daemon-no-chroot change_dir() escape.
- [3] adds the remaining symlink-race-safe do_*_at() wrappers and migrates the affected path-based syscall users.
- [4] closes adjacent bare-path copy/open surfaces. Upstream titles this commit "defence-in-depth", but it is in the same linear v3.4.3 symlink-race fix chain and Debian's rsync 3.4.1+ds1-5+deb13u3 security update carries it as part of the CVE-2026-43619 patch train.
The v3.4.3 security train also has portability and functional follow-ups that were released in v3.4.4, so this CVE-scoped backport includes them under CVE-2026-43619:
- [5] adds the mknodat()/mkfifoat() configure fallback used by the new do_mknod_at() path.
- [6] gates <linux/openat2.h> and openat2() use behind configure/runtime checks so systems without the header or usable syscall fall back safely.
The upstream fixed security release is referenced in [7], the v3.4.4 follow-up release is referenced in [8], and the public advisory is referenced in [9]. The Debian tracker entry is referenced in [10]. Individual backported commit links are also recorded in the embedded patch headers.
[1] https://github.com/RsyncProject/rsync/commit/862fe4eeaf82178324c66e504bf2c6c2e4038f99
[2] https://github.com/RsyncProject/rsync/commit/15d296425609a997f56eb8f9017ff73922f99b20
[3] https://github.com/RsyncProject/rsync/commit/30656c5e358b1c6033f4caf24f3e11b29d25c9f4
[4] https://github.com/RsyncProject/rsync/commit/3cc6a9e8cdcffb637ecdd41dae47833096dca317
[5] https://github.com/RsyncProject/rsync/commit/f3757a470ad3defdb2c4c6ad334e3839f3c7f1ae
[6] https://github.com/RsyncProject/rsync/commit/c14e2258b59a3f16a99c6e83a28660ba438a73e9
[7] https://github.com/RsyncProject/rsync/releases/tag/v3.4.3
[8] https://github.com/RsyncProject/rsync/releases/tag/v3.4.4
[9] https://github.com/RsyncProject/rsync/security/advisories/GHSA-4h9m-w5ff-j735
[10] https://security-tracker.debian.org/tracker/CVE-2026-43619
--- .../rsync/files/CVE-2026-43619_p1.patch | 471 +++++ .../rsync/files/CVE-2026-43619_p2.patch | 205 ++ .../rsync/files/CVE-2026-43619_p3.patch | 1783 +++++++++++++++++ .../rsync/files/CVE-2026-43619_p4.patch | 568 ++++++ .../rsync/files/CVE-2026-43619_p5.patch | 69 + .../rsync/files/CVE-2026-43619_p6.patch | 264 +++ meta/recipes-devtools/rsync/rsync_3.4.1.bb | 6 + 7 files changed, 3366 insertions(+) create mode 100644 meta/recipes-devtools/rsync/files/CVE-2026-43619_p1.patch create mode 100644 meta/recipes-devtools/rsync/files/CVE-2026-43619_p2.patch create mode 100644 meta/recipes-devtools/rsync/files/CVE-2026-43619_p3.patch create mode 100644 meta/recipes-devtools/rsync/files/CVE-2026-43619_p4.patch create mode 100644 meta/recipes-devtools/rsync/files/CVE-2026-43619_p5.patch create mode 100644 meta/recipes-devtools/rsync/files/CVE-2026-43619_p6.patch
meta/recipes-devtools/rsync/files/CVE-2026-43619_p1.patch new file +471 -0
@@ -0,0 +1,471 @@ +From e406eb555ea1fadb7c0dd8ddef3fde1b92771aa7 Mon Sep 17 00:00:00 2001 +From: Andrew Tridgell <[email protected]> +Date: Mon, 4 May 2026 21:53:14 +1000 +Subject: [PATCH] syscall+receiver: secure receiver-side do_chmod against + symlink-race TOCTOU + +CVE-2026-29518's fix routed the receiver's open() through +secure_relative_open(), but every other path-based syscall the +receiver runs on sender-controllable paths is vulnerable to the +same TOCTOU primitive. This commit closes the chmod variant. + +Add do_chmod_at() that opens the parent of fname under +secure_relative_open() and uses fchmodat() against the resulting +dirfd. Gate the secure path on am_daemon && !am_chrooted (the same +gate use_secure_symlinks already uses for the receiver basis-file +open), so non-daemon callers and chrooted daemons keep the original +do_chmod() fast path. + +Migrate the receiver-side do_chmod() call sites in delete.c, +generator.c, rsync.c, and xattrs.c. + +Adds testsuite/chmod-symlink-race.test (with t_chmod_secure helper) +as regression coverage. + +Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]> + +CVE: CVE-2026-43619 +Upstream-Status: Backport [https://github.com/RsyncProject/rsync/commit/862fe4eeaf82178324c66e504bf2c6c2e4038f99] + +Backport Changes: +- Upstream also updates runtests.py and a newer Makefile.in test + helper list that includes simdtest. Wrynose 3.4.1 does not carry + runtests.py and its CHECK_PROGS list has no simdtest entry, so + those test-runner hunks were not applicable. + +(cherry picked from commit 862fe4eeaf82178324c66e504bf2c6c2e4038f99) +Signed-off-by: Ashishkumar Parmar <[email protected]> +--- + Makefile.in | 9 ++- + delete.c | 4 +- + generator.c | 4 +- + rsync.c | 2 +- + syscall.c | 80 ++++++++++++++++++++ + t_chmod_secure.c | 117 ++++++++++++++++++++++++++++++ + t_stub.c | 2 + + testsuite/chmod-symlink-race.test | 68 +++++++++++++++++ + xattrs.c | 6 +- + 9 files changed, 282 insertions(+), 10 deletions(-) + create mode 100644 t_chmod_secure.c + create mode 100755 testsuite/chmod-symlink-race.test + +diff --git a/Makefile.in b/Makefile.in +index 2c5cf996..c30b8876 100644 +--- a/Makefile.in ++++ b/Makefile.in +@@ -62,12 +62,13 @@ TLS_OBJ = tls.o syscall.o util2.o t_stub.o lib/compat.o lib/snprintf.o lib/perms + + # Programs we must have to run the test cases + CHECK_PROGS = rsync$(EXEEXT) tls$(EXEEXT) getgroups$(EXEEXT) getfsdev$(EXEEXT) \ +- testrun$(EXEEXT) trimslash$(EXEEXT) t_unsafe$(EXEEXT) wildtest$(EXEEXT) ++ testrun$(EXEEXT) trimslash$(EXEEXT) t_unsafe$(EXEEXT) t_chmod_secure$(EXEEXT) \ ++ wildtest$(EXEEXT) + + CHECK_SYMLINKS = testsuite/chown-fake.test testsuite/devices-fake.test testsuite/xattrs-hlink.test + + # Objects for CHECK_PROGS to clean +-CHECK_OBJS=tls.o testrun.o getgroups.o getfsdev.o t_stub.o t_unsafe.o trimslash.o wildtest.o ++CHECK_OBJS=tls.o testrun.o getgroups.o getfsdev.o t_stub.o t_unsafe.o t_chmod_secure.o trimslash.o wildtest.o + + # note that the -I. is needed to handle config.h when using VPATH + .c.o: +@@ -183,6 +184,10 @@ T_UNSAFE_OBJ = t_unsafe.o syscall.o util1.o util2.o t_stub.o lib/compat.o lib/sn + t_unsafe$(EXEEXT): $(T_UNSAFE_OBJ) + $(CC) $(CFLAGS) $(LDFLAGS) -o $@ $(T_UNSAFE_OBJ) $(LIBS) + ++T_CHMOD_SECURE_OBJ = t_chmod_secure.o syscall.o util1.o util2.o t_stub.o lib/compat.o lib/snprintf.o lib/wildmatch.o lib/permstring.o ++t_chmod_secure$(EXEEXT): $(T_CHMOD_SECURE_OBJ) ++ $(CC) $(CFLAGS) $(LDFLAGS) -o $@ $(T_CHMOD_SECURE_OBJ) $(LIBS) ++ + .PHONY: conf + conf: configure.sh config.h.in + +diff --git a/delete.c b/delete.c +index 89c1f8d6..ded0ab2a 100644 +--- a/delete.c ++++ b/delete.c +@@ -98,7 +98,7 @@ static enum delret delete_dir_contents(char *fname, uint16 flags) + + strlcpy(p, fp->basename, remainder); + if (!(fp->mode & S_IWUSR) && !am_root && fp->flags & FLAG_OWNED_BY_US) +- do_chmod(fname, fp->mode | S_IWUSR); ++ do_chmod_at(fname, fp->mode | S_IWUSR); + /* Save stack by recursing to ourself directly. */ + if (S_ISDIR(fp->mode)) { + if (delete_dir_contents(fname, flags | DEL_RECURSE) != DR_SUCCESS) +@@ -139,7 +139,7 @@ enum delret delete_item(char *fbuf, uint16 mode, uint16 flags) + } + + if (flags & DEL_NO_UID_WRITE) +- do_chmod(fbuf, mode | S_IWUSR); ++ do_chmod_at(fbuf, mode | S_IWUSR); + + if (S_ISDIR(mode) && !(flags & DEL_DIR_IS_EMPTY)) { + /* This only happens on the first call to delete_item() since +diff --git a/generator.c b/generator.c +index b56fa569..e5aff654 100644 +--- a/generator.c ++++ b/generator.c +@@ -1499,7 +1499,7 @@ static void recv_generator(char *fname, struct file_struct *file, int ndx, + #ifdef HAVE_CHMOD + if (!am_root && (file->mode & S_IRWXU) != S_IRWXU && dir_tweaking) { + mode_t mode = file->mode | S_IRWXU; +- if (do_chmod(fname, mode) < 0) { ++ if (do_chmod_at(fname, mode) < 0) { + rsyserr(FERROR_XFER, errno, + "failed to modify permissions on %s", + full_fname(fname)); +@@ -2111,7 +2111,7 @@ static void touch_up_dirs(struct file_list *flist, int ndx) + continue; + fname = f_name(file, NULL); + if (fix_dir_perms) +- do_chmod(fname, file->mode); ++ do_chmod_at(fname, file->mode); + if (need_retouch_dir_times) { + STRUCT_STAT st; + if (link_stat(fname, &st, 0) == 0 && mtime_differs(&st, file)) { +diff --git a/rsync.c b/rsync.c +index b130aba5..cc46a2f9 100644 +--- a/rsync.c ++++ b/rsync.c +@@ -657,7 +657,7 @@ int set_file_attrs(const char *fname, struct file_struct *file, stat_x *sxp, + + #ifdef HAVE_CHMOD + if (!BITS_EQUAL(sxp->st.st_mode, new_mode, CHMOD_BITS)) { +- int ret = am_root < 0 ? 0 : do_chmod(fname, new_mode); ++ int ret = am_root < 0 ? 0 : do_chmod_at(fname, new_mode); + if (ret < 0) { + rsyserr(FERROR_XFER, errno, + "failed to set permissions on %s", +diff --git a/syscall.c b/syscall.c +index dcecda1a..06d5ed3f 100644 +--- a/syscall.c ++++ b/syscall.c +@@ -281,6 +281,86 @@ int do_chmod(const char *path, mode_t mode) + return code; + return 0; + } ++ ++/* ++ Symlink-race-safe variant of do_chmod() for receiver-side use. ++ ++ Threat model: on a daemon running with "use chroot = no" (the prerequisite ++ for CVE-2026-29518), a local attacker can race a symlink swap of one of ++ the parent directory components of a path the receiver is about to chmod. ++ Because chmod() resolves symlinks at every component, the swap redirects ++ the chmod outside the receiver's confinement. ++ ++ Defence: open the *parent* directory of fname under secure_relative_open() ++ (which uses openat2(RESOLVE_BENEATH) on Linux 5.6+, openat() with ++ O_RESOLVE_BENEATH on FreeBSD 13+ and macOS 15+ (Sequoia), or a per-component ++ O_NOFOLLOW walk elsewhere) and do fchmodat() against that dirfd. A symlink ++ substituted into one of the parent components is then either followed ++ within the tree (legitimate dir-symlinks still work) or rejected by the ++ kernel (escape attempts fail). ++ ++ Final-component handling matches do_chmod(): fchmodat() with flag 0 ++ follows a symlink at the final component, which is the same behaviour as ++ chmod() and matches every current call site (the file being chmod'd is ++ one the receiver itself just created or transferred). For the rare case ++ where the caller wants to chmod a symlink-as-an-object (S_ISLNK in the ++ mode bits), we fall through to do_chmod() which has portability code for ++ that case. ++ ++ Falls back to do_chmod() for absolute paths and for paths with no parent ++ component, where there is nothing to protect against. ++*/ ++int do_chmod_at(const char *fname, mode_t mode) ++{ ++#ifdef AT_FDCWD ++ extern int am_daemon, am_chrooted; ++ char dirpath[MAXPATHLEN]; ++ const char *bname; ++ const char *slash; ++ int dfd, ret, e; ++ size_t dlen; ++ ++ if (dry_run) return 0; ++ RETURN_ERROR_IF_RO_OR_LO; ++ ++ /* Only the daemon-without-chroot case is exposed to the symlink- ++ * race attack: a chroot already confines the receiver, and a ++ * non-daemon rsync runs with the user's own authority so a ++ * symlink they planted can only redirect to files they could ++ * already access. Everywhere else, fall through to plain ++ * do_chmod() to avoid the dirfd-open overhead on every call. */ ++ if (!am_daemon || am_chrooted) ++ return do_chmod(fname, mode); ++ ++ if (!fname || !*fname || *fname == '/' || S_ISLNK(mode)) ++ return do_chmod(fname, mode); ++ ++ slash = strrchr(fname, '/'); ++ if (!slash) ++ return do_chmod(fname, mode); ++ ++ dlen = slash - fname; ++ if (dlen >= sizeof dirpath) { ++ errno = ENAMETOOLONG; ++ return -1; ++ } ++ memcpy(dirpath, fname, dlen); ++ dirpath[dlen] = '\0'; ++ bname = slash + 1; ++ ++ dfd = secure_relative_open(NULL, dirpath, O_RDONLY | O_DIRECTORY, 0); ++ if (dfd < 0) ++ return -1; ++ ++ ret = fchmodat(dfd, bname, mode, 0); ++ e = errno; ++ close(dfd); ++ errno = e; ++ return ret; ++#else ++ return do_chmod(fname, mode); ++#endif ++} + #endif + + int do_rename(const char *old_path, const char *new_path) +diff --git a/t_chmod_secure.c b/t_chmod_secure.c +new file mode 100644 +index 00000000..114dfb2d +--- /dev/null ++++ b/t_chmod_secure.c +@@ -0,0 +1,117 @@ ++/* ++ * Test harness for do_chmod_at(). Confirms the symlink-TOCTOU ++ * primitive used by CVE-2026-29518 (and its incomplete-fix follow-up ++ * for chmod) is closed by do_chmod_at(): a parent directory component ++ * being a symlink that escapes the receiver's confinement must be ++ * rejected, while a parent symlink that resolves *within* the tree ++ * must still work (so legitimate dir-symlinks are not regressed). ++ * ++ * Not linked into rsync itself. ++ * ++ * This program is free software; you can redistribute it and/or modify ++ * it under the terms of the GNU General Public License version 2 as ++ * published by the Free Software Foundation. ++ */ ++ ++#include "rsync.h" ++ ++#include <sys/stat.h> ++ ++int dry_run = 0; ++int am_root = 0; ++int am_sender = 0; ++int read_only = 0; ++int list_only = 0; ++int copy_links = 0; ++int copy_unsafe_links = 0; ++extern int am_daemon, am_chrooted; ++ ++short info_levels[COUNT_INFO], debug_levels[COUNT_DEBUG]; ++ ++static int errs = 0; ++ ++static void check(const char *label, int actual_rc, int expect_ok, ++ const char *path, mode_t expected_mode) ++{ ++ struct stat st; ++ int got_ok = (actual_rc == 0); ++ if (got_ok != expect_ok) { ++ fprintf(stderr, "FAIL [%s]: rc=%d errno=%d (%s), expected %s\n", ++ label, actual_rc, errno, strerror(errno), ++ expect_ok ? "success" : "rejection"); ++ errs++; ++ return; ++ } ++ if (path && stat(path, &st) < 0) { ++ fprintf(stderr, "FAIL [%s]: stat(%s) failed: %s\n", ++ label, path, strerror(errno)); ++ errs++; ++ return; ++ } ++ if (path && (st.st_mode & 07777) != expected_mode) { ++ fprintf(stderr, ++ "FAIL [%s]: %s mode is 0%o, expected 0%o\n", ++ label, path, st.st_mode & 07777, expected_mode); ++ errs++; ++ return; ++ } ++ fprintf(stderr, "OK [%s]\n", label); ++} ++ ++int main(int argc, char **argv) ++{ ++ if (argc != 2) { ++ fprintf(stderr, "usage: %s <module-dir>\n", argv[0]); ++ return 2; ++ } ++ if (chdir(argv[1]) < 0) { ++ perror("chdir"); ++ return 2; ++ } ++ ++ /* Simulate the daemon-without-chroot deployment that do_chmod_at() ++ * defends. With am_daemon=0 or am_chrooted=1 the wrapper falls ++ * through to plain do_chmod() and the symlink-race test would be ++ * meaningless. */ ++ am_daemon = 1; ++ am_chrooted = 0; ++ ++ /* Test layout (all inside the directory we just chdir'd to): ++ * ++ * ./realdir/sentinel -- regular target file ++ * ./inside_link -> realdir -- legitimate dir-symlink within the tree ++ * ./escape_link -> ../trap -- attacker swap, target outside tree ++ * ../trap/sentinel -- the file the attacker wants to alter ++ * ++ * The shell wrapper that calls this helper has set both sentinel ++ * files to mode 0600 so we have a clean baseline to compare. ++ */ ++ ++ /* Scenario A: legitimate parent dir-symlink, chmod must succeed. */ ++ int rc = do_chmod_at("inside_link/sentinel", 0640); ++ check("A: legit dir-symlink within tree", ++ rc, 1, "realdir/sentinel", 0640); ++ ++ /* Scenario B: parent symlink escapes the tree -- chmod must be ++ * rejected and the outside file's mode must be unchanged. */ ++ rc = do_chmod_at("escape_link/sentinel", 0666); ++ check("B: parent symlink escapes tree (the attack)", ++ rc, 0, "../trap/sentinel", 0600); ++ ++ /* Scenario C: plain relative path with no symlink components, ++ * regression check that the safe wrapper doesn't break the ++ * normal case. */ ++ rc = do_chmod_at("realdir/sentinel", 0644); ++ check("C: plain relative path (regression check)", ++ rc, 1, "realdir/sentinel", 0644); ++ ++ /* Scenario D: top-level file, no parent directory component. ++ * Falls back to do_chmod(); should succeed. */ ++ rc = do_chmod_at("topfile", 0640); ++ check("D: top-level file, no parent component", ++ rc, 1, "topfile", 0640); ++ ++ if (errs) ++ fprintf(stderr, "%d failure(s)\n", errs); ++ return errs ? 1 : 0; ++} +diff --git a/t_stub.c b/t_stub.c +index eee92729..63bc144c 100644 +--- a/t_stub.c ++++ b/t_stub.c +@@ -23,6 +23,8 @@ + + int do_fsync = 0; + int inplace = 0; ++int am_daemon = 0; ++int am_chrooted = 0; + int modify_window = 0; + int preallocate_files = 0; + int protect_args = 0; +diff --git a/testsuite/chmod-symlink-race.test b/testsuite/chmod-symlink-race.test +new file mode 100755 +index 00000000..48bbfbb4 +--- /dev/null ++++ b/testsuite/chmod-symlink-race.test +@@ -0,0 +1,68 @@ ++#!/bin/sh ++ ++# Copyright (C) 2026 by Andrew Tridgell ++ ++# This program is distributable under the terms of the GNU GPL (see ++# COPYING). ++ ++# Regression test for the symlink-TOCTOU class of bug applied to ++# chmod() on the receiver side. The CVE-2026-29518 fix used ++# secure_relative_open() for the basis-file open, but every other ++# path-based syscall the receiver runs on sender-controllable paths ++# is vulnerable to the same primitive: a local attacker swaps a ++# symlink into one of the parent directory components between the ++# receiver's check and its act, and the syscall escapes the module. ++# ++# This test exercises the new do_chmod_at() wrapper via the ++# t_chmod_secure helper. The helper sets up two scenarios: ++# - a parent dir-symlink that resolves WITHIN the module tree ++# (legitimate -K-style use, must continue to work) ++# - a parent dir-symlink that escapes the module tree (the ++# attack, must be rejected) ++# plus two regression scenarios (plain relative path, top-level ++# file) that just confirm the safe wrapper doesn't break the ++# normal case. ++# ++# The kernel-enforced "stay below dirfd" path resolution is ++# only available on Linux 5.6+, FreeBSD 13+, and macOS 15+. ++# Skip on platforms that fall back to per-component O_NOFOLLOW ++# (Solaris, OpenBSD, NetBSD, Cygwin); the per-component fallback ++# would also reject the attack but the legitimate dir-symlink ++# scenario would fail there. ++ ++. "$suitedir/rsync.fns" ++ ++case "$(uname -s)" in ++ SunOS|OpenBSD|NetBSD|CYGWIN*) ++ test_skipped "do_chmod_at relies on RESOLVE_BENEATH-equivalent kernel support not available on $(uname -s)" ++ ;; ++esac ++ ++mod="$scratchdir/module" ++trap_outside="$scratchdir/trap" ++rm -rf "$mod" "$trap_outside" ++mkdir -p "$mod/realdir" "$trap_outside" ++ ++# Set up the four file-system objects the helper expects: ++echo bystander > "$mod/realdir/sentinel" ++chmod 0600 "$mod/realdir/sentinel" ++echo target > "$trap_outside/sentinel" ++chmod 0600 "$trap_outside/sentinel" ++ln -s realdir "$mod/inside_link" ++ln -s ../trap "$mod/escape_link" ++echo top > "$mod/topfile" ++chmod 0600 "$mod/topfile" ++ ++"$TOOLDIR/t_chmod_secure" "$mod" || \ ++ test_fail "t_chmod_secure reported failures (see stderr above)" ++ ++# Sanity-check from the shell side too: the outside file's mode must ++# still be 0600 -- the helper checked this, but a second look from ++# the shell guards against a helper-internal stat() bug. ++mode=$(stat -c '%a' "$trap_outside/sentinel" 2>/dev/null \ ++ || stat -f '%Lp' "$trap_outside/sentinel" 2>/dev/null) ++if [ "$mode" != "600" ]; then ++ test_fail "outside sentinel mode changed from 600 to $mode -- chmod escaped the module" ++fi ++ ++exit 0 +diff --git a/xattrs.c b/xattrs.c +index 26e50a6f..4462d63b 100644 +--- a/xattrs.c ++++ b/xattrs.c +@@ -1086,7 +1086,7 @@ int set_xattr(const char *fname, const struct file_struct *file, const char *fna + && !S_ISLNK(sxp->st.st_mode) + #endif + && access(fname, W_OK) < 0 +- && do_chmod(fname, (sxp->st.st_mode & CHMOD_BITS) | S_IWUSR) == 0) ++ && do_chmod_at(fname, (sxp->st.st_mode & CHMOD_BITS) | S_IWUSR) == 0) + added_write_perm = 1; + + ndx = F_XATTR(file); +@@ -1094,7 +1094,7 @@ int set_xattr(const char *fname, const struct file_struct *file, const char *fna + lst = &glst->xa_items; + int return_value = rsync_xal_set(fname, lst, fnamecmp, sxp); + if (added_write_perm) /* remove the temporary write permission */ +- do_chmod(fname, sxp->st.st_mode); ++ do_chmod_at(fname, sxp->st.st_mode); + return return_value; + } + +@@ -1211,7 +1211,7 @@ int set_stat_xattr(const char *fname, struct file_struct *file, mode_t new_mode) + mode = (fst.st_mode & _S_IFMT) | (fmode & ACCESSPERMS) + | (S_ISDIR(fst.st_mode) ? 0700 : 0600); + if (fst.st_mode != mode) +- do_chmod(fname, mode); ++ do_chmod_at(fname, mode); + if (!IS_DEVICE(fst.st_mode)) + fst.st_rdev = 0; /* just in case */ +
meta/recipes-devtools/rsync/files/CVE-2026-43619_p2.patch new file +205 -0
205 lines not on this page · raw message
meta/recipes-devtools/rsync/files/CVE-2026-43619_p3.patch new file +1783 -0
1,783 lines not on this page · raw message
meta/recipes-devtools/rsync/files/CVE-2026-43619_p4.patch new file +568 -0
541 lines not on this page · raw message
meta/recipes-devtools/rsync/files/CVE-2026-43619_p5.patch new file +69 -0
69 lines not on this page · raw message
meta/recipes-devtools/rsync/files/CVE-2026-43619_p6.patch new file +264 -0
264 lines not on this page · raw message
meta/recipes-devtools/rsync/rsync_3.4.1.bb +6 -0
Lines of this file are not shown. View the raw message.
The rest of this diff is not shown. View the raw message.
Signature · 1 line
2.35.6
Part of the diff is not shown here. View the raw message.
Re: [wrynose][PATCH v2 2/6] rsync: Fix CVE-2026-43619
Hi,
Please ignore this patch series. I will send corrected patches again.
Regards,
Ashish.
________________________________
From: Ashishkumar Parmar X (asparmar - E INFOCHIPS PRIVATE LIMITED at Cisco) <[email protected]>
Sent: 17 July 2026 10:50
To: [email protected] <[email protected]>
Cc: Ashishkumar Parmar -X (asparmar - E INFOCHIPS PRIVATE LIMITED at Cisco) <[email protected]>
Subject: [wrynose][PATCH v2 2/6] rsync: Fix CVE-2026-43619
From: Ashishkumar Parmar <[email protected]>
This patch backports the upstream rsync v3.4.3 symlink-race fix chain for CVE-2026-43619. The upstream NEWS entry describes this CVE as symlink races on path-based system calls in "use chroot = no" daemon mode, generalizing CVE-2026-29518 beyond the receiver's open() path.
The backported commits form the following linear upstream chain:
- [1] closes the receiver-side do_chmod() path.
- [2] closes the daemon-no-chroot change_dir() escape.
- [3] adds the remaining symlink-race-safe do_*_at() wrappers and migrates the affected path-based syscall users.
- [4] closes adjacent bare-path copy/open surfaces. Upstream titles this commit "defence-in-depth", but it is in the same linear v3.4.3 symlink-race fix chain and Debian's rsync 3.4.1+ds1-5+deb13u3 security update carries it as part of the CVE-2026-43619 patch train.
The v3.4.3 security train also has portability and functional follow-ups that were released in v3.4.4, so this CVE-scoped backport includes them under CVE-2026-43619:
- [5] adds the mknodat()/mkfifoat() configure fallback used by the new do_mknod_at() path.
- [6] gates <linux/openat2.h> and openat2() use behind configure/runtime checks so systems without the header or usable syscall fall back safely.
The upstream fixed security release is referenced in [7], the v3.4.4 follow-up release is referenced in [8], and the public advisory is referenced in [9]. The Debian tracker entry is referenced in [10]. Individual backported commit links are also recorded in the embedded patch headers.
[1] https://github.com/RsyncProject/rsync/commit/862fe4eeaf82178324c66e504bf2c6c2e4038f99
[2] https://github.com/RsyncProject/rsync/commit/15d296425609a997f56eb8f9017ff73922f99b20
[3] https://github.com/RsyncProject/rsync/commit/30656c5e358b1c6033f4caf24f3e11b29d25c9f4
[4] https://github.com/RsyncProject/rsync/commit/3cc6a9e8cdcffb637ecdd41dae47833096dca317
[5] https://github.com/RsyncProject/rsync/commit/f3757a470ad3defdb2c4c6ad334e3839f3c7f1ae
[6] https://github.com/RsyncProject/rsync/commit/c14e2258b59a3f16a99c6e83a28660ba438a73e9
[7] https://github.com/RsyncProject/rsync/releases/tag/v3.4.3
[8] https://github.com/RsyncProject/rsync/releases/tag/v3.4.4
[9] https://github.com/RsyncProject/rsync/security/advisories/GHSA-4h9m-w5ff-j735
[10] https://security-tracker.debian.org/tracker/CVE-2026-43619
--- .../rsync/files/CVE-2026-43619_p1.patch | 471 +++++ .../rsync/files/CVE-2026-43619_p2.patch | 205 ++ .../rsync/files/CVE-2026-43619_p3.patch | 1783 +++++++++++++++++ .../rsync/files/CVE-2026-43619_p4.patch | 568 ++++++ .../rsync/files/CVE-2026-43619_p5.patch | 69 + .../rsync/files/CVE-2026-43619_p6.patch | 264 +++ meta/recipes-devtools/rsync/rsync_3.4.1.bb | 6 + 7 files changed, 3366 insertions(+) create mode 100644 meta/recipes-devtools/rsync/files/CVE-2026-43619_p1.patch create mode 100644 meta/recipes-devtools/rsync/files/CVE-2026-43619_p2.patch create mode 100644 meta/recipes-devtools/rsync/files/CVE-2026-43619_p3.patch create mode 100644 meta/recipes-devtools/rsync/files/CVE-2026-43619_p4.patch create mode 100644 meta/recipes-devtools/rsync/files/CVE-2026-43619_p5.patch create mode 100644 meta/recipes-devtools/rsync/files/CVE-2026-43619_p6.patch
meta/recipes-devtools/rsync/files/CVE-2026-43619_p1.patch new file +471 -0
@@ -0,0 +1,471 @@ +From e406eb555ea1fadb7c0dd8ddef3fde1b92771aa7 Mon Sep 17 00:00:00 2001 +From: Andrew Tridgell <[email protected]> +Date: Mon, 4 May 2026 21:53:14 +1000 +Subject: [PATCH] syscall+receiver: secure receiver-side do_chmod against + symlink-race TOCTOU + +CVE-2026-29518's fix routed the receiver's open() through +secure_relative_open(), but every other path-based syscall the +receiver runs on sender-controllable paths is vulnerable to the +same TOCTOU primitive. This commit closes the chmod variant. + +Add do_chmod_at() that opens the parent of fname under +secure_relative_open() and uses fchmodat() against the resulting +dirfd. Gate the secure path on am_daemon && !am_chrooted (the same +gate use_secure_symlinks already uses for the receiver basis-file +open), so non-daemon callers and chrooted daemons keep the original +do_chmod() fast path. + +Migrate the receiver-side do_chmod() call sites in delete.c, +generator.c, rsync.c, and xattrs.c. + +Adds testsuite/chmod-symlink-race.test (with t_chmod_secure helper) +as regression coverage. + +Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]> + +CVE: CVE-2026-43619 +Upstream-Status: Backport [https://github.com/RsyncProject/rsync/commit/862fe4eeaf82178324c66e504bf2c6c2e4038f99] + +Backport Changes: +- Upstream also updates runtests.py and a newer Makefile.in test + helper list that includes simdtest. Wrynose 3.4.1 does not carry + runtests.py and its CHECK_PROGS list has no simdtest entry, so + those test-runner hunks were not applicable. + +(cherry picked from commit 862fe4eeaf82178324c66e504bf2c6c2e4038f99) +Signed-off-by: Ashishkumar Parmar <[email protected]> +--- + Makefile.in | 9 ++- + delete.c | 4 +- + generator.c | 4 +- + rsync.c | 2 +- + syscall.c | 80 ++++++++++++++++++++ + t_chmod_secure.c | 117 ++++++++++++++++++++++++++++++ + t_stub.c | 2 + + testsuite/chmod-symlink-race.test | 68 +++++++++++++++++ + xattrs.c | 6 +- + 9 files changed, 282 insertions(+), 10 deletions(-) + create mode 100644 t_chmod_secure.c + create mode 100755 testsuite/chmod-symlink-race.test + +diff --git a/Makefile.in b/Makefile.in +index 2c5cf996..c30b8876 100644 +--- a/Makefile.in ++++ b/Makefile.in +@@ -62,12 +62,13 @@ TLS_OBJ = tls.o syscall.o util2.o t_stub.o lib/compat.o lib/snprintf.o lib/perms + + # Programs we must have to run the test cases + CHECK_PROGS = rsync$(EXEEXT) tls$(EXEEXT) getgroups$(EXEEXT) getfsdev$(EXEEXT) \ +- testrun$(EXEEXT) trimslash$(EXEEXT) t_unsafe$(EXEEXT) wildtest$(EXEEXT) ++ testrun$(EXEEXT) trimslash$(EXEEXT) t_unsafe$(EXEEXT) t_chmod_secure$(EXEEXT) \ ++ wildtest$(EXEEXT) + + CHECK_SYMLINKS = testsuite/chown-fake.test testsuite/devices-fake.test testsuite/xattrs-hlink.test + + # Objects for CHECK_PROGS to clean +-CHECK_OBJS=tls.o testrun.o getgroups.o getfsdev.o t_stub.o t_unsafe.o trimslash.o wildtest.o ++CHECK_OBJS=tls.o testrun.o getgroups.o getfsdev.o t_stub.o t_unsafe.o t_chmod_secure.o trimslash.o wildtest.o + + # note that the -I. is needed to handle config.h when using VPATH + .c.o: +@@ -183,6 +184,10 @@ T_UNSAFE_OBJ = t_unsafe.o syscall.o util1.o util2.o t_stub.o lib/compat.o lib/sn + t_unsafe$(EXEEXT): $(T_UNSAFE_OBJ) + $(CC) $(CFLAGS) $(LDFLAGS) -o $@ $(T_UNSAFE_OBJ) $(LIBS) + ++T_CHMOD_SECURE_OBJ = t_chmod_secure.o syscall.o util1.o util2.o t_stub.o lib/compat.o lib/snprintf.o lib/wildmatch.o lib/permstring.o ++t_chmod_secure$(EXEEXT): $(T_CHMOD_SECURE_OBJ) ++ $(CC) $(CFLAGS) $(LDFLAGS) -o $@ $(T_CHMOD_SECURE_OBJ) $(LIBS) ++ + .PHONY: conf + conf: configure.sh config.h.in + +diff --git a/delete.c b/delete.c +index 89c1f8d6..ded0ab2a 100644 +--- a/delete.c ++++ b/delete.c +@@ -98,7 +98,7 @@ static enum delret delete_dir_contents(char *fname, uint16 flags) + + strlcpy(p, fp->basename, remainder); + if (!(fp->mode & S_IWUSR) && !am_root && fp->flags & FLAG_OWNED_BY_US) +- do_chmod(fname, fp->mode | S_IWUSR); ++ do_chmod_at(fname, fp->mode | S_IWUSR); + /* Save stack by recursing to ourself directly. */ + if (S_ISDIR(fp->mode)) { + if (delete_dir_contents(fname, flags | DEL_RECURSE) != DR_SUCCESS) +@@ -139,7 +139,7 @@ enum delret delete_item(char *fbuf, uint16 mode, uint16 flags) + } + + if (flags & DEL_NO_UID_WRITE) +- do_chmod(fbuf, mode | S_IWUSR); ++ do_chmod_at(fbuf, mode | S_IWUSR); + + if (S_ISDIR(mode) && !(flags & DEL_DIR_IS_EMPTY)) { + /* This only happens on the first call to delete_item() since +diff --git a/generator.c b/generator.c +index b56fa569..e5aff654 100644 +--- a/generator.c ++++ b/generator.c +@@ -1499,7 +1499,7 @@ static void recv_generator(char *fname, struct file_struct *file, int ndx, + #ifdef HAVE_CHMOD + if (!am_root && (file->mode & S_IRWXU) != S_IRWXU && dir_tweaking) { + mode_t mode = file->mode | S_IRWXU; +- if (do_chmod(fname, mode) < 0) { ++ if (do_chmod_at(fname, mode) < 0) { + rsyserr(FERROR_XFER, errno, + "failed to modify permissions on %s", + full_fname(fname)); +@@ -2111,7 +2111,7 @@ static void touch_up_dirs(struct file_list *flist, int ndx) + continue; + fname = f_name(file, NULL); + if (fix_dir_perms) +- do_chmod(fname, file->mode); ++ do_chmod_at(fname, file->mode); + if (need_retouch_dir_times) { + STRUCT_STAT st; + if (link_stat(fname, &st, 0) == 0 && mtime_differs(&st, file)) { +diff --git a/rsync.c b/rsync.c +index b130aba5..cc46a2f9 100644 +--- a/rsync.c ++++ b/rsync.c +@@ -657,7 +657,7 @@ int set_file_attrs(const char *fname, struct file_struct *file, stat_x *sxp, + + #ifdef HAVE_CHMOD + if (!BITS_EQUAL(sxp->st.st_mode, new_mode, CHMOD_BITS)) { +- int ret = am_root < 0 ? 0 : do_chmod(fname, new_mode); ++ int ret = am_root < 0 ? 0 : do_chmod_at(fname, new_mode); + if (ret < 0) { + rsyserr(FERROR_XFER, errno, + "failed to set permissions on %s", +diff --git a/syscall.c b/syscall.c +index dcecda1a..06d5ed3f 100644 +--- a/syscall.c ++++ b/syscall.c +@@ -281,6 +281,86 @@ int do_chmod(const char *path, mode_t mode) + return code; + return 0; + } ++ ++/* ++ Symlink-race-safe variant of do_chmod() for receiver-side use. ++ ++ Threat model: on a daemon running with "use chroot = no" (the prerequisite ++ for CVE-2026-29518), a local attacker can race a symlink swap of one of ++ the parent directory components of a path the receiver is about to chmod. ++ Because chmod() resolves symlinks at every component, the swap redirects ++ the chmod outside the receiver's confinement. ++ ++ Defence: open the *parent* directory of fname under secure_relative_open() ++ (which uses openat2(RESOLVE_BENEATH) on Linux 5.6+, openat() with ++ O_RESOLVE_BENEATH on FreeBSD 13+ and macOS 15+ (Sequoia), or a per-component ++ O_NOFOLLOW walk elsewhere) and do fchmodat() against that dirfd. A symlink ++ substituted into one of the parent components is then either followed ++ within the tree (legitimate dir-symlinks still work) or rejected by the ++ kernel (escape attempts fail). ++ ++ Final-component handling matches do_chmod(): fchmodat() with flag 0 ++ follows a symlink at the final component, which is the same behaviour as ++ chmod() and matches every current call site (the file being chmod'd is ++ one the receiver itself just created or transferred). For the rare case ++ where the caller wants to chmod a symlink-as-an-object (S_ISLNK in the ++ mode bits), we fall through to do_chmod() which has portability code for ++ that case. ++ ++ Falls back to do_chmod() for absolute paths and for paths with no parent ++ component, where there is nothing to protect against. ++*/ ++int do_chmod_at(const char *fname, mode_t mode) ++{ ++#ifdef AT_FDCWD ++ extern int am_daemon, am_chrooted; ++ char dirpath[MAXPATHLEN]; ++ const char *bname; ++ const char *slash; ++ int dfd, ret, e; ++ size_t dlen; ++ ++ if (dry_run) return 0; ++ RETURN_ERROR_IF_RO_OR_LO; ++ ++ /* Only the daemon-without-chroot case is exposed to the symlink- ++ * race attack: a chroot already confines the receiver, and a ++ * non-daemon rsync runs with the user's own authority so a ++ * symlink they planted can only redirect to files they could ++ * already access. Everywhere else, fall through to plain ++ * do_chmod() to avoid the dirfd-open overhead on every call. */ ++ if (!am_daemon || am_chrooted) ++ return do_chmod(fname, mode); ++ ++ if (!fname || !*fname || *fname == '/' || S_ISLNK(mode)) ++ return do_chmod(fname, mode); ++ ++ slash = strrchr(fname, '/'); ++ if (!slash) ++ return do_chmod(fname, mode); ++ ++ dlen = slash - fname; ++ if (dlen >= sizeof dirpath) { ++ errno = ENAMETOOLONG; ++ return -1; ++ } ++ memcpy(dirpath, fname, dlen); ++ dirpath[dlen] = '\0'; ++ bname = slash + 1; ++ ++ dfd = secure_relative_open(NULL, dirpath, O_RDONLY | O_DIRECTORY, 0); ++ if (dfd < 0) ++ return -1; ++ ++ ret = fchmodat(dfd, bname, mode, 0); ++ e = errno; ++ close(dfd); ++ errno = e; ++ return ret; ++#else ++ return do_chmod(fname, mode); ++#endif ++} + #endif + + int do_rename(const char *old_path, const char *new_path) +diff --git a/t_chmod_secure.c b/t_chmod_secure.c +new file mode 100644 +index 00000000..114dfb2d +--- /dev/null ++++ b/t_chmod_secure.c +@@ -0,0 +1,117 @@ ++/* ++ * Test harness for do_chmod_at(). Confirms the symlink-TOCTOU ++ * primitive used by CVE-2026-29518 (and its incomplete-fix follow-up ++ * for chmod) is closed by do_chmod_at(): a parent directory component ++ * being a symlink that escapes the receiver's confinement must be ++ * rejected, while a parent symlink that resolves *within* the tree ++ * must still work (so legitimate dir-symlinks are not regressed). ++ * ++ * Not linked into rsync itself. ++ * ++ * This program is free software; you can redistribute it and/or modify ++ * it under the terms of the GNU General Public License version 2 as ++ * published by the Free Software Foundation. ++ */ ++ ++#include "rsync.h" ++ ++#include <sys/stat.h> ++ ++int dry_run = 0; ++int am_root = 0; ++int am_sender = 0; ++int read_only = 0; ++int list_only = 0; ++int copy_links = 0; ++int copy_unsafe_links = 0; ++extern int am_daemon, am_chrooted; ++ ++short info_levels[COUNT_INFO], debug_levels[COUNT_DEBUG]; ++ ++static int errs = 0; ++ ++static void check(const char *label, int actual_rc, int expect_ok, ++ const char *path, mode_t expected_mode) ++{ ++ struct stat st; ++ int got_ok = (actual_rc == 0); ++ if (got_ok != expect_ok) { ++ fprintf(stderr, "FAIL [%s]: rc=%d errno=%d (%s), expected %s\n", ++ label, actual_rc, errno, strerror(errno), ++ expect_ok ? "success" : "rejection"); ++ errs++; ++ return; ++ } ++ if (path && stat(path, &st) < 0) { ++ fprintf(stderr, "FAIL [%s]: stat(%s) failed: %s\n", ++ label, path, strerror(errno)); ++ errs++; ++ return; ++ } ++ if (path && (st.st_mode & 07777) != expected_mode) { ++ fprintf(stderr, ++ "FAIL [%s]: %s mode is 0%o, expected 0%o\n", ++ label, path, st.st_mode & 07777, expected_mode); ++ errs++; ++ return; ++ } ++ fprintf(stderr, "OK [%s]\n", label); ++} ++ ++int main(int argc, char **argv) ++{ ++ if (argc != 2) { ++ fprintf(stderr, "usage: %s <module-dir>\n", argv[0]); ++ return 2; ++ } ++ if (chdir(argv[1]) < 0) { ++ perror("chdir"); ++ return 2; ++ } ++ ++ /* Simulate the daemon-without-chroot deployment that do_chmod_at() ++ * defends. With am_daemon=0 or am_chrooted=1 the wrapper falls ++ * through to plain do_chmod() and the symlink-race test would be ++ * meaningless. */ ++ am_daemon = 1; ++ am_chrooted = 0; ++ ++ /* Test layout (all inside the directory we just chdir'd to): ++ * ++ * ./realdir/sentinel -- regular target file ++ * ./inside_link -> realdir -- legitimate dir-symlink within the tree ++ * ./escape_link -> ../trap -- attacker swap, target outside tree ++ * ../trap/sentinel -- the file the attacker wants to alter ++ * ++ * The shell wrapper that calls this helper has set both sentinel ++ * files to mode 0600 so we have a clean baseline to compare. ++ */ ++ ++ /* Scenario A: legitimate parent dir-symlink, chmod must succeed. */ ++ int rc = do_chmod_at("inside_link/sentinel", 0640); ++ check("A: legit dir-symlink within tree", ++ rc, 1, "realdir/sentinel", 0640); ++ ++ /* Scenario B: parent symlink escapes the tree -- chmod must be ++ * rejected and the outside file's mode must be unchanged. */ ++ rc = do_chmod_at("escape_link/sentinel", 0666); ++ check("B: parent symlink escapes tree (the attack)", ++ rc, 0, "../trap/sentinel", 0600); ++ ++ /* Scenario C: plain relative path with no symlink components, ++ * regression check that the safe wrapper doesn't break the ++ * normal case. */ ++ rc = do_chmod_at("realdir/sentinel", 0644); ++ check("C: plain relative path (regression check)", ++ rc, 1, "realdir/sentinel", 0644); ++ ++ /* Scenario D: top-level file, no parent directory component. ++ * Falls back to do_chmod(); should succeed. */ ++ rc = do_chmod_at("topfile", 0640); ++ check("D: top-level file, no parent component", ++ rc, 1, "topfile", 0640); ++ ++ if (errs) ++ fprintf(stderr, "%d failure(s)\n", errs); ++ return errs ? 1 : 0; ++} +diff --git a/t_stub.c b/t_stub.c +index eee92729..63bc144c 100644 +--- a/t_stub.c ++++ b/t_stub.c +@@ -23,6 +23,8 @@ + + int do_fsync = 0; + int inplace = 0; ++int am_daemon = 0; ++int am_chrooted = 0; + int modify_window = 0; + int preallocate_files = 0; + int protect_args = 0; +diff --git a/testsuite/chmod-symlink-race.test b/testsuite/chmod-symlink-race.test +new file mode 100755 +index 00000000..48bbfbb4 +--- /dev/null ++++ b/testsuite/chmod-symlink-race.test +@@ -0,0 +1,68 @@ ++#!/bin/sh ++ ++# Copyright (C) 2026 by Andrew Tridgell ++ ++# This program is distributable under the terms of the GNU GPL (see ++# COPYING). ++ ++# Regression test for the symlink-TOCTOU class of bug applied to ++# chmod() on the receiver side. The CVE-2026-29518 fix used ++# secure_relative_open() for the basis-file open, but every other ++# path-based syscall the receiver runs on sender-controllable paths ++# is vulnerable to the same primitive: a local attacker swaps a ++# symlink into one of the parent directory components between the ++# receiver's check and its act, and the syscall escapes the module. ++# ++# This test exercises the new do_chmod_at() wrapper via the ++# t_chmod_secure helper. The helper sets up two scenarios: ++# - a parent dir-symlink that resolves WITHIN the module tree ++# (legitimate -K-style use, must continue to work) ++# - a parent dir-symlink that escapes the module tree (the ++# attack, must be rejected) ++# plus two regression scenarios (plain relative path, top-level ++# file) that just confirm the safe wrapper doesn't break the ++# normal case. ++# ++# The kernel-enforced "stay below dirfd" path resolution is ++# only available on Linux 5.6+, FreeBSD 13+, and macOS 15+. ++# Skip on platforms that fall back to per-component O_NOFOLLOW ++# (Solaris, OpenBSD, NetBSD, Cygwin); the per-component fallback ++# would also reject the attack but the legitimate dir-symlink ++# scenario would fail there. ++ ++. "$suitedir/rsync.fns" ++ ++case "$(uname -s)" in ++ SunOS|OpenBSD|NetBSD|CYGWIN*) ++ test_skipped "do_chmod_at relies on RESOLVE_BENEATH-equivalent kernel support not available on $(uname -s)" ++ ;; ++esac ++ ++mod="$scratchdir/module" ++trap_outside="$scratchdir/trap" ++rm -rf "$mod" "$trap_outside" ++mkdir -p "$mod/realdir" "$trap_outside" ++ ++# Set up the four file-system objects the helper expects: ++echo bystander > "$mod/realdir/sentinel" ++chmod 0600 "$mod/realdir/sentinel" ++echo target > "$trap_outside/sentinel" ++chmod 0600 "$trap_outside/sentinel" ++ln -s realdir "$mod/inside_link" ++ln -s ../trap "$mod/escape_link" ++echo top > "$mod/topfile" ++chmod 0600 "$mod/topfile" ++ ++"$TOOLDIR/t_chmod_secure" "$mod" || \ ++ test_fail "t_chmod_secure reported failures (see stderr above)" ++ ++# Sanity-check from the shell side too: the outside file's mode must ++# still be 0600 -- the helper checked this, but a second look from ++# the shell guards against a helper-internal stat() bug. ++mode=$(stat -c '%a' "$trap_outside/sentinel" 2>/dev/null \ ++ || stat -f '%Lp' "$trap_outside/sentinel" 2>/dev/null) ++if [ "$mode" != "600" ]; then ++ test_fail "outside sentinel mode changed from 600 to $mode -- chmod escaped the module" ++fi ++ ++exit 0 +diff --git a/xattrs.c b/xattrs.c +index 26e50a6f..4462d63b 100644 +--- a/xattrs.c ++++ b/xattrs.c +@@ -1086,7 +1086,7 @@ int set_xattr(const char *fname, const struct file_struct *file, const char *fna + && !S_ISLNK(sxp->st.st_mode) + #endif + && access(fname, W_OK) < 0 +- && do_chmod(fname, (sxp->st.st_mode & CHMOD_BITS) | S_IWUSR) == 0) ++ && do_chmod_at(fname, (sxp->st.st_mode & CHMOD_BITS) | S_IWUSR) == 0) + added_write_perm = 1; + + ndx = F_XATTR(file); +@@ -1094,7 +1094,7 @@ int set_xattr(const char *fname, const struct file_struct *file, const char *fna + lst = &glst->xa_items; + int return_value = rsync_xal_set(fname, lst, fnamecmp, sxp); + if (added_write_perm) /* remove the temporary write permission */ +- do_chmod(fname, sxp->st.st_mode); ++ do_chmod_at(fname, sxp->st.st_mode); + return return_value; + } + +@@ -1211,7 +1211,7 @@ int set_stat_xattr(const char *fname, struct file_struct *file, mode_t new_mode) + mode = (fst.st_mode & _S_IFMT) | (fmode & ACCESSPERMS) + | (S_ISDIR(fst.st_mode) ? 0700 : 0600); + if (fst.st_mode != mode) +- do_chmod(fname, mode); ++ do_chmod_at(fname, mode); + if (!IS_DEVICE(fst.st_mode)) + fst.st_rdev = 0; /* just in case */ +
meta/recipes-devtools/rsync/files/CVE-2026-43619_p2.patch new file +205 -0
205 lines not on this page · raw message
meta/recipes-devtools/rsync/files/CVE-2026-43619_p3.patch new file +1783 -0
1,783 lines not on this page · raw message
meta/recipes-devtools/rsync/files/CVE-2026-43619_p4.patch new file +568 -0
541 lines not on this page · raw message
meta/recipes-devtools/rsync/files/CVE-2026-43619_p5.patch new file +69 -0
69 lines not on this page · raw message
meta/recipes-devtools/rsync/files/CVE-2026-43619_p6.patch new file +264 -0
264 lines not on this page · raw message
meta/recipes-devtools/rsync/rsync_3.4.1.bb +6 -0
Lines of this file are not shown. View the raw message.
The rest of this diff is not shown. View the raw message.
--
2.35.6
Part of the diff is not shown here. View the raw message.
PATCH v2 3/6 rsync: Fix CVE-2026-43618
In reply to Ashishkumar Parmar X (asparmar - E INFOCHIPS PRIVATE LIMITED at Cisco)From: Ashishkumar Parmar <[email protected]>
This patch applies the upstream v3.4.3 backport for CVE-2026-43618. The upstream fix commit is referenced in [1], and the public CVE advisory is referenced in [2].
[1] https://github.com/RsyncProject/rsync/commit/c44c90e9460c666c965446a8c0957f0b9fa4c66a
[2] https://github.com/RsyncProject/rsync/security/advisories/GHSA-g37v-g3gj-pmwq
--- .../rsync/files/CVE-2026-43618.patch | 248 ++++++++++++++++++ meta/recipes-devtools/rsync/rsync_3.4.1.bb | 1 + 2 files changed, 249 insertions(+) create mode 100644 meta/recipes-devtools/rsync/files/CVE-2026-43618.patch
meta/recipes-devtools/rsync/files/CVE-2026-43618.patch new file +248 -0
@@ -0,0 +1,248 @@ +From a93cfc86b6daa2dfd754bcbe261cf01afef056f4 Mon Sep 17 00:00:00 2001 +From: Andrew Tridgell <[email protected]> +Date: Wed, 29 Apr 2026 11:10:59 +1000 +Subject: [PATCH] token: harden compressed-token decoding against integer + overflow + +The receiver's three compressed-token decoders -- +recv_deflated_token (zlib), recv_zstd_token, and +recv_compressed_token (lz4) -- accumulated rx_token (a 32-bit +signed counter) without overflow checking. A malicious sender +could craft a compressed-token stream that walked rx_token past +INT32_MAX, with careful manipulation leaking process memory +contents to the wire (environment variables, passwords, heap +pointers, library pointers -- significantly weakening ASLR +and facilitating further exploitation). + +Cap rx_token at MAX_TOKEN_INDEX = 0x7ffffffe. Fold the +bookkeeping into recv_compressed_token_num() and +recv_compressed_token_run() shared by all three decoders. Reject +negative or out-of-range token values explicitly. Also cap the +simple_recv_token literal-block length at the source: any +wire-supplied length > CHUNK_SIZE is ill-formed (the matching +simple_send_token never writes a chunk larger than CHUNK_SIZE), +so reject before looping on attacker-controlled bytes. + +Reach: an authenticated daemon connection with compression +enabled (the default for protocols >= 30 when both peers +advertise it). Disabling compression on the daemon +("refuse options = compress" in rsyncd.conf) is the available +workaround. + +Reporter: Omar Elsayed (seks99x). + +Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]> + +CVE: CVE-2026-43618 +Upstream-Status: Backport [https://github.com/RsyncProject/rsync/commit/c44c90e9460c666c965446a8c0957f0b9fa4c66a] + +Backport Changes: +- Resolved token.c context differences against Wrynose's 3.4.1 + source, whose three decoder bodies do not all match upstream's + surrounding negative-token validation blocks. +- The Wrynose hunk replaces the older per-decoder token bookkeeping + with the upstream recv_compressed_token_num() and + recv_compressed_token_run() helper calls at the matching zlib, + zstd, and lz4 decoder sites. + +(cherry picked from commit c44c90e9460c666c965446a8c0957f0b9fa4c66a) +Signed-off-by: Ashishkumar Parmar <[email protected]> +--- + receiver.c | 11 +++++- + token.c | 102 ++++++++++++++++++++++++++++++----------------------- + 2 files changed, 67 insertions(+), 46 deletions(-) + +diff --git a/receiver.c b/receiver.c +index 8cf8366b..a487ad5a 100644 +--- a/receiver.c ++++ b/receiver.c +@@ -318,7 +318,12 @@ static int receive_data(int f_in, char *fname_r, int fd_r, OFF_T size_r, + } + } + +- while ((i = recv_token(f_in, &data)) != 0) { ++ while (1) { ++ data = NULL; ++ i = recv_token(f_in, &data); ++ if (i == 0) ++ break; ++ + if (INFO_GTE(PROGRESS, 1)) + show_progress(offset, total_size); + +@@ -326,6 +331,10 @@ static int receive_data(int f_in, char *fname_r, int fd_r, OFF_T size_r, + maybe_send_keepalive(time(NULL), MSK_ALLOW_FLUSH | MSK_ACTIVE_RECEIVER); + + if (i > 0) { ++ if (!data) { ++ rprintf(FERROR, "Invalid literal token with no data [%s]\n", who_am_i()); ++ exit_cleanup(RERR_PROTOCOL); ++ } + if (DEBUG_GTE(DELTASUM, 3)) { + rprintf(FINFO,"data recv %d at %s\n", + i, big_num(offset)); +diff --git a/token.c b/token.c +index c108b3af..02dabd8d 100644 +--- a/token.c ++++ b/token.c +@@ -291,6 +291,14 @@ static int32 simple_recv_token(int f, char **data) + int32 i = read_int(f); + if (i <= 0) + return i; ++ /* simple_send_token caps each literal chunk at CHUNK_SIZE; ++ * reject anything larger so a hostile peer cannot drive the ++ * read_buf below past our static CHUNK_SIZE buffer. */ ++ if (i > CHUNK_SIZE) { ++ rprintf(FERROR, "invalid uncompressed token length %ld [%s]\n", ++ (long)i, who_am_i()); ++ exit_cleanup(RERR_PROTOCOL); ++ } + residue = i; + } + +@@ -493,9 +501,52 @@ static char *cbuf; + static char *dbuf; + + /* for decoding runs of tokens */ ++#define MAX_TOKEN_INDEX ((int32)0x7ffffffe) ++ + static int32 rx_token; + static int32 rx_run; + ++static NORETURN void invalid_compressed_token(void) ++{ ++ rprintf(FERROR, "invalid token number in compressed stream\n"); ++ exit_cleanup(RERR_PROTOCOL); ++} ++ ++static int32 recv_compressed_token_num(int f, int32 flag) ++{ ++ if (flag & TOKEN_REL) { ++ int32 incr = flag & 0x3f; ++ if (rx_token > MAX_TOKEN_INDEX - incr) ++ invalid_compressed_token(); ++ rx_token += incr; ++ flag >>= 6; ++ } else { ++ rx_token = read_int(f); ++ if (rx_token < 0 || rx_token > MAX_TOKEN_INDEX) ++ invalid_compressed_token(); ++ } ++ ++ if (flag & 1) { ++ rx_run = read_byte(f); ++ rx_run += read_byte(f) << 8; ++ if (rx_run <= 0 || rx_token > MAX_TOKEN_INDEX - rx_run) ++ invalid_compressed_token(); ++ recv_state = r_running; ++ } ++ ++ return -1 - rx_token; ++} ++ ++static int32 recv_compressed_token_run(void) ++{ ++ if (rx_run <= 0 || rx_token >= MAX_TOKEN_INDEX) ++ invalid_compressed_token(); ++ ++rx_token; ++ if (--rx_run == 0) ++ recv_state = r_idle; ++ return -1 - rx_token; ++} ++ + /* Receive a deflated token and inflate it */ + static int32 recv_deflated_token(int f, char **data) + { +@@ -586,17 +637,7 @@ static int32 recv_deflated_token(int f, char **data) + } + + /* here we have a token of some kind */ +- if (flag & TOKEN_REL) { +- rx_token += flag & 0x3f; +- flag >>= 6; +- } else +- rx_token = read_int(f); +- if (flag & 1) { +- rx_run = read_byte(f); +- rx_run += read_byte(f) << 8; +- recv_state = r_running; +- } +- return -1 - rx_token; ++ return recv_compressed_token_num(f, flag); + + case r_inflating: + rx_strm.next_out = (Bytef *)dbuf; +@@ -616,10 +657,7 @@ static int32 recv_deflated_token(int f, char **data) + break; + + case r_running: +- ++rx_token; +- if (--rx_run == 0) +- recv_state = r_idle; +- return -1 - rx_token; ++ return recv_compressed_token_run(); + } + } + } +@@ -828,17 +866,7 @@ static int32 recv_zstd_token(int f, char **data) + return 0; + } + /* here we have a token of some kind */ +- if (flag & TOKEN_REL) { +- rx_token += flag & 0x3f; +- flag >>= 6; +- } else +- rx_token = read_int(f); +- if (flag & 1) { +- rx_run = read_byte(f); +- rx_run += read_byte(f) << 8; +- recv_state = r_running; +- } +- return -1 - rx_token; ++ return recv_compressed_token_num(f, flag); + + case r_inflated: /* zstd doesn't get into this state */ + break; +@@ -869,10 +897,7 @@ static int32 recv_zstd_token(int f, char **data) + break; + + case r_running: +- ++rx_token; +- if (--rx_run == 0) +- recv_state = r_idle; +- return -1 - rx_token; ++ return recv_compressed_token_run(); + } + } + } +@@ -992,17 +1017,7 @@ static int32 recv_compressed_token(int f, char **data) + } + + /* here we have a token of some kind */ +- if (flag & TOKEN_REL) { +- rx_token += flag & 0x3f; +- flag >>= 6; +- } else +- rx_token = read_int(f); +- if (flag & 1) { +- rx_run = read_byte(f); +- rx_run += read_byte(f) << 8; +- recv_state = r_running; +- } +- return -1 - rx_token; ++ return recv_compressed_token_num(f, flag); + + case r_inflating: + avail_out = LZ4_decompress_safe(next_in, dbuf, avail_in, size); +@@ -1018,10 +1033,7 @@ static int32 recv_compressed_token(int f, char **data) + break; + + case r_running: +- ++rx_token; +- if (--rx_run == 0) +- recv_state = r_idle; +- return -1 - rx_token; ++ return recv_compressed_token_run(); + } + } + }
meta/recipes-devtools/rsync/rsync_3.4.1.bb +1 -0
@@ -26,6 +26,7 @@ SRC_URI = "https://download.samba.org/pub/${BPN}/src/${BP}.tar.gz \ file://CVE-2026-43619_p4.patch \ file://CVE-2026-43619_p5.patch \ file://CVE-2026-43619_p6.patch \ + file://CVE-2026-43618.patch \ " SRC_URI[sha256sum] = "2924bcb3a1ed8b551fc101f740b9f0fe0a202b115027647cf69850d65fd88c52"
Signature · 1 line
2.35.6
Re: [wrynose][PATCH v2 3/6] rsync: Fix CVE-2026-43618
Hi,
Please ignore this patch series. I will send corrected patches again.
Regards,
Ashish.
________________________________
From: Ashishkumar Parmar X (asparmar - E INFOCHIPS PRIVATE LIMITED at Cisco) <[email protected]>
Sent: 17 July 2026 10:50
To: [email protected] <[email protected]>
Cc: Ashishkumar Parmar -X (asparmar - E INFOCHIPS PRIVATE LIMITED at Cisco) <[email protected]>
Subject: [wrynose][PATCH v2 3/6] rsync: Fix CVE-2026-43618
From: Ashishkumar Parmar <[email protected]>
This patch applies the upstream v3.4.3 backport for CVE-2026-43618. The upstream fix commit is referenced in [1], and the public CVE advisory is referenced in [2].
[1] https://github.com/RsyncProject/rsync/commit/c44c90e9460c666c965446a8c0957f0b9fa4c66a
[2] https://github.com/RsyncProject/rsync/security/advisories/GHSA-g37v-g3gj-pmwq
--- .../rsync/files/CVE-2026-43618.patch | 248 ++++++++++++++++++ meta/recipes-devtools/rsync/rsync_3.4.1.bb | 1 + 2 files changed, 249 insertions(+) create mode 100644 meta/recipes-devtools/rsync/files/CVE-2026-43618.patch
meta/recipes-devtools/rsync/files/CVE-2026-43618.patch new file +248 -0
@@ -0,0 +1,248 @@ +From a93cfc86b6daa2dfd754bcbe261cf01afef056f4 Mon Sep 17 00:00:00 2001 +From: Andrew Tridgell <[email protected]> +Date: Wed, 29 Apr 2026 11:10:59 +1000 +Subject: [PATCH] token: harden compressed-token decoding against integer + overflow + +The receiver's three compressed-token decoders -- +recv_deflated_token (zlib), recv_zstd_token, and +recv_compressed_token (lz4) -- accumulated rx_token (a 32-bit +signed counter) without overflow checking. A malicious sender +could craft a compressed-token stream that walked rx_token past +INT32_MAX, with careful manipulation leaking process memory +contents to the wire (environment variables, passwords, heap +pointers, library pointers -- significantly weakening ASLR +and facilitating further exploitation). + +Cap rx_token at MAX_TOKEN_INDEX = 0x7ffffffe. Fold the +bookkeeping into recv_compressed_token_num() and +recv_compressed_token_run() shared by all three decoders. Reject +negative or out-of-range token values explicitly. Also cap the +simple_recv_token literal-block length at the source: any +wire-supplied length > CHUNK_SIZE is ill-formed (the matching +simple_send_token never writes a chunk larger than CHUNK_SIZE), +so reject before looping on attacker-controlled bytes. + +Reach: an authenticated daemon connection with compression +enabled (the default for protocols >= 30 when both peers +advertise it). Disabling compression on the daemon +("refuse options = compress" in rsyncd.conf) is the available +workaround. + +Reporter: Omar Elsayed (seks99x). + +Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]> + +CVE: CVE-2026-43618 +Upstream-Status: Backport [https://github.com/RsyncProject/rsync/commit/c44c90e9460c666c965446a8c0957f0b9fa4c66a] + +Backport Changes: +- Resolved token.c context differences against Wrynose's 3.4.1 + source, whose three decoder bodies do not all match upstream's + surrounding negative-token validation blocks. +- The Wrynose hunk replaces the older per-decoder token bookkeeping + with the upstream recv_compressed_token_num() and + recv_compressed_token_run() helper calls at the matching zlib, + zstd, and lz4 decoder sites. + +(cherry picked from commit c44c90e9460c666c965446a8c0957f0b9fa4c66a) +Signed-off-by: Ashishkumar Parmar <[email protected]> +--- + receiver.c | 11 +++++- + token.c | 102 ++++++++++++++++++++++++++++++----------------------- + 2 files changed, 67 insertions(+), 46 deletions(-) + +diff --git a/receiver.c b/receiver.c +index 8cf8366b..a487ad5a 100644 +--- a/receiver.c ++++ b/receiver.c +@@ -318,7 +318,12 @@ static int receive_data(int f_in, char *fname_r, int fd_r, OFF_T size_r, + } + } + +- while ((i = recv_token(f_in, &data)) != 0) { ++ while (1) { ++ data = NULL; ++ i = recv_token(f_in, &data); ++ if (i == 0) ++ break; ++ + if (INFO_GTE(PROGRESS, 1)) + show_progress(offset, total_size); + +@@ -326,6 +331,10 @@ static int receive_data(int f_in, char *fname_r, int fd_r, OFF_T size_r, + maybe_send_keepalive(time(NULL), MSK_ALLOW_FLUSH | MSK_ACTIVE_RECEIVER); + + if (i > 0) { ++ if (!data) { ++ rprintf(FERROR, "Invalid literal token with no data [%s]\n", who_am_i()); ++ exit_cleanup(RERR_PROTOCOL); ++ } + if (DEBUG_GTE(DELTASUM, 3)) { + rprintf(FINFO,"data recv %d at %s\n", + i, big_num(offset)); +diff --git a/token.c b/token.c +index c108b3af..02dabd8d 100644 +--- a/token.c ++++ b/token.c +@@ -291,6 +291,14 @@ static int32 simple_recv_token(int f, char **data) + int32 i = read_int(f); + if (i <= 0) + return i; ++ /* simple_send_token caps each literal chunk at CHUNK_SIZE; ++ * reject anything larger so a hostile peer cannot drive the ++ * read_buf below past our static CHUNK_SIZE buffer. */ ++ if (i > CHUNK_SIZE) { ++ rprintf(FERROR, "invalid uncompressed token length %ld [%s]\n", ++ (long)i, who_am_i()); ++ exit_cleanup(RERR_PROTOCOL); ++ } + residue = i; + } + +@@ -493,9 +501,52 @@ static char *cbuf; + static char *dbuf; + + /* for decoding runs of tokens */ ++#define MAX_TOKEN_INDEX ((int32)0x7ffffffe) ++ + static int32 rx_token; + static int32 rx_run; + ++static NORETURN void invalid_compressed_token(void) ++{ ++ rprintf(FERROR, "invalid token number in compressed stream\n"); ++ exit_cleanup(RERR_PROTOCOL); ++} ++ ++static int32 recv_compressed_token_num(int f, int32 flag) ++{ ++ if (flag & TOKEN_REL) { ++ int32 incr = flag & 0x3f; ++ if (rx_token > MAX_TOKEN_INDEX - incr) ++ invalid_compressed_token(); ++ rx_token += incr; ++ flag >>= 6; ++ } else { ++ rx_token = read_int(f); ++ if (rx_token < 0 || rx_token > MAX_TOKEN_INDEX) ++ invalid_compressed_token(); ++ } ++ ++ if (flag & 1) { ++ rx_run = read_byte(f); ++ rx_run += read_byte(f) << 8; ++ if (rx_run <= 0 || rx_token > MAX_TOKEN_INDEX - rx_run) ++ invalid_compressed_token(); ++ recv_state = r_running; ++ } ++ ++ return -1 - rx_token; ++} ++ ++static int32 recv_compressed_token_run(void) ++{ ++ if (rx_run <= 0 || rx_token >= MAX_TOKEN_INDEX) ++ invalid_compressed_token(); ++ ++rx_token; ++ if (--rx_run == 0) ++ recv_state = r_idle; ++ return -1 - rx_token; ++} ++ + /* Receive a deflated token and inflate it */ + static int32 recv_deflated_token(int f, char **data) + { +@@ -586,17 +637,7 @@ static int32 recv_deflated_token(int f, char **data) + } + + /* here we have a token of some kind */ +- if (flag & TOKEN_REL) { +- rx_token += flag & 0x3f; +- flag >>= 6; +- } else +- rx_token = read_int(f); +- if (flag & 1) { +- rx_run = read_byte(f); +- rx_run += read_byte(f) << 8; +- recv_state = r_running; +- } +- return -1 - rx_token; ++ return recv_compressed_token_num(f, flag); + + case r_inflating: + rx_strm.next_out = (Bytef *)dbuf; +@@ -616,10 +657,7 @@ static int32 recv_deflated_token(int f, char **data) + break; + + case r_running: +- ++rx_token; +- if (--rx_run == 0) +- recv_state = r_idle; +- return -1 - rx_token; ++ return recv_compressed_token_run(); + } + } + } +@@ -828,17 +866,7 @@ static int32 recv_zstd_token(int f, char **data) + return 0; + } + /* here we have a token of some kind */ +- if (flag & TOKEN_REL) { +- rx_token += flag & 0x3f; +- flag >>= 6; +- } else +- rx_token = read_int(f); +- if (flag & 1) { +- rx_run = read_byte(f); +- rx_run += read_byte(f) << 8; +- recv_state = r_running; +- } +- return -1 - rx_token; ++ return recv_compressed_token_num(f, flag); + + case r_inflated: /* zstd doesn't get into this state */ + break; +@@ -869,10 +897,7 @@ static int32 recv_zstd_token(int f, char **data) + break; + + case r_running: +- ++rx_token; +- if (--rx_run == 0) +- recv_state = r_idle; +- return -1 - rx_token; ++ return recv_compressed_token_run(); + } + } + } +@@ -992,17 +1017,7 @@ static int32 recv_compressed_token(int f, char **data) + } + + /* here we have a token of some kind */ +- if (flag & TOKEN_REL) { +- rx_token += flag & 0x3f; +- flag >>= 6; +- } else +- rx_token = read_int(f); +- if (flag & 1) { +- rx_run = read_byte(f); +- rx_run += read_byte(f) << 8; +- recv_state = r_running; +- } +- return -1 - rx_token; ++ return recv_compressed_token_num(f, flag); + + case r_inflating: + avail_out = LZ4_decompress_safe(next_in, dbuf, avail_in, size); +@@ -1018,10 +1033,7 @@ static int32 recv_compressed_token(int f, char **data) + break; + + case r_running: +- ++rx_token; +- if (--rx_run == 0) +- recv_state = r_idle; +- return -1 - rx_token; ++ return recv_compressed_token_run(); + } + } + }
meta/recipes-devtools/rsync/rsync_3.4.1.bb +1 -0
@@ -26,6 +26,7 @@ SRC_URI = "https://download.samba.org/pub/${BPN}/src/${BP}.tar.gz \ file://CVE-2026-43619_p4.patch \ file://CVE-2026-43619_p5.patch \ file://CVE-2026-43619_p6.patch \ + file://CVE-2026-43618.patch \ " SRC_URI[sha256sum] = "2924bcb3a1ed8b551fc101f740b9f0fe0a202b115027647cf69850d65fd88c52"
--
2.35.6
PATCH v2 4/6 rsync: Fix CVE-2026-43620
In reply to Ashishkumar Parmar X (asparmar - E INFOCHIPS PRIVATE LIMITED at Cisco)From: Ashishkumar Parmar <[email protected]>
This patch applies the upstream v3.4.3 backport for CVE-2026-43620. The upstream fix commit is referenced in [1], and the public CVE advisory is referenced in [2].
[1] https://github.com/RsyncProject/rsync/commit/0cf200ecbb8baaf58070d825c4fbd892b4a63b69
[2] https://github.com/RsyncProject/rsync/security/advisories/GHSA-28pw-r563-rxvm
--- .../rsync/files/CVE-2026-43620.patch | 122 ++++++++++++++++++ meta/recipes-devtools/rsync/rsync_3.4.1.bb | 1 + 2 files changed, 123 insertions(+) create mode 100644 meta/recipes-devtools/rsync/files/CVE-2026-43620.patch
meta/recipes-devtools/rsync/files/CVE-2026-43620.patch new file +122 -0
@@ -0,0 +1,122 @@ +From 271771f089f5f7acff5cce1ca3242849b0f6a6a0 Mon Sep 17 00:00:00 2001 +From: Andrew Tridgell <[email protected]> +Date: Tue, 5 May 2026 16:48:16 +1000 +Subject: [PATCH] receiver: add parent_ndx<0 guard, mirroring 797e17f + +Commit 797e17f ("fixed an invalid access to files array") added a +parent_ndx < 0 guard to send_files() in sender.c, but the visually- +identical block in recv_files() in receiver.c was not updated. A +malicious rsync:// server can therefore drive any connecting client +into the same out-of-bounds dir_flist->files[-1] read followed by a +file_struct dereference in f_name() one line later. + +Reach: protocol-30+ default (inc_recurse) makes flist.c:2745 set +parent_ndx = -1 on the first received flist when the sender omits a +leading "." entry; rsync.c flist_for_ndx() does not reject ndx == 0 +in that state because the range check evaluates 0 < 0 = false; and +read_ndx_and_attrs() only validates ndx with the ITEM_TRANSFER bit +set, so iflags=ITEM_IS_NEW (or any other non-transfer iflag word) +bypasses the check. + +Apply the same guard receiver-side. Confirmed: the same PoC (a +minimal Python rsyncd that handshakes with CF_INC_RECURSE, sends a +no-leading-"." flist, and emits ndx=0 with ITEM_IS_NEW) crashes +unpatched 3.4.2 with SEGV_MAPERR si_addr=0x4101a-class in the +receiver child; with this guard it exits cleanly with code 2 +(RERR_PROTOCOL). + +The attack surface delta over the sender variant is large: +the original was malicious-client -> daemon, this is +malicious-server -> any rsync client doing a normal rsync:// +or remote-shell pull. + +Reported by Pratham Gupta (alchemy1729). + +Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]> + +CVE: CVE-2026-43620 +Upstream-Status: Backport [https://github.com/RsyncProject/rsync/commit/0cf200ecbb8baaf58070d825c4fbd892b4a63b69] + +(cherry picked from commit 0cf200ecbb8baaf58070d825c4fbd892b4a63b69) +Signed-off-by: Ashishkumar Parmar <[email protected]> +--- + generator.c | 4 ++++ + io.c | 3 +++ + receiver.c | 7 ++++++- + sender.c | 2 ++ + 4 files changed, 15 insertions(+), 1 deletion(-) + +diff --git a/generator.c b/generator.c +index 311e9b78..89f99db4 100644 +--- a/generator.c ++++ b/generator.c +@@ -2146,6 +2146,8 @@ void check_for_finished_files(int itemizing, enum logcode code, int check_redo) + if (send_failed) + ndx = get_hlink_num(); + flist = flist_for_ndx(ndx, "check_for_finished_files.1"); ++ if (ndx < flist->ndx_start) ++ exit_cleanup(RERR_PROTOCOL); + file = flist->files[ndx - flist->ndx_start]; + assert(file->flags & FLAG_HLINKED); + if (send_failed) +@@ -2174,6 +2176,8 @@ void check_for_finished_files(int itemizing, enum logcode code, int check_redo) + + flist = cur_flist; + cur_flist = flist_for_ndx(ndx, "check_for_finished_files.2"); ++ if (ndx < cur_flist->ndx_start) ++ exit_cleanup(RERR_PROTOCOL); + + file = cur_flist->files[ndx - cur_flist->ndx_start]; + if (solo_file) +diff --git a/io.c b/io.c +index bb60eeca..c654a7ba 100644 +--- a/io.c ++++ b/io.c +@@ -1090,6 +1090,9 @@ static void got_flist_entry_status(enum festatus status, int ndx) + { + struct file_list *flist = flist_for_ndx(ndx, "got_flist_entry_status"); + ++ if (ndx < flist->ndx_start) ++ exit_cleanup(RERR_PROTOCOL); ++ + if (remove_source_files) { + active_filecnt--; + active_bytecnt -= F_LENGTH(flist->files[ndx - flist->ndx_start]); +diff --git a/receiver.c b/receiver.c +index a487ad5a..3fa68d71 100644 +--- a/receiver.c ++++ b/receiver.c +@@ -467,7 +467,10 @@ static void handle_delayed_updates(char *local_name) + static void no_batched_update(int ndx, BOOL is_redo) + { + struct file_list *flist = flist_for_ndx(ndx, "no_batched_update"); +- struct file_struct *file = flist->files[ndx - flist->ndx_start]; ++ struct file_struct *file; ++ if (ndx < flist->ndx_start) ++ exit_cleanup(RERR_PROTOCOL); ++ file = flist->files[ndx - flist->ndx_start]; + + rprintf(FERROR_XFER, "(No batched update for%s \"%s\")\n", + is_redo ? " resend of" : "", f_name(file, NULL)); +@@ -604,6 +607,8 @@ int recv_files(int f_in, int f_out, char *local_name) + + if (ndx - cur_flist->ndx_start >= 0) + file = cur_flist->files[ndx - cur_flist->ndx_start]; ++ else if (cur_flist->parent_ndx < 0) ++ exit_cleanup(RERR_PROTOCOL); + else + file = dir_flist->files[cur_flist->parent_ndx]; + fname = local_name ? local_name : f_name(file, fbuf); +diff --git a/sender.c b/sender.c +index 99f431fe..033f87e5 100644 +--- a/sender.c ++++ b/sender.c +@@ -140,6 +140,8 @@ void successful_send(int ndx) + return; + + flist = flist_for_ndx(ndx, "successful_send"); ++ if (ndx < flist->ndx_start) ++ exit_cleanup(RERR_PROTOCOL); + file = flist->files[ndx - flist->ndx_start]; + if (!change_pathname(file, NULL, 0)) + return;
meta/recipes-devtools/rsync/rsync_3.4.1.bb +1 -0
@@ -27,6 +27,7 @@ SRC_URI = "https://download.samba.org/pub/${BPN}/src/${BP}.tar.gz \ file://CVE-2026-43619_p5.patch \ file://CVE-2026-43619_p6.patch \ file://CVE-2026-43618.patch \ + file://CVE-2026-43620.patch \ " SRC_URI[sha256sum] = "2924bcb3a1ed8b551fc101f740b9f0fe0a202b115027647cf69850d65fd88c52"
Signature · 1 line
2.35.6
Re: [wrynose][PATCH v2 4/6] rsync: Fix CVE-2026-43620
Hi,
Please ignore this patch series. I will send corrected patches again.
Regards,
Ashish.
________________________________
From: Ashishkumar Parmar X (asparmar - E INFOCHIPS PRIVATE LIMITED at Cisco) <[email protected]>
Sent: 17 July 2026 10:50
To: [email protected] <[email protected]>
Cc: Ashishkumar Parmar -X (asparmar - E INFOCHIPS PRIVATE LIMITED at Cisco) <[email protected]>
Subject: [wrynose][PATCH v2 4/6] rsync: Fix CVE-2026-43620
From: Ashishkumar Parmar <[email protected]>
This patch applies the upstream v3.4.3 backport for CVE-2026-43620. The upstream fix commit is referenced in [1], and the public CVE advisory is referenced in [2].
[1] https://github.com/RsyncProject/rsync/commit/0cf200ecbb8baaf58070d825c4fbd892b4a63b69
[2] https://github.com/RsyncProject/rsync/security/advisories/GHSA-28pw-r563-rxvm
--- .../rsync/files/CVE-2026-43620.patch | 122 ++++++++++++++++++ meta/recipes-devtools/rsync/rsync_3.4.1.bb | 1 + 2 files changed, 123 insertions(+) create mode 100644 meta/recipes-devtools/rsync/files/CVE-2026-43620.patch
meta/recipes-devtools/rsync/files/CVE-2026-43620.patch new file +122 -0
@@ -0,0 +1,122 @@ +From 271771f089f5f7acff5cce1ca3242849b0f6a6a0 Mon Sep 17 00:00:00 2001 +From: Andrew Tridgell <[email protected]> +Date: Tue, 5 May 2026 16:48:16 +1000 +Subject: [PATCH] receiver: add parent_ndx<0 guard, mirroring 797e17f + +Commit 797e17f ("fixed an invalid access to files array") added a +parent_ndx < 0 guard to send_files() in sender.c, but the visually- +identical block in recv_files() in receiver.c was not updated. A +malicious rsync:// server can therefore drive any connecting client +into the same out-of-bounds dir_flist->files[-1] read followed by a +file_struct dereference in f_name() one line later. + +Reach: protocol-30+ default (inc_recurse) makes flist.c:2745 set +parent_ndx = -1 on the first received flist when the sender omits a +leading "." entry; rsync.c flist_for_ndx() does not reject ndx == 0 +in that state because the range check evaluates 0 < 0 = false; and +read_ndx_and_attrs() only validates ndx with the ITEM_TRANSFER bit +set, so iflags=ITEM_IS_NEW (or any other non-transfer iflag word) +bypasses the check. + +Apply the same guard receiver-side. Confirmed: the same PoC (a +minimal Python rsyncd that handshakes with CF_INC_RECURSE, sends a +no-leading-"." flist, and emits ndx=0 with ITEM_IS_NEW) crashes +unpatched 3.4.2 with SEGV_MAPERR si_addr=0x4101a-class in the +receiver child; with this guard it exits cleanly with code 2 +(RERR_PROTOCOL). + +The attack surface delta over the sender variant is large: +the original was malicious-client -> daemon, this is +malicious-server -> any rsync client doing a normal rsync:// +or remote-shell pull. + +Reported by Pratham Gupta (alchemy1729). + +Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]> + +CVE: CVE-2026-43620 +Upstream-Status: Backport [https://github.com/RsyncProject/rsync/commit/0cf200ecbb8baaf58070d825c4fbd892b4a63b69] + +(cherry picked from commit 0cf200ecbb8baaf58070d825c4fbd892b4a63b69) +Signed-off-by: Ashishkumar Parmar <[email protected]> +--- + generator.c | 4 ++++ + io.c | 3 +++ + receiver.c | 7 ++++++- + sender.c | 2 ++ + 4 files changed, 15 insertions(+), 1 deletion(-) + +diff --git a/generator.c b/generator.c +index 311e9b78..89f99db4 100644 +--- a/generator.c ++++ b/generator.c +@@ -2146,6 +2146,8 @@ void check_for_finished_files(int itemizing, enum logcode code, int check_redo) + if (send_failed) + ndx = get_hlink_num(); + flist = flist_for_ndx(ndx, "check_for_finished_files.1"); ++ if (ndx < flist->ndx_start) ++ exit_cleanup(RERR_PROTOCOL); + file = flist->files[ndx - flist->ndx_start]; + assert(file->flags & FLAG_HLINKED); + if (send_failed) +@@ -2174,6 +2176,8 @@ void check_for_finished_files(int itemizing, enum logcode code, int check_redo) + + flist = cur_flist; + cur_flist = flist_for_ndx(ndx, "check_for_finished_files.2"); ++ if (ndx < cur_flist->ndx_start) ++ exit_cleanup(RERR_PROTOCOL); + + file = cur_flist->files[ndx - cur_flist->ndx_start]; + if (solo_file) +diff --git a/io.c b/io.c +index bb60eeca..c654a7ba 100644 +--- a/io.c ++++ b/io.c +@@ -1090,6 +1090,9 @@ static void got_flist_entry_status(enum festatus status, int ndx) + { + struct file_list *flist = flist_for_ndx(ndx, "got_flist_entry_status"); + ++ if (ndx < flist->ndx_start) ++ exit_cleanup(RERR_PROTOCOL); ++ + if (remove_source_files) { + active_filecnt--; + active_bytecnt -= F_LENGTH(flist->files[ndx - flist->ndx_start]); +diff --git a/receiver.c b/receiver.c +index a487ad5a..3fa68d71 100644 +--- a/receiver.c ++++ b/receiver.c +@@ -467,7 +467,10 @@ static void handle_delayed_updates(char *local_name) + static void no_batched_update(int ndx, BOOL is_redo) + { + struct file_list *flist = flist_for_ndx(ndx, "no_batched_update"); +- struct file_struct *file = flist->files[ndx - flist->ndx_start]; ++ struct file_struct *file; ++ if (ndx < flist->ndx_start) ++ exit_cleanup(RERR_PROTOCOL); ++ file = flist->files[ndx - flist->ndx_start]; + + rprintf(FERROR_XFER, "(No batched update for%s \"%s\")\n", + is_redo ? " resend of" : "", f_name(file, NULL)); +@@ -604,6 +607,8 @@ int recv_files(int f_in, int f_out, char *local_name) + + if (ndx - cur_flist->ndx_start >= 0) + file = cur_flist->files[ndx - cur_flist->ndx_start]; ++ else if (cur_flist->parent_ndx < 0) ++ exit_cleanup(RERR_PROTOCOL); + else + file = dir_flist->files[cur_flist->parent_ndx]; + fname = local_name ? local_name : f_name(file, fbuf); +diff --git a/sender.c b/sender.c +index 99f431fe..033f87e5 100644 +--- a/sender.c ++++ b/sender.c +@@ -140,6 +140,8 @@ void successful_send(int ndx) + return; + + flist = flist_for_ndx(ndx, "successful_send"); ++ if (ndx < flist->ndx_start) ++ exit_cleanup(RERR_PROTOCOL); + file = flist->files[ndx - flist->ndx_start]; + if (!change_pathname(file, NULL, 0)) + return;
meta/recipes-devtools/rsync/rsync_3.4.1.bb +1 -0
@@ -27,6 +27,7 @@ SRC_URI = "https://download.samba.org/pub/${BPN}/src/${BP}.tar.gz \ file://CVE-2026-43619_p5.patch \ file://CVE-2026-43619_p6.patch \ file://CVE-2026-43618.patch \ + file://CVE-2026-43620.patch \ " SRC_URI[sha256sum] = "2924bcb3a1ed8b551fc101f740b9f0fe0a202b115027647cf69850d65fd88c52"
--
2.35.6
PATCH v2 5/6 rsync: Fix CVE-2026-43617
In reply to Ashishkumar Parmar X (asparmar - E INFOCHIPS PRIVATE LIMITED at Cisco)From: Ashishkumar Parmar <[email protected]>
This patch applies the upstream v3.4.3 backport for CVE-2026-43617. The upstream fix commit is referenced in [1], and the public CVE advisory is referenced in [2].
[1] https://github.com/RsyncProject/rsync/commit/c38f20c5ffabacd0c0c483786ab224a36e14bf43
[2] https://github.com/RsyncProject/rsync/security/advisories/GHSA-rjfm-3w2m-jf4f
--- .../rsync/files/CVE-2026-43617.patch | 201 ++++++++++++++++++ meta/recipes-devtools/rsync/rsync_3.4.1.bb | 1 + 2 files changed, 202 insertions(+) create mode 100644 meta/recipes-devtools/rsync/files/CVE-2026-43617.patch
meta/recipes-devtools/rsync/files/CVE-2026-43617.patch new file +201 -0
@@ -0,0 +1,201 @@ +From d811094dc63486a8b6acf23963daccc23e9dbf43 Mon Sep 17 00:00:00 2001 +From: Andrew Tridgell <[email protected]> +Date: Wed, 31 Dec 2025 13:50:35 +1100 +Subject: [PATCH] clientserver: fix hostname ACL bypass when using daemon + chroot + +On an rsync daemon configured with "daemon chroot", the reverse-DNS +lookup of the connecting client was performed *after* the chroot +had been entered. If the chroot did not contain the files glibc +needs for resolution (/etc/resolv.conf, /etc/nsswitch.conf, +/etc/hosts, NSS service modules), the lookup failed and +client_name() returned "UNKNOWN". Hostname-based deny rules +("hosts deny = *.evil.example") therefore could not match, and +an attacker controlling their PTR record could connect from a +hostname the administrator had intended to deny. IP-based ACLs +were unaffected. + +Do the reverse DNS lookup before chroot/setuid; client_name() +caches its result, so the post-chroot call uses the cached value +and hostname-based ACLs work even when DNS is unavailable +post-chroot. + +Adds testsuite/daemon-chroot-acl.test as end-to-end regression +coverage. The test sets up an empty chroot directory, configures +"hosts deny = <localhost-resolved-name>" with daemon chroot, and +asserts the connection is refused with @ERROR access denied. +Uses unshare --user --map-root-user for non-root CAP_SYS_CHROOT; +skips cleanly on non-Linux or when user namespaces aren't +available. + +Reporter: Joshua Rogers (MegaManSec). + +Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]> + +CVE: CVE-2026-43617 +Upstream-Status: Backport [https://github.com/RsyncProject/rsync/commit/c38f20c5ffabacd0c0c483786ab224a36e14bf43] + +Backport Changes: +- Omitted upstream .github/workflows/macos-build.yml because the + Yocto recipe does not carry upstream CI workflow files. That + upstream hunk only adds daemon-chroot-acl to the macOS + RSYNC_EXPECT_SKIPPED list and has no rsync source/runtime effect. + +(cherry picked from commit c38f20c5ffabacd0c0c483786ab224a36e14bf43) +Signed-off-by: Ashishkumar Parmar <[email protected]> +--- + clientserver.c | 22 ++++++ + testsuite/daemon-chroot-acl.test | 111 +++++++++++++++++++++++++++++++ + 2 files changed, 133 insertions(+) + create mode 100644 testsuite/daemon-chroot-acl.test + +diff --git a/clientserver.c b/clientserver.c +index b6eba098..3333aa96 100644 +--- a/clientserver.c ++++ b/clientserver.c +@@ -1310,6 +1310,28 @@ int start_daemon(int f_in, int f_out) + if (lp_proxy_protocol() && !read_proxy_protocol_header(f_in)) + return -1; + ++ /* Do reverse DNS lookup before chroot/setuid. The result is cached, ++ * so the later client_name() call will use this cached value. This ++ * ensures hostname-based ACLs work even when DNS is unavailable ++ * after chroot. ++ * ++ * "reverse lookup" can be set globally OR per-module, so we also ++ * scan each module: a deployment with "reverse lookup = no" in the ++ * global section but "reverse lookup = yes" in a specific module ++ * still triggers a post-chroot lookup at access-check time ++ * (rsync_module() in this file), which would also fail in the ++ * chroot and turn hostname-based deny rules into silent bypasses. */ ++ { ++ int need_reverse = lp_reverse_lookup(-1); ++ int j, num_modules = lp_num_modules(); ++ for (j = 0; !need_reverse && j < num_modules; j++) { ++ if (lp_reverse_lookup(j)) ++ need_reverse = 1; ++ } ++ if (need_reverse) ++ (void)client_name(client_addr(f_in)); ++ } ++ + p = lp_daemon_chroot(); + if (*p) { + log_init(0); /* Make use we've initialized syslog before chrooting. */ +diff --git a/testsuite/daemon-chroot-acl.test b/testsuite/daemon-chroot-acl.test +new file mode 100644 +index 00000000..9d1c1b63 +--- /dev/null ++++ b/testsuite/daemon-chroot-acl.test +@@ -0,0 +1,111 @@ ++#!/bin/sh ++ ++# Copyright (C) 2026 by Andrew Tridgell ++ ++# This program is distributable under the terms of the GNU GPL (see ++# COPYING). ++ ++# Regression test for GHSA-rjfm-3w2m-jf4f: a hostname-based "hosts deny" ++# rule must still match when the daemon performs a 'daemon chroot' and ++# the chroot does not contain the NSS files glibc needs for reverse DNS. ++# ++# Pre-fix, reverse DNS happened *after* the daemon chroot. With an empty ++# chroot the NSS lookup failed, client_name() returned "UNKNOWN", and a ++# deny rule referring to the connecting hostname silently failed to ++# match. ++# ++# Two scenarios are exercised so we can distinguish the case the fix ++# definitely covers from the per-module path that may still be ++# vulnerable: ++# A. global "reverse lookup = yes" (covered by b6abdb4c) ++# B. only module "reverse lookup = yes" (gap to verify) ++ ++. "$suitedir/rsync.fns" ++ ++case `uname -s` in ++Linux*) ;; ++*) test_skipped "test is Linux-specific (uses chroot+unshare)" ;; ++esac ++ ++# We need CAP_SYS_CHROOT. Re-exec under a user namespace if not root. ++if ! chroot / /bin/true 2>/dev/null; then ++ if [ -z "$RSYNC_UNSHARED" ] && unshare --user --map-root-user true 2>/dev/null; then ++ echo "Re-running under unshare --user --map-root-user..." ++ RSYNC_UNSHARED=1 exec unshare --user --map-root-user "$SHELL_PATH" $RUNSHFLAGS "$0" ++ fi ++ test_skipped "need CAP_SYS_CHROOT (root or unshare --user --map-root-user)" ++fi ++ ++# We need 127.0.0.1 to reverse-resolve to a real hostname while NSS is ++# still working (i.e. before the daemon's chroot). The daemon will ++# look that name up itself as part of its hostname-based ACL check; ++# we then deny that name and assert the connection is rejected. ++client_hostname=`getent hosts 127.0.0.1 2>/dev/null | awk 'NR==1 {print $2}'` ++if [ -z "$client_hostname" ] || [ "$client_hostname" = "127.0.0.1" ]; then ++ test_skipped "no reverse DNS for 127.0.0.1" ++fi ++ ++chrootdir="$scratchdir/chroot" ++rm -rf "$chrootdir" ++mkdir -p "$chrootdir/modroot" ++echo "from chroot" > "$chrootdir/modroot/file1" ++ ++conf="$scratchdir/test-rsyncd.conf" ++logfile="$scratchdir/rsyncd.log" ++ ++write_conf() { ++ cat >"$conf" <<EOF ++use chroot = no ++log file = $logfile ++daemon chroot = $chrootdir ++reverse lookup = $1 ++hosts deny = $client_hostname ++max verbosity = 4 ++ ++[chrootmod] ++ path = /modroot ++ read only = yes ++ reverse lookup = $2 ++EOF ++} ++ ++# Run a transfer and return 0 if the daemon refused with @ERROR access ++# denied (the expected outcome when the deny rule matches). ++run_check() { ++ label="$1" ++ ++ rm -f "$logfile" ++ rm -rf "$todir" ++ mkdir -p "$todir" ++ ++ out="$scratchdir/run.out" ++ ++ RSYNC_CONNECT_PROG="$RSYNC --config=$conf --daemon" \ ++ $RSYNC -av localhost::chrootmod/ "$todir/" >"$out" 2>&1 ++ rc=$? ++ ++ echo "----- $label (rsync exit $rc):" ++ cat "$out" ++ echo "----- daemon log:" ++ [ -f "$logfile" ] && cat "$logfile" ++ echo "-----" ++ ++ grep -q '@ERROR.*access denied' "$out" ++} ++ ++# Scenario A: global reverse lookup. Covered by b6abdb4c. ++write_conf yes yes ++if ! run_check "Scenario A (global reverse lookup = yes)"; then ++ test_fail "Scenario A: hostname deny rule was bypassed" ++fi ++ ++# Scenario B: only the per-module reverse-lookup setting is enabled. ++# The b6abdb4c fix only pre-warms client_name()'s cache when the ++# global setting is on, so the post-chroot lookup in this path may ++# still produce "UNKNOWN" and bypass the deny rule. ++write_conf no yes ++if ! run_check "Scenario B (per-module reverse lookup only)"; then ++ test_fail "Scenario B: hostname deny rule was bypassed (per-module reverse lookup with daemon chroot still has the bypass)" ++fi ++ ++exit 0
meta/recipes-devtools/rsync/rsync_3.4.1.bb +1 -0
@@ -28,6 +28,7 @@ SRC_URI = "https://download.samba.org/pub/${BPN}/src/${BP}.tar.gz \ file://CVE-2026-43619_p6.patch \ file://CVE-2026-43618.patch \ file://CVE-2026-43620.patch \ + file://CVE-2026-43617.patch \ " SRC_URI[sha256sum] = "2924bcb3a1ed8b551fc101f740b9f0fe0a202b115027647cf69850d65fd88c52"
Signature · 1 line
2.35.6
Re: [wrynose][PATCH v2 5/6] rsync: Fix CVE-2026-43617
Hi,
Please ignore this patch series. I will send corrected patches again.
Regards,
Ashish.
________________________________
From: Ashishkumar Parmar X (asparmar - E INFOCHIPS PRIVATE LIMITED at Cisco) <[email protected]>
Sent: 17 July 2026 10:50
To: [email protected] <[email protected]>
Cc: Ashishkumar Parmar -X (asparmar - E INFOCHIPS PRIVATE LIMITED at Cisco) <[email protected]>
Subject: [wrynose][PATCH v2 5/6] rsync: Fix CVE-2026-43617
From: Ashishkumar Parmar <[email protected]>
This patch applies the upstream v3.4.3 backport for CVE-2026-43617. The upstream fix commit is referenced in [1], and the public CVE advisory is referenced in [2].
[1] https://github.com/RsyncProject/rsync/commit/c38f20c5ffabacd0c0c483786ab224a36e14bf43
[2] https://github.com/RsyncProject/rsync/security/advisories/GHSA-rjfm-3w2m-jf4f
--- .../rsync/files/CVE-2026-43617.patch | 201 ++++++++++++++++++ meta/recipes-devtools/rsync/rsync_3.4.1.bb | 1 + 2 files changed, 202 insertions(+) create mode 100644 meta/recipes-devtools/rsync/files/CVE-2026-43617.patch
meta/recipes-devtools/rsync/files/CVE-2026-43617.patch new file +201 -0
@@ -0,0 +1,201 @@ +From d811094dc63486a8b6acf23963daccc23e9dbf43 Mon Sep 17 00:00:00 2001 +From: Andrew Tridgell <[email protected]> +Date: Wed, 31 Dec 2025 13:50:35 +1100 +Subject: [PATCH] clientserver: fix hostname ACL bypass when using daemon + chroot + +On an rsync daemon configured with "daemon chroot", the reverse-DNS +lookup of the connecting client was performed *after* the chroot +had been entered. If the chroot did not contain the files glibc +needs for resolution (/etc/resolv.conf, /etc/nsswitch.conf, +/etc/hosts, NSS service modules), the lookup failed and +client_name() returned "UNKNOWN". Hostname-based deny rules +("hosts deny = *.evil.example") therefore could not match, and +an attacker controlling their PTR record could connect from a +hostname the administrator had intended to deny. IP-based ACLs +were unaffected. + +Do the reverse DNS lookup before chroot/setuid; client_name() +caches its result, so the post-chroot call uses the cached value +and hostname-based ACLs work even when DNS is unavailable +post-chroot. + +Adds testsuite/daemon-chroot-acl.test as end-to-end regression +coverage. The test sets up an empty chroot directory, configures +"hosts deny = <localhost-resolved-name>" with daemon chroot, and +asserts the connection is refused with @ERROR access denied. +Uses unshare --user --map-root-user for non-root CAP_SYS_CHROOT; +skips cleanly on non-Linux or when user namespaces aren't +available. + +Reporter: Joshua Rogers (MegaManSec). + +Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]> + +CVE: CVE-2026-43617 +Upstream-Status: Backport [https://github.com/RsyncProject/rsync/commit/c38f20c5ffabacd0c0c483786ab224a36e14bf43] + +Backport Changes: +- Omitted upstream .github/workflows/macos-build.yml because the + Yocto recipe does not carry upstream CI workflow files. That + upstream hunk only adds daemon-chroot-acl to the macOS + RSYNC_EXPECT_SKIPPED list and has no rsync source/runtime effect. + +(cherry picked from commit c38f20c5ffabacd0c0c483786ab224a36e14bf43) +Signed-off-by: Ashishkumar Parmar <[email protected]> +--- + clientserver.c | 22 ++++++ + testsuite/daemon-chroot-acl.test | 111 +++++++++++++++++++++++++++++++ + 2 files changed, 133 insertions(+) + create mode 100644 testsuite/daemon-chroot-acl.test + +diff --git a/clientserver.c b/clientserver.c +index b6eba098..3333aa96 100644 +--- a/clientserver.c ++++ b/clientserver.c +@@ -1310,6 +1310,28 @@ int start_daemon(int f_in, int f_out) + if (lp_proxy_protocol() && !read_proxy_protocol_header(f_in)) + return -1; + ++ /* Do reverse DNS lookup before chroot/setuid. The result is cached, ++ * so the later client_name() call will use this cached value. This ++ * ensures hostname-based ACLs work even when DNS is unavailable ++ * after chroot. ++ * ++ * "reverse lookup" can be set globally OR per-module, so we also ++ * scan each module: a deployment with "reverse lookup = no" in the ++ * global section but "reverse lookup = yes" in a specific module ++ * still triggers a post-chroot lookup at access-check time ++ * (rsync_module() in this file), which would also fail in the ++ * chroot and turn hostname-based deny rules into silent bypasses. */ ++ { ++ int need_reverse = lp_reverse_lookup(-1); ++ int j, num_modules = lp_num_modules(); ++ for (j = 0; !need_reverse && j < num_modules; j++) { ++ if (lp_reverse_lookup(j)) ++ need_reverse = 1; ++ } ++ if (need_reverse) ++ (void)client_name(client_addr(f_in)); ++ } ++ + p = lp_daemon_chroot(); + if (*p) { + log_init(0); /* Make use we've initialized syslog before chrooting. */ +diff --git a/testsuite/daemon-chroot-acl.test b/testsuite/daemon-chroot-acl.test +new file mode 100644 +index 00000000..9d1c1b63 +--- /dev/null ++++ b/testsuite/daemon-chroot-acl.test +@@ -0,0 +1,111 @@ ++#!/bin/sh ++ ++# Copyright (C) 2026 by Andrew Tridgell ++ ++# This program is distributable under the terms of the GNU GPL (see ++# COPYING). ++ ++# Regression test for GHSA-rjfm-3w2m-jf4f: a hostname-based "hosts deny" ++# rule must still match when the daemon performs a 'daemon chroot' and ++# the chroot does not contain the NSS files glibc needs for reverse DNS. ++# ++# Pre-fix, reverse DNS happened *after* the daemon chroot. With an empty ++# chroot the NSS lookup failed, client_name() returned "UNKNOWN", and a ++# deny rule referring to the connecting hostname silently failed to ++# match. ++# ++# Two scenarios are exercised so we can distinguish the case the fix ++# definitely covers from the per-module path that may still be ++# vulnerable: ++# A. global "reverse lookup = yes" (covered by b6abdb4c) ++# B. only module "reverse lookup = yes" (gap to verify) ++ ++. "$suitedir/rsync.fns" ++ ++case `uname -s` in ++Linux*) ;; ++*) test_skipped "test is Linux-specific (uses chroot+unshare)" ;; ++esac ++ ++# We need CAP_SYS_CHROOT. Re-exec under a user namespace if not root. ++if ! chroot / /bin/true 2>/dev/null; then ++ if [ -z "$RSYNC_UNSHARED" ] && unshare --user --map-root-user true 2>/dev/null; then ++ echo "Re-running under unshare --user --map-root-user..." ++ RSYNC_UNSHARED=1 exec unshare --user --map-root-user "$SHELL_PATH" $RUNSHFLAGS "$0" ++ fi ++ test_skipped "need CAP_SYS_CHROOT (root or unshare --user --map-root-user)" ++fi ++ ++# We need 127.0.0.1 to reverse-resolve to a real hostname while NSS is ++# still working (i.e. before the daemon's chroot). The daemon will ++# look that name up itself as part of its hostname-based ACL check; ++# we then deny that name and assert the connection is rejected. ++client_hostname=`getent hosts 127.0.0.1 2>/dev/null | awk 'NR==1 {print $2}'` ++if [ -z "$client_hostname" ] || [ "$client_hostname" = "127.0.0.1" ]; then ++ test_skipped "no reverse DNS for 127.0.0.1" ++fi ++ ++chrootdir="$scratchdir/chroot" ++rm -rf "$chrootdir" ++mkdir -p "$chrootdir/modroot" ++echo "from chroot" > "$chrootdir/modroot/file1" ++ ++conf="$scratchdir/test-rsyncd.conf" ++logfile="$scratchdir/rsyncd.log" ++ ++write_conf() { ++ cat >"$conf" <<EOF ++use chroot = no ++log file = $logfile ++daemon chroot = $chrootdir ++reverse lookup = $1 ++hosts deny = $client_hostname ++max verbosity = 4 ++ ++[chrootmod] ++ path = /modroot ++ read only = yes ++ reverse lookup = $2 ++EOF ++} ++ ++# Run a transfer and return 0 if the daemon refused with @ERROR access ++# denied (the expected outcome when the deny rule matches). ++run_check() { ++ label="$1" ++ ++ rm -f "$logfile" ++ rm -rf "$todir" ++ mkdir -p "$todir" ++ ++ out="$scratchdir/run.out" ++ ++ RSYNC_CONNECT_PROG="$RSYNC --config=$conf --daemon" \ ++ $RSYNC -av localhost::chrootmod/ "$todir/" >"$out" 2>&1 ++ rc=$? ++ ++ echo "----- $label (rsync exit $rc):" ++ cat "$out" ++ echo "----- daemon log:" ++ [ -f "$logfile" ] && cat "$logfile" ++ echo "-----" ++ ++ grep -q '@ERROR.*access denied' "$out" ++} ++ ++# Scenario A: global reverse lookup. Covered by b6abdb4c. ++write_conf yes yes ++if ! run_check "Scenario A (global reverse lookup = yes)"; then ++ test_fail "Scenario A: hostname deny rule was bypassed" ++fi ++ ++# Scenario B: only the per-module reverse-lookup setting is enabled. ++# The b6abdb4c fix only pre-warms client_name()'s cache when the ++# global setting is on, so the post-chroot lookup in this path may ++# still produce "UNKNOWN" and bypass the deny rule. ++write_conf no yes ++if ! run_check "Scenario B (per-module reverse lookup only)"; then ++ test_fail "Scenario B: hostname deny rule was bypassed (per-module reverse lookup with daemon chroot still has the bypass)" ++fi ++ ++exit 0
meta/recipes-devtools/rsync/rsync_3.4.1.bb +1 -0
@@ -28,6 +28,7 @@ SRC_URI = "https://download.samba.org/pub/${BPN}/src/${BP}.tar.gz \ file://CVE-2026-43619_p6.patch \ file://CVE-2026-43618.patch \ file://CVE-2026-43620.patch \ + file://CVE-2026-43617.patch \ " SRC_URI[sha256sum] = "2924bcb3a1ed8b551fc101f740b9f0fe0a202b115027647cf69850d65fd88c52"
--
2.35.6
PATCH v2 6/6 rsync: Fix CVE-2026-45232
In reply to Ashishkumar Parmar X (asparmar - E INFOCHIPS PRIVATE LIMITED at Cisco)From: Ashishkumar Parmar <[email protected]>
This patch applies the upstream v3.4.3 backport for CVE-2026-45232. The upstream fix commit is referenced in [1], and the public CVE advisory is referenced in [2].
[1] https://github.com/RsyncProject/rsync/commit/a5fc5ebe7a8ef1aa72f6e344599f97fd4427ecba
[2] https://github.com/RsyncProject/rsync/security/advisories/GHSA-8f85-j2cv-59m8
--- .../rsync/files/CVE-2026-45232.patch | 240 ++++++++++++++++++ meta/recipes-devtools/rsync/rsync_3.4.1.bb | 1 + 2 files changed, 241 insertions(+) create mode 100644 meta/recipes-devtools/rsync/files/CVE-2026-45232.patch
meta/recipes-devtools/rsync/files/CVE-2026-45232.patch new file +240 -0
@@ -0,0 +1,240 @@ +From de47c801a8083f1913e149c6ee6d0e864b728995 Mon Sep 17 00:00:00 2001 +From: Andrew Tridgell <[email protected]> +Date: Wed, 13 May 2026 20:35:35 +1000 +Subject: [PATCH] socket: reject over-long proxy response line + +fixes a one byte stack overflow when using RSYNC_PROXY with a +malicious proxy. + +Reach: only when RSYNC_PROXY is set and a malicious or MITM'd +proxy returns the pathological response. The byte written is +always '\0' and the attacker doesn't choose the offset, so impact +is corruption of one adjacent stack byte and possible later +misbehaviour or crash -- no information disclosure beyond the +existing rprintf of buffer contents. + +Reported by Aisle Research via Michal Ruprich + +CVE: CVE-2026-45232 +Upstream-Status: Backport [https://github.com/RsyncProject/rsync/commit/a5fc5ebe7a8ef1aa72f6e344599f97fd4427ecba] + +(cherry picked from commit a5fc5ebe7a8ef1aa72f6e344599f97fd4427ecba) +Signed-off-by: Ashishkumar Parmar <[email protected]> +--- + socket.c | 30 +++-- + testsuite/proxy-response-line-too-long.test | 128 ++++++++++++++++++++ + 2 files changed, 145 insertions(+), 13 deletions(-) + create mode 100755 testsuite/proxy-response-line-too-long.test + +diff --git a/socket.c b/socket.c +index c2075adf..6a8f6f4a 100644 +--- a/socket.c ++++ b/socket.c +@@ -47,21 +47,23 @@ static struct sigaction sigact; + + static int sock_exec(const char *prog); + ++#define PROXY_BUF_SIZE 1024 ++ + /* Establish a proxy connection on an open socket to a web proxy by using the + * CONNECT method. If proxy_user and proxy_pass are not NULL, they are used to + * authenticate to the proxy using the "Basic" proxy-authorization protocol. */ + static int establish_proxy_connection(int fd, char *host, int port, char *proxy_user, char *proxy_pass) + { +- char *cp, buffer[1024]; +- char *authhdr, authbuf[1024]; ++ char *cp, buffer[PROXY_BUF_SIZE + 1]; ++ char *authhdr, authbuf[PROXY_BUF_SIZE + 1]; + int len; + + if (proxy_user && proxy_pass) { +- stringjoin(buffer, sizeof buffer, ++ stringjoin(buffer, PROXY_BUF_SIZE, + proxy_user, ":", proxy_pass, NULL); + len = strlen(buffer); + +- if ((len*8 + 5) / 6 >= (int)sizeof authbuf - 3) { ++ if ((len*8 + 5) / 6 >= PROXY_BUF_SIZE - 3) { + rprintf(FERROR, + "authentication information is too long\n"); + return -1; +@@ -74,14 +76,14 @@ static int establish_proxy_connection(int fd, char *host, int port, char *proxy_ + authhdr = ""; + } + +- len = snprintf(buffer, sizeof buffer, "CONNECT %s:%d HTTP/1.0%s%s\r\n\r\n", host, port, authhdr, authbuf); +- assert(len > 0 && len < (int)sizeof buffer); ++ len = snprintf(buffer, PROXY_BUF_SIZE, "CONNECT %s:%d HTTP/1.0%s%s\r\n\r\n", host, port, authhdr, authbuf); ++ assert(len > 0 && len < PROXY_BUF_SIZE); + if (write(fd, buffer, len) != len) { + rsyserr(FERROR, errno, "failed to write to proxy"); + return -1; + } + +- for (cp = buffer; cp < &buffer[sizeof buffer - 1]; cp++) { ++ for (cp = buffer; cp < &buffer[PROXY_BUF_SIZE - 1]; cp++) { + if (read(fd, cp, 1) != 1) { + rsyserr(FERROR, errno, "failed to read from proxy"); + return -1; +@@ -90,11 +92,13 @@ static int establish_proxy_connection(int fd, char *host, int port, char *proxy_ + break; + } + +- if (*cp != '\n') +- cp++; +- *cp-- = '\0'; +- if (*cp == '\r') +- *cp = '\0'; ++ if (cp == &buffer[PROXY_BUF_SIZE - 1]) { ++ rprintf(FERROR, "proxy response line too long\n"); ++ return -1; ++ } ++ *cp = '\0'; ++ if (cp > buffer && cp[-1] == '\r') ++ cp[-1] = '\0'; + if (strncmp(buffer, "HTTP/", 5) != 0) { + rprintf(FERROR, "bad response from proxy -- %s\n", + buffer); +@@ -110,7 +114,7 @@ static int establish_proxy_connection(int fd, char *host, int port, char *proxy_ + } + /* throw away the rest of the HTTP header */ + while (1) { +- for (cp = buffer; cp < &buffer[sizeof buffer - 1]; cp++) { ++ for (cp = buffer; cp < &buffer[PROXY_BUF_SIZE]; cp++) { + if (read(fd, cp, 1) != 1) { + rsyserr(FERROR, errno, + "failed to read from proxy"); +diff --git a/testsuite/proxy-response-line-too-long.test b/testsuite/proxy-response-line-too-long.test +new file mode 100755 +index 00000000..7f55c43b +--- /dev/null ++++ b/testsuite/proxy-response-line-too-long.test +@@ -0,0 +1,128 @@ ++#!/bin/sh ++ ++# Copyright (C) 2026 by Andrew Tridgell ++ ++# This program is distributable under the terms of the GNU GPL (see ++# COPYING). ++ ++# Regression test for the off-by-one stack OOB write in ++# establish_proxy_connection() in socket.c when a malicious or ++# man-in-the-middle HTTP proxy returns a first response line of ++# 1023+ bytes without a '\n' terminator. ++# ++# Pre-fix: the read loop walked buffer[0..sizeof-2] one byte at a ++# time, then post-loop logic did "if (*cp != '\n') cp++; *cp-- = ++# '\0';". If no newline arrived before the loop filled the buffer, ++# cp was left at &buffer[sizeof-1] (never written by the loop), ++# *cp held stale stack bytes, and cp++ pushed cp one past the array. ++# The null-termination then wrote one byte out of bounds on the ++# stack. AddressSanitizer reports stack-buffer-overflow at the ++# null-termination site. ++# ++# Post-fix: the bound-exhaustion case is detected by position and ++# rejected with an "proxy response line too long" message, so no ++# OOB write occurs and rsync exits with a non-signal status. ++ ++. "$suitedir/rsync.fns" ++ ++command -v python3 >/dev/null 2>&1 || test_skipped "python3 not available" ++ ++workdir="$scratchdir/workdir" ++mkdir -p "$workdir" ++cd "$workdir" ++ ++port_file="$workdir/port" ++proxy_log="$workdir/proxy.log" ++ ++# A minimal TCP listener: binds to an ephemeral port on 127.0.0.1, ++# writes the chosen port to $port_file *before* accept() so the test ++# can synchronise without a sleep, accepts one connection, reads ++# until end-of-headers or 64 KiB, sends exactly 1023 bytes of 'X' ++# with no '\n', then closes. ++python3 - "$port_file" >"$proxy_log" 2>&1 <<'PYEOF' & ++import socket, sys, os ++port_file = sys.argv[1] ++s = socket.socket(socket.AF_INET, socket.SOCK_STREAM) ++s.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1) ++s.bind(("127.0.0.1", 0)) ++port = s.getsockname()[1] ++tmp = port_file + ".tmp" ++with open(tmp, "w") as fp: ++ fp.write("%d\n" % port) ++os.rename(tmp, port_file) # atomic visibility to the shell side ++s.listen(1) ++conn, _ = s.accept() ++conn.settimeout(5) ++try: ++ data = b"" ++ while b"\r\n\r\n" not in data and len(data) < 65536: ++ chunk = conn.recv(8192) ++ if not chunk: ++ break ++ data += chunk ++except socket.timeout: ++ pass ++conn.sendall(b"X" * 1023) # exactly the buffer-1 trigger size ++try: ++ conn.shutdown(socket.SHUT_RDWR) ++except OSError: ++ pass ++conn.close() ++s.close() ++PYEOF ++proxy_pid=$! ++ ++# Wait up to ~10s for the listener to publish its port. ++i=0 ++while [ ! -s "$port_file" ] && [ $i -lt 10 ]; do ++ sleep 1 ++ i=$((i + 1)) ++done ++ ++if [ ! -s "$port_file" ]; then ++ kill "$proxy_pid" 2>/dev/null ++ cat "$proxy_log" >&2 2>/dev/null ++ test_fail "proxy listener never published a port" ++fi ++ ++port=`cat "$port_file"` ++case "$port" in ++ *[!0-9]*|"") kill "$proxy_pid" 2>/dev/null; test_fail "bogus port from listener: '$port'" ;; ++esac ++ ++# Run rsync through the malicious proxy. Any rsync:// URL works: ++# the proxy intercepts the CONNECT and never forwards anywhere. ++rsync_err="$workdir/rsync.err" ++ ++# rsync MUST exit non-zero here (the proxy is misbehaving). ++# Use `|| status=$?` so we capture the real exit code under `sh -e`; ++# `if ! cmd; then status=$?` would only ever see 0 because the `!` ++# is the last command before `$?`. ++status=0 ++RSYNC_PROXY="127.0.0.1:$port" \ ++ $RSYNC rsync://example.invalid:873/whatever/ "$workdir/out/" \ ++ >/dev/null 2>"$rsync_err" || status=$? ++ ++# Reap the listener. ++wait "$proxy_pid" 2>/dev/null || true ++ ++# 1. rsync must not have crashed (SIGSEGV/SIGABRT report >= 128). ++if [ "$status" -ge 128 ]; then ++ cat "$rsync_err" >&2 ++ test_fail "rsync killed by signal (status=$status) -- possible stack OOB regression" ++fi ++ ++# 2. rsync must have actually exited non-zero (i.e. saw the bad proxy). ++if [ "$status" -eq 0 ]; then ++ cat "$rsync_err" >&2 ++ test_fail "rsync returned success despite malformed proxy response" ++fi ++ ++# 3. The new error message must appear. ++if ! grep -q "proxy response line too long" "$rsync_err"; then ++ cat "$rsync_err" >&2 ++ test_fail "expected 'proxy response line too long' in rsync stderr" ++fi ++ ++echo "OK: over-long proxy response line rejected cleanly without crashing" ++exit 0
meta/recipes-devtools/rsync/rsync_3.4.1.bb +1 -0
@@ -29,6 +29,7 @@ SRC_URI = "https://download.samba.org/pub/${BPN}/src/${BP}.tar.gz \ file://CVE-2026-43618.patch \ file://CVE-2026-43620.patch \ file://CVE-2026-43617.patch \ + file://CVE-2026-45232.patch \ " SRC_URI[sha256sum] = "2924bcb3a1ed8b551fc101f740b9f0fe0a202b115027647cf69850d65fd88c52"
Signature · 1 line
2.35.6
Re: [wrynose][PATCH v2 6/6] rsync: Fix CVE-2026-45232
Hi,
Please ignore this patch series. I will send corrected patches again.
Regards,
Ashish.
________________________________
From: Ashishkumar Parmar X (asparmar - E INFOCHIPS PRIVATE LIMITED at Cisco) <[email protected]>
Sent: 17 July 2026 10:50
To: [email protected] <[email protected]>
Cc: Ashishkumar Parmar -X (asparmar - E INFOCHIPS PRIVATE LIMITED at Cisco) <[email protected]>
Subject: [wrynose][PATCH v2 6/6] rsync: Fix CVE-2026-45232
From: Ashishkumar Parmar <[email protected]>
This patch applies the upstream v3.4.3 backport for CVE-2026-45232. The upstream fix commit is referenced in [1], and the public CVE advisory is referenced in [2].
[1] https://github.com/RsyncProject/rsync/commit/a5fc5ebe7a8ef1aa72f6e344599f97fd4427ecba
[2] https://github.com/RsyncProject/rsync/security/advisories/GHSA-8f85-j2cv-59m8
--- .../rsync/files/CVE-2026-45232.patch | 240 ++++++++++++++++++ meta/recipes-devtools/rsync/rsync_3.4.1.bb | 1 + 2 files changed, 241 insertions(+) create mode 100644 meta/recipes-devtools/rsync/files/CVE-2026-45232.patch
meta/recipes-devtools/rsync/files/CVE-2026-45232.patch new file +240 -0
@@ -0,0 +1,240 @@ +From de47c801a8083f1913e149c6ee6d0e864b728995 Mon Sep 17 00:00:00 2001 +From: Andrew Tridgell <[email protected]> +Date: Wed, 13 May 2026 20:35:35 +1000 +Subject: [PATCH] socket: reject over-long proxy response line + +fixes a one byte stack overflow when using RSYNC_PROXY with a +malicious proxy. + +Reach: only when RSYNC_PROXY is set and a malicious or MITM'd +proxy returns the pathological response. The byte written is +always '\0' and the attacker doesn't choose the offset, so impact +is corruption of one adjacent stack byte and possible later +misbehaviour or crash -- no information disclosure beyond the +existing rprintf of buffer contents. + +Reported by Aisle Research via Michal Ruprich + +CVE: CVE-2026-45232 +Upstream-Status: Backport [https://github.com/RsyncProject/rsync/commit/a5fc5ebe7a8ef1aa72f6e344599f97fd4427ecba] + +(cherry picked from commit a5fc5ebe7a8ef1aa72f6e344599f97fd4427ecba) +Signed-off-by: Ashishkumar Parmar <[email protected]> +--- + socket.c | 30 +++-- + testsuite/proxy-response-line-too-long.test | 128 ++++++++++++++++++++ + 2 files changed, 145 insertions(+), 13 deletions(-) + create mode 100755 testsuite/proxy-response-line-too-long.test + +diff --git a/socket.c b/socket.c +index c2075adf..6a8f6f4a 100644 +--- a/socket.c ++++ b/socket.c +@@ -47,21 +47,23 @@ static struct sigaction sigact; + + static int sock_exec(const char *prog); + ++#define PROXY_BUF_SIZE 1024 ++ + /* Establish a proxy connection on an open socket to a web proxy by using the + * CONNECT method. If proxy_user and proxy_pass are not NULL, they are used to + * authenticate to the proxy using the "Basic" proxy-authorization protocol. */ + static int establish_proxy_connection(int fd, char *host, int port, char *proxy_user, char *proxy_pass) + { +- char *cp, buffer[1024]; +- char *authhdr, authbuf[1024]; ++ char *cp, buffer[PROXY_BUF_SIZE + 1]; ++ char *authhdr, authbuf[PROXY_BUF_SIZE + 1]; + int len; + + if (proxy_user && proxy_pass) { +- stringjoin(buffer, sizeof buffer, ++ stringjoin(buffer, PROXY_BUF_SIZE, + proxy_user, ":", proxy_pass, NULL); + len = strlen(buffer); + +- if ((len*8 + 5) / 6 >= (int)sizeof authbuf - 3) { ++ if ((len*8 + 5) / 6 >= PROXY_BUF_SIZE - 3) { + rprintf(FERROR, + "authentication information is too long\n"); + return -1; +@@ -74,14 +76,14 @@ static int establish_proxy_connection(int fd, char *host, int port, char *proxy_ + authhdr = ""; + } + +- len = snprintf(buffer, sizeof buffer, "CONNECT %s:%d HTTP/1.0%s%s\r\n\r\n", host, port, authhdr, authbuf); +- assert(len > 0 && len < (int)sizeof buffer); ++ len = snprintf(buffer, PROXY_BUF_SIZE, "CONNECT %s:%d HTTP/1.0%s%s\r\n\r\n", host, port, authhdr, authbuf); ++ assert(len > 0 && len < PROXY_BUF_SIZE); + if (write(fd, buffer, len) != len) { + rsyserr(FERROR, errno, "failed to write to proxy"); + return -1; + } + +- for (cp = buffer; cp < &buffer[sizeof buffer - 1]; cp++) { ++ for (cp = buffer; cp < &buffer[PROXY_BUF_SIZE - 1]; cp++) { + if (read(fd, cp, 1) != 1) { + rsyserr(FERROR, errno, "failed to read from proxy"); + return -1; +@@ -90,11 +92,13 @@ static int establish_proxy_connection(int fd, char *host, int port, char *proxy_ + break; + } + +- if (*cp != '\n') +- cp++; +- *cp-- = '\0'; +- if (*cp == '\r') +- *cp = '\0'; ++ if (cp == &buffer[PROXY_BUF_SIZE - 1]) { ++ rprintf(FERROR, "proxy response line too long\n"); ++ return -1; ++ } ++ *cp = '\0'; ++ if (cp > buffer && cp[-1] == '\r') ++ cp[-1] = '\0'; + if (strncmp(buffer, "HTTP/", 5) != 0) { + rprintf(FERROR, "bad response from proxy -- %s\n", + buffer); +@@ -110,7 +114,7 @@ static int establish_proxy_connection(int fd, char *host, int port, char *proxy_ + } + /* throw away the rest of the HTTP header */ + while (1) { +- for (cp = buffer; cp < &buffer[sizeof buffer - 1]; cp++) { ++ for (cp = buffer; cp < &buffer[PROXY_BUF_SIZE]; cp++) { + if (read(fd, cp, 1) != 1) { + rsyserr(FERROR, errno, + "failed to read from proxy"); +diff --git a/testsuite/proxy-response-line-too-long.test b/testsuite/proxy-response-line-too-long.test +new file mode 100755 +index 00000000..7f55c43b +--- /dev/null ++++ b/testsuite/proxy-response-line-too-long.test +@@ -0,0 +1,128 @@ ++#!/bin/sh ++ ++# Copyright (C) 2026 by Andrew Tridgell ++ ++# This program is distributable under the terms of the GNU GPL (see ++# COPYING). ++ ++# Regression test for the off-by-one stack OOB write in ++# establish_proxy_connection() in socket.c when a malicious or ++# man-in-the-middle HTTP proxy returns a first response line of ++# 1023+ bytes without a '\n' terminator. ++# ++# Pre-fix: the read loop walked buffer[0..sizeof-2] one byte at a ++# time, then post-loop logic did "if (*cp != '\n') cp++; *cp-- = ++# '\0';". If no newline arrived before the loop filled the buffer, ++# cp was left at &buffer[sizeof-1] (never written by the loop), ++# *cp held stale stack bytes, and cp++ pushed cp one past the array. ++# The null-termination then wrote one byte out of bounds on the ++# stack. AddressSanitizer reports stack-buffer-overflow at the ++# null-termination site. ++# ++# Post-fix: the bound-exhaustion case is detected by position and ++# rejected with an "proxy response line too long" message, so no ++# OOB write occurs and rsync exits with a non-signal status. ++ ++. "$suitedir/rsync.fns" ++ ++command -v python3 >/dev/null 2>&1 || test_skipped "python3 not available" ++ ++workdir="$scratchdir/workdir" ++mkdir -p "$workdir" ++cd "$workdir" ++ ++port_file="$workdir/port" ++proxy_log="$workdir/proxy.log" ++ ++# A minimal TCP listener: binds to an ephemeral port on 127.0.0.1, ++# writes the chosen port to $port_file *before* accept() so the test ++# can synchronise without a sleep, accepts one connection, reads ++# until end-of-headers or 64 KiB, sends exactly 1023 bytes of 'X' ++# with no '\n', then closes. ++python3 - "$port_file" >"$proxy_log" 2>&1 <<'PYEOF' & ++import socket, sys, os ++port_file = sys.argv[1] ++s = socket.socket(socket.AF_INET, socket.SOCK_STREAM) ++s.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1) ++s.bind(("127.0.0.1", 0)) ++port = s.getsockname()[1] ++tmp = port_file + ".tmp" ++with open(tmp, "w") as fp: ++ fp.write("%d\n" % port) ++os.rename(tmp, port_file) # atomic visibility to the shell side ++s.listen(1) ++conn, _ = s.accept() ++conn.settimeout(5) ++try: ++ data = b"" ++ while b"\r\n\r\n" not in data and len(data) < 65536: ++ chunk = conn.recv(8192) ++ if not chunk: ++ break ++ data += chunk ++except socket.timeout: ++ pass ++conn.sendall(b"X" * 1023) # exactly the buffer-1 trigger size ++try: ++ conn.shutdown(socket.SHUT_RDWR) ++except OSError: ++ pass ++conn.close() ++s.close() ++PYEOF ++proxy_pid=$! ++ ++# Wait up to ~10s for the listener to publish its port. ++i=0 ++while [ ! -s "$port_file" ] && [ $i -lt 10 ]; do ++ sleep 1 ++ i=$((i + 1)) ++done ++ ++if [ ! -s "$port_file" ]; then ++ kill "$proxy_pid" 2>/dev/null ++ cat "$proxy_log" >&2 2>/dev/null ++ test_fail "proxy listener never published a port" ++fi ++ ++port=`cat "$port_file"` ++case "$port" in ++ *[!0-9]*|"") kill "$proxy_pid" 2>/dev/null; test_fail "bogus port from listener: '$port'" ;; ++esac ++ ++# Run rsync through the malicious proxy. Any rsync:// URL works: ++# the proxy intercepts the CONNECT and never forwards anywhere. ++rsync_err="$workdir/rsync.err" ++ ++# rsync MUST exit non-zero here (the proxy is misbehaving). ++# Use `|| status=$?` so we capture the real exit code under `sh -e`; ++# `if ! cmd; then status=$?` would only ever see 0 because the `!` ++# is the last command before `$?`. ++status=0 ++RSYNC_PROXY="127.0.0.1:$port" \ ++ $RSYNC rsync://example.invalid:873/whatever/ "$workdir/out/" \ ++ >/dev/null 2>"$rsync_err" || status=$? ++ ++# Reap the listener. ++wait "$proxy_pid" 2>/dev/null || true ++ ++# 1. rsync must not have crashed (SIGSEGV/SIGABRT report >= 128). ++if [ "$status" -ge 128 ]; then ++ cat "$rsync_err" >&2 ++ test_fail "rsync killed by signal (status=$status) -- possible stack OOB regression" ++fi ++ ++# 2. rsync must have actually exited non-zero (i.e. saw the bad proxy). ++if [ "$status" -eq 0 ]; then ++ cat "$rsync_err" >&2 ++ test_fail "rsync returned success despite malformed proxy response" ++fi ++ ++# 3. The new error message must appear. ++if ! grep -q "proxy response line too long" "$rsync_err"; then ++ cat "$rsync_err" >&2 ++ test_fail "expected 'proxy response line too long' in rsync stderr" ++fi ++ ++echo "OK: over-long proxy response line rejected cleanly without crashing" ++exit 0
meta/recipes-devtools/rsync/rsync_3.4.1.bb +1 -0
@@ -29,6 +29,7 @@ SRC_URI = "https://download.samba.org/pub/${BPN}/src/${BP}.tar.gz \ file://CVE-2026-43618.patch \ file://CVE-2026-43620.patch \ file://CVE-2026-43617.patch \ + file://CVE-2026-45232.patch \ " SRC_URI[sha256sum] = "2924bcb3a1ed8b551fc101f740b9f0fe0a202b115027647cf69850d65fd88c52"
--
2.35.6
Hi,
Please ignore this patch series. I will send corrected patches again.
Regards,
Ashish.
________________________________
From: Ashishkumar Parmar X (asparmar - E INFOCHIPS PRIVATE LIMITED at Cisco) <[email protected]>
Sent: 17 July 2026 10:50
To: [email protected] <[email protected]>
Cc: Ashishkumar Parmar -X (asparmar - E INFOCHIPS PRIVATE LIMITED at Cisco) <[email protected]>
Subject: [wrynose][PATCH v2 0/6] rsync: Fix multiple CVEs
From: Ashishkumar Parmar <[email protected]>
Backport the upstream rsync security fixes for CVE-2026-29518, CVE-2026-43617, CVE-2026-43618, CVE-2026-43619, CVE-2026-43620, and CVE-2026-45232 to Wrynose.
Changes in v2:
- Updated the CVE-2026-43619 backport to restore the upstream commit description and explanatory comments for the openat2/Android fallback handling.
- Clarified the Wrynose-specific Backport Changes metadata.
- No functional changes were made to the other five CVE patches.
Ashishkumar Parmar (6):
rsync: Fix CVE-2026-29518
rsync: Fix CVE-2026-43619
rsync: Fix CVE-2026-43618
rsync: Fix CVE-2026-43620
rsync: Fix CVE-2026-43617
rsync: Fix CVE-2026-45232
--
2.35.6