linux/poky.git/scripts, branch scarthgap

wic: filemap: use separate fd for SEEK_HOLE probes

2026-06-10T13:35:20+00:00

While working on splitting-out wic from oe-core, on my openSUSE Leap 16.0 machine, the moment I split wic out, 2 oe-selftests always failed with 100% reproducibility: - wic.ModifyTests.test_wic_cp_ext - wic.Wic2.test_expand_mbr_image In both cases the symptom is the same: the filesystem has inode tables that are completely zeroed out. Both issues are linked together to the same underlying fault. FilemapSeek._get_ranges() is a generator. Due to the nature of finding each hole/data extent one at a time using the lseek() system call, it calls os.lseek() on a raw file descriptor, then yields, then the caller, sparse_copy(), calls file.seek() + file.read() on a Python BufferedReader wrapping that same fd — then the generator resumes and calls os.lseek() again. This interleaving of raw os.lseek() and buffered I/O on the same fd is undefined behaviour from Python's perspective. The BufferedReader tracks its own idea of the fd's position and buffer contents; os.lseek() changes the position behind its back. This can corrupt its internal state and cause read() to return stale/zero data. This code, however, has existed in wic since it was written, so why was it not noticed before? It turns out this bug was being masked by a number of implementation details that changed, especially when wic was split out for oe-core. These changes conspired together to cause the bug to be triggered. One of the root causes of this bug is that Python 3.14 increased the default buffer size from 8KB to 128KB[1]. With 8 KB buffers, read()s either go through the direct-read path leaving the buffer empty, or if it fills in 8KB chunks the buffer is fully drained. Either way, with a small buffer, read()s do a real raw seek. No fast path. No corruption. With a 128KB buffer, however, a much larger window exists where BufferedReader.seek() can take the fast-path after the raw file descriptor has already been repositioned by os.lseek() in the generator. With the smaller buffer, this window was too narrow to hit in practice. This is fixed by opening a second file object in FilemapSeek.__init__() dedicated to SEEK_DATA/SEEK_HOLE probes, leaving the data-reading handle (self._f_image) untouched. This explains why the corruption is deterministic and tied to specific block boundaries, why it only manifests with the split-out version using Python 3.14 (on systems that are using Python versions less than 3.14 on the host), and why using a separate file descriptor for reading bypasses the issue entirely. This is not an intermittent bug. For a more detailed explanation including log files, in-depth analysis, and a standalone Python reproducer, please see the linked bugzilla entry. Fixes: [YOCTO #16197] [1] https://github.com/python/cpython/commit/b1b4f9625c5f2a6b2c32bc5ee91c9fef3894b5e6 b1b4f9625c5f ("gh-117151: IO performance improvement, increase io.DEFAULT_BUFFER_SIZE to 128k (GH-118144)") AI-Generated: codex/claude-opus-4.6 (xhigh) (From OE-Core rev: 37a45219dd204b07bad40576fefccb2cf85b255c) Signed-off-by: Trevor Woerner Signed-off-by: Richard Purdie (cherry picked from commit 481969844385f2fa40a1230ca50253ec4ff516cd) Signed-off-by: Yoann Congal Signed-off-by: Paul Barker

scripts/install-buildtools: Update to 5.0.18

2026-06-10T13:35:20+00:00

Update to the 5.0.18 release of the 5.0 series for buildtools (From OE-Core rev: 7b2955ae65d4060aaa8f3eda9f3583750606a646) Signed-off-by: Yoann Congal Signed-off-by: Paul Barker

devtool: Disable gpg signing when setting up source tree repos

2026-05-12T20:31:34+00:00

This stops 'devtool modify foo' from failing with an error message like ERROR: Execution of 'git -c user.name=\"OpenEmbedded\" -c user.email=\"oe.patch@oe\" commit -q -m "Initial commit from upstream at version 1.90.0"' failed with exit code 128: error: cannot run ssh-keygen: No such file or directory error: fatal: failed to write commit object when GPG signing is enabled in the git configuration. (cherry picked from commit b5c84b07b87eafb4f68f7662b6cf26d8b73e3247) (From OE-Core rev: bbe0df71933174d8becc52184cd235277f10a141) Signed-off-by: Moritz Haase Signed-off-by: Richard Purdie Signed-off-by: Yoann Congal Signed-off-by: Paul Barker

selftest/scripts: Update old git protocol references

2026-05-04T12:57:32+00:00

git protocol accesses to our infrastructure are currently struggling and this has highlighted a number of places we're making those obsolete access forms. Update them to use https instead of the git protocol since it is preferred and more reliable. The devtool test needed quoting to handle the ';' in the url. The -f option to devtool also shows a deprecation warning so remove that. There were internal references to git protocol urls inside the nested git submodules test report, which means those repos need updating to use new git revisions. (From OE-Core rev: cbb3e323b74d4351c772a9bcd553008c31a220f0) Signed-off-by: Richard Purdie (cherry picked from commit 1ceba42623c5187d2f5a100d6a523abcdc75d34e) Signed-off-by: Yoann Congal Signed-off-by: Paul Barker

scripts: Default to https git protocol for YP/OE repos

2026-05-04T12:57:32+00:00

Following up on commit 139102a73d41 ("recipes: Default to https git protocol where possible"), > The recommendation from server maintainers is that the https protocol > is both faster and more reliable than the dedicated git protocol at this point. > Switch to it where possible. (From OE-Core rev: b9de796f797844b035e50e22c4f939fc90e176f8) Signed-off-by: Yoann Congal Signed-off-by: Richard Purdie (cherry picked from commit 74dd04549624b1d3ba6d4febe51fa9ba1c86944c) Signed-off-by: Paul Barker

Revert "scripts/install-buildtools: Update to 5.0.16"

2026-03-26T09:53:17+00:00

This reverts commit 917c382c412766b40779788ee2411ac23358f0ae. I mechanically did the update but should not have: this was already raised here: Re: [OE-core] Wrong version of buildtools files uploaded for 5.0.16. https://lore.kernel.org/all/DH2QPDTPQT3Q.1TJYOVQ7PIU7B@smile.fr/ Trying to fetch the 5.0.16 buildtools installer results in an error: $ ./scripts/install-buildtools INFO: Fetching buildtools installer ERROR: Could not download file from https://downloads.yoctoproject.org/releases/yocto/yocto-5.0.16/buildtools/x86_64-buildtools-extended-nativesdk-standalone-5.0.16.sh See https://downloads.yoctoproject.org/releases/yocto/yocto-5.0.16/RELEASENOTES: > --------------- > Known Issues > --------------- > The poky DISTRO_VERSION was incorrectly left at 5.0.15. This is a minor issue, if a workaround is needed please cherry-pick: > - poky commit 06210079b2e10d6d3fb943afe87864267e329821, or > - meta-yocto commit 03f93c769ec99e5086e492d8145eb308a718e8d3. (From OE-Core rev: 41597b5260fb5ca811d0fb4ae7e65246d61734eb) Signed-off-by: Yoann Congal Signed-off-by: Paul Barker

oe-setup-build: Fix typo

2026-03-25T17:34:13+00:00

A variable was mistyped in an error message resulting in this error: NameError: name 'tempalte_name' is not defined. Did you mean: 'template_name'? (From OE-Core rev: 2b69c249ab3e2fc0ffb6196080fe35266ee91bd3) Signed-off-by: Ryan Eatmon Signed-off-by: Richard Purdie (cherry picked from commit 275a3c015d37729c3b0c9cc4395d50ea2f210f02) Signed-off-by: Clayton Casciato Signed-off-by: Yoann Congal Signed-off-by: Paul Barker

improve_kernel_cve_report: add option to read debugsources.zstd

2026-03-25T17:34:13+00:00

Adding option to be able to import debugsources.zstd directly. The linux-yocto-debugsources.zstd is generated in every build and does not require any additional configuration. In contrast, SPDX_INCLUDE_COMPILED_SOURCES needs to be explicitly added and increases build time. (From OE-Core rev: e8e61a97bf0e88c45a323353e865a5654a500966) Signed-off-by: Daniel Turull Signed-off-by: Mathieu Dubois-Briand (cherry picked from commit c84a8958f30bbb982656ddcbe7476f6f81e1a6fb) Signed-off-by: Himanshu Jadon Signed-off-by: Yoann Congal Signed-off-by: Paul Barker

improve_kernel_cve_report: do not use custom version

2026-03-25T17:34:13+00:00

When using the version specified in cve-summary.json, we need to remove the suffix containing the custom version to match the versions from the CVEs. This patch truncates the version from cve-summary.json to use only the base version of the kernel. This is only applicable for kernels where the user has added their own version. (From OE-Core rev: 66968d81d55de7afdf139eac56a460d04c383a6c) Signed-off-by: Daniel Turull Signed-off-by: Antonin Godard Signed-off-by: Richard Purdie (cherry picked from commit 3942d40e96989268e8d1030f9d8c3859044d9635) Signed-off-by: Himanshu Jadon Signed-off-by: Yoann Congal Signed-off-by: Paul Barker

improve_kernel_cve_report: do not override backported-patch

2026-03-25T17:34:13+00:00

If the user has a CVE_STATUS for their own backported patch, the backport takes priority over upstream vulnerable versions. (From OE-Core rev: d317e2a52bd29a772de9bcd751f5b0e03277bd77) Signed-off-by: Daniel Turull Signed-off-by: Antonin Godard Signed-off-by: Richard Purdie (cherry picked from commit 0beef05be119ea465ba06553a42edea03dfc9fd3) Signed-off-by: Himanshu Jadon Signed-off-by: Yoann Congal Signed-off-by: Paul Barker