Branch: refs/heads/webkitglib/2.54
Home: https://github.com/WebKit/WebKit
Commit: cf77eb790e037850c99dc6ac8d6f7f0be440805c
https://github.com/WebKit/WebKit/commit/cf77eb790e037850c99dc6ac8d6f7f0be440805c
Author: Pawel Lampe <[email protected]>
Date: 2026-10-13 (Tue, 13 Oct 2026)
Changed paths:
M Source/WebCore/testing/Internals.cpp
Log Message:
-----------
Cherry-pick 323555@main (026e0af143fd).
https://bugs.webkit.org/show_bug.cgi?id=327227
[GLib] Damage tests fail when device scale != 1
https://bugs.webkit.org/show_bug.cgi?id=327227
Reviewed by Carlos Garcia Campos.
This change makes internals returning frame damage history rectangles
in CSS pixel coordinates rather than device pixel coordinates as the
tests under platform/glib/damage/ assume CSS pixel coordinates.
Canonical link: https://commits.webkit.org/323555@main
Canonical link: https://commits.webkit.org/317695.435@webkitglib/2.54
Commit: 2d23be4449df0ee4a073cedecaee7a96beb244a4
https://github.com/WebKit/WebKit/commit/2d23be4449df0ee4a073cedecaee7a96beb244a4
Author: Patrick Griffis <[email protected]>
Date: 2026-10-13 (Tue, 13 Oct 2026)
Changed paths:
M Source/WebKit/Shared/Extensions/wpe/WebExtensionUtilitiesWPE.cpp
Log Message:
-----------
Cherry-pick 323394@main (dc9edbe4aca6).
https://bugs.webkit.org/show_bug.cgi?id=326922
[WPE] Fix possible critical getting screen scales
https://bugs.webkit.org/show_bug.cgi?id=326922
Reviewed by Carlos Garcia Campos.
It is possible for a screen to be NULL, for example background views
in web extensions.
* Source/WebKit/Shared/Extensions/wpe/WebExtensionUtilitiesWPE.cpp:
(WebKit::availableScreenScales):
Canonical link: https://commits.webkit.org/323394@main
Canonical link: https://commits.webkit.org/317695.436@webkitglib/2.54
Commit: 7f6917fe82b9c1821b86c5151c490236487fe4fe
https://github.com/WebKit/WebKit/commit/7f6917fe82b9c1821b86c5151c490236487fe4fe
Author: Michael Catanzaro <[email protected]>
Date: 2026-10-13 (Tue, 13 Oct 2026)
Changed paths:
M Source/WebKit/UIProcess/gtk/SystemSettingsManagerProxyGtk.cpp
Log Message:
-----------
Cherry-pick 323359@main (de320ba3ff90).
https://bugs.webkit.org/show_bug.cgi?id=327072
REGRESSION(316709@main): [GTK] prefers-color-scheme doesn't update when the
system color scheme changes until restart
https://bugs.webkit.org/show_bug.cgi?id=327072
Reviewed by Patrick Griffis.
Now that we listen to different settings, we also need to listen to
different settings change notifications.
Canonical link: https://commits.webkit.org/323359@main
Canonical link: https://commits.webkit.org/317695.437@webkitglib/2.54
Commit: aa01e65b2817c990a2a1142cbb6ad221713843ac
https://github.com/WebKit/WebKit/commit/aa01e65b2817c990a2a1142cbb6ad221713843ac
Author: Claudio Saavedra <[email protected]>
Date: 2026-10-13 (Tue, 13 Oct 2026)
Changed paths:
A LayoutTests/http/wpt/crypto/export-rsa-key-crash.any-expected.txt
A LayoutTests/http/wpt/crypto/export-rsa-key-crash.any.html
A LayoutTests/http/wpt/crypto/export-rsa-key-crash.any.js
A LayoutTests/http/wpt/crypto/export-rsa-key-crash.any.worker-expected.txt
A LayoutTests/http/wpt/crypto/export-rsa-key-crash.any.worker.html
M Source/WebCore/crypto/gcrypt/GCryptUtilities.cpp
Log Message:
-----------
Cherry-pick 323448@main (e6cec05c0d9b).
https://bugs.webkit.org/show_bug.cgi?id=327077
[GCrypt] Exporting an RSA key with a zero component crashes
https://bugs.webkit.org/show_bug.cgi?id=327077
Reviewed by Javier Fernandez.
mpiSignedData() reads the first byte of the MPI data to decide whether
a sign byte is needed, but the data for zero is empty, so exporting an
RSA key with, for example, a zero public exponent as SPKI or PKCS#8
crashed. Encode zero as a single zero byte, as DER requires.
Test: http/wpt/crypto/export-rsa-key-crash.any.html
* LayoutTests/http/wpt/crypto/export-rsa-key-crash.any-expected.txt: Added.
* LayoutTests/http/wpt/crypto/export-rsa-key-crash.any.html: Added.
* LayoutTests/http/wpt/crypto/export-rsa-key-crash.any.js: Added.
* LayoutTests/http/wpt/crypto/export-rsa-key-crash.any.worker-expected.txt:
Added.
* LayoutTests/http/wpt/crypto/export-rsa-key-crash.any.worker.html: Added.
* Source/WebCore/crypto/gcrypt/GCryptUtilities.cpp:
(WebCore::mpiSignedData):
Canonical link: https://commits.webkit.org/323448@main
Canonical link: https://commits.webkit.org/317695.438@webkitglib/2.54
Commit: 31913c7a454349ab9e6ebd47763c8d8a4ad18087
https://github.com/WebKit/WebKit/commit/31913c7a454349ab9e6ebd47763c8d8a4ad18087
Author: Claudio Saavedra <[email protected]>
Date: 2026-10-13 (Tue, 13 Oct 2026)
Changed paths:
M LayoutTests/platform/glib/TestExpectations
A
LayoutTests/platform/glib/imported/w3c/web-platform-tests/WebCryptoAPI/sign_verify/eddsa_small_order_points.https.any-expected.txt
A
LayoutTests/platform/glib/imported/w3c/web-platform-tests/WebCryptoAPI/sign_verify/eddsa_small_order_points.https.any.worker-expected.txt
M Source/WebCore/crypto/gcrypt/CryptoAlgorithmEd25519GCrypt.cpp
Log Message:
-----------
Cherry-pick 323450@main (183d2f8db43a).
https://bugs.webkit.org/show_bug.cgi?id=280672
[GCrypt] Ed25519 verify() crashes with a small-order public key
https://bugs.webkit.org/show_bug.cgi?id=280672
Reviewed by Javier Fernandez.
gcry_pk_verify() aborts the process when the Ed25519 public key is a
point of order 1, 2 or 4, such as the identity. Reject small-order public
keys and signature R values before calling it, as the WebCrypto spec
requires anyway, so verification returns false instead.
Tests 3 and 6 of eddsa_small_order_points, which use mixed-order points,
still fail: libgcrypt gives the opposite result for them.
* LayoutTests/platform/glib/TestExpectations:
*
LayoutTests/platform/glib/imported/w3c/web-platform-tests/WebCryptoAPI/sign_verify/eddsa_small_order_points.https.any-expected.txt:
Added.
*
LayoutTests/platform/glib/imported/w3c/web-platform-tests/WebCryptoAPI/sign_verify/eddsa_small_order_points.https.any.worker-expected.txt:
Added.
* Source/WebCore/crypto/gcrypt/CryptoAlgorithmEd25519GCrypt.cpp:
(WebCore::isSmallOrderPoint):
(WebCore::verifyEd25519):
Canonical link: https://commits.webkit.org/323450@main
Canonical link: https://commits.webkit.org/317695.439@webkitglib/2.54
Commit: c2a683a7867efd01960897d1e83cce84d4f913b8
https://github.com/WebKit/WebKit/commit/c2a683a7867efd01960897d1e83cce84d4f913b8
Author: Claudio Saavedra <[email protected]>
Date: 2026-10-13 (Tue, 13 Oct 2026)
Changed paths:
M
LayoutTests/platform/glib/imported/w3c/web-platform-tests/WebCryptoAPI/derive_bits_keys/pbkdf2.https.any.worker_5001-6000-expected.txt
R
LayoutTests/platform/glib/imported/w3c/web-platform-tests/WebCryptoAPI/derive_bits_keys/pbkdf2.https.any.worker_6001-7000-expected.txt
M
LayoutTests/platform/glib/imported/w3c/web-platform-tests/WebCryptoAPI/derive_bits_keys/pbkdf2.https.any.worker_7001-8000-expected.txt
M
LayoutTests/platform/glib/imported/w3c/web-platform-tests/WebCryptoAPI/derive_bits_keys/pbkdf2.https.any_5001-6000-expected.txt
R
LayoutTests/platform/glib/imported/w3c/web-platform-tests/WebCryptoAPI/derive_bits_keys/pbkdf2.https.any_6001-7000-expected.txt
M
LayoutTests/platform/glib/imported/w3c/web-platform-tests/WebCryptoAPI/derive_bits_keys/pbkdf2.https.any_7001-8000-expected.txt
M Source/WebCore/crypto/gcrypt/CryptoAlgorithmPBKDF2GCrypt.cpp
Log Message:
-----------
Cherry-pick 323449@main (d4d7ebca935e).
https://bugs.webkit.org/show_bug.cgi?id=327074
[GCrypt] PBKDF2 fails with an empty password
https://bugs.webkit.org/show_bug.cgi?id=327074
Reviewed by Javier Fernandez.
gcry_kdf_derive() rejects a null passphrase, which is what an empty
Vector yields. Pass a non-null pointer instead.
*
LayoutTests/platform/glib/imported/w3c/web-platform-tests/WebCryptoAPI/derive_bits_keys/pbkdf2.https.any_5001-6000-expected.txt:
Updated.
*
LayoutTests/platform/glib/imported/w3c/web-platform-tests/WebCryptoAPI/derive_bits_keys/pbkdf2.https.any_6001-7000-expected.txt:
Removed.
*
LayoutTests/platform/glib/imported/w3c/web-platform-tests/WebCryptoAPI/derive_bits_keys/pbkdf2.https.any_7001-8000-expected.txt:
Updated.
*
LayoutTests/platform/glib/imported/w3c/web-platform-tests/WebCryptoAPI/derive_bits_keys/pbkdf2.https.any.worker_5001-6000-expected.txt:
Updated.
*
LayoutTests/platform/glib/imported/w3c/web-platform-tests/WebCryptoAPI/derive_bits_keys/pbkdf2.https.any.worker_6001-7000-expected.txt:
Removed.
*
LayoutTests/platform/glib/imported/w3c/web-platform-tests/WebCryptoAPI/derive_bits_keys/pbkdf2.https.any.worker_7001-8000-expected.txt:
Updated.
* Source/WebCore/crypto/gcrypt/CryptoAlgorithmPBKDF2GCrypt.cpp:
(WebCore::gcryptDeriveBits):
Canonical link: https://commits.webkit.org/323449@main
Canonical link: https://commits.webkit.org/317695.440@webkitglib/2.54
Commit: 9ec40f7ecd58843f36b1315241d8e6a358cbb791
https://github.com/WebKit/WebKit/commit/9ec40f7ecd58843f36b1315241d8e6a358cbb791
Author: Claudio Saavedra <[email protected]>
Date: 2026-10-13 (Tue, 13 Oct 2026)
Changed paths:
A
LayoutTests/crypto/subtle/rsa-oaep-import-key-decrypt-leading-zeros-expected.txt
A LayoutTests/crypto/subtle/rsa-oaep-import-key-decrypt-leading-zeros.html
M Source/WebCore/crypto/gcrypt/CryptoAlgorithmRSA_OAEPGCrypt.cpp
Log Message:
-----------
Cherry-pick 323285@main (f065a754d536).
https://bugs.webkit.org/show_bug.cgi?id=326989
[GCrypt] RSA-OAEP decrypt drops leading zero bytes of the plain text
https://bugs.webkit.org/show_bug.cgi?id=326989
Reviewed by Patrick Griffis.
libgcrypt returns the decrypted plain text as a byte string, but it was
read back through mpiData(), which parses it as an integer and drops
leading zero bytes. Decrypting or unwrapping data that starts with 0x00
returned a shorter result. Copy the bytes directly instead.
Test: crypto/subtle/rsa-oaep-import-key-decrypt-leading-zeros.html
*
LayoutTests/crypto/subtle/rsa-oaep-import-key-decrypt-leading-zeros-expected.txt:
Added.
* LayoutTests/crypto/subtle/rsa-oaep-import-key-decrypt-leading-zeros.html:
Added.
* Source/WebCore/crypto/gcrypt/CryptoAlgorithmRSA_OAEPGCrypt.cpp:
(WebCore::gcryptDecrypt):
Canonical link: https://commits.webkit.org/323285@main
Canonical link: https://commits.webkit.org/317695.441@webkitglib/2.54
Commit: a2b6863af2683d6bf5cd0ce9b3ce4ad5539aecc4
https://github.com/WebKit/WebKit/commit/a2b6863af2683d6bf5cd0ce9b3ce4ad5539aecc4
Author: Claudio Saavedra <[email protected]>
Date: 2026-10-13 (Tue, 13 Oct 2026)
Changed paths:
M Tools/Scripts/build-webkit
Log Message:
-----------
Cherry-pick 323187@main (ac0cb5443431).
https://bugs.webkit.org/show_bug.cgi?id=326843
[CMake] Report ccache hits in the build-webkit summary
https://bugs.webkit.org/show_bug.cgi?id=326843
Reviewed by Justin Michaud.
There is no easy way to tell how much of a build was served by ccache:
the global statistics accumulate across builds, and resetting them would
discard data the user may want to keep.
When ccache is in use, point CCACHE_STATSLOG to a fresh log in the
product directory, so that every compiler invocation of this build
records its result there, and summarize it in the "is now built"
message, splitting hits between local and remote storage and mentioning
remote storage errors when there are any. A CCACHE_STATSLOG already set
by the user is left untouched and no summary is printed.
* Tools/Scripts/build-webkit:
(setUpCCacheStatsLog):
(ccacheSummary):
(writeCongrats):
Canonical link: https://commits.webkit.org/323187@main
Canonical link: https://commits.webkit.org/317695.442@webkitglib/2.54
Commit: 9cbbf052922ea2b9336caaa1ea0ceaea18424033
https://github.com/WebKit/WebKit/commit/9cbbf052922ea2b9336caaa1ea0ceaea18424033
Author: Fujii Hironori <[email protected]>
Date: 2026-10-13 (Tue, 13 Oct 2026)
Changed paths:
M Tools/glib/glib_test_runner.py
Log Message:
-----------
Cherry-pick 323311@main (9c50e091cb1b).
https://bugs.webkit.org/show_bug.cgi?id=327076
[GLib] run-api-tests aborts with UnicodeDecodeError when a UTF-8 character
in test stderr is split across 4096-byte reads
https://bugs.webkit.org/show_bug.cgi?id=327076
Reviewed by Claudio Saavedra.
GLibTestRunner._read_from_stderr() decoded each 4096-byte chunk read
from stderr separately, so a multibyte UTF-8 character split across two
chunks raised UnicodeDecodeError and aborted the whole run. This happens
on the WPE ARM64 bot when the Swift runtime backtracer dumps memory after
a test crashes, because the dump contains many 2-byte UTF-8 characters.
Collect the chunks as bytes and decode them once at the end. Use
errors='replace' so that invalid UTF-8 in test output never aborts the
test runner.
* Tools/glib/glib_test_runner.py:
(GLibTestRunner._read_from_stderr):
Canonical link: https://commits.webkit.org/323311@main
Canonical link: https://commits.webkit.org/317695.443@webkitglib/2.54
Compare: https://github.com/WebKit/WebKit/compare/934d1b1a92a6...9cbbf052922e
To unsubscribe from these emails, change your notification settings at
https://github.com/WebKit/WebKit/settings/notifications
