https://bugzilla.suse.com/show_bug.cgi?id=1212101 https://bugzilla.suse.com/show_bug.cgi?id=1212101#c2 Paul Tannington changed: What |Removed |Added ---------------------------------------------------------------------------- Flags|needinfo?(paul.pgp-7@gmx.co | |m) | --- Comment #2 from Paul Tannington ---

Do you have more information?

Unsure of what other information, or where to obtain it, would be useful :( The machine in question is quite elderly hardware: Operating System: openSUSE Tumbleweed 20230605 KDE Plasma Version: 5.27.5 KDE Frameworks Version: 5.106.0 Qt Version: 5.15.9 Kernel Version: 6.3.4-1-default (64-bit) Graphics Platform: X11 Processors: 2 × AMD Athlon(tm) 64 X2 Dual Core Processor 5600+ Memory: 3.8 GiB of RAM Graphics Processor: NV84 Manufacturer: Gigabyte Technology Co., Ltd. Product Name: GA-MA770-DS3 (I have a second machine which is hardware identical, also running TW20230605. Updating FF 113.0.2 -> 114.0 also results in the immediate crash. Reverting both machines to FF 113.0.2 using a saved 113 profile and FF starts as normal.) "journalctl -f" yields the following upon attempting to launch FF: Jun 07 13:51:27 Orion-15.openSUSE systemd[1095]: Started Firefox - Web Browser. Jun 07 13:51:28 Orion-15.openSUSE kmozillahelper[2148]: kf.i18n: KLocalizedString: Using an empty domain, fix the code. msgid: "Mozilla Firefox" msgid_plural: "" msgctxt: "" Jun 07 13:51:29 Orion-15.openSUSE firefox[2111]: Theme parsing error: gtk-dark.css:1:50: Failed to import: Error opening file /usr/share/themes/Breeze-Dark/gtk-3.20/gtk.css: No such file or directory Jun 07 13:51:29 Orion-15.openSUSE firefox[2111]: Theme parsing error: gtk-dark.css:1:50: Failed to import: Error opening file /usr/share/themes/Breeze-Dark/gtk-3.20/gtk.css: No such file or directory Jun 07 13:51:29 Orion-15.openSUSE firefox[2111]: Theme parsing error: gtk.css:68:35: The style property GtkButton:child-displacement-x is deprecated and shouldn't be used anymore. It will be removed in a future version Jun 07 13:51:29 Orion-15.openSUSE firefox[2111]: Theme parsing error: gtk.css:69:35: The style property GtkButton:child-displacement-y is deprecated and shouldn't be used anymore. It will be removed in a future version Jun 07 13:51:29 Orion-15.openSUSE firefox[2111]: Theme parsing error: gtk.css:71:36: The style property GtkCheckMenuItem:indicator-size is deprecated and shouldn't be used anymore. It will be removed in a future version Jun 07 13:51:29 Orion-15.openSUSE firefox[2111]: Theme parsing error: gtk.css:73:46: The style property GtkScrolledWindow:scrollbars-within-bevel is deprecated and shouldn't be used anymore. It will be removed in a future version Jun 07 13:51:29 Orion-15.openSUSE firefox[2111]: Theme parsing error: gtk.css:76:30: The style property GtkExpander:expander-size is deprecated and shouldn't be used anymore. It will be removed in a future version Jun 07 13:51:29 Orion-15.openSUSE plasmashell[2111]: ATTENTION: default value of option mesa_glthread overridden by environment. Jun 07 13:51:29 Orion-15.openSUSE plasmashell[2111]: ATTENTION: default value of option mesa_glthread overridden by environment. Jun 07 13:51:29 Orion-15.openSUSE plasmashell[2111]: ATTENTION: default value of option mesa_glthread overridden by environment. Jun 07 13:51:29 Orion-15.openSUSE plasmashell[2111]: ExceptionHandler::GenerateDump cloned child 2174 Jun 07 13:51:29 Orion-15.openSUSE plasmashell[2111]: ExceptionHandler::SendContinueSignalToChild sent continue signal to child Jun 07 13:51:29 Orion-15.openSUSE plasmashell[2174]: ExceptionHandler::WaitForContinueSignal waiting for continue signal... Jun 07 13:51:30 Orion-15.openSUSE plasmashell[1221]: file:///usr/share/plasma/plasmoids/org.kde.plasma.taskmanager/contents/ui/Task.qml:286: Unable to assign [undefined] to QString Jun 07 13:51:38 Orion-15.openSUSE systemd[1095]: app-firefox-e557337bebaa4a8283013fac9b8f22d0.scope: Consumed 2.750s CPU time. What other information would be useful? -- You are receiving this mail because: You are the assignee for the bug.

https://bugzilla.suse.com/show_bug.cgi?id=1212101 https://bugzilla.suse.com/show_bug.cgi?id=1212101#c4 --- Comment #4 from Paul Tannington --- (In reply to hui from comment #3)

Basics like: - output when startet from terminal?

Provides nothing over and above that shown in comment #2 from "journalctl -f" and starting FF paul@Orion-15:~$ firefox (firefox:1455): Gtk-WARNING **: 18:19:56.281: Theme parsing error: gtk-dark.css:1:50: Failed to import: Error opening file /usr/share/themes/Breeze-Dark/gtk-3.20/gtk.css: No such file or directory (firefox:1455): Gtk-WARNING **: 18:19:56.346: Theme parsing error: gtk-dark.css:1:50: Failed to import: Error opening file /usr/share/themes/Breeze-Dark/gtk-3.20/gtk.css: No such file or directory (firefox:1455): Gtk-WARNING **: 18:19:56.350: Theme parsing error: gtk.css:68:35: The style property GtkButton:child-displacement-x is deprecated and shouldn't be used anymore. It will be removed in a future version (firefox:1455): Gtk-WARNING **: 18:19:56.351: Theme parsing error: gtk.css:69:35: The style property GtkButton:child-displacement-y is deprecated and shouldn't be used anymore. It will be removed in a future version (firefox:1455): Gtk-WARNING **: 18:19:56.351: Theme parsing error: gtk.css:71:36: The style property GtkCheckMenuItem:indicator-size is deprecated and shouldn't be used anymore. It will be removed in a future version (firefox:1455): Gtk-WARNING **: 18:19:56.351: Theme parsing error: gtk.css:73:46: The style property GtkScrolledWindow:scrollbars-within-bevel is deprecated and shouldn't be used anymore. It will be removed in a future version (firefox:1455): Gtk-WARNING **: 18:19:56.351: Theme parsing error: gtk.css:76:30: The style property GtkExpander:expander-size is deprecated and shouldn't be used anymore. It will be removed in a future version ATTENTION: default value of option mesa_glthread overridden by environment. ATTENTION: default value of option mesa_glthread overridden by environment. ATTENTION: default value of option mesa_glthread overridden by environment. ExceptionHandler::GenerateDump cloned child ExceptionHandler::WaitForContinueSignal waiting for continue signal... 1506 ExceptionHandler::SendContinueSignalToChild sent continue signal to child paul@Orion-15:~$ I think the Gtk warnings about "No such file or directory" are rather a red herring, as the referenced file is not present on my Leap 15.4 systems on which FF 114 is working without this problem.

- Troubleshooting mode? https://support.mozilla.org/en-US/kb/diagnose-firefox-issues-using- troubleshoot-mode

As I wrote in the initial post, FF is crashing before it even displays it's GUI, it goes straight to the "Mozilla Crash Reporter" dialogue. Starting FF from the command line with "firefox -safe-mode" also immediately crashes. The "real" information of use is probably the crash dump file produced which is submitted to Mozilla via the "Mozilla Crash Reporter" - it's an approx 500k binary file... -- You are receiving this mail because: You are the assignee for the bug.

https://bugzilla.suse.com/show_bug.cgi?id=1212101 https://bugzilla.suse.com/show_bug.cgi?id=1212101#c11 --- Comment #11 from Paul Tannington --- (In reply to Paul Tannington from comment #10)

(In reply to Wolfgang Rosenauer from comment #8)

...

Did Firefox create a crashreport which was sent to mozilla for any of you?

Yes, also submitted.

https://crash-stats.mozilla.org/report/index/ec7a98ca-eef4-4ee7-879a-309e802... -- You are receiving this mail because: You are the assignee for the bug.

https://bugzilla.suse.com/show_bug.cgi?id=1212101 https://bugzilla.suse.com/show_bug.cgi?id=1212101#c13 Markus Elfring changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |Markus.Elfring@web.de --- Comment #13 from Markus Elfring --- (In reply to Paul Tannington from comment #0) I became also curious when the program can be properly started again after the software package “MozillaFirefox 114.0-2.1”. -- You are receiving this mail because: You are the assignee for the bug.

https://bugzilla.suse.com/show_bug.cgi?id=1212101 https://bugzilla.suse.com/show_bug.cgi?id=1212101#c15 --- Comment #15 from Paul Tannington --- (In reply to Andreas Stieger from comment #14)

Possibly https://bugzilla.mozilla.org/show_bug.cgi?id=1837201

I don't believe that is the reason for the crash reported here: I had seen that bug report, however looking at the linked (Mozilla) bug reports the reason for the crash is either "EXCEPTION_BREAKPOINT" or "EXC_BAD_ACCESS / KERN_INVALID_ADDRESS". Whereas in this (Mozilla) bug report instance, link in comment #11, all reports give the reason as "SIGILL / ILL_ILLOPN". Also, interestingly, disregarding 5 duplicated reports, 16 out of 17 list the OS as "openSUSE Tumbleweed" (the 17th being a generic "Linux"). 16/17 seems a little more than coincidence maybe? -- You are receiving this mail because: You are the assignee for the bug.

https://bugzilla.suse.com/show_bug.cgi?id=1212101 https://bugzilla.suse.com/show_bug.cgi?id=1212101#c17 --- Comment #17 from Andreas Stieger --- Confirm still crashing on start-up with 114.0.1 -- You are receiving this mail because: You are the assignee for the bug.

https://bugzilla.suse.com/show_bug.cgi?id=1212101 https://bugzilla.suse.com/show_bug.cgi?id=1212101#c19 --- Comment #19 from Paul Tannington --- Let's get the ball rolling again... Submitted Crash Report to Mozilla: https://crash-stats.mozilla.org/report/index/78361b89-e993-4de7-a59e-d808a02... As with the 114.0 crashes, the reason given in the Crash Report is "SIGILL / ILL_ILLOPN" -- You are receiving this mail because: You are the assignee for the bug.

https://bugzilla.suse.com/show_bug.cgi?id=1212101 https://bugzilla.suse.com/show_bug.cgi?id=1212101#c20 Jazz changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |jayjayjazz@gmail.com --- Comment #20 from Jazz --- Also happening on my Tumbleweed machine: https://crash-stats.mozilla.org/report/index/699d1007-3129-4c78-b137-6bc3002... - KDE: yes - nvidia: no -- You are receiving this mail because: You are the assignee for the bug.

https://bugzilla.suse.com/show_bug.cgi?id=1212101 https://bugzilla.suse.com/show_bug.cgi?id=1212101#c21 Nikolai Nikolaevskii changed: What |Removed |Added ---------------------------------------------------------------------------- Flags| |needinfo? CC| |kaykaykay123@gmail.com --- Comment #21 from Nikolai Nikolaevskii --- Users with troubles - please try to use Firefox tar from Mozilla: https://www.mozilla.org/en-US/firefox/all/#product-desktop-release I have some suspicions. FF 114 & 114.0.1 works OK on Leap 15.4. -- You are receiving this mail because: You are the assignee for the bug.

https://bugzilla.suse.com/show_bug.cgi?id=1212101 https://bugzilla.suse.com/show_bug.cgi?id=1212101#c24 Paul Tannington changed: What |Removed |Added ---------------------------------------------------------------------------- Flags|needinfo? | --- Comment #24 from Paul Tannington --- (In reply to Paul Tannington from comment #22)

(In reply to Nikolai Nikolaevskii from comment #21)

...

Users with troubles - please try to use Firefox tar from Mozilla: https://www.mozilla.org/en-US/firefox/all/#product-desktop-release ...

Tried that already, FF provided by "firefox-114.0.1.tar.bz2" still crashes in the same manner on my TW system.

... Further to that, I've just retried once more to confirm: Using "https://ftp.mozilla.org/pub/firefox/releases/114.0.1/linux-x86_64/en-GB/fire..." (MD5: 4ab794c5200c65f67cab6f9db1340d78) FF Immediately crashes at start-up in the same manner. Mozilla Crash Report: https://crash-stats.mozilla.org/report/index/81f45207-9b1e-44d8-b727-7594a02... (In reply to Martin Sirringhaus from comment #23)

I just contacted upstream, and they don't have the symbols for those builds yet. Which is why the crash-reports are more or less empty (all red).

They'll scrape them now and hopefully are able to reprocess the crash-reports to contain actual stack traces and proper info, soon.

Excellent, many thanks. We will await the outcome. -- You are receiving this mail because: You are the assignee for the bug.

https://bugzilla.suse.com/show_bug.cgi?id=1212101 https://bugzilla.suse.com/show_bug.cgi?id=1212101#c26 Ilgaz Öcal changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |ilgaz@ilgaz.gen.tr --- Comment #26 from Ilgaz Öcal --- I have Tumbleweed installed on Powerbook 5.1 (core2 duo/nvidia 9400/nouveau) and HP i5/HD5500. I run zypper dup on both systems. On core2 duo/nv9400 openSUSE Firefox 114.0.1-1.1 crashes (wayland/kde) however stock Firefox 114.0.1 from getfirefox.com runs fine. Stock Firefox says: "ATTENTION: default value of option mesa_glthread overridden by environment." On i5/Intel HD5500, openSUSE Tumbleweed stock Firefox launched fine. -- You are receiving this mail because: You are the assignee for the bug.

https://bugzilla.suse.com/show_bug.cgi?id=1212101 https://bugzilla.suse.com/show_bug.cgi?id=1212101#c28 Andreas Stieger changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |sachse@gmx.net --- Comment #28 from Andreas Stieger --- *** Bug 1212384 has been marked as a duplicate of this bug. *** -- You are receiving this mail because: You are the assignee for the bug.

https://bugzilla.suse.com/show_bug.cgi?id=1212101 Andreas Stieger changed: What |Removed |Added ---------------------------------------------------------------------------- Severity|Critical |Major Priority|P2 - High |P5 - None -- You are receiving this mail because: You are the assignee for the bug.

https://bugzilla.suse.com/show_bug.cgi?id=1212101 https://bugzilla.suse.com/show_bug.cgi?id=1212101#c33 Victor hck changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |victorhck@opensuse.org --- Comment #33 from Victor hck --- Same issue here after update Tumbleweed+Plasma yesterday. I tried also run firefox from Xfce desktop and get the same issue. My machine specification here: https://paste.opensuse.org/pastes/416b637cd3d8 Error info when I run firefox or firefox -safe-mode command form terminal: ExceptionHandler::GenerateDump cloned child 5285 ExceptionHandler::SendContinueSignalToChild sent continue signal to child ExceptionHandler::WaitForContinueSignal waiting for continue signal... Report content that I send to Mozilla: AdapterDeviceID: 0x0106 AdapterDriverVendor: mesa/crocus AdapterDriverVersion: 23.1.2.0 AdapterVendorID: 0x8086 AvailablePageFile: 752132096 AvailablePhysicalMemory: 2446344192 AvailableSwapMemory: 4292866048 AvailableVirtualMemory: 4856406016 BackgroundTaskMode: 0 BuildID: 20230608214645 CrashTime: 1686823682 DesktopEnvironment: kde HeadlessMode: 0 InstallTime: 1686773525 IsWayland: 0 Notes: FP(D00-L1000-W0000000-T010) WR? WR+ ProductID: {ec8030f7-c20a-464f-9b0e-13a3a9e97384} ProductName: Firefox ReleaseChannel: release SafeMode: 0 SecondsSinceLastCrash: 44 StartupCrash: 1 StartupTime: 1686823680 SubmittedFrom: Client Throttleable: 1 TotalPageFile: 12417294336 TotalPhysicalMemory: 8124428288 UptimeTS: 1.65052749 Vendor: Mozilla Version: 114.0.1 This report also contains technical information about the state of the application when it crashed. -- You are receiving this mail because: You are the assignee for the bug.

https://bugzilla.suse.com/show_bug.cgi?id=1212101 https://bugzilla.suse.com/show_bug.cgi?id=1212101#c35 --- Comment #35 from Nikolai Nikolaevskii --- ILL as gcc or code optimiser have changed instruction. With that change code works faster on compatible CPUs by using AVX, avoiding penalties for hopping between ordinary and VEX coding scheme https://en.wikipedia.org/wiki/VEX_prefix -- You are receiving this mail because: You are the assignee for the bug.

https://bugzilla.suse.com/show_bug.cgi?id=1212101 https://bugzilla.suse.com/show_bug.cgi?id=1212101#c40 --- Comment #40 from Paul Tannington --- (In reply to Wolfgang Rosenauer from comment #38)

Could some people here try https://download.opensuse.org/repositories/mozilla/openSUSE_Tumbleweed/ x86_64/MozillaFirefox-114.0.1-3.2.x86_64.rpm either by direct download or using the mozilla repository?

There should be a small change in compile optimizations in there which could make a difference but as I do not have any machine which seems to be affected I have no idea if it does. Please let us know.

Version : 114.0.1-3.2 Build Time : Sat 17 Jun 2023 02:01:52 BST Install Time : Sat 17 Jun 2023 08:50:20 BST paul@Orion-15:~$ firefox -d gdb ... ... Thread 1 "firefox" received signal SIGILL, Illegal instruction. skvx::Vec<4, float>::VecStorage () at /usr/src/debug/firefox-114.0.1/gfx/skia/skia/src/base/SkVx.h:274 274 using VecStorage::VecStorage; (gdb) bt 10 #0 skvx::Vec<4, float>::VecStorage () at /usr/src/debug/firefox-114.0.1/gfx/skia/skia/src/base/SkVx.h:274 #1 0x00007ffff01a095b in operator*<4, float, float> () at /usr/src/debug/firefox-114.0.1/gfx/skia/skia/src/base/SkVx.h:539 #2 map_rect_affine () at /usr/src/debug/firefox-114.0.1/gfx/skia/skia/src/core/SkM44.cpp:154 #3 0x00007ffff01a1851 in SkMatrixPriv::MapRect () at /usr/src/debug/firefox-114.0.1/gfx/skia/skia/src/core/SkM44.cpp:220 #4 SkCanvas::computeDeviceClipBounds () at /usr/src/debug/firefox-114.0.1/gfx/skia/skia/src/core/SkCanvas.cpp:1709 #5 0x00007ffff01a24a4 in SkCanvas::init () at /usr/src/debug/firefox-114.0.1/gfx/skia/skia/src/core/SkCanvas.cpp:425 #6 0x00007ffff00ec8ce in SkCanvas::SkCanvas () at /usr/src/debug/firefox-114.0.1/gfx/skia/skia/src/core/SkCanvas_Raster.cpp:31 #7 SkSurface_Raster::onNewCanvas () at /usr/src/debug/firefox-114.0.1/gfx/skia/skia/src/image/SkSurface_Raster.cpp:76 #8 0x00007fffed94bc9e in SkSurface_Base::getCachedCanvas () at /usr/src/debug/firefox-114.0.1/gfx/skia/skia/src/image/SkSurface_Base.h:210 #9 SkSurface::getCanvas () at /usr/src/debug/firefox-114.0.1/gfx/skia/skia/src/image/SkSurface.cpp:80 (More stack frames follow...) ... ... -- You are receiving this mail because: You are the assignee for the bug.

https://bugzilla.suse.com/show_bug.cgi?id=1212101 Lee Seymour changed: What |Removed |Added ---------------------------------------------------------------------------- Alias| |fleamour CC| |gtx.swift@gmail.com -- You are receiving this mail because: You are the assignee for the bug.

https://bugzilla.suse.com/show_bug.cgi?id=1212101 https://bugzilla.suse.com/show_bug.cgi?id=1212101#c45 --- Comment #45 from Christopher Yeleighton --- [ 2121s] + echo 'file /home/abuild/rpmbuild/BUILD/obj/toolkit/library/build/libxul.so [ 2121s] disassemble '\''skvx::Vec<4, float>::VecStorage(float)'\'' [ 2121s] quit' [ 2121s] + gdb -batch -x no-avx.gdb [ 2127s] + grep -F vshufps VecStorage.s [ 2127s] 0x0000000003cbfda0 <+0>: vshufps $0x0,%xmm0,%xmm0,%xmm0 [ 2127s] + false -- You are receiving this mail because: You are the assignee for the bug.

https://bugzilla.suse.com/show_bug.cgi?id=1212101 https://bugzilla.suse.com/show_bug.cgi?id=1212101#c47 --- Comment #47 from Paul Tannington --- Quote from post 10 of mozilla bug report https://bugzilla.mozilla.org/show_bug.cgi?id=1838323 "My offhand guess as to why our official builds work and these distro builds don't is they're building with AVX flags in the binary. This causes SkVx to include them in the build and use them regardless. Official builds on x86 are only built with SSE2, with higher SIMD levels only explicitly enabled selectively in various files. I would encourage the distro build maintainers to fix this on their end by not supplying AVX flags to the build. Acceleration for higher levels of SIMD is still used in Mozilla builds, but again, we only selectively enable that in various parts of the moz.build where it is actually safe and properly gated by CPU feature-level checks. There is no nice way to work around this inside our code, but the downstream build fixes on the distro end should be simple." -- You are receiving this mail because: You are the assignee for the bug.

https://bugzilla.suse.com/show_bug.cgi?id=1212101 https://bugzilla.suse.com/show_bug.cgi?id=1212101#c55 Edward Davis changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |ed.davis@davishome.org --- Comment #55 from Edward Davis --- (In reply to Wolfgang Rosenauer from comment #54)

There is another build with a different build change: https://download.opensuse.org/repositories/mozilla/openSUSE_Tumbleweed/ x86_64/MozillaFirefox-114.0.1-5.1.x86_64.rpm

Please try and report if it changes anything.

I upgraded my laptop to this version and it opens OK now. My laptop is an ASUS with an Intel(R) Celeron(R) N4500 @ 1.10GHz Running tumbleweed:20230617 and firefox failed to open with the 114 update from the other day. Thanks, Edward -- You are receiving this mail because: You are the assignee for the bug.

https://bugzilla.suse.com/show_bug.cgi?id=1212101 https://bugzilla.suse.com/show_bug.cgi?id=1212101#c59 --- Comment #59 from OBSbugzilla Bot --- This is an autogenerated message for OBS integration: This bug (1212101) was mentioned in https://build.opensuse.org/request/show/1093874 Factory / MozillaFirefox -- You are receiving this mail because: You are the assignee for the bug.

https://bugzilla.suse.com/show_bug.cgi?id=1212101 https://bugzilla.suse.com/show_bug.cgi?id=1212101#c74 --- Comment #74 from Victor hck --- (In reply to Wolfgang Rosenauer from comment #54)

There is another build with a different build change: https://download.opensuse.org/repositories/mozilla/openSUSE_Tumbleweed/ x86_64/MozillaFirefox-114.0.1-5.1.x86_64.rpm

Please try and report if it changes anything.

Downloades the rpm with wget and updated with sudo rpm -U MozillaFirefox-114.0.1-5.1.x86_64.rpm and now works fine again! Thanks a lot for your work! Greetings! -- You are receiving this mail because: You are the assignee for the bug.

https://bugzilla.suse.com/show_bug.cgi?id=1212101 https://bugzilla.suse.com/show_bug.cgi?id=1212101#c76 --- Comment #76 from Christopher Yeleighton --- (In reply to Jiri Slaby from comment #46)

(In reply to Christopher Yeleighton from comment #45)

...

[ 2121s] + echo 'file /home/abuild/rpmbuild/BUILD/obj/toolkit/library/build/libxul.so [ 2121s] disassemble '\''skvx::Vec<4, float>::VecStorage(float)'\'' [ 2121s] quit' [ 2121s] + gdb -batch -x no-avx.gdb [ 2127s] + grep -F vshufps VecStorage.s [ 2127s] 0x0000000003cbfda0 <+0>: vshufps $0x0,%xmm0,%xmm0,%xmm0

This should be a "x86-64-v3" library residing in /usr/lib64/glibc-hwcaps/x86-64-v3/. Something is wrong with how libxul is built.

There is nothing wrong, I just checked the result in the build tree. -- You are receiving this mail because: You are the assignee for the bug.

https://bugzilla.suse.com/show_bug.cgi?id=1212101 https://bugzilla.suse.com/show_bug.cgi?id=1212101#c78 --- Comment #78 from Christopher Yeleighton --- (In reply to Richard Biener from comment #70)

Now I don't think we "clone" the comdats that were built with different ISA flags to make them prevail. But I might be wrong. I know we promote them local when possible, but I don't think that helps here since it's the linker deciding which one prevails. I also don't think we diagnose mismatched optimization/target flags on cgraph nodes we merge.

Upstream could solve this problem by putting AVX-enabled instantiations into a separate namespace. -- You are receiving this mail because: You are the assignee for the bug.

https://bugzilla.suse.com/show_bug.cgi?id=1212101 https://bugzilla.suse.com/show_bug.cgi?id=1212101#c82 Lee Salzman changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |lsalzman@mozilla.com Flags| |needinfo?(rguenther@suse.co | |m) --- Comment #82 from Lee Salzman --- (In reply to Richard Biener from comment #79)

Btw, the code actually instantiates skvx::Vec<> stuff via

namespace avx {

template <typename T> static void memsetT(T buffer[], T value, int count) {

static constexpr int N = 32 / sizeof(T);

static_assert(N > 0, "T is too big for memsetT");

skvx::Vec wideValue(value); while (count >= N) {

wideValue.store(buffer); buffer += N; count -= N; }

while (count --> 0) { *buffer++ = value; } }

inline void memset16(uint16_t buffer[], uint16_t value, int count) { memsetT(buffer, value, count); } ... }

and in SkOpts_avx.cpp:

namespace SkOpts { void Init_avx() { memset16 = avx::memset16; ...

and it expects everything to be fully optimized, eliminating all used template instantiations. But at the time we stream out for LTO that does not have happened.

So I don't see how we can avoid this issue besides inventing some "DWIM" mechanism here.

I would suggest to add -fno-lto to all TUs that are built with extra machine specific flags as workaround.

Richard, would just adding -fno-lto to the flags for the SkOpts_foo.cpp files be sufficient to work around this on your end for GCC builds? I am willing to add this to Firefox's moz.builds to just make this easy. If GCC will let us selectively disable LTO on certain files like that, I am on board. -- You are receiving this mail because: You are the assignee for the bug.

https://bugzilla.suse.com/show_bug.cgi?id=1212101 https://bugzilla.suse.com/show_bug.cgi?id=1212101#c86 --- Comment #86 from OBSbugzilla Bot --- This is an autogenerated message for OBS integration: This bug (1212101) was mentioned in https://build.opensuse.org/request/show/1097045 Factory / gcc13 -- You are receiving this mail because: You are the assignee for the bug.