firefox-115.9.0 - Assertion failure: i < NUM_TEST_EV...

[lyallp]

The only reason I am building chromium is because firefox is failing on my system, despite my best efforts, I cannot get it to run without crashing behind the scenes.

This includes masking the two latest versions and completely re-building my system, including kernel.

Split off of Chromium build time 36 hours?? after post 8826729. —Chiitoo

[lyallp]

I gave seamonkey a really quick try - no go, it does not work with the nextcloud login, possibly I didn't have some setting set, but I could not be bothered checking.

[eccerr0r]

firefox-bin also not working?

I haven't had a machine not run firefox properly in quite a while... or at least I don't use any websites that fail on the multiple machines...

[lyallp]

firefox-bin works fine, damnit!

Re-building www-client/firefox-115.9.0, maybe having firefox-bin work on my profile has 'fixed' something in my profile?

I doubt this because I have tried resetting firefox and creating a new profile, rebuilding the system top to bottom, masking a couple of the later releases, on the assumption that it worked before.

I have just updated the kernel to 6.6.30, re-installed nvidia-drivers (as usual) and rebooted, so I will report back.

Obviously, I wold prefer firefox to chromium as it's 6 hours less compilation time.

I also prefer to build my own, rather than use -bin packages, don't ask me why, I just do... It's why I started with Gentoo just over 20 years ago!

[lyallp]

Firefox-115.9.0 does not work, at least for me....
sigh....

Code: Select all

Initializing context 0x7fd355e0e3a1 surface (nil) on display 0x7fd347ada400
GL_VENDOR: NVIDIA Corporation
mVendor: NVIDIA
GL_RENDERER: NVIDIA GeForce RTX 3060/PCIe/SSE2
mRenderer: Unknown
mIsMesa: 0
[Parent 10277, Main Thread] WARNING: 'mIndex >= Count()', file /tmp/portage/www-client/firefox-115.9.0/work/firefox-115.9.0/xpcom/ds/nsStringEnumerator.cpp:202
[Parent 10277, Main Thread] WARNING: 'mIndex >= Count()', file /tmp/portage/www-client/firefox-115.9.0/work/firefox-115.9.0/xpcom/ds/nsStringEnumerator.cpp:202
[Parent 10277, Main Thread] WARNING: 'mIndex >= Count()', file /tmp/portage/www-client/firefox-115.9.0/work/firefox-115.9.0/xpcom/ds/nsStringEnumerator.cpp:202
[Socket 10469, Main Thread] WARNING: 'NS_FAILED(rv)', file /tmp/portage/www-client/firefox-115.9.0/work/firefox-115.9.0/netwerk/protocol/http/nsHttpHandler.cpp:371
[Child 10513, Main Thread] WARNING: Fallback to FallbackRenderer: file /tmp/portage/www-client/firefox-115.9.0/work/firefox-115.9.0/dom/ipc/BrowserChild.cpp:2688
[Child 10513, Main Thread] WARNING: Fallback to FallbackRenderer: file /tmp/portage/www-client/firefox-115.9.0/work/firefox-115.9.0/dom/ipc/BrowserChild.cpp:2688
[Child 10513, Main Thread] WARNING: Fallback to FallbackRenderer: file /tmp/portage/www-client/firefox-115.9.0/work/firefox-115.9.0/dom/ipc/BrowserChild.cpp:2688
[Child 10513, Main Thread] WARNING: Fallback to FallbackRenderer: file /tmp/portage/www-client/firefox-115.9.0/work/firefox-115.9.0/dom/ipc/BrowserChild.cpp:2688
[Child 10513, Main Thread] WARNING: Fallback to FallbackRenderer: file /tmp/portage/www-client/firefox-115.9.0/work/firefox-115.9.0/dom/ipc/BrowserChild.cpp:2688
[WARN  webrender::device::gl] Missing optimized shader source for gpu_cache_update
[Parent 10277, Main Thread] WARNING: NS_ENSURE_TRUE(mPresShell) failed: file /tmp/portage/www-client/firefox-115.9.0/work/firefox-115.9.0/layout/generic/nsFrameSelection.cpp:1711
Assertion failure: i < NUM_TEST_EV_ROOTS (Could not find built-in EV root), at /tmp/portage/www-client/firefox-115.9.0/work/firefox-115.9.0/security/certverifier/ExtendedValidation.cpp:1282
#01: ???[/usr/lib64/firefox/libxul.so +0xfda1d48]
#02: ???[/usr/lib64/firefox/libxul.so +0xfe4294d]
#03: ???[/usr/lib64/firefox/libxul.so +0x6e98ae9]
#04: ???[/usr/lib64/firefox/libxul.so +0x6e8d7cd]
#05: ???[/usr/lib64/firefox/libxul.so +0x6e95c5a]
#06: ???[/usr/lib64/firefox/libxul.so +0x7c87b7f]
#07: ???[/usr/lib64/firefox/libxul.so +0x7ba919c]
#08: ???[/usr/lib64/firefox/libxul.so +0x7ba9119]
#09: ???[/usr/lib64/firefox/libxul.so +0x7ba90d6]
#10: ???[/usr/lib64/firefox/libxul.so +0x6e87cb4]
#11: ???[/usr/lib64/libnspr4.so +0x29273]
#12: ???[/usr/lib64/libc.so.6 +0x88341]
#13: ???[/usr/lib64/libc.so.6 +0xfae1c]
#14: ??? (???:???)

Program /usr/lib64/firefox/firefox (pid = 10277) received signal 11.
Stack:
#01: ???[/usr/lib64/firefox/libxul.so +0x10383102]
#02: ???[/usr/lib64/firefox/libxul.so +0x11c3cf51]
#03: ???[/usr/lib64/libc.so.6 +0x3a2f0]
#04: ???[/usr/lib64/firefox/libxul.so +0xfda1d56]
#05: ???[/usr/lib64/firefox/libxul.so +0xfe4294d]
#06: ???[/usr/lib64/firefox/libxul.so +0x6e98ae9]
#07: ???[/usr/lib64/firefox/libxul.so +0x6e8d7cd]
#08: ???[/usr/lib64/firefox/libxul.so +0x6e95c5a]
#09: ???[/usr/lib64/firefox/libxul.so +0x7c87b7f]
#10: ???[/usr/lib64/firefox/libxul.so +0x7ba919c]
#11: ???[/usr/lib64/firefox/libxul.so +0x7ba9119]
#12: ???[/usr/lib64/firefox/libxul.so +0x7ba90d6]
#13: ???[/usr/lib64/firefox/libxul.so +0x6e87cb4]
#14: ???[/usr/lib64/libnspr4.so +0x29273]
#15: ???[/usr/lib64/libc.so.6 +0x88341]
#16: ???[/usr/lib64/libc.so.6 +0xfae1c]
#17: ??? (???:???)
Sleeping for 300 seconds.

[e8root]

Did you try to build without hwaccel USE flag?
What are the USE flags anyways?

Mine are

Code: Select all

USE="X clang jumbo-build lto openh264 pgo pulseaudio system-av1 system-harfbuzz system-icu system-jpeg system-libevent system-libvpx system-png system-webp wayland
-dbus -debug -eme-free -geckodriver -gmp-autoupdate -hardened -hwaccel -jack -libproxy (-selinux) -sndio -telemetry (-valgrind) -wifi"

For reference Seamonkey

Code: Select all

USE="chatzilla custom-cflags custom-optimization dbus gmp-autoupdate ipc pulseaudio startup-notification system-av1 system-harfbuzz system-icu system-jpeg system-libevent system-libvpx system-png system-sqlite webrtc
-jack (-lto) (-selinux) -test -wifi"

Regarding Seamonkey its just classic open source Mozilla Browser with modern developments pulled from Firefox - apparently only things which are actually needed to run modern websites and not bloatware. It supports classic extensions only and lots of other classic stuff including rendering modes which on Windows allows it to be used with MacType to get Linux's FreeType2 font rendering. I'll need to compare it (read: pixel peep) under Linux itself to see if there is any difference between these things if any (might be only related to Windows) and maybe do it with Windows without MacType as even with modern Win11 ClearType (which isn't that bad) it might look better than DirectWrite - because DirectWrite allows for subpixel font precision which doesn't play well with hinting. Win11 uses less hinting that what it used to in the past...

...anyways, Seamonkey is modern Mozilla Browser that builds in a fraction of time it takes Firefox to build which itself takes fraction of time it takes to build Chromium.

There is less choice for extensions unfotunately but there is modern build of uBlock so its fine https://github.com/UCyborg/uBlock-for-f ... y/releases

BTW. I had blurry fonts on Chromium vs FF/Seamonkey and even Chrome but I tinkered with flags and resolves the issue.
Flags needed were:
Vulkan - Enabled
Default ANGLE Vulkan - Enabled
Vulkan fron ANGLE - Disabled
Smooth Scrolling - Disabled - oh wait... not it but still needed to use Chromium
Now from what I can see text looks the same as on FF/Seamonkey. Version www-client/chromium-124.0.6367.155::gentoo

[lyallp]

Code: Select all

Installed versions:  115.9.0(esr)(16:42:25 15/05/24)(X clang dbus debug gmp-autoupdate openh264 pulseaudio screencast system-harfbuzz system-icu system-jpeg system-libevent system-libvpx system-png system-webp wayland -eme-free -geckodriver -hardened -hwaccel -jack -libproxy -lto -pgo -selinux -sndio -system-av1 -system-python-libs -wifi CPU_FLAGS_ARM="-neon" L10N="-ach -af -an -ar -ast -az -be -bg -bn -br -bs -ca -ca-valencia -cak -cs -cy -da -de -dsb -el -en-CA -en-GB -eo -es-AR -es-CL -es-ES -es-MX -et -eu -fa -ff -fi -fr -fur -fy -ga -gd -gl -gn -gu -he -hi -hr -hsb -hu -hy -ia -id -is -it -ja -ka -kab -kk -km -kn -ko -lij -lt -lv -mk -mr -ms -my -nb -ne -nl -nn -oc -pa -pl -pt-BR -pt-PT -rm -ro -ru -sc -sco -si -sk -sl -son -sq -sr -sv -szl -ta -te -th -tl -tr -trs -uk -ur -uz -vi -xh -zh-CN -zh-TW")

[lyallp]

Woot!

www-client/firefox-115.10.0 works!

There are multiple use flag changes in this build, including the removal of

• debug
• clang

and the addition of

• system-av1

The added bonus is Firefox only takes 25 minutes to build, rather than 9 hours which Chromium takes to build!

[Chiitoo]

Split off of Chromium build time 36 hours?? after post 8826729. —Chiitoo

Looking at the 'Assertion failure: i < NUM_TEST_EV_ROOTS', I suspect the 'debug' USE-flag was in play indeed, but it seems the trace has no useful info:

Code: Select all

debug            : Enable extra debug codepaths, like asserts and extra output. If you want to get meaningful backtraces see
                        https://wiki.gentoo.org/wiki/Project:Quality_Assurance/Backtraces

This flag is generally not meant for what you probably were thinking (I know it seems like an obvious thing to do when trying to troubleshoot something).

Some ebuilds provide a debug USE flag. While some mistakenly use it to provide debug information and play with compiler flags when it is enabled, that is not its purpose.

If you're trying to debug a reproducible crash, you want to leave this USE flag alone, as it'll be building a different source than what you had before. It is more efficient to get first a backtrace without changing the code, by simply emitting symbol information, and just afterward enable debug features to track the issue further down.

Debug features that are enabled by the USE flag include assertions, debug logs on screen, debug files, leak detection and extra-safe operations (such as scrubbing memory before use). Some of them might be taxing, especially for complex software or software where performance is an important issue.

For these reasons, please exercise caution when enabling the debug USE flag, and only consider it a last-chance card.

[lyallp]

Yeah, I take that on board.

I was trying to diagnose and generate decent back traces for Firefox, after it was generating about 10-15 core dumps per day. I tried the debug flag Plus I changed the CFLAGS and LDFLAGS to generate debug information and rebuilt my entire system, in the hope of having a good stack trace. This, in turn caused compile times to get significantly worse. Then, firefox stopped working and I ended up building Chromium....A cascade of cock ups...

Curiously enough, initial problem seems to have gone away, as an almost guaranteed way of generating a failure was to have 2 videos running at the same time in different windows.

I tried this now and firefox didn't crash.

So, somewhere along the line, the original problem has gone away after weeks of trying to get firefox to even run, let alone crash.

[Chiitoo]

Did any of those builds actually have symbol names and such in the dumps before going for USE="debug"?

Could shed more light on what was going on, but it's indeed great that things got back into working order!

[lyallp]

My first point of call was to use gdb on the original dumps, very little info.

So, then, I updated /etc/portage/make.conf to have CFLAGS include -ggdb3 and exclude any -O2 or -O3

I also then updated LDFLAGS to include -g

Then, I emerge --empty-tree --newuse @system @world

I expected a minor performance penalty for this.

Core dumps still did not give a huge amount of detail. I figured I had missed something like 'strip' in the build process.

I still had no joy, then I tried the 'debug' use flag, that's when things broke.

I tried masking the latest couple of versions of Firefox, thinking it was working before, maybe it was an update, to no avail.

I lost sight of the fact that I had set the 'debug' USE flag.

I then tried to build chromium as an alternative browser, whilst I continued looking at firefox, but after 3 and a half DAYS of compilation, I gave up.

It was suggested that global debug symbols where slowing down compilation, so I reverted out the CFLAG/LDFLAG updates and did another emerge --empty-tree --newuse @system @world, which took 30 hours

A rebuild of firefox and it still failed.

All of this happened at the same time as grub changed (root= parm changed content), my automated kernel build and install script screwed my boot, and I had to figure out why my system was not booting, so, I was attempting to get my system booting again and build chromium and fix firefox... sigh...

I then updated my firefox use flags, as described, and that's when things started working, no more core files (yet). I think there has been a firefox update since this whole episode began but the update happened after I had enabled the debug USE flag, which was my catastrophe point.

[Chiitoo]

Quite the adventure fer sure.

My first point of call was to use gdb on the original dumps, very little info.

So, then, I updated /etc/portage/make.conf to have CFLAGS include -ggdb3 and exclude any -O2 or -O3

Generally I think this should be enough (not forgetting CXXFLAGS and '-nostrip' feature), and they should only be needed for the package where the symbols are wanted.

For example, if a Qt 5 application crashes showing some qtgui or/and qtcore files in the backtrace, then only the packages providing those will need to be built with the debug build configuration (still no USE="debug" needed for this).

I don't think I ever did this with Firefox personally, so not sure what could have been missing.

If it happens again, I'd definitely recommend adding a post here (or maybe a new topic), so we might get to know if there is a certain combination of things that causes it.

Hopefully it will stay away though!