[SOLVED] Pale Moon crashes when I try to log in to Yahoo
Linux - SoftwareThis forum is for Software issues.
Having a problem installing a new program? Want to know which application is best for the job? Post your question in this forum.
Notices
Welcome to LinuxQuestions.org, a friendly and active Linux Community.
You are currently viewing LQ as a guest. By joining our community you will have the ability to post topics, receive our newsletter, use the advanced search, subscribe to threads and access many other special features. Registration is quick, simple and absolutely free. Join our community today!
Note that registered members see fewer ads, and ContentLink is completely disabled once you log in.
If you have any problems with the registration process or your account login, please contact us. If you need to reset your password, click here.
Having a problem logging in? Please visit this page to clear all LQ-related cookies.
Get a virtual cloud desktop with the Linux distro that you want in less than five minutes with Shells! With over 10 pre-installed distros to choose from, the worry-free installation life is here! Whether you are a digital nomad or just looking for flexibility, Shells can put your Linux machine on the device that you want to use.
Exclusive for LQ members, get up to 45% off per month. Click here for more info.
I have a Q33 chip, which (according to Gentoo) is second generation and needs i915. I tried following the Gentoo link, but I didn't really understand what I was doing and the kernel build crashed. I don't think I'll go any further down that route.
There are some other things I'd like to try, like building Pale Moon in a different distro. I have both Crux and Debian on this machine. Maybe I can get the build to go to completion in one of them.
@Emerson
Could you please build me a Pale Moon binary without system cairo, i.e. one that uses its own built-in cairo library. According to the LFS devs, FF has a bug that sometimes causes it to crash during rendering when it uses the system cairo. It's possible that PM has the same bug. Worth trying anyway.
Looking at your .mozconfig system cairo already was disabled, this is what I used for last build:
Code:
# If you have a multicore machine, all cores will be used by default.
# If desired, you can reduce the number of cores used, e.g. to 1, by
# uncommenting the next line and setting a valid number of CPU cores.
#mk_add_options MOZ_MAKE_FLAGS="-j1"
# If you have installed dbus-glib, comment out this line:
ac_add_options --disable-dbus
# If you have installed dbus-glib, and you have installed (or will install)
# wireless-tools, and you wish to use geolocation web services, comment out
# this line
ac_add_options --disable-necko-wifi
# API Keys for geolocation APIs - necko-wifi (above) is required for MLS
# Uncomment the following line if you wish to use Mozilla Location Service
#ac_add_options --with-mozilla-api-keyfile=$PWD/mozilla-key
# Uncomment the following line if you wish to use Google's geolocaton API
# (needed for use with saved maps with Google Maps)
#ac_add_options --with-google-api-keyfile=$PWD/google-key
# Uncomment these lines if you have installed optional dependencies
#ac_add_options --enable-system-hunspell
#ac_add_options --enable-startup-notification
# Uncomment the following option if you have not installed PulseAudio
ac_add_options --disable-pulseaudio
# and uncomment this if you installed alsa-lib instead of PulseAudio
ac_add_options --enable-alsa
# If you have installed GConf, comment out this line
ac_add_options --disable-gconf
# Comment out following options if you have not installed
# recommended dependencies:
ac_add_options --enable-system-sqlite
#ac_add_options --with-system-libevent
#ac_add_options --with-system-libvpx
ac_add_options --with-system-nspr
ac_add_options --with-system-nss
#ac_add_options --with-system-icu
# If you are going to apply the patch for system graphite
# and system harfbuzz, uncomment these lines:
#ac_add_options --with-system-graphite2
#ac_add_options --with-system-harfbuzz
# Stripping is now enabled by default.
# Uncomment these lines if you need to run a debug
#ac_add_options --disable-strip
#ac_add_options --disable-install-strip
# The BLFS editors recommend not changing anything below this line:
ac_add_options --prefix=/usr
ac_add_options --enable-application=browser
ac_add_options --disable-crashreporter
ac_add_options --disable-updater
ac_add_options --disable-tests
# Optimization for size is broken with gcc7
ac_add_options --enable-optimize="-O2"
ac_add_options --enable-official-branding
ac_add_options --enable-safe-browsing
ac_add_options --enable-url-classifier
# From firefox-40, using system cairo causes firefox to crash
# frequently when it is doing background rendering in a tab.
#ac_add_options --enable-system-cairo
ac_add_options --enable-system-ffi
ac_add_options --enable-system-pixman
ac_add_options --with-pthreads
ac_add_options --with-system-bz2
# ac_add_options --with-system-jpeg
ac_add_options --with-system-png
ac_add_options --with-system-zlib
mk_add_options MOZ_OBJDIR=@TOPSRCDIR@/firefox-build-dir
NEWS: I just tried running the most recent PM that Emerson built for me in Crux. Of course it then links to the Crux libraries, not the LFS ones. And I got Yahoo and was able to log in successfully! No crash, but I got some warning messages as shown below:
libGL error: failed to open drm device: Permission denied
libGL error: failed to load driver: i915
ATTENTION: default value of option force_s3tc_enable overridden by environment.
The third one is familiar but the first two are not. Possibly these libGL functions which don't work in Crux were causing the problem in LFS. What do you think?
I've been out most of the day. No, I wasn't a member of the video group when I did that first test. I remembered that while I was walking the dog, so I added myself after I came home, then tried again. This time the GL warning messages didn't occur but I could still log into Yahoo with PM. So that wasn't the deciding factor.
On Crux, Xorg uses the intel-915 driver. modesetting unloads itself for some reason:
Code:
grep -i modesetting /var/log/Xorg.0.log
[ 45.288] (==) Matched modesetting as autoconfigured driver 2
[ 45.449] (II) LoadModule: "modesetting"
[ 45.449] (II) Loading /usr/lib/xorg/modules/drivers/modesetting_drv.so
[ 45.454] (II) Module modesetting: vendor="X.Org Foundation"
[ 45.523] (II) modesetting: Driver for Modesetting Kernel Drivers: kms
[ 45.610] (WW) Falling back to old probe method for modesetting
[ 45.613] (II) UnloadModule: "modesetting"
[ 45.613] (II) Unloading modesetting
Another thing we may have to consider is that I have FF installed on Crux, and PM was linking to some of the FF libraries in /usr/lib (for example libmozalloc). I'm surprised they were compatible, because the PM team have largely rewritten the program; it's not really a variant of FF any more. But I suppose there is a possibility that they are more reliable than PM's own.
But one thing we do know now: Your version of PM is probably OK. It's a problem with the environment.
I'm going to rebuild LFS mesa3d with the software rasteriser in place. It means building llvm, which takes forever, but I need to make sure that I've covered all bases.
You may want to test with javascript disabled, it may be a workaround for Yahoo. I'm sure there are extensions for Pale Moon for this.
Yes, they have their own version of noscript, but I've already tried with this in place and it doesn't help.
There's a version of PM for Crux, in one of the private repositories. I'm going to see if I can build it there. If it works, I might be able to port it to LFS, using the same mozconfig file. Crux uses gcc version 6 compared with LFS's version 7.
I'd experiment with modesetting/intel drivers. One of them may work. The driver you wish to load must be in your system indeed, then you can specify which one to use in xorg.conf or equivalent in xorg.conf.d/*.
OK, next installment of this thrilling saga! The Pale Moon that I built in Crux has now been transported to LFS. I had to install a few extra libraries, including gtk3. The program runs, but it still won't let me log in to Yahoo, whereas it did let me do that on Crux. So it must have something to do with the environment in which the program is running. It's a bug that only comes out in certain circumstances but I suppose it is a bug all the same because programs ought to be more resilient.
Actually now that I have all these dependencies installed, I could just build FF and have done with it, but I'd like to go a bit further with PM because I really like the interface. First, I need to know if I can build it in LFS using the code and mozconfig from Crux. And then I need to try it with a complete mesa build including the software rasteriser. Maybe I should compare the two Xorg logs too to see if there is any difference in the way X works in these two systems. Lots of work, lots of head-scratching. Hope I learn something.
Yesterday I tried three different build techniques using gcc-7:
1) make -f client.mk (recommended by LFS)
2) python mach build (used by Crux)
3) configure and make (Emerson's method).
They all failed. But then I checked what version of gcc my old LFS-8.0 was using, and it was gcc-6, like in Crux. So today I tried building the program in that system and the build went to completion without any errors. That means that the build failures must be due to an incompatibility between Pale Moon source (specifically the provided mozalloc.h header) and gcc-7. The header contains things that gcc-6 accepts as valid but gcc-7 spits out. I leave it to the community to decide whether that counts as a PM bug or a gcc bug.
Unfortunately the version of PM that got built was totally unstable. It crashed at every attempt to interact with a page. I tried building it with both the Crux mozconfig and the one provided by LFS for FF, and neither worked. Which is odd because the version I built on Crux is stable when unpacked and run in LFS-8.1. Possibly in LFS-8.0 there is some kind of interference with libraries installed by FF.
What I'd like to try next is to build in LFS-8.1 but with the gcc-6 from LFS-8.0. Hopefully I can do that by setting the environmental variable CC to /mnt/oldlfs/usr/bin/gcc. Then I can see if that is more stable.
LinuxQuestions.org is looking for people interested in writing
Editorials, Articles, Reviews, and more. If you'd like to contribute
content, let us know.
