Summary: | net-libs/webkit-gtk segmentation faults a lot on hppa | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Jeroen Roovers (RETIRED) <jer> |
Component: | Current packages | Assignee: | Gentoo Linux Gnome Desktop Team <gnome> |
Status: | RESOLVED UPSTREAM | ||
Severity: | normal | CC: | hppa |
Priority: | High | ||
Version: | unspecified | ||
Hardware: | All | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- | |
Bug Depends on: | |||
Bug Blocks: | 272160 | ||
Attachments: |
gdb output [hppa]
gdb output using -O0 and with USE=-pango |
Description
Jeroen Roovers (RETIRED)
2009-06-21 12:46:49 UTC
Created attachment 195338 [details]
gdb output [hppa]
Created attachment 195420 [details]
gdb output using -O0 and with USE=-pango
I suspect this is blocking the keywording for you. Marking blocker Have you tried to report this to upstream (if you are using 1.1.10) Can you try with webkit-gtk-1.1.11 ? (bumped into gnome overlay) (In reply to comment #4) > Have you tried to report this to upstream (if you are using 1.1.10) Generally that is what package maintenance is about. (In reply to comment #5) > Can you try with webkit-gtk-1.1.11 ? (bumped into gnome overlay) No. Arch maintainers don't use overlays. :) (In reply to comment #7) > (In reply to comment #5) > > Can you try with webkit-gtk-1.1.11 ? (bumped into gnome overlay) > > No. Arch maintainers don't use overlays. :) > layman -a gnome && emerge -av1 webkit-gtk The reason why it's not in tree is because it needs libsoup-2.27.4, if we know that 1.1.11 fixes the problem, we can try to backport the fix. If it doesn't, we can report the problem upstream. @hppa: ping, can you please try webkit-gtk-1.1.14 in overlay? (It's not in-tree due to deps on libsoup-2.27) (In reply to comment #9) > @hppa: ping, can you please try webkit-gtk-1.1.14 in overlay? (It's not in-tree > due to deps on libsoup-2.27) Do you have any inkling that this version does indeed address some of the platform issues raised here? I am not going to test yet another version just because it's higher. I am not going to play around with overlays on a system intended to test and mark stable with either, so maybe another hppa dev can play around with these "unstable" builds. To date I haven't seen any progress on odd-arch work in both major webkit forks (gtk and qt) so I am not going to hold my breath. :) Hello Jeroen Are you still affected by this bug with 1.1.15.4? If you are still affected, and the same backtrace applies, please confirm and I will report this to upstream, maybe they will prefer to get it tested with latest development version 1.1.21 but, if you still prefer to not try that "dev" versions... no problem :-) I'll be testing it shortly. 1.1.15.4 does sometimes segfault but is more prone to loop (endlessly?) somewhere in JavaScriptCore/wtf/dtoa.cpp (See also bug #239441). Pages that run minimal or no JavaScript are usually fine, so this is definitely an improvement, but we aren't there yet. (In reply to comment #13) > 1.1.15.4 does sometimes segfault Well, webkit-gtk is still a bit crashy on all platforms, it's improved on each new release but... :-/ > but is more prone to loop (endlessly?) > somewhere in > JavaScriptCore/wtf/dtoa.cpp > > (See also bug #239441). This is a problem I haven't suffered, I will ask upstream but, do you have a testcase for reproducing the problem (at least on hppa)? I mean that would be really interesting to know what webpage tends to cause the loop Thanks (In reply to comment #14) > (In reply to comment #13) > > 1.1.15.4 does sometimes segfault > > Well, webkit-gtk is still a bit crashy on all platforms, it's improved on each > new release but... :-/ You have no idea how disproportionately crashy it is in HPPA... 9 times out of 10 it crashes on loading a page. 999 out of a thousand times it crashes when you click somewhere on the page that (cor)responds to a JavaScript event. > > but is more prone to loop (endlessly?) > > somewhere in > > JavaScriptCore/wtf/dtoa.cpp > > > > (See also bug #239441). > > This is a problem I haven't suffered, I will ask upstream but, do you have a > testcase for reproducing the problem (at least on hppa)? I mean that would be > really interesting to know what webpage tends to cause the loop Just about any page that loads JavaScript. I added a comment about hangs affecting also to webkitgtk on https://bugs.webkit.org/show_bug.cgi?id=34037#c3 About the crashes, were you able to get a recent backtrace with 1.1.15.4 that we could send upstream? Or are they similar to comment #1 provided one? Thanks https://bugs.webkit.org/show_bug.cgi?id=34037#c5 : "The plan is to drop support for HP-UX on PA-RISC in Qt 4.7 [1] so I can only assume that Linux on PA-RISC is equally unsupported, meaning this is not high priority for us, but patches are welcome!" I am not sure then if this will ever be solved :-( HP-UX is not Linux, and the author of that comment appears unaware of that fact that HP-UX popularly runs on IA64 as well as PARISC. See bug #239441 comment 10 for more info. This needs to be handled directly with upstream if anyone is still interested |