| Summary: | www-client/chromium: random crashes on hardened environment | ||
|---|---|---|---|
| Product: | Gentoo Linux | Reporter: | Agostino Sarubbo <ago> |
| Component: | Current packages | Assignee: | Chromium Project <chromium> |
| Status: | RESOLVED UPSTREAM | ||
| Severity: | normal | ||
| Priority: | Normal | ||
| Version: | unspecified | ||
| Hardware: | All | ||
| OS: | Linux | ||
| See Also: | http://code.google.com/p/chromium/issues/detail?id=94472 | ||
| Whiteboard: | |||
| Package list: | Runtime testing required: | --- | |
|
Description
Agostino Sarubbo
2011-07-12 16:54:08 UTC
Okay, could you please get a stacktrace? (http://code.google.com/p/chromium/wiki/LinuxDebugging may be helpful, and I can give you more assistance if needed) Also, are there some specific steps to reliably reproduce the issue? I noticed you wrote it happens totally randomly in the report, but I'd just like to make sure there is no obvious correlation. (In reply to comment #1) > Also, are there some specific steps to reliably reproduce the issue? no Atm, I have modified chromium ebuild, so filtering flag like -fomit-frame-pointer and replace -O3 with -O2 so, -O3 should be a cause of crashing on hardened. After using my custom ebuild I didn't see other crashes, so I test it for another days. To do a correct backtrace I must recompile with the original ebuild, I'll do also it. (In reply to comment #2) > Atm, I have modified chromium ebuild, so filtering flag like > -fomit-frame-pointer and replace -O3 with -O2 so, -O3 should be a cause of > crashing on hardened. Okay, so does it mean -O3 is causing the crashes? No, also with -O3 dropped there is also a problem. Anyway i didn't see anything with gdb and the problem is not reproducible every time. It is probably related to flash interaction. That's not enough to fix anything. It'd be interesting if you could get a core dump file. If can be interesting, I see this error from gdb: V8 error: V8 is no longer usable (v8::V8::SetGlobalGCPrologueCallback()). Current memory usage: 0 MB When it crashes i see always this warning (In reply to comment #6) > V8 error: V8 is no longer usable (v8::V8::SetGlobalGCPrologueCallback()). Sorry, this is still not enough. Could you obtain a stack trace or something? (gdb) bt
#0 0x4b8fa416 in __kernel_vsyscall ()
#1 0x492e7556 in epoll_wait () at ../sysdeps/unix/syscall-template.S:82
#2 0x49a3c6a2 in epoll_dispatch (base=0x135d2480, tv=0x0) at epoll.c:404
#3 0x49a2885a in event_base_loop (base=0x135d2480, flags=1) at event.c:1562
#4 0x112932d1 in base::MessagePumpLibevent::Run (this=0x13600440, delegate=0x47fd8008)
at base/message_pump_libevent.cc:260
#5 0x112c1f4b in MessageLoop::RunInternal (this=0x47fd8008) at base/message_loop.cc:443
#6 0x112c2149 in RunHandler (this=0xfffffffc) at base/message_loop.cc:416
#7 MessageLoop::Run (this=0xfffffffc) at base/message_loop.cc:340
#8 0x112f9d99 in base::Thread::Run (this=0x135adc30, message_loop=0x47fd8008)
at base/threading/thread.cc:128
#9 0x112f9b8e in base::Thread::ThreadMain (this=0x135adc30) at base/threading/thread.cc:163
#10 0x112f9815 in base::(anonymous namespace)::ThreadFunc (params=0x135aadb0)
at base/threading/platform_thread_posix.cc:58
#11 0x4b8cdb4c in start_thread (arg=0x47fd8b70) at pthread_create.c:301
#12 0x492e6b6e in clone () at ../sysdeps/unix/sysv/linux/i386/clone.S:133
(gdb) bt
#0 0x4ac21416 in __kernel_vsyscall ()
#1 0x48603bbc in __poll (fds=0x48690e4c, nfds=4, timeout=1677) at ../sysdeps/unix/sysv/linux/poll.c:87
#2 0x48dd22c7 in g_poll (fds=0x16b99600, nfds=4, timeout=1677) at gpoll.c:132
#3 0x48dc45d0 in g_main_context_poll (context=0x141262d0, block=<value optimized out>, dispatch=1,
self=0x1411f4e0) at gmain.c:3405
#4 g_main_context_iterate (context=0x141262d0, block=<value optimized out>, dispatch=1, self=0x1411f4e0)
at gmain.c:3087
#5 0x48dc48d1 in g_main_context_iteration (context=0x141262d0, may_block=1) at gmain.c:3155
#6 0x11e5a9b7 in base::MessagePumpGtk::RunOnce (this=0x14158180, context=0x141262d0,
block=<value optimized out>) at base/message_pump_gtk.cc:41
#7 0x11e59aee in base::MessagePumpGlib::RunWithDispatcher (this=0x14158180, delegate=0x14115b00,
dispatcher=0x0) at base/message_pump_glib.cc:206
#8 0x11e59c6c in base::MessagePumpGlib::Run (this=0x14158180, delegate=0x14115b00)
at base/message_pump_glib.cc:298
#9 0x11e16f4b in MessageLoop::RunInternal (this=0x14115b00) at base/message_loop.cc:443
#10 0x11e17048 in RunHandler (this=0xfffffdfc, dispatcher=0x4) at base/message_loop.cc:416
#11 MessageLoopForUI::Run (this=0xfffffdfc, dispatcher=0x4) at base/message_loop.cc:823
#12 0x11abd8a2 in RunUIMessageLoop (this=0x1411f0f0) at chrome/browser/browser_main.cc:245
#13 ChromeBrowserMainParts::TemporaryContinue (this=0x1411f0f0) at chrome/browser/browser_main.cc:1840
#14 0x13397c86 in BrowserMain (parameters=...) at content/browser/browser_main.cc:311
#15 0x11d6675b in RunNamedProcessTypeMain (argc=1, argv=0x5a94a8f4, delegate=0x5a94a7d0)
at content/app/content_main.cc:292
#16 content::ContentMain (argc=1, argv=0x5a94a8f4, delegate=0x5a94a7d0) at content/app/content_main.cc:482
@Pawel, are them enough?
(In reply to comment #8) > @Pawel, are them enough? Yes, they're good. I noticed you've reported that upstream, which is good (however, you really should try to reproduce the crashes with Google Chrome). I replied in the upstream bug report. (In reply to comment #9) > (however, you really should try to reproduce the crashes with Google Chrome). Is not possible, the problem is chromium compiled with hardened gcc and there aren't sources of google chrome, so, imho is no-sense try it (In reply to comment #10) > (In reply to comment #9) > > (however, you really should try to reproduce the crashes with Google Chrome). > Is not possible, the problem is chromium compiled with hardened gcc and there > aren't sources of google chrome, so, imho is no-sense try it It makes no sense to argue with people trying to do something with your bug reports. Please go install www-client/google-chrome and see if it crashes, just as asked. If it doesn't after *trying* it, it's still very valuable information. Please Just Do It, it'll make it easier for upstream to further process the bug report. (In reply to comment #11) > It makes no sense to argue with people trying to do something with your bug > reports. Please go install www-client/google-chrome and see if it crashes, just > as asked. If it doesn't after *trying* it, it's still very valuable > information. Please Just Do It, it'll make it easier for upstream to further > process the bug report. Before do unnecessary and vague attempts you should take a look on what causes this problem. So, for me, chromium in not hardened environment works perfectly as for all I guess. The problem is there when I compile with hardened gcc, so, flags like: -fno-strict-overflow -fPIE -fstack-protector-all can annoying in some way. Now, imho, google-chrome is not compiled with that flags and cannot reproduce this issue. Anyway, as I said, after have tried it, google-chrome works without a problem(s). |