Closed Bug 319236 Opened 20 years ago Closed 16 years ago

Talkback initialization crashes when /dev/zero can't be opened RDWR

Categories

(Core Graveyard :: Talkback Client, defect)

1.8 Branch
x86
Linux
defect
Not set
critical

Tracking

(Not tracked)

RESOLVED INVALID

People

(Reporter: dothebart, Assigned: jay)

References

Details

(Keywords: crash)

Attachments

(1 file)

User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.12) Gecko/20050922 Firefox/1.0.7 (Debian package 1.0.7-1) Build Identifier: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.12) Gecko/20050922 Firefox/1.0.7 (Debian package 1.0.7-1) <- not this one, the fresh downloaded 1.5 crashes. freshly installed firefox in a local dir of my sid box, and it keeps crashing. i removed .phoenix, .mozilla and .firefox, started over again, nothing changes. Reproducible: Always Steps to Reproduce: 1. export LD_LIBRARY_PATH=/home/willi/local/firefox 2. ./firefox Actual Results: crashes. here's a backtrace in gdb: (gdb) bt #0 0xb7e2cd38 in pthread_mutex_init () from /lib/tls/libpthread.so.0 #1 0xb6b9d243 in __FCDeleteKey () from /home/willi/local/firefox/extensions/talkback@mozilla.org/components/talkback/talkback.so #2 0xb6b894d1 in ?? () from /home/willi/local/firefox/extensions/talkback@mozilla.org/components/talkback/talkback.so #3 0xb6b8d735 in ?? () from /home/willi/local/firefox/extensions/talkback@mozilla.org/components/talkback/talkback.so #4 0xb6b8d85f in __FCInitialize () from /home/willi/local/firefox/extensions/talkback@mozilla.org/components/talkback/talkback.so #5 0xb73da44d in NSGetModule () from /home/willi/local/firefox/extensions/talkback@mozilla.org/components/libqfaservices.so #6 0xb73da83a in NSGetModule () from /home/willi/local/firefox/extensions/talkback@mozilla.org/components/libqfaservices.so #7 0xb73da877 in FCInitialize () from /home/willi/local/firefox/extensions/talkback@mozilla.org/components/libqfaservices.so #8 0xb73d9f78 in nsQfaServicesImpl::~nsQfaServicesImpl () from /home/willi/local/firefox/extensions/talkback@mozilla.org/components/libqfaservices.so #9 0xb73da00a in nsQfaServicesImpl::~nsQfaServicesImpl () from /home/willi/local/firefox/extensions/talkback@mozilla.org/components/libqfaservices.so #10 0x0851e240 in nsXPTCVariant::Init () #11 0x0807a6d9 in ?? () #12 0x08a89558 in ?? () #13 0x00000000 in ?? ()
it asks if it should import anything, and i say no. then it closes the window, and short arfter it segfaults.
usually talkback triggers as a signal handler to report crashes...
Severity: normal → critical
Keywords: crash
Version: 1.0 Branch → 1.5 Branch
the new debian Sid package comes up stable. maybe this was an interfering between system firefox 1.0.* packages and the localy unpackaged firefox? should this happen? shouldn't firefox primarily take it's own files instead of system files?
the problem still is here with debian sid. meanwhile i flushed an reconstructed my profile. The SID iceweasel runs flawlessly. does the firefox shellwrapper set LD_LIBRARY_PATH's?
thanks, that's actually useful. you could just delete /home/willi/local/firefox/extensions/talkback@mozilla.org/components/talkback/talkback.so if you want to work around this. The problem is visible here: open("/dev/zero", O_RDWR) = -1 EACCES (Permission denied) mmap2(NULL, 1056768, PROT_READ|PROT_WRITE, MAP_PRIVATE, -1, 0) = -1 EBADF (Bad file descriptor) close(-1) = -1 EBADF (Bad file descriptor) --- SIGSEGV (Segmentation fault) @ 0 (0) --- timeless@boffo:/tmp$ ls -al /dev/zero crw-rw-rw- 1 root sys 1, 5 1994-07-17 16:46 /dev/zero I have no idea why your /dev/zero is not writable but i'm fairly certain that's the problem (i'd love confirmation).
Assignee: nobody → jay
Component: General → Talkback Client
Product: Firefox → Core
QA Contact: general → chofmann
Summary: Firefox crashes on startup → Talkback initialization crashes when /dev/zero can't be opened RDWR
Whiteboard: DUPEME
Version: 1.5.0.x Branch → 1.8 Branch
anyway, java and mplayer have code to recognize such misconfigured systems. our code is of course not perfect (it should handle your misconfigured system with a bit more grace), but you should really fix your system.
hm, yes, you're right: crw-rw---- 1 root root 1, 5 Mar 2 20:11 /dev/zero but.. udev on /dev type tmpfs (rw,mode=0755) shouldn't that be done by bootscripts? its public writable on my other systems though... exiting with an error would indeed be a bit more polite than just segfaulting... A bug squishing facility shouldn't introduce crashes imho ;-)
as it happens, talkback is end of lifed, so it doesn't matter much. you can simply install a minefield nightly and experience breakpad. this bug may or may not be fixed depending on whether or not an engineer is assigned to work on it.
Product: Core → Core Graveyard
Talkback is totally gone now.
Status: UNCONFIRMED → RESOLVED
Closed: 16 years ago
Resolution: --- → INVALID
Whiteboard: DUPEME
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: