Red Hat Bugzilla – Bug 79491
Sessions shut down for "Out of Memory" error
Last modified: 2007-04-18 12:49:00 EDT
From Bugzilla Helper: User-Agent: Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1) Description of problem: During the operation of Red Hat 8.0, my open sessions will suddenly start to close down for what "appears" to be an out of memory issue. I am running 512MB of DDRRAM and am not running any memory intensive applications. I have seen some postings in Linux groups that refer to an issue with the vmalloc.c file. Is this the underlying problem? Version-Release number of selected component (if applicable): How reproducible: Sometimes Steps to Reproduce: 1. Run Linux (sorry that's the only thing that I can think of) 2. 3. Actual Results: Sessions begin to shut down for out of memory errors. Expected Results: Sessions should not shut down. Additional info: I receive the following message: Dec 5 07:15:52 syscon kernel: Out of Memory: Killed process 1436 (gnome- terminal). Dec 5 07:15:53 syscon kernel: Trying to vfree() nonexistent vm area (e02bd000) Dec 5 07:15:53 syscon kernel: d8513ea4 d859a180 00000025 c012bd4b d859a180 dc4b9940 00000000 d8512000 Dec 5 07:15:53 syscon kernel: 00000009 c01193f9 dc4b9940 00200202 dc4b9940 c011e006 dc4b9940 c245a2dc Dec 5 07:15:53 syscon kernel: d8512000 00000000 d8513f30 00000009 c012436c 00000009 c0124554 00000009 Dec 5 07:15:53 syscon kernel: Call Trace: [<c012bd4b>] do_no_page [kernel] 0xeb (0xd8513eb0)) Dec 5 07:15:53 syscon kernel: [<c01193f9>] mmput [kernel] 0x39 (0xd8513ec8)) Dec 5 07:15:53 syscon kernel: [<c011e006>] do_exit [kernel] 0xa6 (0xd8513ed8)) Dec 5 07:15:53 syscon kernel: [<c012436c>] sig_exit [kernel] 0xac (0xd8513ef4)) Dec 5 07:15:53 syscon kernel: [<c0124554>] dequeue_signal [kernel] 0x64 (0xd8513efc)) Dec 5 07:15:53 syscon kernel: [<c0108ea7>] do_signal [kernel] 0x1f7 (0xd8513f14)) Dec 5 07:15:53 syscon kernel: [<c01249c1>] deliver_signal [kernel] 0x31 (0xd8513f68)) Dec 5 07:15:53 syscon kernel: [<c0109c00>] do_general_protection [kernel] 0x0 (0xd8513fa0)) Dec 5 07:15:53 syscon kernel: [<c01168c0>] do_page_fault [kernel] 0x0 (0xd8513fb8)) Dec 5 07:15:54 syscon kernel: [<c0109148>] signal_return [kernel] 0x14 (0xd8513fc0)) Dec 5 07:15:55 syscon kernel:
Ick - it is the behaviour of the kernel to salvage a running system by nixing off processes to free memory. This is to prevent complete system instability. Could you please provide the output of your /proc/meminfo after having the system up and running for a while?
Run xdpyinfo, check whether you have a RENDER extension. If not, gnome-terminal and other GNOME apps leak memory. The workaround is IIRC to use konsole or xterm instead of gnome-terminal. Note that other apps reportedly suffer from the leaks too, but gnome-terminal is usually the most severe.
'kernel: Trying to vfree() nonexistent vm area (e02bd000)' is the start of the kernel running into problems. That shouldn't normally happen. Are you 100% sure that the memory in the system is reliable? If so, and you see the same behaviour with different memory modules in place as well, please re-open and submit the entire 'oops' message. Thanks.
Memory modules have been swapped and replaced. The intermitent problem remains. We've switched to KDE from GNOME on the advice of Miloslav and are waiting to see if this reolves the problem. We have not been able to capture the output of /proc/meminfo yet.
does the system pass hardware test like memtes86 http://www.memtest86.com/ or cpuburn http://users.ev1.net/~redelm ?
Switching from GNOME to KDE resolved problem. There is a meory leak in GNOME desktop.