[Bug 998850] New: Kernel slab eats all memory
http://bugzilla.opensuse.org/show_bug.cgi?id=998850 Bug ID: 998850 Summary: Kernel slab eats all memory Classification: openSUSE Product: openSUSE Tumbleweed Version: Current Hardware: Other OS: Other Status: NEW Severity: Normal Priority: P5 - None Component: Kernel Assignee: kernel-maintainers@forge.provo.novell.com Reporter: P.Suetterlin@royac.iac.es QA Contact: qa-bugs@suse.de Found By: --- Blocker: --- I have a new Skylake laptop (Thinkpad T460p) running an up-to-date Tumbleweed with kernel 4.7.2-2-default I noticed that the system started swapping despite 16GB RAM and little mem usage. A look at /proc/meminfo revealed that the memory is used by the kernel slab: Slab: 12410408 kB SReclaimable: 83752 kB SUnreclaim: 12326656 kB and slabtop shows one process permanently acquiring new pages: OBJS ACTIVE USE OBJ SIZE SLABS OBJ/SLAB CACHE SIZE NAME 3065838 3065838 100% 4.00K 3065838 1 12263352K kmalloc-4096 I'm not really sure where else to look or what other information to post, but will attach everything you request.... -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=998850
http://bugzilla.opensuse.org/show_bug.cgi?id=998850#c1
Takashi Iwai
http://bugzilla.opensuse.org/show_bug.cgi?id=998850 http://bugzilla.opensuse.org/show_bug.cgi?id=998850#c3 --- Comment #3 from Peter Sütterlin
http://bugzilla.opensuse.org/show_bug.cgi?id=998850 http://bugzilla.opensuse.org/show_bug.cgi?id=998850#c4 --- Comment #4 from Peter Sütterlin
http://bugzilla.opensuse.org/show_bug.cgi?id=998850 http://bugzilla.opensuse.org/show_bug.cgi?id=998850#c5 --- Comment #5 from Peter Sütterlin
http://bugzilla.opensuse.org/show_bug.cgi?id=998850 http://bugzilla.opensuse.org/show_bug.cgi?id=998850#c8 --- Comment #8 from Peter Sütterlin
http://bugzilla.opensuse.org/show_bug.cgi?id=998850 http://bugzilla.opensuse.org/show_bug.cgi?id=998850#c10 --- Comment #10 from Peter Sütterlin
http://bugzilla.opensuse.org/show_bug.cgi?id=998850 http://bugzilla.opensuse.org/show_bug.cgi?id=998850#c12 Peter Sütterlin
http://bugzilla.opensuse.org/show_bug.cgi?id=998850 http://bugzilla.opensuse.org/show_bug.cgi?id=998850#c13 --- Comment #13 from Peter Sütterlin
http://bugzilla.opensuse.org/show_bug.cgi?id=998850 http://bugzilla.opensuse.org/show_bug.cgi?id=998850#c16 Peter Sütterlin
(In reply to Peter Sütterlin from comment #12)
In any case it is not a Tumbleweed-bug, so I guess it can be closed.
I wouldn't be so sure about it. If the plugin (which is not a tumbleweed package then, IIUC?) was leaking its own memory, then sure. But if it's causing a memory leak in *kernel* memory, it's suspicious and probably a kernel (video driver?) bug.
Indeed, the plugin is not a Tumbleweed package, but compiled from source myself, and the package is basically unmaintained since 2003..... It is however not doing any memory allocation by itself (at least greping for alloc does not give any result).
Does the slab memory get freed when you kill/restart X?
No, it stays occupied, even when switching the system to emergency mode.
Sorry for the wrong alarm (but at least I have learned quite something about kernel tracing, which is highly appreciated!)
A trace with full stack (see my previous comment) could still be useful.
Lets see if I can make some sense of it: echo bytes_alloc==4096 > /sys/kernel/debug/tracing/events/kmem/kmem_cache_alloc/filter echo 1 > /sys/kernel/debug/tracing/options/stacktrace echo 1 > /sys/kernel/debug/tracing/events/kmem/kmem_cache_alloc/enable cat /sys/kernel/debug/tracing/trace > stacktrace If this is correct, the output is in the next attachment. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=998850 http://bugzilla.opensuse.org/show_bug.cgi?id=998850#c17 --- Comment #17 from Peter Sütterlin
http://bugzilla.opensuse.org/show_bug.cgi?id=998850 http://bugzilla.opensuse.org/show_bug.cgi?id=998850#c21 --- Comment #21 from Peter Sütterlin
http://bugzilla.opensuse.org/show_bug.cgi?id=998850 http://bugzilla.opensuse.org/show_bug.cgi?id=998850#c22 --- Comment #22 from Peter Sütterlin
participants (1)
-
bugzilla_noreply@novell.com