http://bugzilla.suse.com/show_bug.cgi?id=1027456
http://bugzilla.suse.com/show_bug.cgi?id=1027456#c5
Luis Rodriguez changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |lurodriguez@suse.com
--- Comment #5 from Luis Rodriguez ---
Created attachment 716112
--> http://bugzilla.suse.com/attachment.cgi?id=716112&action=edit
fix kmemleak splat on alloc_thread_stack_node()
You can test this patch. I can reproduce this with :
echo clear > /sys/kernel/debug/kmemleak
echo scan > /sys/kernel/debug/kmemleak
echo scan > /sys/kernel/debug/kmemleak
cat /sys/kernel/debug/kmemleak
Can you reproduce as well with this?
Some others cannot reproduce this so this may not be a general thing but very
closely related to a kernel configuration. If so, what aspects of a kernel
configuration make this possible ?
Once you can reproduce the next steps before we consider this upstream is for
us is to configure a system with kdump (Documentation/kdump/kdump.txt) and
force a BUG_ON() on this path so a vmcore is generated. In turn we'd use this
vmcore with crash [0] to inspect the process that crashed to see if we can
verify if the process still owns its stack. I have not had a chance for this
later step yet.
For discussion upstream refer to lkml [1].
[0] https://people.redhat.com/anderson/crash_whitepaper/
[1]
https://lkml.kernel.org/r/CAB=NE6VZXq3y-3pfouYTBUco2Cq2xqoLZrgDFdVx+_=_=SwG_...
--
You are receiving this mail because:
You are on the CC list for the bug.