PAGE FAULT in kernel mode at the vm_page_alloc() call

[Sibananda Sahu]

Hi,



We were doing internal testing using mrsas driver for certain feature on
FreeBSD-10-release.



While test execution ,we have observed  frequent kernel panic in same code
path on FreeBSD 10. Please find more detail about test case and symptoms.



1.        Create more than 64 VD using latest mrsas feature which supports
Extended VD. (This issue may not be related to this, but just wanted to
highlight exact steps)

All 240 VD is able to detect in “camcontrol”.

2.       Using LSI StorCli utility do Full Init on all 240 Virtual Drives
those are connected behind an LSI controller.



Eventually, we observe kernel panic with below backtraces.  It is a random
behavior but seems to be occurring at the same place all the time, i.e.
around vm_page_alloc() call.

--

KDB: stack backtrace:

#0 0xffffffff808e7dd0 at kdb_backtrace+0x60

#1 0xffffffff808af8b5 at panic+0x155

#2 0xffffffff80c8e692 at trap_fatal+0x3a2

#3 0xffffffff80c8e969 at trap_pfault+0x2c9

#4 0xffffffff80c8e0f6 at trap+0x5e6

#5 0xffffffff80c75392 at calltrap+0x8

#6 0xffffffff80b1d1e9 at vm_page_alloc+0x389

#7 0xffffffff80b0f9fe at kmem_back+0xde

#8 0xffffffff80b0f8f6 at kmem_malloc+0x66

#9 0xffffffff80b08c6b at uma_large_malloc+0x4b

#10 0xffffffff80898c33 at malloc+0x103

#11 0xffffffff8030eaf5 at enc_daemon+0x1c5

#12 0xffffffff8088198a at fork_exit+0x9a

#13 0xffffffff80c758ce at fork_trampoline+0xe



--

Here I have attached the crash dump text file to grep more info.

If you need more information from us, with some more config or debug
parameters enabled or something else, please let me know.

I can do some experiment to root cause the issue. I also check on web, but
there was not clear direction on this type of kernel panic ..but looks like
many folks encountered similar issues.



Also, I verified same behavior on latest FreeBSD-11-Current as well.
(Revision 273072)



Thanks