|
Description
|
cores will be copied to /net/coresvr.sfbay/export/cores4/bugid.
panic[cpu1]/thread=fffffffec1d16280:
BAD TRAP: type=e (#pf Page fault) rp=ffffff0004522b80 addr=28 occurred in module
"genunix" due to a NULL pointer dereference
tc_aio_read:
#pf Page fault
Bad kernel fault at addr=0x28
pid=3080, pc=0xfffffffffb93744d, sp=0xffffff0004522c70, eflags=0x10207
cr0: 8005003b<pg,wp,ne,et,ts,mp,pe> cr4: 6f8<xmme,fxsr,pge,mce,pae,pse,de>
cr2: 28 cr3: 3f11b000 cr8: c
rdi: fffffffeceb73028 rsi: fffffffec6bf9e00 rdx: fffffffec1d16280
rcx: 0 r8: 0 r9: 0
rax: 0 rbx: 1 rbp: ffffff0004522c80
r10: fffffffec1e5ca20 r11: 4 r12: 16
r13: fffffffeceb73060 r14: 100000 r15: fffffffeceb73000
fsb: 0 gsb: fffffffec1692800 ds: 4b
es: 4b fs: 0 gs: 1c3
trp: e err: 2 rip: fffffffffb93744d
cs: 30 rfl: 10207 rsp: ffffff0004522c70
ss: 38
ffffff0004522a60 unix:die+c8 ()
ffffff0004522b70 unix:trap+135b ()
ffffff0004522b80 unix:cmntrap+e9 ()
ffffff0004522c80 genunix:aio_deq+1d ()
ffffff0004522e40 kaio:aiorw+4b1 ()
ffffff0004522e80 kaio:kaio+212 ()
ffffff0004522ec0 genunix:syscall_ap+8f ()
ffffff0004522f10 unix:brand_sys_syscall32+1a3 ()
syncing file systems...
4
1
done
dumping to /dev/dsk/c2t0d0s1, offset 431030272, content: kernel
Mail from an SC engineer who looked at the cores.
============================================================================
This looks like a Solaris aio bug. I'd file a bug against Solaris.
The console panic trace seems to suggest a brandz syscall is triggering the panic, but closer inspection shows that it is generic aio call:
> ::status
debugging crash dump vmcore.1 (64-bit) from pkuda1
operating system: 5.11 snv_64a (i86pc)
panic message:
BAD TRAP: type=e (#pf Page fault) rp=ffffff00041d6b80 addr=28 occurred in module
"genunix" due to a NULL pointer dereference
dump content: kernel pages only
> $C
ffffff00041d6c80 aio_deq+0x1d(fffffffed1aec028, fffffffec64f9a80)
ffffff00041d6e40 aiorw+0x4b1(b, 8047510, 1, 1)
ffffff00041d6e80 kaio+0x212(fffffffec1c7f918, ffffff00041d6eb8)
ffffff00041d6ec0 syscall_ap+0x8f()
ffffff00041d6f10 sys_syscall32+0x101()
The issue is noted during pxfs related automated stress tests. Since we don't do any tests without cluster s/w installed, we cannot say if it happens outside of cluster as well.
Running the pxfs tests from scate should result in panic at some point.
|
