On Thu, Feb 16, 2006 at 03:15:14PM -0500, Kris Kennaway wrote: > Peter Holm's stress test gave me this on an SMP machine running fresh > 7.0: > > Fatal trap 12: page fault while in kernel mode > cpuid = 0; apic id = 00 > fault virtual address = 0xdeadc0de > fault code = supervisor write, page not present > instruction pointer = 0x20:0xc0681633 > stack pointer = 0x28:0xf3bbeb88 > frame pointer = 0x28:0xf3bbeb88 > code segment = base 0x0, limit 0xfffff, type 0x1b > = DPL 0, pres 1, def32 1, gran 1 > processor eflags = interrupt enabled, resume, IOPL = 0 > current process = 24 (irq17: fxp0) > [thread pid 24 tid 100020 ] > Stopped at trash_dtor+0x10: movl $0xdeadc0de,0(%edx) > db> wh > Tracing pid 24 tid 100020 td 0xcc47b1a0 > trash_dtor(deadc0de,800,0,f3bbebb8,c05295bb) at trash_dtor+0x10 > trash_init(deadc0de,800,1,7f,35) at trash_init+0x20 > mb_zinit_pack(ccb7e100,100,1,85f,f3bbebec) at mb_zinit_pack+0x50 > uma_zalloc_bucket(c1057000,1,c073d432,75d,0) at uma_zalloc_bucket+0x1f1 > uma_zalloc_arg(c1057000,f3bbec4c,1,1,c072a086) at uma_zalloc_arg+0x38e > fxp_add_rfabuf(cc546000,cc54604c,2,61a,cc546014) at fxp_add_rfabuf+0x35 > fxp_intr_body(cc546000,cc53c000,40,ffffffff,cc53c000) at fxp_intr_body+0x115 > fxp_intr(cc546000,f3bbecdc,c052aa10,c07f5c90,1) at fxp_intr+0xcf > ithread_execute_handlers(cc4cccd8,cc477700,c0725a19,2f9,cc47b1a0) at > ithread_execute_handlers+0x10e > ithread_loop(cc539960,f3bbed38,c0725807,31a,cc539960) at ithread_loop+0x78 > fork_exit(c051cdfa,cc539960,f3bbed38) at fork_exit+0xc5 > fork_trampoline() at fork_trampoline+0x8 > --- trap 0x1, eip = 0, esp = 0xf3bbed6c, ebp = 0 --- > db>
Peter Holm has also seen this panic involving the fxp driver: http://people.freebsd.org/~pho/stress/log/cons186.html When I ran the stress test on two machines with em they did not panic after about twice as long under load. Kris
pgpbPh6yOgefw.pgp
Description: PGP signature
