On Friday 22 March 2002 14:22, Nathan Straz wrote:
> Stupid questions first:
>
> What kernel are you running?
2.4.9-31SGI-XFS1.1 + some plaform patches we swear by... (or at)
> What compiler did you use?
gcc-2.96-101
> Do you know how to reproduce this?
Generate lots of I/O I don't mean to be flipant but I was expecting the IO
subsystem to buckle under the strain
> Can you compile the kernel with kdb and get more info?
I had expected kdb, but after reading the .spec file saw that the "prep" does
not apply the patch, so the next build will include kdb.
I am not concerned about these being an XFS bug. It has more to do with the
handling of "bottom halves" and the __global_cli() call, as can be seen by
the following:
wait_on_irq, CPU 0:
irq: 1 [ 0 0 0 0 0 0 1 0 1 1 1 0 0 0 0 0 ]
bh: 1 [ 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 ]
Stack dumps:
CPU 0:
Call Trace: [<e0000000044974d0>] sp=0xe0000000d8c37b00 bsp=0xe0000000d8c31418
decoded to show_stack [kernel] 0x50
[<e000000004494030>] sp=0xe0000000d8c37cc0 bsp=0xe0000000d8c313e0
decoded to show [kernel] 0x310
[<e000000004494560>] sp=0xe0000000d8c37cc0 bsp=0xe0000000d8c31390
decoded to __global_cli [kernel] 0x3e0
[<e00000000473e160>] sp=0xe0000000d8c37cc0 bsp=0xe0000000d8c31350
decoded to rs_timer [kernel] 0xa0
[<e000000004502110>] sp=0xe0000000d8c37cc0 bsp=0xe0000000d8c312e0
decoded to __run_timers [kernel] 0x290
[<e000000004503100>] sp=0xe0000000d8c37cc0 bsp=0xe0000000d8c312b0
decoded to run_all_timers [kernel] 0xc0
[<e0000000044f8f90>] sp=0xe0000000d8c37cc0 bsp=0xe0000000d8c31290
decoded to bh_action [kernel] 0x110
[<e0000000044f8bb0>] sp=0xe0000000d8c37cc0 bsp=0xe0000000d8c31240
decoded to tasklet_hi_action [kernel] 0x250
[<e0000000044f80e0>] sp=0xe0000000d8c37cc0 bsp=0xe0000000d8c311c0
decoded to do_softirq [kernel] 0x140
[<e000000004496e80>] sp=0xe0000000d8c37cc0 bsp=0xe0000000d8c31198
decoded to ia64_handle_irq [kernel] 0x100
[<e000000004491da0>] sp=0xe0000000d8c37cc0 bsp=0xe0000000d8c31198
decoded to ia64_leave_kernel [kernel] 0x0
[<e0000000045494d0>] sp=0xe0000000d8c37e60 bsp=0xe0000000d8c31110
decoded to sys_lseek [kernel] 0x110
[<e000000004491d80>] sp=0xe0000000d8c37e60 bsp=0xe0000000d8c31110
decoded to ia64_ret_from_syscall [kernel] 0x0
|