Re: Kernel BUG at NULL Pointer Dereference

[fche at redhat dot com (Frank Ch. Eigler)]

Hi -

insfocus wrote:

>    I compiled the latest kernel from Linus's git, and the latest systemtap.

(By the way, generally such a configuration works for me.)


Thanks for the log:

> Dmesg Log:
>
> [22816.202933] BUG: unable to handle kernel NULL pointer dereference
> at 00000000000000ca
> [22816.203027] IP: [<ffffffff812e90b8>] strcmp+0x8/0x30
> [22816.203080] PGD 10bf57067 PUD 11204b067 PMD 0
> [22816.203130] Oops: 0000 [#1] SMP
> [22816.203166] Modules linked in:
> [...]
> [22816.203982] CPU 2
> [22816.204006] Pid: 9337, comm: stapio Tainted: G           O
> 3.7.0-rc6+ #1 LENOVO 2356BG6/2356BG6
> [22816.204082] RIP: 0010:[<ffffffff812e90b8>]  [<ffffffff812e90b8>]
> strcmp+0x8/0x30
> [22816.204151] RSP: 0018:ffff88013297be68  EFLAGS: 00010206
> [22816.204197] RAX: ffff88013801aa80 RBX: 000000000000000c RCX: 0000000000000000
> [22816.204257] RDX: 0000000000000000 RSI: ffffffffa068a728 RDI: 00000000000000ca
> [22816.204318] RBP: ffff88013297be68 R08: ffff8800532c3a30 R09: 00007fb03ce06700
> [22816.204378] R10: 0000000000000000 R11: 0000000000000000 R12: 00000000000000ca
> [22816.204438] R13: 0000000000000000 R14: 0000000000000008 R15: 0000000000000000
> [22816.204499] FS:  00007fb043a35740(0000) GS:ffff88013e280000(0000)
> knlGS:0000000000000000
> [22816.204567] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> [22816.204616] CR2: 00000000000000ca CR3: 000000010bf1a000 CR4: 00000000001407e0
> [22816.204676] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
> [22816.204736] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
> [22816.204797] Process stapio (pid: 9337, threadinfo ffff88013297a000,
> task ffff880137bf2e40)
> [22816.204866] Stack:
> [22816.204885]  ffff88013297bef8 ffffffffa067f53a 00007fffffffeffd
> 0000000000000002
> [22816.204959]  ffff8800532c3a00 0000000000000002 ffff88013297bec8
> ffffffff81284bcc
> [22816.205032]  ffffffff8105d562 000000000000000c ffff8800532c3a00
> 0000000000000000
> [22816.205105] Call Trace:
> [22816.205138]  [<ffffffffa067f53a>] _stp_ctl_write_cmd+0x32a/0x1250
> [stap_8eed450ffabfaa1bce874c07b2dd847e_14_9337]
> [...]


There aren't many strcmp's downstream of _stp_ctl_write_cmd, and those
ones don't look like they take parameters that could be NULL (or 0xc).
Could you try a few things?

- collect & share stap-report output
- build a new copy of the stap script with -BCONFIG_DEBUG_INFO=y, so
  that it may be disassembled around the point of crash?
- run the stap script with -DDEBUG_TRANS=2 -DDEBUG_SYMS=2


- FChE