Re: 2.4.27: md RAID1 oops on alpha revisited

From: Ivan Kokshaysky (ink_at_jurassic.park.msu.ru)
Date: 09/28/04

  • Next message: Christoph Lameter: "Re: [RFC] Posix compliant behavior of CLOCK_PROCESS/THREAD_CPUTIME_ID"
    Date:	Tue, 28 Sep 2004 23:24:34 +0400
    To: Ingvar Hagelund <ingvar@linpro.no>
    
    

    On Tue, Sep 28, 2004 at 01:44:24AM +0200, Ingvar Hagelund wrote:
    > We have a Compaq Alphaserver DS10 466 MHz running Debian Woody with a
    > self compiled 2.4.27 from kernel.org. We have not been able to make it
    > run stable on md RAID1. It always crashes in less than an hour
    > uptime, presumely while stressing the RAID code. Running on single
    > disks, it's rock stable.

    The problem seems to be in qlogic isp1020 driver, not in the RAID code.
    I've seen exactly the same oops report, but that had happened while
    writing to SCSI tape.

    > Code; fffffc000050da00 <isp1020_intr_handler+300/420>
    > 14: 49 00 40 d3 bsr ra,13c <_PC+0x13c> fffffc000050db28 <isp1020_return_status+8/100>
    > Code; fffffc000050da04 <isp1020_intr_handler+304/420> <=====
    > 18: 40 02 09 b0 stl v0,576(s0) <=====

    It's the line 1052 in drivers/scsi/qlogicisp.c:

                    if (sts->hdr.entry_type == ENTRY_STATUS)
                            Cmnd->result = isp1020_return_status(sts);

    The "Cmnd" pointer happened to be a NULL somehow...

    Ivan.
    -
    To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
    the body of a message to majordomo@vger.kernel.org
    More majordomo info at http://vger.kernel.org/majordomo-info.html
    Please read the FAQ at http://www.tux.org/lkml/


  • Next message: Christoph Lameter: "Re: [RFC] Posix compliant behavior of CLOCK_PROCESS/THREAD_CPUTIME_ID"

    Relevant Pages

    • Re: JFS resize=0 problem in 2.6.0
      ... it should not break on top of the raid code. ... These bugfixes are in 2.6.1-pre1 I ... mounted filesystems. ... send the line "unsubscribe linux-kernel" in ...
      (Linux-Kernel)
    • Re: 2.4.27: md RAID1 oops on alpha revisited
      ... > The problem seems to be in qlogic isp1020 driver, not in the RAID code. ... So if it was a RAID1 bug, ... it does not affect RAID5. ... send the line "unsubscribe linux-kernel" in ...
      (Linux-Kernel)