RE: Latest Adaptec AACRAID driver doesn't compile on latest kernel

From: Martin Drab (
Date: 08/02/05

  • Next message: Christoph Hellwig: "Re: Latest Adaptec AACRAID driver doesn't compile on latest kernel"
    Date:	Tue, 2 Aug 2005 14:31:55 +0200 (CEST)
    To: "Salyzyn, Mark" <>

    >> -----Original Message-----
    >> From: Martin Drab []
    >> Sent: Monday, August 01, 2005 6:31 PM
    >> To: Salyzyn, Mark
    >> Subject: Latest Adaptec AACRAID driver doesn't compile on latest kernel
    >> Hi, Mark,
    >> because of the following kernel patch issued in 2.6.13-rc4:
    >> -------------
    >> commit 8d115f845a0bd59cd263e791f739964f42b7b0e8
    >> Author: Christoph Hellwig <>
    >> Date: Sun Jun 19 13:42:05 2005 +0200
    >> [SCSI] remove scsi_cmnd->state
    >> We never look at it except for the old megaraid driver that abuses it
    >> for sending internal commands. That usage can be fixed easily because
    >> those internal commands are single-threaded by a mutex and we can easily
    >> use a completion there.
    >> Signed-off-by: James Bottomley <>
    >> -------------
    >> the Adaptec AACRAID driver 1.1.5-2405 that you sent me is unable to
    >> compile. Attached patch fixes the problem by simply removing the
    >> condition where it is used. It seems to be just some kind of debug info
    >> guard. And since there doesn't seem to be any other way to obtain the
    >> state we would just have to suffice with this, or remove the debug info
    >> printing entirely.
    >> Martin

    On Tue, 2 Aug 2005, Salyzyn, Mark wrote:

    > Thanks MArtin, I have this change in 1.1.5-2408 wrapped with an ifdef
    > for kernels 2.6.13 and higher to permit compile and pending an
    > investigation to determine a suitable substitute in the 2.6.13+ tree.

    Is it downloadable somewhere? If not, could you please send it to me?
    > The check is done because when commands are completed, they still reside
    > on the device queue and this filters out those that are on the road to
    > completion. I am not sure this is still the case in 2.6.13. It is not a
    > debug check; it is a means to determine that all the commands have
    > quiesced to the adapter. We do this because several conditions that
    > cause the tripping of the SCSI bus timeout with this adapter are a
    > result of the adapter performing some RAID recovery, flush or build
    > tasks and stuffing the adapter with commands during these operations
    > only serves to exasperate the situation and cause devices to go offline.
    > I have referred to this condition as Adapter reticence.
    > The side effect of this change may be that the reset handler will wait
    > the full 60 seconds for commands to complete in the adapter rather than
    > exiting as soon as they have all completed. This will still serve the
    > purpose to alleviate this situation. However, as this code is extending
    > the 60 second SCSI timeout, we get nervous that this change will result
    > in increased number of servers dropping their network connections.

    Hmm. I see. So perhaps you may ask someone (James, Christoph, ...?) to
    revert that patch, because the scsi_cmnd->state is needed here for this
    purpose. It was removed just because nobody else was using it (at least
    thats what the patch description says). So, if you explain that it is
    necessary for this purpose, it could either be reverted, or you may be
    told of a way to get around it. (That's a result of that the driver didn't
    yet make it into the mainline. Before it does, mainline may change to a
    level that it would no longer be possible to apply it. :()

    To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
    the body of a message to
    More majordomo info at
    Please read the FAQ at

  • Next message: Christoph Hellwig: "Re: Latest Adaptec AACRAID driver doesn't compile on latest kernel"

    Relevant Pages

    • [long] Oubtfuls questions on test199 (from Redbook)???
      ... A system administrator has just moved an Ethernet adapter to another ... The netstat and ifconfig commands show the interface just as ... The adapter speed has changed to 10 Mbps. ... Kbytes allocated to mbufs ...
    • Re: SqlDataAdapter Questions
      ... No 'default' commands are generated by the adapter. ... If you don't use the Update method, then yes, there is no point in defining ... UpdateCommand, DeleteCommand, and InsertCommand. ...
    • Re: Updated dataset back to the database
      ... "So even if the adapter is disposed and you can't change that - ... Ryan, Would you provide a sample code for this scenario? ... > process is done for each of the other types of commands - insert and update. ... > configured correctly or not disposing of the first one and using it again - ...
    • Re: Networking Blues, can someone make sense of this?
      ... >>> It seems you made a mistake when typing the commands I gave you. ... >> There are no entries in the list. ... >> Ethernet adapter Local Area Connection 4: ... I recommend that you now do this on Jenn: ...
    • Re: Getting back to the last position in the code.
      ... XEmacs push the mark first (like search commands) all by themselves. ... Text insertion commands also push a mark at the opposite side of the ... Adrian Aichner ...