Christoph Lameter <[email protected]> reported this
compile error in lk 2.5.10-dj1:
> ide-scsi.c:837: unknown field `abort' specified in initializer
> ide-scsi.c:837: warning: initialization from incompatible pointer type
> ide-scsi.c:838: unknown field `reset' specified in initializer
> ide-scsi.c:838: warning: initialization from incompatible pointer type
Below is a patch which attempts to do the right thing:
it wires up the scsi new eh handling and attempts to do
a device reset.
It has been tested and oopses in start_request() inside
ide.c when a device reset is issued :-) Since the previous
ide-scsi logic just ignored scsi error handling, it isn't
really a whole lot worse. There is a "fix me" at the
appropriate point.
Doug Gilbert
--- linux/drivers/scsi/ide-scsi.c Sat Apr 27 14:52:08 2002
+++ linux/drivers/scsi/ide-scsi.c2510dj1hack Sat Apr 27 17:36:07 2002
@@ -804,14 +804,20 @@
return 0;
}
-int idescsi_abort (Scsi_Cmnd *cmd)
+/* try to do correct thing for scsi subsystem's new eh */
+int idescsi_device_reset (Scsi_Cmnd *cmd)
{
- return SCSI_ABORT_SNOOZE;
-}
+ ide_drive_t *drive = idescsi_drives[cmd->target];
+ struct request req;
-int idescsi_reset (Scsi_Cmnd *cmd, unsigned int resetflags)
-{
- return SCSI_RESET_SUCCESS;
+ ide_init_drive_cmd(&req);
+ req.flags = REQ_SPECIAL;
+/* FIX ME, the next executable line causes on oops in lk 2.5.10-dj1
+ * [code copied from ide-cd's ide_cdrom_reset(), does it work?]
+ */
+ ide_do_drive_cmd(drive, &req, ide_wait);
+
+ return SUCCESS;
}
int idescsi_bios (Disk *disk, kdev_t dev, int *parm)
@@ -834,8 +840,8 @@
info: idescsi_info,
ioctl: idescsi_ioctl,
queuecommand: idescsi_queue,
- abort: idescsi_abort,
- reset: idescsi_reset,
+ eh_device_reset_handler:
+ idescsi_device_reset,
bios_param: idescsi_bios,
can_queue: 10,
this_id: -1,
Uz.ytkownik Douglas Gilbert napisa?:
> Christoph Lameter <[email protected]> reported this
> compile error in lk 2.5.10-dj1:
>
>>ide-scsi.c:837: unknown field `abort' specified in initializer
>>ide-scsi.c:837: warning: initialization from incompatible pointer type
>>ide-scsi.c:838: unknown field `reset' specified in initializer
>>ide-scsi.c:838: warning: initialization from incompatible pointer type
>
>
> Below is a patch which attempts to do the right thing:
> it wires up the scsi new eh handling and attempts to do
> a device reset.
>
> It has been tested and oopses in start_request() inside
> ide.c when a device reset is issued :-) Since the previous
> ide-scsi logic just ignored scsi error handling, it isn't
> really a whole lot worse. There is a "fix me" at the
> appropriate point.
>
> Doug Gilbert
Well the "hidden" queueing of some commands on the
device request queue is what is getting you here. It's subject
to be gone, so this problem will be resolved.