Re: [Qemu-devel] [sneak preview] major scsi overhaul

[Hannes Reinecke <hare@suse.de>]

Gerd Hoffmann wrote:
> On 11/26/09 12:20, Hannes Reinecke wrote:
>> Gerd Hoffmann wrote:
>>> /me looks at drivers/scsi/sd.c:sd_done()
>>>
>>> I can't see any sane way to tell linux that the request was too big.
>>>
>> residuals is the key:
>>
>> drivers/scsi/scsi.c:scsi_finish_command()
>>
>>
>>     good_bytes = scsi_bufflen(cmd);
>>          if (cmd->request->cmd_type != REQ_TYPE_BLOCK_PC) {
>>         int old_good_bytes = good_bytes;
>>         drv = scsi_cmd_to_driver(cmd);
>>         if (drv->done)
>>             good_bytes = drv->done(cmd);
> 
> drv->done() actually is sd_done() mentioned above.
> 
> sd_done() tries to figure how many sectors it actually got for serious
> errors.  I don't feel signaling "medium error" for the first sector
> behind our limit just because we'd like to have smaller requests.
> 
>>         /*
>>          * USB may not give sense identifying bad sector and
>>          * simply return a residue instead, so subtract off the
>>          * residue if drv->done() error processing indicates no
>>          * change to the completion length.
>>          */
>>         if (good_bytes == old_good_bytes)
>>             good_bytes -= scsi_get_resid(cmd);
> 
> Poor mans bad sector identification.  Same issue as above IMHO.  On top
> of that I wouldn't expect all other guest OSes having the same quirk in
> there to handle usb disks.
> 
Sigh. Communication seems to be tricky here.

Please, just ignore scsi_finish_command(), it's not doing anything
useful here.
Do skip this and study the next snipped, scsi_io_completion():

drivers/scsi/scsi_lib.c:scsi_io_completion()

	/*
	 * A number of bytes were successfully read.  If there
	 * are leftovers and there is some kind of error
	 * (result != 0), retry the rest.
	 */
	if (scsi_end_request(cmd, error, good_bytes, result == 0) == NULL)
		return;

scsi_end_request is being called with the number of bytes _actually processed_,
which might be less than the number of bytes requested.
And the remainder will be retried by the upper layers.

Cheers,

Hannes
-- 
Dr. Hannes Reinecke		      zSeries & Storage
hare@suse.de			      +49 911 74053 688
SUSE LINUX Products GmbH, Maxfeldstr. 5, 90409 Nürnberg
GF: Markus Rex, HRB 16746 (AG Nürnberg)