On 17/10/18 15:00, Matthew Wilcox wrote:
> On Wed, Oct 17, 2018 at 02:49:50PM +0200, Johannes Thumshirn wrote:
>> On 17/10/18 14:34, YueHaibing wrote:
>>> - srb_tag = le32_to_cpu(scsiqp->srb_tag);
>>> scp = scsi_host_find_tag(boardp->shost, scsiqp->srb_tag);
>>
>> Shouldn't this be:
>> scp = scsi_host_find_tag(boardp->shost,
>> le32_to_cpu(scsiqp->srb_tag));
>
> I don't think so. Look at how scsiqp->srb_tag is set:
>
> adv_build_req(struct asc_board *boardp, struct scsi_cmnd *scp,
> ADV_SCSI_REQ_Q **adv_scsiqpp)
> {
> + u32 srb_tag = scp->request->tag;
> ...
> + scsiqp->srb_tag = srb_tag;
>
> If we're not converting it to le32 on the way in, we probably don't want
> to convert it back on the way out.
Yeah right, didn't check the submission path. It was just the removed
srb_tag = le32_to_cpu(scsiqp->srb_tag);
that triggered here.
> Please don't make me power up my PA-RISC machine with an Advansys card
> in it, just to find out ...
Now that your saying it...
--
Johannes Thumshirn SUSE Labs
jthumshirn@xxxxxxx +49 911 74053 689
SUSE LINUX GmbH, Maxfeldstr. 5, 90409 Nürnberg
GF: Felix Imendörffer, Jane Smithard, Graham Norton
HRB 21284 (AG Nürnberg)
Key fingerprint = EC38 9CAB C2C4 F25D 8600 D0D0 0393 969D 2D76 0850
