On Oct 28, 2008, at 7:59 AM, Boaz Harrosh wrote:
Boaz Harrosh wrote:
Boaz Harrosh wrote:
...
[snip]
...
if the attached file is the code you tested then it is wrong look
here:
+
+ if (service->srv_reply.residual) {
+ service->req->data_len = 0;
+ service->req->next_rq->data_len = service->srv_reply.residual;
+ } else {
+ service->req->data_len = 0;
+ service->req->next_rq->data_len = 0;
+ }
+
Move above to after the blk_end_bidi_request call
+ blk_end_bidi_request(service->req, 0, blk_rq_bytes(service->req),
+ service->req->next_rq ? blk_rq_bytes(service->req-
>next_rq) : 0);
You must call blk_end_bidi_request before you change service->req-
>data_len
to hold the residual (or 0). Otherwise you damage the request.
+ service->req->end_io(service->req, 0);
Hmm, on re inspection req->end_io(...) called here has the same
problem.
Are you sure it's needed?
No you do not call req->end_io(..) directly. It eventual gets called
by blk_end_bidi_request() inside end_that_request_last(). (Once
all byte are completed)
Yes, that is correct. Thank you for the correction - end_io() call has
removed.
Other than setting req->errors field and req->next_rq->errors, there
are not
many things to do in the callback prior to calling
blk_end_bidi_request().
With the clean-up, updated patch will be followed.
Thank you,
Seokmann
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
