提交 3765138a 编写于 作者: C Christof Schmitt 提交者: James Bottomley

[SCSI] zfcp: Fix request list handling in error path

Fix the handling of the request list in the error path:
 - Use irqsave for the lock as in the good path.
 - Before removing the request, check if it is still in the list, a
   call to dismiss_all might have changed the list in between.
 - zfcp_qdio_send does not change the queue counters on failure,
   trying revert something is wrong, so remove this.
Signed-off-by: NChristof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: NSwen Schillig <swen@vnet.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>
上级 88f2a977
......@@ -770,7 +770,6 @@ static struct zfcp_fsf_req *zfcp_fsf_req_create(struct zfcp_adapter *adapter,
static int zfcp_fsf_req_send(struct zfcp_fsf_req *req)
{
struct zfcp_adapter *adapter = req->adapter;
struct zfcp_qdio_queue *req_q = &adapter->req_q;
unsigned long flags;
int idx;
......@@ -780,19 +779,15 @@ static int zfcp_fsf_req_send(struct zfcp_fsf_req *req)
list_add_tail(&req->list, &adapter->req_list[idx]);
spin_unlock_irqrestore(&adapter->req_list_lock, flags);
req->qdio_outb_usage = atomic_read(&req_q->count);
req->qdio_outb_usage = atomic_read(&adapter->req_q.count);
req->issued = get_clock();
if (zfcp_qdio_send(req)) {
/* Queues are down..... */
del_timer(&req->timer);
spin_lock(&adapter->req_list_lock);
zfcp_reqlist_remove(adapter, req);
spin_unlock(&adapter->req_list_lock);
/* undo changes in request queue made for this request */
atomic_add(req->sbal_number, &req_q->count);
req_q->first -= req->sbal_number;
req_q->first += QDIO_MAX_BUFFERS_PER_Q;
req_q->first %= QDIO_MAX_BUFFERS_PER_Q; /* wrap */
spin_lock_irqsave(&adapter->req_list_lock, flags);
/* lookup request again, list might have changed */
if (zfcp_reqlist_find_safe(adapter, req))
zfcp_reqlist_remove(adapter, req);
spin_unlock_irqrestore(&adapter->req_list_lock, flags);
zfcp_erp_adapter_reopen(adapter, 0, 116, req);
return -EIO;
}
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册