xen-blkfront: use a different scatterlist for each request
authorRoger Pau Monne <roger.pau@citrix.com>
Thu, 2 May 2013 08:58:50 +0000 (10:58 +0200)
committerKonrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Wed, 8 May 2013 12:46:51 +0000 (08:46 -0400)
In blkif_queue_request blkfront iterates over the scatterlist in order
to set the segments of the request, and in blkif_completion blkfront
iterates over the raw request, which makes it hard to know the exact
position of the source and destination memory positions.

This can be solved by allocating a scatterlist for each request, that
will be keep until the request is finished, allowing us to copy the
data back to the original memory without having to iterate over the
raw request.

Oracle-Bug: 16660413 - LARGE ASYNCHRONOUS READS APPEAR BROKEN ON 2.6.39-400
CC: stable@vger.kernel.org
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reported-and-Tested-by: Anne Milicia <anne.milicia@oracle.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
drivers/block/xen-blkfront.c

index 82d63d5b1750d1ac8c54f7c366edceae36367a9a..bac8cf31319b6b269840bd03583679c0cb4aec78 100644 (file)
@@ -76,6 +76,7 @@ struct blk_shadow {
        struct request *request;
        struct grant **grants_used;
        struct grant **indirect_grants;
+       struct scatterlist *sg;
 };
 
 struct split_bio {
@@ -113,7 +114,6 @@ struct blkfront_info
        enum blkif_state connected;
        int ring_ref;
        struct blkif_front_ring ring;
-       struct scatterlist *sg;
        unsigned int evtchn, irq;
        struct request_queue *rq;
        struct work_struct work;
@@ -438,7 +438,7 @@ static int blkif_queue_request(struct request *req)
                       req->nr_phys_segments > BLKIF_MAX_SEGMENTS_PER_REQUEST);
                BUG_ON(info->max_indirect_segments &&
                       req->nr_phys_segments > info->max_indirect_segments);
-               nseg = blk_rq_map_sg(req->q, req, info->sg);
+               nseg = blk_rq_map_sg(req->q, req, info->shadow[id].sg);
                ring_req->u.rw.id = id;
                if (nseg > BLKIF_MAX_SEGMENTS_PER_REQUEST) {
                        /*
@@ -469,7 +469,7 @@ static int blkif_queue_request(struct request *req)
                        }
                        ring_req->u.rw.nr_segments = nseg;
                }
-               for_each_sg(info->sg, sg, nseg, i) {
+               for_each_sg(info->shadow[id].sg, sg, nseg, i) {
                        fsect = sg->offset >> 9;
                        lsect = fsect + (sg->length >> 9) - 1;
 
@@ -914,8 +914,6 @@ static void blkif_free(struct blkfront_info *info, int suspend)
        }
        BUG_ON(info->persistent_gnts_c != 0);
 
-       kfree(info->sg);
-       info->sg = NULL;
        for (i = 0; i < BLK_RING_SIZE; i++) {
                /*
                 * Clear persistent grants present in requests already
@@ -953,6 +951,8 @@ free_shadow:
                info->shadow[i].grants_used = NULL;
                kfree(info->shadow[i].indirect_grants);
                info->shadow[i].indirect_grants = NULL;
+               kfree(info->shadow[i].sg);
+               info->shadow[i].sg = NULL;
        }
 
        /* No more gnttab callback work. */
@@ -979,12 +979,9 @@ static void blkif_completion(struct blk_shadow *s, struct blkfront_info *info,
                             struct blkif_response *bret)
 {
        int i = 0;
-       struct bio_vec *bvec;
-       struct req_iterator iter;
-       unsigned long flags;
+       struct scatterlist *sg;
        char *bvec_data;
        void *shared_data;
-       unsigned int offset = 0;
        int nseg;
 
        nseg = s->req.operation == BLKIF_OP_INDIRECT ?
@@ -997,19 +994,16 @@ static void blkif_completion(struct blk_shadow *s, struct blkfront_info *info,
                 * than PAGE_SIZE, we have to keep track of the current offset,
                 * to be sure we are copying the data from the right shared page.
                 */
-               rq_for_each_segment(bvec, s->request, iter) {
-                       BUG_ON((bvec->bv_offset + bvec->bv_len) > PAGE_SIZE);
-                       if (bvec->bv_offset < offset)
-                               i++;
-                       BUG_ON(i >= nseg);
+               for_each_sg(s->sg, sg, nseg, i) {
+                       BUG_ON(sg->offset + sg->length > PAGE_SIZE);
                        shared_data = kmap_atomic(
                                pfn_to_page(s->grants_used[i]->pfn));
-                       bvec_data = bvec_kmap_irq(bvec, &flags);
-                       memcpy(bvec_data, shared_data + bvec->bv_offset,
-                               bvec->bv_len);
-                       bvec_kunmap_irq(bvec_data, &flags);
+                       bvec_data = kmap_atomic(sg_page(sg));
+                       memcpy(bvec_data   + sg->offset,
+                              shared_data + sg->offset,
+                              sg->length);
+                       kunmap_atomic(bvec_data);
                        kunmap_atomic(shared_data);
-                       offset = bvec->bv_offset + bvec->bv_len;
                }
        }
        /* Add the persistent grant into the list of free grants */
@@ -1656,10 +1650,6 @@ static int blkfront_setup_indirect(struct blkfront_info *info)
                                                  xen_blkif_max_segments);
                segs = info->max_indirect_segments;
        }
-       info->sg = kzalloc(sizeof(info->sg[0]) * segs, GFP_KERNEL);
-       if (info->sg == NULL)
-               goto out_of_memory;
-       sg_init_table(info->sg, segs);
 
        err = fill_grant_buffer(info, (segs + INDIRECT_GREFS(segs)) * BLK_RING_SIZE);
        if (err)
@@ -1669,26 +1659,29 @@ static int blkfront_setup_indirect(struct blkfront_info *info)
                info->shadow[i].grants_used = kzalloc(
                        sizeof(info->shadow[i].grants_used[0]) * segs,
                        GFP_NOIO);
+               info->shadow[i].sg = kzalloc(sizeof(info->shadow[i].sg[0]) * segs, GFP_NOIO);
                if (info->max_indirect_segments)
                        info->shadow[i].indirect_grants = kzalloc(
                                sizeof(info->shadow[i].indirect_grants[0]) *
                                INDIRECT_GREFS(segs),
                                GFP_NOIO);
                if ((info->shadow[i].grants_used == NULL) ||
+                       (info->shadow[i].sg == NULL) ||
                     (info->max_indirect_segments &&
                     (info->shadow[i].indirect_grants == NULL)))
                        goto out_of_memory;
+               sg_init_table(info->shadow[i].sg, segs);
        }
 
 
        return 0;
 
 out_of_memory:
-       kfree(info->sg);
-       info->sg = NULL;
        for (i = 0; i < BLK_RING_SIZE; i++) {
                kfree(info->shadow[i].grants_used);
                info->shadow[i].grants_used = NULL;
+               kfree(info->shadow[i].sg);
+               info->shadow[i].sg = NULL;
                kfree(info->shadow[i].indirect_grants);
                info->shadow[i].indirect_grants = NULL;
        }