From efd1535270c1deb0487527bf0c3c827301a69c93 Mon Sep 17 00:00:00 2001 From: Bob Liu Date: Tue, 7 Jun 2016 10:43:15 -0400 Subject: [PATCH 1/2] xen-blkfront: don't call talk_to_blkback when already connected to blkback Sometimes blkfront may twice receive blkback_changed() notification (XenbusStateConnected) after migration, which will cause talk_to_blkback() to be called twice too and confuse xen-blkback. The flow is as follow: blkfront blkback blkfront_resume() > talk_to_blkback() > Set blkfront to XenbusStateInitialised front changed() > Connect() > Set blkback to XenbusStateConnected blkback_changed() > Skip talk_to_blkback() because frontstate == XenbusStateInitialised > blkfront_connect() > Set blkfront to XenbusStateConnected ----- And here we get another XenbusStateConnected notification leading to: ----- blkback_changed() > because now frontstate != XenbusStateInitialised talk_to_blkback() is also called again > blkfront state changed from XenbusStateConnected to XenbusStateInitialised (Which is not correct!) front_changed(): > Do nothing because blkback already in XenbusStateConnected Now blkback is in XenbusStateConnected but blkfront is still in XenbusStateInitialised - leading to no disks. Poking of the XenbusStateConnected state is allowed (to deal with block disk change) and has to be dealt with. The most likely cause of this bug are custom udev scripts hooking up the disks and then validating the size. Signed-off-by: Bob Liu Signed-off-by: Konrad Rzeszutek Wilk --- drivers/block/xen-blkfront.c | 15 ++++++++++++++- 1 file changed, 14 insertions(+), 1 deletion(-) diff --git a/drivers/block/xen-blkfront.c b/drivers/block/xen-blkfront.c index ca13df854639..6ba8891e8efe 100644 --- a/drivers/block/xen-blkfront.c +++ b/drivers/block/xen-blkfront.c @@ -2485,10 +2485,23 @@ static void blkback_changed(struct xenbus_device *dev, break; case XenbusStateConnected: - if (dev->state != XenbusStateInitialised) { + /* + * talk_to_blkback sets state to XenbusStateInitialised + * and blkfront_connect sets it to XenbusStateConnected + * (if connection went OK). + * + * If the backend (or toolstack) decides to poke at backend + * state (and re-trigger the watch by setting the state repeatedly + * to XenbusStateConnected (4)) we need to deal with this. + * This is allowed as this is used to communicate to the guest + * that the size of disk has changed! + */ + if ((dev->state != XenbusStateInitialised) && + (dev->state != XenbusStateConnected)) { if (talk_to_blkback(dev, info)) break; } + blkfront_connect(info); break; From 2a6f71ad99cabe436e70c3f5fcf58072cb3bc07f Mon Sep 17 00:00:00 2001 From: Bob Liu Date: Tue, 31 May 2016 16:59:17 +0800 Subject: [PATCH 2/2] xen-blkfront: fix resume issues after a migration After a migrate to another host (which may not have multiqueue support), the number of rings (block hardware queues) may be changed and the ring info structure will also be reallocated. This patch fixes two related bugs: * call blk_mq_update_nr_hw_queues() to make blk-core know the number of hardware queues have been changed. * Don't store rinfo pointer to hctx->driver_data, because rinfo may be reallocated so use hctx->queue_num to get the rinfo structure instead. Signed-off-by: Bob Liu Signed-off-by: Konrad Rzeszutek Wilk --- drivers/block/xen-blkfront.c | 20 ++++++++------------ 1 file changed, 8 insertions(+), 12 deletions(-) diff --git a/drivers/block/xen-blkfront.c b/drivers/block/xen-blkfront.c index 6ba8891e8efe..2e6d1e9c3345 100644 --- a/drivers/block/xen-blkfront.c +++ b/drivers/block/xen-blkfront.c @@ -874,8 +874,12 @@ static int blkif_queue_rq(struct blk_mq_hw_ctx *hctx, const struct blk_mq_queue_data *qd) { unsigned long flags; - struct blkfront_ring_info *rinfo = (struct blkfront_ring_info *)hctx->driver_data; + int qid = hctx->queue_num; + struct blkfront_info *info = hctx->queue->queuedata; + struct blkfront_ring_info *rinfo = NULL; + BUG_ON(info->nr_rings <= qid); + rinfo = &info->rinfo[qid]; blk_mq_start_request(qd->rq); spin_lock_irqsave(&rinfo->ring_lock, flags); if (RING_FULL(&rinfo->ring)) @@ -901,20 +905,9 @@ out_busy: return BLK_MQ_RQ_QUEUE_BUSY; } -static int blk_mq_init_hctx(struct blk_mq_hw_ctx *hctx, void *data, - unsigned int index) -{ - struct blkfront_info *info = (struct blkfront_info *)data; - - BUG_ON(info->nr_rings <= index); - hctx->driver_data = &info->rinfo[index]; - return 0; -} - static struct blk_mq_ops blkfront_mq_ops = { .queue_rq = blkif_queue_rq, .map_queue = blk_mq_map_queue, - .init_hctx = blk_mq_init_hctx, }; static int xlvbd_init_blk_queue(struct gendisk *gd, u16 sector_size, @@ -950,6 +943,7 @@ static int xlvbd_init_blk_queue(struct gendisk *gd, u16 sector_size, return PTR_ERR(rq); } + rq->queuedata = info; queue_flag_set_unlocked(QUEUE_FLAG_VIRT, rq); if (info->feature_discard) { @@ -2149,6 +2143,8 @@ static int blkfront_resume(struct xenbus_device *dev) return err; err = talk_to_blkback(dev, info); + if (!err) + blk_mq_update_nr_hw_queues(&info->tag_set, info->nr_rings); /* * We have to wait for the backend to switch to