From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: From: Greg Kroah-Hartman To: linux-kernel@vger.kernel.org Cc: Greg Kroah-Hartman , stable@vger.kernel.org, Keith Busch , Christoph Hellwig , Jens Axboe Subject: [PATCH 4.2 105/110] NVMe: Fix memory leak on retried commands Date: Fri, 6 Nov 2015 11:19:52 -0800 Message-Id: <20151106191709.450437186@linuxfoundation.org> In-Reply-To: <20151106191703.247930828@linuxfoundation.org> References: <20151106191703.247930828@linuxfoundation.org> MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-15 Sender: linux-kernel-owner@vger.kernel.org List-ID: 4.2-stable review patch. If anyone has any objections, please let me know. ------------------ From: Keith Busch commit 0dfc70c33409afc232ef0b9ec210535dfbf9bc61 upstream. Resources are reallocated for requeued commands, so unmap and release the iod for the failed command. It's a pretty bad memory leak and causes a kernel hang if you remove a drive because of a busy dma pool. You'll get messages spewing like this: nvme 0000:xx:xx.x: dma_pool_destroy prp list 256, ffff880420dec000 busy and lock up pci and the driver since removal never completes while holding a lock. Signed-off-by: Keith Busch Reviewed-by: Christoph Hellwig Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman --- drivers/block/nvme-core.c | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-) --- a/drivers/block/nvme-core.c +++ b/drivers/block/nvme-core.c @@ -597,6 +597,7 @@ static void req_completion(struct nvme_q struct nvme_iod *iod = ctx; struct request *req = iod_get_private(iod); struct nvme_cmd_info *cmd_rq = blk_mq_rq_to_pdu(req); + bool requeue = false; u16 status = le16_to_cpup(&cqe->status) >> 1; @@ -605,12 +606,13 @@ static void req_completion(struct nvme_q && (jiffies - req->start_time) < req->timeout) { unsigned long flags; + requeue = true; blk_mq_requeue_request(req); spin_lock_irqsave(req->q->queue_lock, flags); if (!blk_queue_stopped(req->q)) blk_mq_kick_requeue_list(req->q); spin_unlock_irqrestore(req->q->queue_lock, flags); - return; + goto release_iod; } if (req->cmd_type == REQ_TYPE_DRV_PRIV) { if (cmd_rq->ctx == CMD_CTX_CANCELLED) @@ -631,7 +633,7 @@ static void req_completion(struct nvme_q dev_warn(nvmeq->dev->dev, "completing aborted command with status:%04x\n", status); - + release_iod: if (iod->nents) { dma_unmap_sg(nvmeq->dev->dev, iod->sg, iod->nents, rq_data_dir(req) ? DMA_TO_DEVICE : DMA_FROM_DEVICE); @@ -644,7 +646,8 @@ static void req_completion(struct nvme_q } nvme_free_iod(nvmeq->dev, iod); - blk_mq_complete_request(req); + if (likely(!requeue)) + blk_mq_complete_request(req); } /* length is in bytes. gfp flags indicates whether we may sleep. */