NVMe: Cancel outstanding IOs on queue deletion

[platform/adaptation/renesas_rcar/renesas_kernel.git] / drivers / block / nvme.c

diff --git a/drivers/block/nvme.c b/drivers/block/nvme.c
index 0ba6b7c..f9ad514 100644 (file)
--- a/drivers/block/nvme.c
+++ b/drivers/block/nvme.c
@@ -868,11 +868,51 @@ static int nvme_set_features(struct nvme_dev *dev, unsigned fid,
        return nvme_submit_admin_cmd(dev, &c, result);
 }
 
+/**
+ * nvme_cancel_ios - Cancel outstanding I/Os
+ * @queue: The queue to cancel I/Os on
+ * @timeout: True to only cancel I/Os which have timed out
+ */
+static void nvme_cancel_ios(struct nvme_queue *nvmeq, bool timeout)
+{
+       int depth = nvmeq->q_depth - 1;
+       struct nvme_cmd_info *info = nvme_cmd_info(nvmeq);
+       unsigned long now = jiffies;
+       int cmdid;
+
+       for_each_set_bit(cmdid, nvmeq->cmdid_data, depth) {
+               void *ctx;
+               nvme_completion_fn fn;
+               static struct nvme_completion cqe = {
+                       .status = cpu_to_le16(NVME_SC_ABORT_REQ) << 1,
+               };
+
+               if (timeout && !time_after(now, info[cmdid].timeout))
+                       continue;
+               dev_warn(nvmeq->q_dmadev, "Cancelling I/O %d\n", cmdid);
+               ctx = cancel_cmdid(nvmeq, cmdid, &fn);
+               fn(nvmeq->dev, ctx, &cqe);
+       }
+}
+
+static void nvme_free_queue_mem(struct nvme_queue *nvmeq)
+{
+       dma_free_coherent(nvmeq->q_dmadev, CQ_SIZE(nvmeq->q_depth),
+                               (void *)nvmeq->cqes, nvmeq->cq_dma_addr);
+       dma_free_coherent(nvmeq->q_dmadev, SQ_SIZE(nvmeq->q_depth),
+                                       nvmeq->sq_cmds, nvmeq->sq_dma_addr);
+       kfree(nvmeq);
+}
+
 static void nvme_free_queue(struct nvme_dev *dev, int qid)
 {
        struct nvme_queue *nvmeq = dev->queues[qid];
        int vector = dev->entry[nvmeq->cq_vector].vector;
 
+       spin_lock_irq(&nvmeq->q_lock);
+       nvme_cancel_ios(nvmeq, false);
+       spin_unlock_irq(&nvmeq->q_lock);
+
        irq_set_affinity_hint(vector, NULL);
        free_irq(vector, nvmeq);
 
@@ -882,11 +922,7 @@ static void nvme_free_queue(struct nvme_dev *dev, int qid)
                adapter_delete_cq(dev, qid);
        }
 
-       dma_free_coherent(nvmeq->q_dmadev, CQ_SIZE(nvmeq->q_depth),
-                               (void *)nvmeq->cqes, nvmeq->cq_dma_addr);
-       dma_free_coherent(nvmeq->q_dmadev, SQ_SIZE(nvmeq->q_depth),
-                                       nvmeq->sq_cmds, nvmeq->sq_dma_addr);
-       kfree(nvmeq);
+       nvme_free_queue_mem(nvmeq);
 }
 
 static struct nvme_queue *nvme_alloc_queue(struct nvme_dev *dev, int qid,
@@ -982,7 +1018,7 @@ static __devinit struct nvme_queue *nvme_create_queue(struct nvme_dev *dev,
 
 static int __devinit nvme_configure_admin_queue(struct nvme_dev *dev)
 {
-       int result;
+       int result = 0;
        u32 aqa;
        u64 cap;
        unsigned long timeout;
@@ -1012,17 +1048,22 @@ static int __devinit nvme_configure_admin_queue(struct nvme_dev *dev)
        timeout = ((NVME_CAP_TIMEOUT(cap) + 1) * HZ / 2) + jiffies;
        dev->db_stride = NVME_CAP_STRIDE(cap);
 
-       while (!(readl(&dev->bar->csts) & NVME_CSTS_RDY)) {
+       while (!result && !(readl(&dev->bar->csts) & NVME_CSTS_RDY)) {
                msleep(100);
                if (fatal_signal_pending(current))
-                       return -EINTR;
+                       result = -EINTR;
                if (time_after(jiffies, timeout)) {
                        dev_err(&dev->pci_dev->dev,
                                "Device not ready; aborting initialisation\n");
-                       return -ENODEV;
+                       result = -ENODEV;
                }
        }
 
+       if (result) {
+               nvme_free_queue_mem(nvmeq);
+               return result;
+       }
+
        result = queue_request_irq(dev, nvmeq, "nvme admin");
        dev->queues[0] = nvmeq;
        return result;
@@ -1226,26 +1267,6 @@ static const struct block_device_operations nvme_fops = {
        .compat_ioctl   = nvme_ioctl,
 };
 
-static void nvme_timeout_ios(struct nvme_queue *nvmeq)
-{
-       int depth = nvmeq->q_depth - 1;
-       struct nvme_cmd_info *info = nvme_cmd_info(nvmeq);
-       unsigned long now = jiffies;
-       int cmdid;
-
-       for_each_set_bit(cmdid, nvmeq->cmdid_data, depth) {
-               void *ctx;
-               nvme_completion_fn fn;
-               static struct nvme_completion cqe = { .status = cpu_to_le16(NVME_SC_ABORT_REQ) << 1, };
-
-               if (!time_after(now, info[cmdid].timeout))
-                       continue;
-               dev_warn(nvmeq->q_dmadev, "Timing out I/O %d\n", cmdid);
-               ctx = cancel_cmdid(nvmeq, cmdid, &fn);
-               fn(nvmeq->dev, ctx, &cqe);
-       }
-}
-
 static void nvme_resubmit_bios(struct nvme_queue *nvmeq)
 {
        while (bio_list_peek(&nvmeq->sq_cong)) {
@@ -1277,7 +1298,7 @@ static int nvme_kthread(void *data)
                                spin_lock_irq(&nvmeq->q_lock);
                                if (nvme_process_cq(nvmeq))
                                        printk("process_cq did something\n");
-                               nvme_timeout_ios(nvmeq);
+                               nvme_cancel_ios(nvmeq, true);
                                nvme_resubmit_bios(nvmeq);
                                spin_unlock_irq(&nvmeq->q_lock);
                        }
@@ -1539,8 +1560,6 @@ static int nvme_dev_remove(struct nvme_dev *dev)
        list_del(&dev->node);
        spin_unlock(&dev_list_lock);
 
-       /* TODO: wait all I/O finished or cancel them */
-
        list_for_each_entry_safe(ns, next, &dev->namespaces, list) {
                list_del(&ns->list);
                del_gendisk(ns->disk);
@@ -1576,15 +1595,33 @@ static void nvme_release_prp_pools(struct nvme_dev *dev)
        dma_pool_destroy(dev->prp_small_pool);
 }
 
-/* XXX: Use an ida or something to let remove / add work correctly */
-static void nvme_set_instance(struct nvme_dev *dev)
+static DEFINE_IDA(nvme_instance_ida);
+
+static int nvme_set_instance(struct nvme_dev *dev)
 {
-       static int instance;
-       dev->instance = instance++;
+       int instance, error;
+
+       do {
+               if (!ida_pre_get(&nvme_instance_ida, GFP_KERNEL))
+                       return -ENODEV;
+
+               spin_lock(&dev_list_lock);
+               error = ida_get_new(&nvme_instance_ida, &instance);
+               spin_unlock(&dev_list_lock);
+       } while (error == -EAGAIN);
+
+       if (error)
+               return -ENODEV;
+
+       dev->instance = instance;
+       return 0;
 }
 
 static void nvme_release_instance(struct nvme_dev *dev)
 {
+       spin_lock(&dev_list_lock);
+       ida_remove(&nvme_instance_ida, dev->instance);
+       spin_unlock(&dev_list_lock);
 }
 
 static int __devinit nvme_probe(struct pci_dev *pdev,
@@ -1617,7 +1654,10 @@ static int __devinit nvme_probe(struct pci_dev *pdev,
        pci_set_drvdata(pdev, dev);
        dma_set_mask(&pdev->dev, DMA_BIT_MASK(64));
        dma_set_coherent_mask(&pdev->dev, DMA_BIT_MASK(64));
-       nvme_set_instance(dev);
+       result = nvme_set_instance(dev);
+       if (result)
+               goto disable;
+
        dev->entry[0].vector = pdev->irq;
 
        result = nvme_setup_prp_pools(dev);