On 10/03/2026 02:40, Benjamin Marzinski wrote:
+static int sd_mpath_probe(struct scsi_disk *sdkp)
+{
+       struct scsi_device *sdp = sdkp->device;
+       struct scsi_mpath_device *scsi_mpath_dev = sdp->scsi_mpath_dev;
+       struct device *dma_dev = sdp->host->dma_dev;
+       struct scsi_mpath_head *scsi_mpath_head =
+                               scsi_mpath_dev->scsi_mpath_head;
+       struct sd_mpath_disk *sd_mpath_disk;
+       struct mpath_head *mpath_head = scsi_mpath_head->mpath_head;
+       struct queue_limits lim;
+       struct gendisk *disk;
+       int error;
+
+       /*
+        * sd_mpath_disks_list is kept locked if no disk found.
+        * Otherwise an extra reference is taken.
+        */
Again, I personally think the logic is easier to follow when all the
locking isn't split over multiple functions.

Sure, but I am considering removing the mpath_disk structure, so things may change here anyway. Removing mpath_disk should simplify things for the nvme driver transition.


+       sd_mpath_disk = sd_mpath_find_disk(sdp);
+       if (sd_mpath_disk) {
+               mutex_lock(&sd_mpath_disk->lock);
+               sd_mpath_disk->disk_count++;
+               mutex_unlock(&sd_mpath_disk->lock);
+               goto found;
+       }
+
+       sd_mpath_disk = kzalloc(sizeof(*sd_mpath_disk), GFP_KERNEL);
+       if (!sd_mpath_disk) {
+               error = -ENOMEM;
+               goto out_unlock;
+       }
+
+       sd_mpath_disk->scsi_mpath_head = scsi_mpath_head;
+       device_initialize(&sd_mpath_disk->dev);
+       mutex_init(&sd_mpath_disk->lock);
+       sd_mpath_disk->dev.class = &sd_mpath_disk_class;
+
+       blk_set_stacking_limits(&lim);
+       lim.dma_alignment = 3;
+       lim.features |= BLK_FEAT_IO_STAT | BLK_FEAT_NOWAIT |
+               BLK_FEAT_POLL | BLK_FEAT_ATOMIC_WRITES;
+
+       sd_mpath_disk->mpath_disk = mpath_alloc_head_disk(&lim,
+                                               dev_to_node(dma_dev));
+       if (!sd_mpath_disk->mpath_disk) {
+               error = -ENOMEM;
+               goto out_free_disk;
+       }
+       disk = sd_mpath_disk->mpath_disk->disk;
+       mpath_get_head(mpath_head); /* undone in mpath_free_disk() */
+
+       sd_mpath_disk->mpath_disk->mpath_head = mpath_head;
+       sd_mpath_disk->mpath_disk->parent = &sd_mpath_disk->dev;
+
+       error = ida_alloc(&sd_index_ida, GFP_KERNEL);
+       if (error < 0) {
+               sdev_printk(KERN_WARNING, sdp, "sd_probe: memory exhausted.\n");
+               goto out_put_disk;
+       }
+       sd_mpath_disk->disk_index = error;
+       error = sd_format_disk_name("sd", sd_mpath_disk->disk_index,
+                               disk->disk_name, DISK_NAME_LEN);
+       if (error)
+               goto out_free_index;
+
+       error = dev_set_name(&sd_mpath_disk->dev, "%s",
+                               dev_name(&scsi_mpath_head->dev));
+       if (error)
+               goto out_free_index;
+
+       /* undone in sd_mpath_disk_release() */
+       scsi_mpath_get_head(scsi_mpath_head);
+
+       error = device_add(&sd_mpath_disk->dev);
+       if (error) {
+               put_device(&sd_mpath_disk->dev);
+               goto out_unlock;
We should clean up when we fail here, instead of just unlocking without
fully setting things up.

I think that the release function is called from put_device(), which should do the class tidy up. Something similar is done in sd_probe() for the disk_dev.

Thanks,
John


Reply via email to