[PATCH] nvme: fix potential memory leaks in nvme_cdev_add

Chaitanya Kulkarni Chaitanya.Kulkarni at wdc.com
Fri May 21 14:23:08 PDT 2021


On 5/21/21 00:45, Guoqing Jiang wrote:
> We need to call put_device if cdev_device_add failed, otherwise
> kmemleak has below report.
>
> [<0000000024c71758>] kmem_cache_alloc_trace+0x233/0x480
> [<00000000ad2813ed>] device_add+0x7ff/0xe10
> [<0000000035bc54c4>] cdev_device_add+0x72/0xa0
> [<000000006c9aa1e8>] nvme_cdev_add+0xa9/0xf0 [nvme_core]
> [<000000003c4d492d>] nvme_mpath_set_live+0x251/0x290 [nvme_core]
> [<00000000889a58da>] nvme_mpath_add_disk+0x268/0x320 [nvme_core]
> [<00000000192e7161>] nvme_alloc_ns+0x669/0xac0 [nvme_core]
> [<000000007a1a6041>] nvme_validate_or_alloc_ns+0x156/0x280 [nvme_core]
> [<000000003a763c35>] nvme_scan_work+0x221/0x3c0 [nvme_core]
> [<000000009ff10706>] process_one_work+0x5cf/0xb10
> [<000000000644ee25>] worker_thread+0x7a/0x680
> [<00000000285ebd2f>] kthread+0x1c6/0x210
> [<00000000e297c6ea>] ret_from_fork+0x22/0x30
>
> Fixes: 2637baed7801 ("nvme: introduce generic per-namespace chardev")
> Signed-off-by: Guoqing Jiang <jiangguoqing at kylinos.cn>
>

Looks good.

Reviewed-by: Chaitanya Kulkarni <chaitanya.kulkarni at wdc.com>





More information about the Linux-nvme mailing list