* [PATCH v3] scsi: sd: fix missing put_disk() when device_add(&disk_dev) fails
@ 2026-03-30 1:49 Yang Xiuwei
2026-03-30 2:12 ` Ming Lei
0 siblings, 1 reply; 3+ messages in thread
From: Yang Xiuwei @ 2026-03-30 1:49 UTC (permalink / raw)
To: James Bottomley, Martin K . Petersen
Cc: linux-scsi, Ming Lei, Bart Van Assche, John Garry, stable,
Yang Xiuwei
If device_add(&sdkp->disk_dev) fails, put_device() runs
scsi_disk_release(), which frees the scsi_disk but leaves the gendisk
referenced. The device_add_disk() error path in sd_probe() calls
put_disk(gd); call put_disk(gd) here to mirror that cleanup.
Fixes: 265dfe8ebbab ("scsi: sd: Free scsi_disk device via put_device()")
Cc: stable@vger.kernel.org
Reviewed-by: John Garry <john.g.garry@oracle.com>
Signed-off-by: Yang Xiuwei <yangxiuwei@kylinos.cn>
---
v2: Add Fixes: and Cc: stable; add a short note on how the issue was found.
v3: Commit message and subject refined per review; add Reviewed-by from John Garry.
drivers/scsi/sd.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/drivers/scsi/sd.c b/drivers/scsi/sd.c
index 628a1d0a74ba..aba22060fcd5 100644
--- a/drivers/scsi/sd.c
+++ b/drivers/scsi/sd.c
@@ -4018,6 +4018,7 @@ static int sd_probe(struct scsi_device *sdp)
error = device_add(&sdkp->disk_dev);
if (error) {
put_device(&sdkp->disk_dev);
+ put_disk(gd);
goto out;
}
--
2.25.1
^ permalink raw reply related [flat|nested] 3+ messages in thread
* Re: [PATCH v3] scsi: sd: fix missing put_disk() when device_add(&disk_dev) fails
2026-03-30 1:49 [PATCH v3] scsi: sd: fix missing put_disk() when device_add(&disk_dev) fails Yang Xiuwei
@ 2026-03-30 2:12 ` Ming Lei
2026-03-30 3:06 ` Yang Xiuwei
0 siblings, 1 reply; 3+ messages in thread
From: Ming Lei @ 2026-03-30 2:12 UTC (permalink / raw)
To: Yang Xiuwei
Cc: James Bottomley, Martin K . Petersen, linux-scsi, Bart Van Assche,
John Garry, stable
On Mon, Mar 30, 2026 at 09:49:52AM +0800, Yang Xiuwei wrote:
> If device_add(&sdkp->disk_dev) fails, put_device() runs
> scsi_disk_release(), which frees the scsi_disk but leaves the gendisk
> referenced. The device_add_disk() error path in sd_probe() calls
> put_disk(gd); call put_disk(gd) here to mirror that cleanup.
>
> Fixes: 265dfe8ebbab ("scsi: sd: Free scsi_disk device via put_device()")
> Cc: stable@vger.kernel.org
> Reviewed-by: John Garry <john.g.garry@oracle.com>
> Signed-off-by: Yang Xiuwei <yangxiuwei@kylinos.cn>
> ---
> v2: Add Fixes: and Cc: stable; add a short note on how the issue was found.
> v3: Commit message and subject refined per review; add Reviewed-by from John Garry.
>
> drivers/scsi/sd.c | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/drivers/scsi/sd.c b/drivers/scsi/sd.c
> index 628a1d0a74ba..aba22060fcd5 100644
> --- a/drivers/scsi/sd.c
> +++ b/drivers/scsi/sd.c
> @@ -4018,6 +4018,7 @@ static int sd_probe(struct scsi_device *sdp)
> error = device_add(&sdkp->disk_dev);
> if (error) {
> put_device(&sdkp->disk_dev);
> + put_disk(gd);
> goto out;
> }
Another fix is to clear `sdkp` and `goto out_put`:
- clearing `sdkp` because the device is released already, and this way is `memory safe`
- `goto out_put` can release disk centrally
Thanks,
Ming
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH v3] scsi: sd: fix missing put_disk() when device_add(&disk_dev) fails
2026-03-30 2:12 ` Ming Lei
@ 2026-03-30 3:06 ` Yang Xiuwei
0 siblings, 0 replies; 3+ messages in thread
From: Yang Xiuwei @ 2026-03-30 3:06 UTC (permalink / raw)
To: James Bottomley, Martin K . Petersen
Cc: linux-scsi, Ming Lei, Bart Van Assche, John Garry, stable
Hi Ming,
On Mon, Mar 30, 2026 at 10:12:51AM +0800, Ming Lei wrote:
> Another fix is to clear `sdkp` and `goto out_put`:
>
> - clearing `sdkp` because the device is released already, and this way is `memory safe`
> - `goto out_put` can release disk centrally
>
Thank you for the suggestion.
As I understand it, the behaviour is the same as in v3: once
put_device(&sdkp->disk_dev) has run, scsi_disk_release() frees the
scsi_disk, so following with put_disk(gd) and goto out avoids the gendisk
leak and also avoids falling through to out_free, where kfree(sdkp) would
no longer be valid.
For now I am inclined to keep the v3 form for simplicity, but I am happy to
switch to your sdkp = NULL + goto out_put approach if you or the
maintainers would find that clearer or more consistent.
Thanks,
Yang Xiuwei
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2026-03-30 3:06 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2026-03-30 1:49 [PATCH v3] scsi: sd: fix missing put_disk() when device_add(&disk_dev) fails Yang Xiuwei
2026-03-30 2:12 ` Ming Lei
2026-03-30 3:06 ` Yang Xiuwei
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox