gb_vibrator_probe() maps any device_create() failure to -EINVAL. This loses the real errno returned by the driver core, such as -ENOMEM, and makes probe failures harder to diagnose correctly.
Return PTR_ERR(dev) instead so callers receive the actual failure reason while preserving the existing cleanup path.
Signed-off-by: Alfie Varghese alfievarghese22@gmail.com --- drivers/staging/greybus/vibrator.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/staging/greybus/vibrator.c b/drivers/staging/greybus/vibrator.c index 0ec4d317c..763c234fb 100644 --- a/drivers/staging/greybus/vibrator.c +++ b/drivers/staging/greybus/vibrator.c @@ -161,7 +161,7 @@ static int gb_vibrator_probe(struct gb_bundle *bundle, dev = device_create(&vibrator_class, &bundle->dev, MKDEV(0, 0), vib, "vibrator%d", vib->minor); if (IS_ERR(dev)) { - retval = -EINVAL; + retval = PTR_ERR(dev); goto err_ida_remove; } vib->dev = dev;
Gentle ping on this patch from May 28.
Apologies for the accidental duplicate send; a second copy went out shortly after this one by mistake. This thread is the canonical one to review.
Happy to resend or rework if anything is needed.
On Thu, Jun 18, 2026 at 04:01:32AM +0000, Alfie Varghese wrote:
Gentle ping on this patch from May 28.
What patch? I have no context here, sorry.
Apologies for the accidental duplicate send; a second copy went out shortly after this one by mistake. This thread is the canonical one to review.
If you sent two different ones, always properly version them so we know which one should be reviewed.
And its the merge window right now, nothing can happen until after -rc1 is out, thanks,
greg k-h