| Message ID | 20200905125836.GF183976@mwanda (mailing list archive) |
|---|---|
| State | Accepted |
| Headers | show |
| Series | scsi: libsas: Fix error path in sas_notify_lldd_dev_found() | expand |
在 2020/9/5 20:58, Dan Carpenter 写道: > In sas_notify_lldd_dev_found(), if we can't find a device, then it seems > like the wrong thing to mark the device as found and to increment the > reference count. None of the callers ever drop the reference in that > situation. > > Fixes: 735f7d2fedf5 ("[SCSI] libsas: fix domain_device leak") > Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> > --- > drivers/scsi/libsas/sas_discover.c | 3 ++- > 1 file changed, 2 insertions(+), 1 deletion(-) > > diff --git a/drivers/scsi/libsas/sas_discover.c b/drivers/scsi/libsas/sas_discover.c > index cd7c7d269f6f..d0f9e90e3279 100644 > --- a/drivers/scsi/libsas/sas_discover.c > +++ b/drivers/scsi/libsas/sas_discover.c > @@ -182,10 +182,11 @@ int sas_notify_lldd_dev_found(struct domain_device *dev) > pr_warn("driver on host %s cannot handle device %016llx, error:%d\n", > dev_name(sas_ha->dev), > SAS_ADDR(dev->sas_addr), res); > + return res; > } > set_bit(SAS_DEV_FOUND, &dev->state); > kref_get(&dev->kref); > - return res; > + return 0; > } > > > Hi Dan, thanks for finding this, Reviewed-by: Jason Yan <yanaijie@huawei.com>
On 05/09/2020 13:58, Dan Carpenter wrote: > In sas_notify_lldd_dev_found(), if we can't find a device, nit: the callback is for the LLDD is to allocate resources, device context etc., for that domain_device, and not find the device. The device has been found at this point. > then it seems > like the wrong thing to mark the device as found and to increment the > reference count. None of the callers ever drop the reference in that > situation. > > Fixes: 735f7d2fedf5 ("[SCSI] libsas: fix domain_device leak") > Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> > --- > drivers/scsi/libsas/sas_discover.c | 3 ++- > 1 file changed, 2 insertions(+), 1 deletion(-) > > diff --git a/drivers/scsi/libsas/sas_discover.c b/drivers/scsi/libsas/sas_discover.c > index cd7c7d269f6f..d0f9e90e3279 100644 > --- a/drivers/scsi/libsas/sas_discover.c > +++ b/drivers/scsi/libsas/sas_discover.c > @@ -182,10 +182,11 @@ int sas_notify_lldd_dev_found(struct domain_device *dev) > pr_warn("driver on host %s cannot handle device %016llx, error:%d\n", > dev_name(sas_ha->dev), > SAS_ADDR(dev->sas_addr), res); > + return res; > } > set_bit(SAS_DEV_FOUND, &dev->state); > kref_get(&dev->kref); > - return res; > + return 0; This looks ok. > } Thanks, John > > >
On Sat, 5 Sep 2020 15:58:36 +0300, Dan Carpenter wrote: > In sas_notify_lldd_dev_found(), if we can't find a device, then it seems > like the wrong thing to mark the device as found and to increment the > reference count. None of the callers ever drop the reference in that > situation. Applied to 5.9/scsi-fixes, thanks! [1/1] scsi: libsas: Fix error path in sas_notify_lldd_dev_found() https://git.kernel.org/mkp/scsi/c/fdcb7900d9ab
diff --git a/drivers/scsi/libsas/sas_discover.c b/drivers/scsi/libsas/sas_discover.c index cd7c7d269f6f..d0f9e90e3279 100644 --- a/drivers/scsi/libsas/sas_discover.c +++ b/drivers/scsi/libsas/sas_discover.c @@ -182,10 +182,11 @@ int sas_notify_lldd_dev_found(struct domain_device *dev) pr_warn("driver on host %s cannot handle device %016llx, error:%d\n", dev_name(sas_ha->dev), SAS_ADDR(dev->sas_addr), res); + return res; } set_bit(SAS_DEV_FOUND, &dev->state); kref_get(&dev->kref); - return res; + return 0; }
In sas_notify_lldd_dev_found(), if we can't find a device, then it seems like the wrong thing to mark the device as found and to increment the reference count. None of the callers ever drop the reference in that situation. Fixes: 735f7d2fedf5 ("[SCSI] libsas: fix domain_device leak") Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> --- drivers/scsi/libsas/sas_discover.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)