bus/vdev: vdev_cleanup checks dev->device.driver
Checks
Commit Message
The vdev_probe calls driver->probe, which may fail
and dev->device.driver will still be NULL.
In vdev_cleanup, drv = container_of(dev->device.driver) returns !NULL,
then drv->remove will trigger Segmentation fault.
Fix it by checking dev->device.driver first.
Log:
Thread 1 "dpdk-test" received signal SIGSEGV, Segmentation fault.
0x00000000012c484d in vdev_cleanup ()
Signed-off-by: Zhangfei Gao <zhangfei.gao@linaro.org>
---
drivers/bus/vdev/vdev.c | 3 +++
1 file changed, 3 insertions(+)
Comments
On Wed, Oct 19, 2022 at 12:55 PM Zhangfei Gao <zhangfei.gao@linaro.org> wrote:
>
> The vdev_probe calls driver->probe, which may fail
> and dev->device.driver will still be NULL.
>
> In vdev_cleanup, drv = container_of(dev->device.driver) returns !NULL,
> then drv->remove will trigger Segmentation fault.
> Fix it by checking dev->device.driver first.
>
> Log:
> Thread 1 "dpdk-test" received signal SIGSEGV, Segmentation fault.
> 0x00000000012c484d in vdev_cleanup ()
I suspect you hit this issue when running some crypto autotest.
Can you confirm?
The commit that introduced the issue should be mentionned, with a
Fixes: tag, like:
Fixes: 1cab1a40ea9b ("bus: cleanup devices on shutdown")
>
> Signed-off-by: Zhangfei Gao <zhangfei.gao@linaro.org>
> ---
> drivers/bus/vdev/vdev.c | 3 +++
> 1 file changed, 3 insertions(+)
>
> diff --git a/drivers/bus/vdev/vdev.c b/drivers/bus/vdev/vdev.c
> index f5b43f1930..fbdaf68380 100644
> --- a/drivers/bus/vdev/vdev.c
> +++ b/drivers/bus/vdev/vdev.c
> @@ -577,6 +577,9 @@ vdev_cleanup(void)
> const struct rte_vdev_driver *drv;
> int ret = 0;
>
> + if (dev->device.driver == NULL)
> + continue;
> +
> drv = container_of(dev->device.driver, const struct rte_vdev_driver, driver);
>
> if (drv == NULL || drv->remove == NULL)
If dev->device.driver != NULL, then drv won't be NULL.
Hi, David
On Wed, 19 Oct 2022 at 19:39, David Marchand <david.marchand@redhat.com> wrote:
>
> On Wed, Oct 19, 2022 at 12:55 PM Zhangfei Gao <zhangfei.gao@linaro.org> wrote:
> >
> > The vdev_probe calls driver->probe, which may fail
> > and dev->device.driver will still be NULL.
> >
> > In vdev_cleanup, drv = container_of(dev->device.driver) returns !NULL,
> > then drv->remove will trigger Segmentation fault.
> > Fix it by checking dev->device.driver first.
> >
> > Log:
> > Thread 1 "dpdk-test" received signal SIGSEGV, Segmentation fault.
> > 0x00000000012c484d in vdev_cleanup ()
>
> I suspect you hit this issue when running some crypto autotest.
> Can you confirm?
Yes, I am testing uadk crypto autotest on x86,
Since there is no hardware, so probe fail and return (which is expected)
But quit dpdk-test triggered Segmentation fault
$ sudo dpdk-test --vdev=crypto_uadk --log-level=6
vdev_probe(): failed to initialize crypto_uadk device
EAL: Bus (vdev) probe failed.
TELEMETRY: No legacy callbacks, legacy socket not created
RTE>>quit
Segmentation fault
>
>
> The commit that introduced the issue should be mentionned, with a
> Fixes: tag, like:
> Fixes: 1cab1a40ea9b ("bus: cleanup devices on shutdown")
OK, will update
Thanks David.
>
> >
> > Signed-off-by: Zhangfei Gao <zhangfei.gao@linaro.org>
> > ---
> > drivers/bus/vdev/vdev.c | 3 +++
> > 1 file changed, 3 insertions(+)
> >
> > diff --git a/drivers/bus/vdev/vdev.c b/drivers/bus/vdev/vdev.c
> > index f5b43f1930..fbdaf68380 100644
> > --- a/drivers/bus/vdev/vdev.c
> > +++ b/drivers/bus/vdev/vdev.c
> > @@ -577,6 +577,9 @@ vdev_cleanup(void)
> > const struct rte_vdev_driver *drv;
> > int ret = 0;
> >
> > + if (dev->device.driver == NULL)
> > + continue;
> > +
> > drv = container_of(dev->device.driver, const struct rte_vdev_driver, driver);
> >
> > if (drv == NULL || drv->remove == NULL)
>
> If dev->device.driver != NULL, then drv won't be NULL.
>
>
> --
> David Marchand
>
On Wed, 19 Oct 2022 at 19:39, David Marchand <david.marchand@redhat.com> wrote:
>
> On Wed, Oct 19, 2022 at 12:55 PM Zhangfei Gao <zhangfei.gao@linaro.org> wrote:
> >
> > The vdev_probe calls driver->probe, which may fail
> > and dev->device.driver will still be NULL.
> >
> > In vdev_cleanup, drv = container_of(dev->device.driver) returns !NULL,
> > then drv->remove will trigger Segmentation fault.
> > Fix it by checking dev->device.driver first.
> >
> > Log:
> > Thread 1 "dpdk-test" received signal SIGSEGV, Segmentation fault.
> > 0x00000000012c484d in vdev_cleanup ()
>
> I suspect you hit this issue when running some crypto autotest.
> Can you confirm?
>
>
> The commit that introduced the issue should be mentionned, with a
> Fixes: tag, like:
> Fixes: 1cab1a40ea9b ("bus: cleanup devices on shutdown")
>
> >
> > Signed-off-by: Zhangfei Gao <zhangfei.gao@linaro.org>
> > ---
> > drivers/bus/vdev/vdev.c | 3 +++
> > 1 file changed, 3 insertions(+)
> >
> > diff --git a/drivers/bus/vdev/vdev.c b/drivers/bus/vdev/vdev.c
> > index f5b43f1930..fbdaf68380 100644
> > --- a/drivers/bus/vdev/vdev.c
> > +++ b/drivers/bus/vdev/vdev.c
> > @@ -577,6 +577,9 @@ vdev_cleanup(void)
> > const struct rte_vdev_driver *drv;
> > int ret = 0;
> >
> > + if (dev->device.driver == NULL)
> > + continue;
> > +
> > drv = container_of(dev->device.driver, const struct rte_vdev_driver, driver);
> >
> > if (drv == NULL || drv->remove == NULL)
>
> If dev->device.driver != NULL, then drv won't be NULL.
Sorry, I miss this, add log to be clear
$ sudo dpdk-test --vdev=crypto_uadk --log-level=6
vdev_probe_all_drivers dev->device.driver=(nil) ret=-19
vdev_probe(): failed to initialize crypto_uadk device
EAL: Bus (vdev) probe failed.
TELEMETRY: No legacy callbacks, legacy socket not created
RTE>>quit
vdev_cleanup dev->device.driver=(nil)
vdev_cleanup drv=0xfffffffffffffff0
Segmentation fault
>
>
> --
> David Marchand
>
@@ -577,6 +577,9 @@ vdev_cleanup(void)
const struct rte_vdev_driver *drv;
int ret = 0;
+ if (dev->device.driver == NULL)
+ continue;
+
drv = container_of(dev->device.driver, const struct rte_vdev_driver, driver);
if (drv == NULL || drv->remove == NULL)