On Sep 22 10:08, Klaus Jensen wrote:
> From: Klaus Jensen <k.jensen@xxxxxxxxxxx>
>
> Devices that actually honor the Bus Master Enable bit and considers the
> resulting failure to read/write a fatal error, often ends up getting
> disabled by the kernel when running block/011.
>
> Remove the device and rescan the pci bus after completing the test to
> make sure we bring the device back up if required.
>
> Cc: Keith Busch <kbusch@xxxxxxxxxx>
> Cc: Johannes Thumshirn <johannes.thumshirn@xxxxxxx>
> Signed-off-by: Klaus Jensen <k.jensen@xxxxxxxxxxx>
> ---
> tests/block/011 | 5 +++++
> 1 file changed, 5 insertions(+)
>
> diff --git a/tests/block/011 b/tests/block/011
> index 4f331b4a7522..f1306ae3896f 100755
> --- a/tests/block/011
> +++ b/tests/block/011
> @@ -22,6 +22,7 @@ test_device() {
> echo "Running ${TEST_NAME}"
>
> pdev="$(_get_pci_dev_from_blkdev)"
> + sysfs="/sys/bus/pci/devices/${pdev}"
>
> if _test_dev_is_rotational; then
> size="32m"
> @@ -42,4 +43,8 @@ test_device() {
> done
>
> echo "Test complete"
> +
> + # bring the device back up if it was disabled by the kernel
> + echo 1 > "${sysfs}/remove"
> + echo 1 > /sys/bus/pci/rescan
> }
> --
> 2.28.0
>
Hmfh. Looks like this at least needs a wait for TEST_DEV being back or
something. Please disregard this patch for now as I try out some other
work arounds.
Attachment:
signature.asc
Description: PGP signature
