On Fri, Nov 29, 2019 at 08:26:36PM +0000, Martin Wilck wrote:
> From: Martin Wilck <mwilck@xxxxxxxx>
>
> Since 45235022da99, the firmware is shut down early in the controller
> shutdown process. This causes commands sent to the firmware (such as LOGO)
> to hang forever. Eventually one or more timeouts will be triggered.
> Move the stopping of the firmware until after sessions have terminated.
>
> Fixes: 45235022da99 ("scsi: qla2xxx: Fix driver unload by shutting down chip")
> Signed-off-by: Martin Wilck <mwilck@xxxxxxxx>
> ---
> drivers/scsi/qla2xxx/qla_os.c | 21 ++++++++++-----------
> 1 file changed, 10 insertions(+), 11 deletions(-)
>
> diff --git a/drivers/scsi/qla2xxx/qla_os.c b/drivers/scsi/qla2xxx/qla_os.c
> index 43d0aa0..0cc127d 100644
> --- a/drivers/scsi/qla2xxx/qla_os.c
> +++ b/drivers/scsi/qla2xxx/qla_os.c
> @@ -3710,6 +3710,16 @@ qla2x00_remove_one(struct pci_dev *pdev)
> }
> qla2x00_wait_for_hba_ready(base_vha);
>
> + qla2x00_wait_for_sess_deletion(base_vha);
> +
> + /*
> + * if UNLOAD flag is already set, then continue unload,
> + * where it was set first.
> + */
> + if (test_bit(UNLOADING, &base_vha->dpc_flags))
> + return;
> +
> + set_bit(UNLOADING, &base_vha->dpc_flags);
Hi Martin,
Moving qla2x00_wait_for_sess_deletion up ensures hw->wq is flushed
before shutdown is done.
But I think we need to set UNLOADING bit earlier to break-up async
discovery chain. The comment just above qla2x00_wait_for_sess_deletion
definition mentions that.
Thanks,
Roman
