On Wed, 2019-01-16 at 08:35 +-0100, Hannes Reinecke wrote: +AD4 If the device is +AF8-not+AF8 removable we should not start the event +AD4 poller as the media will not go away. Having the event poller running +AD4 will block the open() call as it will try to flush outstanding events, +AD4 which it can't if the device is in state 'BLOCKED'. So the open() call +AD4 will be stalled until the device state changed, which might be quite +AD4 some time depending on the transport. +AD4 +AD4 Signed-off-by: Hannes Reinecke +ADw-hare+AEA-suse.com+AD4 +AD4 --- +AD4 drivers/scsi/sd.c +AHw 3 +-+-+- +AD4 1 file changed, 3 insertions(+-) +AD4 +AD4 diff --git a/drivers/scsi/sd.c b/drivers/scsi/sd.c +AD4 index a1a44f52e0e8..521f0a384446 100644 +AD4 --- a/drivers/scsi/sd.c +AD4 +-+-+- b/drivers/scsi/sd.c +AD4 +AEAAQA -1539,6 +-1539,9 +AEAAQA static unsigned int sd+AF8-check+AF8-events(struct gendisk +ACo-disk, unsigned int clearing) +AD4 return 0+ADs +AD4 +AD4 sdp +AD0 sdkp-+AD4-device+ADs +AD4 +- if (+ACE-sdp-+AD4-removable) +AD4 +- return 0+ADs +AD4 +- +AD4 SCSI+AF8-LOG+AF8-HLQUEUE(3, sd+AF8-printk(KERN+AF8-INFO, sdkp, +ACI-sd+AF8-check+AF8-events+AFw-n+ACI))+ADs +AD4 +AD4 /+ACo Hi Hannes, Although this patch looks fine to me, wouldn't it be a better approach to cause the events checker not to submit any SCSI commands to blocked devices? That approach should improve the sd behavior for both removable and non-removable devices. Thanks, Bart.
