Am 13.02.24 um 15:28 schrieb Martin K. Petersen:
Julian,
I am cross-posting this to scsi list, maybe someone there would be
able to understand what is going on.
I'm going to copy you on a patch I have in my development tree. I would
appreciate if you could try that out.
Thanks!
Hi Martin,
Thanks for responding. I tried your patch but it, unfortunately, did not
help with the issue I and Tasos were having. Not on its own, and not
with LaCie:Rugged FW USB3:0x2000 added to /proc/scsi/device_info.
Just to make sure we are talking about the same problem: prior to
c92a6b5d6335, the drive would be usable pretty much immediately after
connecting - in the logs this manifests by the following rows appearing:
Feb 13 18:59:42 kernel: sdb: sdb1 sdb2
Feb 13 18:59:42 kernel: sd 8:0:0:0: [sdb] Attached SCSI disk
After c92a6b5d6335, one or more restarts are required (below is from
Fedora's 6.7.4 with your patch applied on top):
Feb 13 19:01:41 kernel: sd 8:0:0:0: [sdb] tag#13 uas_eh_abort_handler 0
uas-tag 1 inflight: IN
Feb 13 19:01:41 kernel: sd 8:0:0:0: [sdb] tag#13 CDB: Inquiry 12 01 b9
00 ff 00
Feb 13 19:01:41 kernel: scsi host8: uas_eh_device_reset_handler start
Feb 13 19:01:41 napoleon2 kernel: usb 2-3: reset SuperSpeed USB device
number 3 using xhci_hcd
Feb 13 19:01:41 kernel: scsi host8: uas_eh_device_reset_handler success
Feb 13 19:02:11 kernel: scsi host8: uas_eh_device_reset_handler start
Feb 13 19:02:11 kernel: sd 8:0:0:0: [sdb] tag#16 uas_zap_pending 0
uas-tag 1 inflight:
Feb 13 19:02:11 kernel: sd 8:0:0:0: [sdb] tag#16 CDB: Inquiry 12 01 b9
00 ff 00
Feb 13 19:02:11 kernel: usb 2-3: reset SuperSpeed USB device number 3
using xhci_hcd
Feb 13 19:02:11 kernel: scsi host8: uas_eh_device_reset_handler success
Feb 13 19:02:41 kernel: scsi host8: uas_eh_device_reset_handler start
Feb 13 19:02:41 kernel: sd 8:0:0:0: [sdb] tag#17 uas_zap_pending 0
uas-tag 1 inflight:
Feb 13 19:02:41 kernel: sd 8:0:0:0: [sdb] tag#17 CDB: Inquiry 12 01 b9
00 ff 00
Feb 13 19:02:42 kernel: usb 2-3: reset SuperSpeed USB device number 3
using xhci_hcd
Feb 13 19:02:42 kernel: scsi host8: uas_eh_device_reset_handler success
Feb 13 19:03:12 kernel: scsi host8: uas_eh_device_reset_handler start
Feb 13 19:03:12 kernel: sd 8:0:0:0: [sdb] tag#28 uas_zap_pending 0
uas-tag 1 inflight:
Feb 13 19:03:12 kernel: sd 8:0:0:0: [sdb] tag#28 CDB: Inquiry 12 01 b9
00 ff 00
Feb 13 19:03:12 kernel: usb 2-3: reset SuperSpeed USB device number 3
using xhci_hcd
Feb 13 19:03:12 kernel: scsi host8: uas_eh_device_reset_handler success
Four resets are needed with the current kernel-6.7.4. With early 5.19
kernels on and after 7fb019c46eee it needed one reset, with c92a6b5d6335
and with e60ac0b9e445 multiple resets were needed too. As 30 seconds
need to pass between the resets, it does take a while until the drive
can be used.
I did not test other commits between c92a6b5d6335 and c92a6b5d6335 as I
was bisecting and only tested the commits suggested by git.
6.7.4 I tested with bare metal Fedora 39, 5.19 with a Fedora 36 VM.
If this was already clear, please feel free to ignore.
In case it is relevant: my drive (but not Tasos') is patched to use
US_FL_NO_REPORT_OPCODES | US_FL_NO_SAME.
Best regards,
Julian
