CDROMEJECT (or START_STOP(2)) completely disables USB flash stick

CDROMEJECT (or START_STOP(2)) completely disables USB flash stick

[Andrey Borzenkov]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

While working on some HAL issues I hit some not clear - to me at least - 
behavior. I was testing HAL Eject method that under Linux effectively 
calls 'eject /dev/block'. Now for USB stick - at least, that I have, this 
results in "ejecting medium" without apparently any way to get it back (the 
only way is to replug stick).

UEVENT[1152175788.114199] remove@/block/sda/sda1
ACTION=remove
DEVPATH=/block/sda/sda1
SUBSYSTEM=block
SEQNUM=1332
MINOR=1
MAJOR=8
PHYSDEVPATH=/devices/pci0000:00/0000:00:02.0/usb1/1-1/1-1:1.0/host1/target1:0:0/1:0:0:0
PHYSDEVBUS=scsi
PHYSDEVDRIVER=sd

In this state I do see device:

{pts/0}% sudo sdparm -a /dev/sda
    /dev/sda:           128MB             2.00
{pts/0}% cat /proc/scsi/scsi
Attached devices:
Host: scsi1 Channel: 00 Id: 00 Lun: 00
  Vendor:          Model: 128MB            Rev: 2.00
  Type:   Direct-Access                    ANSI SCSI revision: 02

but any attempt to get partition back results in

{pts/0}% LC_ALL=C sudo blockdev --rereadpt /dev/sda
/dev/sda: No medium found

I tried 'sdparm -C start' without much success as well:

{pts/0}% LC_ALL=C sudo sdparm -v -C start /dev/sda
    /dev/sda:           128MB             2.00
    Start stop unit command: 1b 00 00 00 01 00
{pts/0}% LC_ALL=C sudo blockdev --rereadpt /dev/sda
/dev/sda: No medium found

Well, I just try to understand if this is a expected (and reasonable) 
behaviour or a bug in hardware or possibly software?

Thank you

- -andrey


-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.3 (GNU/Linux)

iD8DBQFErNN2R6LMutpd94wRAg1DAJ9EITRTZCm/ej5PXzxAK71s7sUCVACgiNFx
KXOuNAwwhPKYcogaK6LgMZM=
=MHVX
-----END PGP SIGNATURE-----

Re: CDROMEJECT (or START_STOP(2)) completely disables USB flash stick

[Jan Engelhardt]

>
>While working on some HAL issues I hit some not clear - to me at least - 
>behavior. I was testing HAL Eject method that under Linux effectively 
>calls 'eject /dev/block'. Now for USB stick - at least, that I have, this 
>results in "ejecting medium" without apparently any way to get it back (the 
>only way is to replug stick).

I suppose eject does the same as `sdparm -C eject`, and yes, it does not 
seem like you can reload an USB stick with the opposite of eject, `sdparm 
-C load` (at least these are the commands that work for CDROM devices).

>{pts/0}% LC_ALL=C sudo blockdev --rereadpt /dev/sda
>/dev/sda: No medium found
>
>I tried 'sdparm -C start' without much success as well:
>
>{pts/0}% LC_ALL=C sudo sdparm -v -C start /dev/sda
>    /dev/sda:           128MB             2.00
>    Start stop unit command: 1b 00 00 00 01 00
>{pts/0}% LC_ALL=C sudo blockdev --rereadpt /dev/sda
>/dev/sda: No medium found
>
>Well, I just try to understand if this is a expected (and reasonable) 
>behaviour or a bug in hardware or possibly software?

I cannot close the tray of the CDROM drive with -C start (or -C ready), so 
I doubt the same will be true for USB sticks.

Maybe all of this is a limitation of USB sticks - I got to check it with 
real USB harddisks or hotswap SCA disks.


Jan Engelhardt
--