Re: USB regression in 10.1

[Michael van Elst <mlelstv%serpens.de@localhost>]

On Sat, Jan 25, 2025 at 08:13:38AM +0700, Robert Elz wrote:
> 
>   | Now it is simply skipped with the same result.
> 
> Not quite the same - previously once the drive was ready, wedge
> discovery would happen.

At attach time a TEST UNIT READY command is sent to the drive.

If it returns OK, the drive parameters are retrieved and shown
Otherwise a message ("drive offline") or ("unformatted media")
is reported.

Then wedge discovery was done once and failed if the drive
was offline. Now this is simply skipped.

Nothing happens later "once the drive was ready".

At open time, a TEST UNIT READY is done again, and drive
parameters are retrieved when necessary. This actually
makes it possible to use the device itself if it became
online in between. But there is and was no wedge discovery
at that time.


What happens is a side effect. Wedge discovery itself opens the
device. Besides the description above, this will also issue a START
UNIT command to the drive when the (now second) TEST UNIT READY
check failed (as well as a PREVENT MEDIUM REMOVAL command for a
removable device).

When START UNIT succeeds, it waits until the disk has started
and then it usually is in online state.

The attach code may also issue a START UNIT command, but this is
only done as the very first thing when the device has set
PQUIRK_START (that quirk is used for an Apple iPod).

When wedge discovery was added, this effectively made all
disks have the equivalent of PQUIRK_START.

With the change that's no longer true and we are back to the original
behaviour where a disk stays stopped until it is used (and then
you need to discover wedges manually).


Greetings,
-- 
                                Michael van Elst
Internet: mlelstv%serpens.de@localhost
                                "A potential Snark may lurk in every tree."