Re: USB regression in 10.1

[mlelstv%serpens.de@localhost (Michael van Elst)]

kre%munnari.OZ.AU@localhost (Robert Elz) writes:

>    Date:        Fri, 24 Jan 2025 18:22:14 +0100
>    From:        Rhialto <rhialto%falu.nl@localhost>
>    Message-ID:  <Z5PMRnJv5_gmc_FW%falu.nl@localhost>

>  | I notice a smallish but annoying regression in 10.1 compared to 10.0.

>I have seen this too (in HEAD rather than 10.x, not that that should matter)

>  | The only thing I found so far was to unplug and replug the USB cable.

>For me, I just do

>	dkctl sdN makewedges

>after the drive has had time to spin up.   It is annoying, the way
>things worked before was better, but it isn't completely unmanageable.



When a disk attaches offline, the wedge discovery routine
previously failed and reported errors. Now it is simply
skipped with the same result.

There is a small chance that a disk starts offline, and a
few milliseconds later, it would successfully allow reading.
Maybe this disk behaves that way.

The real issue here is that discovery starts at attach time
in opposite to disklabel partitions, that always exist and
get defined at open time. You can easily open a partition
again once a disk becomes online.

With wedges, there is no device to open when the discovery
failed (or was skipped). You need to trigger discovery
manually using 'makewedges' before a wedge is created and
you can open it.

The only way to automate that process is to poll 'offline'
disks regularly, but that doesn't solve subsequent problems,
e.g. delaying an automatic mount.