Re: removing certain umass (cd/dvd?) devices appears to lock up the USB bus on 14.2/14.3 (at least)

From: David Cross <david_at_crossfamilyweb.com>
Date: Wed, 13 Aug 2025 04:33:15 UTC
I am upgrading one of my machines to 15.0 now; though the last 2 times I 
tried to repro on 14.3 it failed (after I had posted the initial example 
of a failure on 14.3).  I will be trying a few more times on 14.3 with 
different configs, this machine should be on 15.0 by tomorrow AM.)

On 8/12/25 16:12, Warner Losh wrote:
> Hey David,
>
> Did you get a chance to try the USB sticks with -current yet?
>
> Warner
>
> On Thu, Aug 7, 2025 at 12:54 PM Warner Losh <imp@bsdimp.com> wrote:
>
>     Maybe you could try 15.0. I've done a lot to fix these things in 15.0.
>     If it still happens there, we'll talk about changes. I've not MFC'd
>     the changes I've made since they are kinda extensive. On the other
>     hand, there's been a lot of reports of 'works better for me' with only
>     one regression. On the other hand, it's not entirely clear this is in
>     the class of problems I solved.
>
>     Sadly, I suspect that we'll need to have hardware in the right hands
>     to solve this completely, especially since the reproduction appears to
>     be hit or miss. Is this a pure DVD drive? Is it portable or desktop?
>     If it's the one I found on ebay (https://www.ebay.com/itm/396752411077
>     though there's others), looks in budget for me to obtain to help if
>     there's a reliable recipe to reproduce and it's not the "fault" of
>     your host USB chipset.
>
>     Warner
>
>     On Thu, Aug 7, 2025 at 12:43 PM David Cross
>     <david@crossfamilyweb.com> wrote:
>     >
>     > I have a USB DVD drive that when I unplug (even when empty and
>     > unmounted) appears to stop the entire USB stack on 14.2/14.3.
>     There are
>     > no errors (that I can see) in the dmesg, it *appears* to detach, but
>     > then no other USB events happen.
>     >
>     >
>     > I originally thought it was just the port that it was attached
>     to, but
>     > in recent experimentation I attached it to a USB hub, then
>     detached it
>     > from the hub, and even other ports on the host stopped responding.
>     > Currently connected devices appear to continue to work, but if
>     removed
>     > nothing will reattach to those ports.
>     >
>     >
>     > This has now happened across a couple of different machines with
>     this
>     > device, so I don't think it is host specific behavior
>     >
>     >
>     > When trying to run 'usbconfig -v - l' to get some kind of
>     diagnostics it
>     > just hangs; C-t yields the following:
>     >
>     >
>     >  > load: 0.12  cmd: usbconfig 52496 [USB config SX lock] 3.12r 0.00u
>     > 0.00s 0% 2280k
>     >
>     >
>     > Of course in the interests of a fresh reproduction for this
>     email, it
>     > just worked; I am continuing to attempt to reproduce. Suggestions of
>     > things I should try when I am successful to diagnose the issue
>     so we can
>     > fix it?
>     >
>     > The relevant USB attach messages are:
>     >
>     > ugen1.10: <Hitachi-LG Data Storage Inc Portable Super Multi
>     Drive> at usbus1
>     > umass0 on uhub4
>     > umass0: <Hitachi-LG Data Storage Inc Portable Super Multi Drive,
>     class
>     > 0/0, rev 2.00/0.00, addr 14> on usbus1
>     > umass0:  8070i (ATAPI) over Bulk-Only; quirks = 0x0000
>     > umass0:1:0: Attached to scbus1
>     > cd0 at umass-sim0 bus 0 scbus1 target 0 lun 0
>     > cd0: <HL-DT-ST DVDRAM SP80NB80 RC00> Removable CD-ROM SCSI device
>     > cd0: 40.000MB/s transfers
>     > cd0: 0MB (1 0 byte sectors)
>     > cd0: quirks=0x10<10_BYTE_ONLY>
>     >
>     > Detach messages are:
>     >
>     > ugen1.10: <Hitachi-LG Data Storage Inc Portable Super Multi
>     Drive> at
>     > usbus1 (disconnected)
>     > umass0: at uhub4, port 2, addr 14 (disconnected)
>     > cd0 at umass-sim0 bus 0 scbus1 target 0 lun 0
>     > cd0: <HL-DT-ST DVDRAM SP80NB80 RC00>  detached
>     > (cd0:umass-sim0:0:0:0): Periph destroyed
>     > umass0: detached
>     >
>     >
>