Current-Users archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: Hang/reboot with -current and umidi on i386



On Sat, 4 Feb 2012, Tom Ivar Helbekkmo wrote:

> Iain Hibbert <plunky%rya-online.net@localhost> writes:
>
> > On Tue, 31 Jan 2012, Tom Ivar Helbekkmo wrote:
> >
> >> It's not just umidi.  The silly midi over PC beeper thing has related
> >> trouble, for one.  I'm not going to worry about that - a quick "no midi*
> >> at pcppi?" in my kernel config felt like a more proper solution...  :)
> >
> > Can you expand on that a bit?  I don't have any midi hardware, and the
> > midi at pcppi does seem to work for me (i386) though sounds rather tinny
> > (perhaps it is expected to be that?)
>
> While I was debugging this stuff, one of the last problems I was seeing
> was this kernel output:
>
> vrelel: missing VOP_CLOSE(): vnode @ 0xce72000c, flags (0x10<MPSAFE>)
>         tag VT_NON(0), type VCHR(4), usecount 1, writecount 0, holdcount 0
>         freelisthd 0x0, mount 0x0, data 0x0 lock 0xce7200b0

that leaves me with no clue since I don't know much about vnodes :)

> This would happen when closing a MIDI interface after reading from it,
> and the interface would then quit working: it would disappear from the
> list given by "midiplay -l".  When I was sure I'd solved that, and the
> message still appeared after reading from /dev/music, that had me
> stumped (I couldn't understand why going through /dev/music should cause
> it, when directly reading from the raw interfaces didn't) until I
> noticed that it was the pcppi MIDI interface that was causing it, and
> disappearing from the "midiplay -l" output.

/dev/music and /dev/sequencer is provided by src/sys/dev/sequencer.c so
perhaps that is in there rather than the lower layer which doesn't really
handle the vnodes.. the only VOP_OPEN is in midiseq_open() and, as per the
complaint, I don't see a VOP_CLOSE but I don't know where that should be..

> I haven't looked at it any closer, though.

if you could construct a set of commands that would produce that error
surely somebody might be able to work that out..

iain


Home | Main Index | Thread Index | Old Index