kern/47077: [dM] audio(4) disagrees with the code

[Mouse <mouse%Rodents-Montreal.ORG@localhost>]

>Number:         47077
>Category:       kern
>Synopsis:       [dM] audio(4) disagrees with the code
>Confidential:   no
>Severity:       non-critical
>Priority:       medium
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Sun Oct 14 20:00:01 +0000 2012
>Originator:     Mouse
>Release:        NetBSD 5.1
>Organization:
        Dis-
>Environment:
        Apparently present on many recent versions; noticed on

System: NetBSD VAIO-Video.Rodents-Montreal.ORG 5.1 NetBSD 5.1 (MOUSEGEN) #3: 
Fri Oct 5 06:02:53 EDT 2012 
mouse%VAIO-51.Rodents-Montreal.ORG@localhost:/home/mouse/kbuild/MOUSEGEN i386
Architecture: i386
Machine: i386
>Description:
        audio(4) says (in part), of the blocksize field in an
        audio_info_t, that "Upon return from an AUDIO_SETINFO call, the
        actual blocksize set is returned in this field.".  This does
        not appear to actually be the case; AUDIO_SETINFO appears to
        leave the blocksize field unchanged.

        jmcneill says this appears to have been broken "for over 15
        years", pointing to the delta between 1.38 and 1.39 of
        sys/dev/audio.c, presumably referring to, if I'm reading cvsweb
        text right, the removal of line 1686 of the older version.
        This change is dated 1997-03-20.
>How-To-Repeat:
        Try to take advantage of the functionality that sentence
        promises.  Notice it doesn't work.

        See my tech-kern message
        <201210140805.EAA08506%Sparkle.Rodents-Montreal.ORG@localhost> (at
        ftp.rodents-montreal.org:/mouse/misc/AUDIO_SETINFO.2012-10-14
        among other places) for a test program and sample run of it.
>Fix:
        As a workaround, call AUDIO_GETINFO or AUDIO_GETBUFINFO after
        doing AUDIO_SETINFO.

        Since it's a useful facility, I'd rather see the code repaired.
        But, since the breakage is apparently well over a decade old,
        it seems almost nobody cares, so it might be more appropriate
        to treat this as a doc-bug instead and rip that promise out of
        the documentation.

        I haven't created patches for either.

        /~\ The ASCII                             Mouse
        \ / Ribbon Campaign
         X  Against HTML                mouse%rodents-montreal.org@localhost
        / \ Email!           7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B