kern/38636: ld(4) is now totally broken with ataraid(4) (ld.c 1.57 botch?)

["Greg A. Woods" <woods%planix.com@localhost>]

>Number:         38636
>Category:       kern
>Synopsis:       ld(4) is now totally broken with ataraid(4) (ld.c 1.57 botch?)
>Confidential:   no
>Severity:       critical
>Priority:       high
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Mon May 12 12:30:00 +0000 2008
>Originator:     Greg A. Woods
>Release:        NetBSD 4.99.62 2008/05/11
>Organization:
Planix, Inc.; Toronto, Ontario; Canada
>Environment:
System: NetBSD 4.99.62 (GENERIC)
Architecture: i386
Machine: i386
>Description:
This is from my recent post to current-users.

With 4.99.62 2008/05/08 and on, and perhaps from sometime earlier too (I
haven't had a working -current on this machine in a few days but I think
I saw this briefly in one of the near-working versions), I'm now seeing
some weirdness with the ld(4) device that should be attaching to the
ataraid(4) device, namely the device name doesn't appear in dmesg any
more (except for the first line) and there's a weird "error = 19"
showing up too.

Given the "rnd:  attached..." line from the #ifdef RND_VERBOSE code I
happen to have turned on in rnd(4) it looks as if the device name is an
empty string.  Normally it shows something like this: "rnd: wd0 attached
as an entropy source (collecting)"

However the device name does still show up in the "RAIDframe:" and
"opendisk:"  lines.

This is likely due to some botch with ld.c's revision 1.57, though I
don't understand the problem at all.

ataraid0: found 1 RAID volume
ld0 at ataraid0 vendtype 1 unit 0: Adaptec ATA RAID-1 array
: 186 GB, 24321 cyl, 255 head, 63 sec, 512 bytes/sect x 390721536 sectors
rnd:  attached as an entropy source (collecting)
: unable to compute pdev, error = 19
RAIDframe: Searching for RAID components...
RAIDframe: wd1d: failed to open raw device: 16
RAIDframe: wd2d: failed to open raw device: 16
RAIDframe: ld0d: failed to open raw device: 19
RAIDframe: NOTICE: no RAID sets found, no root RAID possible...
opendisk: can't open dev wd1 (16)
opendisk: can't open dev wd2 (16)
opendisk: can't open dev ld0 (19)
opendisk: can't open dev wd1 (16)
opendisk: can't open dev wd2 (16)
opendisk: can't open dev ld0 (19)
opendisk: can't open dev wd1 (16)
opendisk: can't open dev wd2 (16)
opendisk: can't open dev ld0 (19)


Finally I can't actually access it from userland any more either:

        16:49 [146] # disklabel ld0     
        disklabel: /dev/rld0d: Operation not supported by device

This may also be fallout from the same problem given the way
device_xname() gets used everywhere now.

>How-To-Repeat:

        observe dmesg on some system with ld(4)

>Fix:

        unknown