Who/what broke -current kernel?

[Peter Jeremy]

On Fri, May 13, 2005 at 12:45:15PM +0400, Andrey Chernov wrote:
>On Fri, May 13, 2005 at 11:49:48AM +0400, Andrey Chernov wrote:
>> I'll try, but I doubt it helps, if directory is not found.
>> It looks line someone damage ufs code or softupdates (I have them).
>
>I try bootverbose and it not difers from old (i.e. working) kernel 
>output.

OK.  Elsewhere you've wondered if it is a low-level bug in ATAmkIII.
If you believe it may be a low-level bug, you could try:
1) Get a USB memory stick, floppy disk or CD-ROM
2) Build a UFS[2] filesystem on it
3) Copy /sbin/init into an sbin subdirectory on the device
4) Use "boot -a" and specify that device as the root filesystem.

This won't work (because there's nothing for init to do if it starts)
but it will let you determine whether the bug is in the UFS[2] code
(exec still fails) or in the ATA driver (the exec works - init starts
but dies because there's nothing else in root).

Your other option is a binary search through -current and/or a session
with kgdb or ddb.

Peter