re: NetBSD-current/sparc install ISO no longer boots

[matthew green <mrg%eterna.com.au@localhost>]

> Hi all,
> 
> The NetBSD-current SPARC install ISO no longer installs; it panics
> during boot, at least on the qemu-based TNF automated test server.
> The problem was introduced some time in the last few days.  Maybe it
> is related to the switch of the sparc port to GCC 4.5, but it's hard
> to tell because the switch was made at a point in time when the sparc
> build was already broken...
> 
> The panic looks like this:
> 
>    Jumping to entry point 00004000 for type 00000005...
>    >> NetBSD/sparc Secondary Boot, Revision 1.15 (Sat Aug 20 12:43:27 UTC 
> 2011)
>    Booting netbsd
>    2981676+82236=0x2ec4b8
>    OBP version 3, revision 2.25 (plugin rev 2)
>    qemu: fatal: Trap 0x29 while interrupts disabled, Error state
>    pc: f0006010  npc: f0006014
>    General Registers:
>    %g0-7: 00000000 00000200 00000000 00000001 00000034 00000fff f0006000 
> ffd37170
> 
>    Current Register Window:
>    %o0-7: 00000000 00000004 00000001 ffd32d98 ffd32e00 00000004 f01eb4c0 
> ffd07880
>    %l0-7: ffd7275c f0012ed8 f0012edc f01eb4f0 f01eb4ee f01cbef8 00000001 
> f02e034f
>    %i0-7: 04000fc5 ffd3713c 00000004 f01eb594 f01eb598 f01bc144 f01eb540 
> f0012ecc
> 
>    Floating Point Registers:
>    %f00: 000000000.000000 000000000.000000 000000000.000000 000000000.000000
>    %f04: 000000000.000000 000000000.000000 000000000.000000 000000000.000000
>    %f08: 000000000.000000 000000000.000000 000000000.000000 000000000.000000
>    %f12: 000000000.000000 000000000.000000 000000000.000000 000000000.000000
>    %f16: 000000000.000000 000000000.000000 000000000.000000 000000000.000000
>    %f20: 000000000.000000 000000000.000000 000000000.000000 000000000.000000
>    %f24: 000000000.000000
> 
> Full log at
> 
>   
> http://releng.netbsd.org/b5reports/sparc/build/2011.08.20.06.00.14/install.log
> 
> and additional logs via the "install: failed" links around
> 
>   
> http://releng.netbsd.org/b5reports/sparc/commits-2011.08.html#2011.08.20.06.00.14
> 
> I am currently redoing some of the builds with a work-around for bin/45269
> in place; this may help narrow things down a little bit.

i'm pretty sure that this is related to GCC 4.5 change.  the same tree
built with GCC 4.1 gets beyond the kernel init.

it dies after loading the kernel while starting up.  trap 0x29 is a
data fault, though the pc indicates it is handling a text fault.

more investigation is pending.


.mrg.