Re: newfs seems stuck in loop

[Hans Rosenfeld <rosenfeld%grumpf.hope-2000.org@localhost>]

Hi,

On Sun, May 02, 2010 at 10:22:11PM +0200, Rhialto wrote:
> Well, I started a sort-of-binary search, and I found that making the
> partition 1 700 000 sectors long worked, 2 000 000 sectors failed, and
> when I tried 1 900 000 sectors, the kernel paniced(!). I had several
> overlapping partitions in the disklabel, all starting at 1489266, with
> different sizes, as markers for the binary search.
> 
> panic: Segv in kernel mode: pc 800d307c addr 47
> Stopped in pid 199.1 (newfs) at netbsd:trap+0x298: movl $2,-52(fp)
> db> bt
> panic: Segv in kernel mode: pc %x addr %x
> Stack traceback :
> ... trap+...
> ... trap type=0x4c code=0x47 pc=800d307c psl=0xc00008
> ... vfs_busy+0xe(0xffffffff,0x10,0x80148e20)
> ... sys_getvfsstat+0x5e(0x803c5c98,0x84abdf60,0x84abdf58)
> ... syscall+0xdc(0x84abdfb4)
> 
> When I continued trying after a reboot, it didn't do it again at that
> size. 1 811 250 worked, 1 822 500 failed, and it paniced again at 1 817
> 250 sectors, again with a segv in kernel mode, but in sh and with a
> different backtrace. I would think maybe my microVAX's poor brain may be
> getting a little old.
> 
> pmap_page_protect_long
> uvm_anfree
> amap_wipeout
> amap_unref_detach
> uvmspace_free
> uvm_proc_exit
> exit1
> sigexit
> postsig
> trap
> 
> In emulation I have tried to reproduce any of this, but no "luck".

FYI, today I tried to get 5.1_RC1 installed onto a RZ26 (1GB) disk in 2
different 3100m76 and one m30. Had no luck, all kinds of strange things
happened:
- newfs died with "illegal instruction"
- machine check 24
- kernel segv 

I have no idea whats causing this. The same install kernel works
flawlessly on a VLC.


-- 
%SYSTEM-F-ANARCHISM, The operating system has been overthrown