Next mips bugs to examine...

Martin Husemann

unread,

Jun 27, 2010, 4:20:48 PM6/27/10

to

Hi folks,

after Simon fixed the emulation return address earlier today, most of
mips works again. New kernel with older userland seems to work completely
(i.e. I haven't managed to run into any problem there yet), but current
userland has at least two major breakages:

- sshd does not work (it claims hosts_access() would deny all access)
- getty does not work (I tracked this to setjmp() returning imediately,
which makes it think it timed out)

Another mystery, but unrelated, is why the emulation fix was needed not
only for my alchemy system (where I expected it), but also for sgimips,
where I did not expect any emulated instructions (R5000 and R10000 cpus).

Martin

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-...@muc.de

Martin Husemann

unread,

Jun 27, 2010, 4:41:48 PM6/27/10

to

On Sun, Jun 27, 2010 at 10:20:47PM +0200, Martin Husemann wrote:
> Another mystery, but unrelated, is why the emulation fix was needed not
> only for my alchemy system (where I expected it), but also for sgimips,
> where I did not expect any emulated instructions (R5000 and R10000 cpus).

And, being curios, I checked:

cpu0 at mainbus0: MIPS R5000 CPU (0x2321) Rev. 2.1 with built-in FPU Rev. 1.0
cpu0: 48 TLB entries, 16MB max page size
cpu0: 32KB/32B 2-way set-associative L1 Instruction cache
cpu0: 32KB/32B 2-way set-associative write-back L1 Data cache
cpu0: 1024KB/32B direct-mapped write-through L2 Unified cache

but:

MachEmulateFP+18 (4442f800,cba83f60,1000002c,cba83f60) ra 801a0d74 sz 24
MachEmulateInst+100 (4442f800,cba83f60,1000002c,cba83f60) ra 80233eac sz 80
trap+6ac (4442f800,1000002c,7dea4560,7dea459c) ra 8019f7b0 sz 120
mips3_UserGenException+d8 (4442f800,1000002c,7dea4560,7dea459c) ra 0 sz 0
User-level: pid 2.1

Unfortunately gdb does not work well right now and cba83f60 seems to be in
some shared library, so I don't realy know what the instruction was.

Michael

unread,

Jun 29, 2010, 11:54:40 AM6/29/10

to

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hello,

On Jun 27, 2010, at 4:20 PM, Martin Husemann wrote:

> after Simon fixed the emulation return address earlier today, most of
> mips works again. New kernel with older userland seems to work
> completely
> (i.e. I haven't managed to run into any problem there yet)

Not here - csh and tcsh at least still crash more or less immediately
while sh seems to work just fine. That's with a userland from roughly
around 5.99.18.

have fun
Michael

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7 (Darwin)

iQEVAwUBTCoXQcpnzkX8Yg2nAQJIcgf+OvLaqe9NnVle8wj5VTG/BxZqMwnd0hxx
1KUWetWwhmKwQXZ5IoysOUIB+pKoVXBbql7Ki/O1Rq7Nqsb4yaavzagZWJk/7WdY
zcqI0opZDsmnYGceTYV9tCIbGcJXMJZwHIYZbYFOeZbfWyQaxK61geXDiLa1Ma+E
GHxj/uuNMs/buOfPYaSAO3/HcWigbssM1Ublz7an0QkMmAS4QPkt7L+vZx0UsFs2
HOSdtoNpjUeQOJnkcZ7/WCHViE9YDHOT2iDaKV7tuQie3Or1iSiv3saS3JFuvWrp
paLrsWEbTPLFN9Zt11Sx33BAWSBPqD9QXFKOiZWn2cxhAzs08nJaHQ==
=hklX
-----END PGP SIGNATURE-----

David Holland

unread,

Jun 29, 2010, 10:29:48 PM6/29/10

to

On Tue, Jun 29, 2010 at 11:54:40AM -0400, Michael wrote:
> Not here - csh and tcsh at least still crash more or less immediately
> while sh seems to work just fine. That's with a userland from roughly
> around 5.99.18.

Check sigprocmask() and friends; csh is very sensitive to that for
some reason.

--
David A. Holland
dhol...@netbsd.org