Rafal Boni | 6 Mar 2003 10:16
Picon

Old ld_elf.so and 1.6P == dead dynamic binaries? (worked with 1.6O)

Folks:
	I'm trying to get my sgimips hardware up-to-date and came across
	this earlier this evening:

	I was running a 1.6O kernel and an older ld_elf.so (though I can't
	say of what exact date as I blew it away in order to get the box
	up and running and didn't note the date on it).  

	I rebuilt a 1.6P kernel as of ~ 1:50 EST, 3/06/2003 to add UNIONFS
	so I could get my userland build going again (needed to union-mount
	some more space to complete the build 8-); after rebooting with the
	new kernel, all dynamic executables I ran would dump core.

	Extracting a base.tgz I had cross-built on 2/28 from my i386 fixed
	things up to run again, but it was strange that things that had no
	issue on 1.6O completely fell over with 1.6P.  Any clues?  I do have
	one or two more sgimips boxes to upgrade, so I'll try and keep my
	eyes better peeled next time (and/or take notes) to see if this is
	just some glitch I hit or a more serious incompatibility.

	The kernels:
/netbsd
	NetBSD 1.6P (BABYLON_SOUND_SYSTEM) #0: Thu Mar  6 02:57:56 EST 2003

/netbsd.old
	NetBSD 1.6O (BABYLON_SOUND_SYSTEM) #0: Tue Feb 25 23:24:39 EST 2003

Any ideas appreciated,
--rafal

(Continue reading)

Ilpo Ruotsalainen | 6 Mar 2003 10:31
Picon
Picon
Favicon

Re: Old ld_elf.so and 1.6P == dead dynamic binaries? (worked with 1.6O)

On Thu Mar 06 2003 at 04:16:17 -0500, Rafal Boni wrote:
> 	I rebuilt a 1.6P kernel as of ~ 1:50 EST, 3/06/2003 to add UNIONFS
> 	so I could get my userland build going again (needed to union-mount
> 	some more space to complete the build 8-); after rebooting with the
> 	new kernel, all dynamic executables I ran would dump core.

1.6 userland, 1.6N kernel, same result (sgimips here too). Didn't have
time to investigate, replacing userland with some releng snapshot solved
the issue.

--

-- 
Ilpo Ruotsalainen - <lonewolf <at> iki.fi> - http://www.iki.fi/lonewolf/

Mark Hoekstra | 6 Mar 2003 23:55

Challenge S netbootproblem, SCSI?

Hello Everyone,

Well, I'm new to this list, new to SGI-hardware and new to 
netbsd-sgimips(but I run netbsd-sparc for a while now...) ;-). I've gotten 
myself a Challenge S R5000SC180 and I'm trying to netboot it and install 
netbsd on it.

I got everything sorted out as far as I could find online and in the 
archives, but I haven't found an answer to this problem:

Whenever I try to boot from netbsd-INDY_INSTALL (I also tried 
netbsd-GENERIC_INDY, same problem)

                          Running

                            Starting up the 
system...

                To perform system maintenance instead, press 
<Esc>
zs channel 0 had address 0xbfbd9830
  [ no symbols available ]
IOC rev 1, machine Indy (Guiness), board rev 
3
Timer calibration, got 900000 cycles (900000, 900000, 
900000)
CPU clock speed = 180.00Mhz
Copyright (c) 1996, 1997, 1998, 1999, 2000, 2001, 
2002
     The NetBSD Foundation, Inc.  All rights 
(Continue reading)

Rafal Boni | 7 Mar 2003 17:43

Re: Challenge S netbootproblem, SCSI?

In message <5.2.0.9.0.20030306230326.02819068 <at> getmail.hoekstra.com>, you write:

-> Hello Everyone,
-> 
-> Well, I'm new to this list, new to SGI-hardware and new to 
-> netbsd-sgimips(but I run netbsd-sparc for a while now...) ;-). I've gotten 
-> myself a Challenge S R5000SC180 and I'm trying to netboot it and install 
-> netbsd on it.
-> 
-> I got everything sorted out as far as I could find online and in the 
-> archives, but I haven't found an answer to this problem:
-> 
-> Whenever I try to boot from netbsd-INDY_INSTALL (I also tried 
-> netbsd-GENERIC_INDY, same problem)
-> 
[...]
-> wdsc0 at hpc0 offset 0x44000: WD33C93B SCSI, rev=0, target 7
-> scsibus2 at wdsc0: 8 targets, 8 luns per target
[...]
-> scsibus2: waiting 2 seconds for devices to settle...
-> MSGOUT Failed!
-> wd33c93_intr: LCI asr:c0 csr:1f
-> probe(wdsc0:0:1:0): wdsc0: timed out; asr=0x20 [acb 0x88be2000 (flags 0x1, 
-> dleft 20)], <state 5, nexus 0x88be2000, resid 20, msg(q 80,o 0)>
-> probe(wdsc0:0:1:0): ABORT in timeout: csr=0xff, asr=0x20
-> probe(wdsc0:0:1:0): sending ABORT command
-> probe(wdsc0:0:1:0): Resetting bus: WD33C93B SCSI, rev=0, target 7
-> probe(wdsc0:0:1:0): wdsc0: timed out; asr=0x00 [acb 0x88be2000 (flags 0x41, 
-> dleft 20)], <state 8, nexus 0x88be2000, resid 20, msg(q 80,o 0)>
-> probe(wdsc0:0:1:0): ABORT in timeout: csr=0x01, asr=0x00
(Continue reading)

Mark Hoekstra | 7 Mar 2003 19:12

Re: Challenge S netbootproblem, SCSI?


>-> wd33c93_wait: TIMEO  <at> 877 with asr=x0 csr=x1
>
>I'd make sure that your SCSI bus is correctly terminated, etc.  However,
>there are some known (?) bugs IIRC with timeout handling in the wd33c93
>SCSI driver.  Unfortunately, I forgot all the little SCSI clue I once had,
>so I don't even have good diags for you to try.  Does `hinv' from the PROM
>report the drive OK, or does IRIX see it if you have an IRIX installation?
>
>--rafal

Yes, I guess it was pulled in working order from a serverroom. It boots 
IRIX as long as i don't put netBSD over it. I can't do anything with the 
IRIX-install as i don't have the rootpass... But anyway, I guess it's 
correctly terminated and such...

I also read this on this mailinglist:

http://mail-index.netbsd.org/port-sgimips/2001/11/14/0004.html

..seems there's no solution since? Maybe any one a clue for a working 
(older) kernel for the Challenge S? I don't mind the mezz-board and my 
phobos G100 card won't work(for now) ;-)

mark 

Wayne Knowles | 7 Mar 2003 21:45
Picon

Re: Challenge S netbootproblem, SCSI?

On Fri, 7 Mar 2003, Mark Hoekstra wrote:

> Yes, I guess it was pulled in working order from a serverroom. It boots
> IRIX as long as i don't put netBSD over it. I can't do anything with the
> IRIX-install as i don't have the rootpass... But anyway, I guess it's
> correctly terminated and such...
>
> I also read this on this mailinglist:
>
> http://mail-index.netbsd.org/port-sgimips/2001/11/14/0004.html
>
> ..seems there's no solution since? Maybe any one a clue for a working
> (older) kernel for the Challenge S? I don't mind the mezz-board and my
> phobos G100 card won't work(for now) ;-)

I believe the problem described above is related to a problem with the
SCSI Bus and Crontroller not being reset correctly by NetBSD during
startup.  This was resolved several days later in rev 1.3 of wdsc.c

There may be a problem with the DMA or Interrupt handling caused by
multiple HPC3's in you challenge S.
Target 1 was found, which is a healthy sign, but I suspect that it isn't
starting the sync negotiation correctly, which makes me suspect
interrupts are not being delivered correcly.

Will have a look over the sources to see if I can spot any problems.

In the meantime, could you send the output of a SCSI bus probe

--

-- 
(Continue reading)

systems engineer | 8 Mar 2003 00:09

Re: Challenge S netbootproblem, SCSI?

> so I don't even have good diags for you to try.  Does `hinv' from the PROM
> report the drive OK, or does IRIX see it if you have an IRIX installation?

nb: indy, for some reason, uses 0 as the initiator target, so setting a
drive to id 0 (as would be common practice in other systems) freaks the
system out. prom hinv will report this as disk on all targets.

systems engineer | 8 Mar 2003 00:11

Re: Challenge S netbootproblem, SCSI?

> ..seems there's no solution since? Maybe any one a clue for a working 
> (older) kernel for the Challenge S? I don't mind the mezz-board and my 
> phobos G100 card won't work(for now) ;-)

i meant to get back to the list on this one:

the 33c95 on challenge mezz cards (used in the big iron as well as the S)
is undocumented, difficult to program, and requires microcode. it is
surrounded by enough weird glue logic that implementing a driver from
scratch would be "nearly impossible". sigh. no wide scsi for me.

Mark Hoekstra | 8 Mar 2003 01:20

Re: Challenge S netbootproblem, SCSI?


>
>In the meantime, could you send the output of a SCSI bus probe
>
>--
>Wayne Knowles                   NetBSD/mipsco port maintainer
>wdk <at> netbsd.org                  http://www.netbsd.org

Well, here you go, this comes by when I boot it(this one is from 
netbsd-INDY_INSTALL):

wdsc0 at hpc0 offset 0x44000: WD33C93B SCSI, rev=0, target 
7
scsibus2 at wdsc0: 8 targets
dsclock0 at hpc0 offset 0x60000
hpc1 at gio0 addr 0x1fb00000: SGI HPC3
zsc at hpc1 offset 0x59830 not 
configured
sq at hpc1 offset 0x54000 not 
configured
wdsc at hpc1 offset 0x44000 not 
configured
dsclock at hpc1 offset 0x60000 not 
configured
biomask 07 netmask 07 ttymask 0f clockmask 
bf
scsibus2: waiting 2 seconds for devices to 
settle...
MSGOUT Failed!
wd33c93_intr: LCI asr:c0 csr:1f
(Continue reading)

Rafal Boni | 8 Mar 2003 04:12
Picon

Re: Challenge S netbootproblem, SCSI?

In message <5.2.0.9.0.20030308011238.0283f800 <at> getmail.hoekstra.com>, you write:

->  >> hinv
->                     System: IP22
->                  Processor: 180 Mhz R5000, with FPU
->       Primary I-cache size: 32 Kbytes
->       Primary D-cache size: 32 Kbytes
->       Secondary cache size: 512 Kbytes
->                Memory size: 64 Mbytes
->                  SCSI Disk: scsi(0)disk(1)

Are you trying to boot 1.6, or something more recent?  I'm not sure 1.6 was
ever tested on a R5kSC, but I think the workarounds for lack of secondary
there may have worked well enough... If you're running -current, I'm quite
sure it won't work as in CVS right now both because people have reported
issues with R5kSC Indys and because I still have a bunch of R5k cache code
which I haven't checked in (I should make time to get that checked in RSN).

--rafal

----
Rafal Boni                                                     rafal <at> attbi.com
  We are all worms.  But I do believe I am a glowworm.  -- Winston Churchill


Gmane