SILO problems

SILO problems

[Matt Cole]

Ok, I'm ready to cry uncle on this one.

I have a Sun Ultra 10.  The machine was given to me without a hard 
drive, so I slapped an old 40GB Maxtor in there that I had sitting 
around and compiled an entire Linux setup from source.  The machine 
initially came with a password set in the NVRAM, so everything that I've 
done to it so far has been accomplished by allowing it to netboot off of 
my other computer.  Since then, I've pulled the old "pull the NVRAM chip 
out, put it back in backwards, turn the power on for 2 seconds, then put 
the NVRAM chip back in the right way" trick to get rid of the password.

Anywho, I'm having the damnedest time getting SILO to work.  I think 
I've tried just about every trick that Google knows of.  Initially, when 
I turned the machine on, I would get "the file just loaded does not 
appear to be executable".  I since ran SILO with -t to install it on 
/dev/hda1, and now I get:

Sun Ultra 5/10 UPA/PCI (UltraSPARC-IIi 440MHz), Keyboard Present
OpenBoot 3.31, 256MB (50 ns) memory installed, Serial #16661092.
Ethernet address 8:0:20:fe:3a:64, Host ID: 80fe3a64

Boot device: disk  File and args:

...and at this point the machine freezes.  It still responds to Stop+A, 
but any attempts to do "boot disk0", "boot disk:3", "boot disk:c", etc., 
etc., yield the same result.

Here's the dump of my partition table from fdisk:

Disk /dev/hda (Sun disk label): 64 heads, 32 sectors, 38170 cylinders
Units = cylinders of 2048 * 512 bytes

   Device Flag    Start       End    Blocks   Id  System
/dev/hda1   r         1        11     10240   83  Linux native
/dev/hda2  u         12     36241  37098496   83  Linux native
/dev/hda3             0     38170  39086080    5  Whole disk
/dev/hda4  u      36242     38170   1974272   82  Linux swap

/dev/hda1 is a small ext2 partition that has the SILO boot loaders, my 
kernel, and silo.conf.  Hopefully, the others should be obvious.  Any 
suggestions anyone has would be appreciated!

Thanks,
Matt Cole

Re: SILO problems

[Meelis Roos]

> Boot device: disk  File and args:
> 
> ...and at this point the machine freezes.  It still responds to Stop+A, 
> but any attempts to do "boot disk0", "boot disk:3", "boot disk:c", etc., 
> etc., yield the same result.

I assume you have tried to reinstall silo intio the boot block already?

I can not provide a good answer, but having struggled with similar 
problem once, I suggest backup your filesystem contents, wipe the 
beginning of the disk with dd if=/dez/zero, create an empty partition 
and reinstall. This has solved one silo problem and problems with 
partitioning and installer/parted.

-- 
Meelis Roos (mroos@linux.ee)

Re: SILO problems

[David Miller]

From: Matt Cole <mikaey@neb.rr.com>
Date: Sat, 07 Feb 2009 02:07:23 -0600

> Anywho, I'm having the damnedest time getting SILO to work.  I think
> I've tried just about every trick that Google knows of.  Initially,
> when I turned the machine on, I would get "the file just loaded does
> not appear to be executable".  I since ran SILO with -t to install
> it on /dev/hda1, and now I get:
>
> Sun Ultra 5/10 UPA/PCI (UltraSPARC-IIi 440MHz), Keyboard Present
> OpenBoot 3.31, 256MB (50 ns) memory installed, Serial #16661092.
> Ethernet address 8:0:20:fe:3a:64, Host ID: 80fe3a64
> 
> Boot device: disk  File and args:
>
> ...and at this point the machine freezes.  It still responds to
> Stop+A, but any attempts to do "boot disk0", "boot disk:3", "boot
> disk:c", etc., etc., yield the same result.

If you're going to do that you have to boot from that specific
partition.  For example, if you put the master bootblock on
/dev/hda1 you probably need to boot from "disk:a" or similar.

> Disk /dev/hda (Sun disk label): 64 heads, 32 sectors, 38170 cylinders
> Units = cylinders of 2048 * 512 bytes
> 
>    Device Flag    Start       End    Blocks   Id  System
> /dev/hda1   r         1        11     10240   83  Linux native
> /dev/hda2  u         12     36241  37098496   83  Linux native
> /dev/hda3             0     38170  39086080    5  Whole disk
> /dev/hda4  u      36242     38170   1974272   82  Linux swap

Hmmm, looking at mine:

   Device Flag    Start       End    Blocks   Id  System
/dev/hda1             0        12     96390    1  Boot
/dev/hda2            12     14219 114117727+  83  Linux native
/dev/hda3             0     14593 117218272+   5  Whole disk
/dev/hda4         14219     14593   3004155   82  Linux swap

I notice that I have none of those funny flags set, and my
boot partition starts at zero and is of type "1" or Sun partition
type "Boot".

Perhaps one of those two things makes a difference.

Re: SILO problems

[Tom "spot" Callaway]

On 2009-02-07 at 5:32:52 -0500, David Miller <davem@davemloft.net> wrote:
> I notice that I have none of those funny flags set, and my
> boot partition starts at zero and is of type "1" or Sun partition
> type "Boot".
> 
> Perhaps one of those two things makes a difference.

Your boot partition needs to start on 0, or silo can't be loaded by the
OBP. Hit this the hard way debugging the Fedora installer. :)

~spot

Re: SILO problems

[Matt Cole]

Tom "spot" Callaway wrote:
> On 2009-02-07 at 5:32:52 -0500, David Miller <davem@davemloft.net> wrote:
>   
>> I notice that I have none of those funny flags set, and my
>> boot partition starts at zero and is of type "1" or Sun partition
>> type "Boot".
>>
>> Perhaps one of those two things makes a difference.
>>     
>
> Your boot partition needs to start on 0, or silo can't be loaded by the
> OBP. Hit this the hard way debugging the Fedora installer. :)
>
> ~spot
>
>   
Thanks for the great responses everyone.

Meelie, I've reinstalled silo.  Several times.  Haven't tried zeroing 
out the beginning of the disk yet, but that will be in the list of 
things to try.

David, the flags I had set on my partitions were the mountable (r) flag, 
and the read-only (u) flag.  fdisk set these for me automatically.  I 
initially didn't bother to change it.  I went back and unset all the 
flags, no effect.

David and Tom, one of the problems I've noticed is that, with having the 
boot partition start on cylinder 0, that creating anything on that 
partition results in overwriting the disklabel.  David, I noticed your 
fdisk readout has your boot partition starting on cylinder 0, is this 
just a dummy partition?  Or how did you get around this problem?

Well, at any rate, I repartitioned the drive with the following:

Disk /dev/sdb (Sun disk label): 64 heads, 32 sectors, 38170 cylinders
Units = cylinders of 2048 * 512 bytes

   Device Flag    Start       End    Blocks   Id  System
/dev/sdb1             0         1      1024    1  Boot
/dev/sdb2             1        11     10240   83  Linux native
/dev/sdb3             0     38170  39086080    5  Whole disk
/dev/sdb4            11     34343  35155968   83  Linux native
/dev/sdb5         34343     38170   3918848   82  Linux swap

I'm in the process of copying everything back over to the drive.  I'll 
let you know how it goes.

Thanks,
Matt Cole

Re: SILO problems

[David Miller]

From: Matt Cole <mikaey@neb.rr.com>
Date: Sat, 07 Feb 2009 20:29:14 -0600

> David and Tom, one of the problems I've noticed is that, with having
> the boot partition start on cylinder 0, that creating anything on
> that partition results in overwriting the disklabel.  David, I
> noticed your fdisk readout has your boot partition starting on
> cylinder 0, is this just a dummy partition?  Or how did you get
> around this problem?

Most filesystems do not write their superblock into the initial
bytes of the partition, they leave it alone.

You can't put things like RAID and swap partitions there, because
those will in fact zap out your boot block and disk label.

Re: SILO problems

[Matt Cole]

David Miller wrote:
> From: Matt Cole <mikaey@neb.rr.com>
> Date: Sat, 07 Feb 2009 20:29:14 -0600
>
>   
>> David and Tom, one of the problems I've noticed is that, with having
>> the boot partition start on cylinder 0, that creating anything on
>> that partition results in overwriting the disklabel.  David, I
>> noticed your fdisk readout has your boot partition starting on
>> cylinder 0, is this just a dummy partition?  Or how did you get
>> around this problem?
>>     
>
> Most filesystems do not write their superblock into the initial
> bytes of the partition, they leave it alone.
>
> You can't put things like RAID and swap partitions there, because
> those will in fact zap out your boot block and disk label.
> --
> To unsubscribe from this list: send the line "unsubscribe sparclinux" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>
>   
Well I'll be damned...so you're right.  But still no luck.  I'm 
wondering if there's a deeper problem...you'd think if it was a problem 
with the first stage loader trying to find the second stage loader, I'd 
at least see the "S" printed, but I get nothing.  It would appear it's 
trying to execute something, however, because if I do Stop+A and issue 
any sort of "boot" command, it reboots the machine.  You don't suppose 
any of it would have to do that I downloaded the pre-compiled loaders, 
instead of compiling them from source?  (I tried, the second stage 
loader won't compile on my machine for some reason)

Thanks,
Matt Cole

SILO problems

[Jurij Smakov]

[-- Attachment #1: Type: text/plain, Size: 1900 bytes --]

Hi,

While preparing an update for Debian's SILO package, I ran into a 
problem: the version built from the current git tree [0] fails to boot 
on my SunBlade 1000 box with the following messages:

SUNW,Sun-Blade-1000 (2 X UltraSPARC-III) , No Keyboard
Copyright 1998-2004 Sun Microsystems, Inc.  All rights reserved.
OpenBoot 4.16.4, 2048 MB memory installed, Serial #51831895.
Ethernet address 0:3:ba:16:e4:57, Host ID: 8316e457.



Rebooting with command: boot                                          
Boot device: disk  File and args: 
SILO Version 1.4.14
boot: 
Allocated 64 Megs of memory at 0x40000000 for kernel
Uncompressing image...
     ERROR: Last Trap: Fast Data Access MMU Miss

Error -256 
{0} ok 

I have confirmed that the "silo: move second to make room for larger 
kernel" commit [1] is causing the breakage, excluding this commit 
makes the system boot normally:

Rebooting with command: boot                                          
Boot device: disk  File and args: 
SILO Version 1.4.14
boot: 
Allocated 64 Megs of memory at 0x40000000 for kernel
Uncompressing image...
Loaded kernel version 2.6.32
Loading initial ramdisk (7746912 bytes at 0x7F000000 phys, 0x40C00000 virt)...
\
[    0.000000] PROMLIB: Sun IEEE Boot Prom 'OBP 4.16.4 2004/12/18 05:18'
[...]

For now I just plan to exclude this patch from Debian package.

On a related note, I had to apply the attached patch (Signed-off-by: 
Jurij Smakov <jurij@debian.org>) to make SILO build with 2.6.32 kernel 
headers (which, apparently, are defining _LINUX_TYPES_H again).

[0] http://git.kernel.org/?p=linux/kernel/git/davem/silo.git;a=summary
[1] http://git.kernel.org/?p=linux/kernel/git/davem/silo.git;a=commit;h=71816c5699b32bab03e57be6768a562d9568e33b

Best regards,
-- 
Jurij Smakov                                           jurij@wooyd.org
Key: http://www.wooyd.org/pgpkey/                      KeyID: C99E03CC

[-- Attachment #2: silo-linux-types.patch --]
[-- Type: text/x-diff, Size: 763 bytes --]

diff -aur a/include/ext2fs/ext2fs.h b/include/ext2fs/ext2fs.h
--- a/include/ext2fs/ext2fs.h	2010-01-30 13:40:56.000000000 +0000
+++ b/include/ext2fs/ext2fs.h	2010-01-30 13:47:54.000000000 +0000
@@ -39,7 +39,7 @@
  */
 #define EXT2_LIB_CURRENT_REV	0
 
-#if defined(HAVE_SYS_TYPES_H) && !defined(_LINUX_TYPES_H)
+#if defined(HAVE_SYS_TYPES_H)
 #include <sys/types.h>
 #endif
 
diff -aur a/include/stringops.h b/include/stringops.h
--- a/include/stringops.h	2010-01-30 13:40:56.000000000 +0000
+++ b/include/stringops.h	2010-01-30 13:47:25.000000000 +0000
@@ -7,9 +7,7 @@
 #ifndef __SIZE_TYPE__
 #define __SIZE_TYPE__ long unsigned int
 #endif
-#ifndef _LINUX_TYPES_H
 typedef __SIZE_TYPE__ size_t;
-#endif
 
 /* stringops1.c */
 char *strcpy(char *, const char *);

Re: SILO problems

[David Miller]

From: Jurij Smakov <jurij@wooyd.org>
Date: Mon, 1 Feb 2010 21:30:40 +0000

> While preparing an update for Debian's SILO package, I ran into a 
> problem: the version built from the current git tree [0] fails to boot 
> on my SunBlade 1000 box with the following messages:

Thanks for reporting these issues Jurij, I'll look into
either fixing or reverting that buggy change and also
integrating your build fix patch.

Thanks again!

Re: SILO problems

[David Miller]

From: Jurij Smakov <jurij@wooyd.org>
Date: Mon, 1 Feb 2010 21:30:40 +0000

> I have confirmed that the "silo: move second to make room for larger 
> kernel" commit [1] is causing the breakage, excluding this commit 
> makes the system boot normally:
 ...
> On a related note, I had to apply the attached patch (Signed-off-by: 
> Jurij Smakov <jurij@debian.org>) to make SILO build with 2.6.32 kernel 
> headers (which, apparently, are defining _LINUX_TYPES_H again).

I've commited the revert and your build fix to the GIT tree
on kernel.org, thanks!