View previous topic :: View next topic |
Author |
Message |
joakim_tosteberg n00b
Joined: 17 Sep 2005 Posts: 6 Location: Linkoping, Sweden
|
Posted: Tue Nov 28, 2006 7:38 am Post subject: SILO doesn't work |
|
|
I recently updated several programs on my UltraSparc10, As gcc was among the upgraded programs (frpm 4.1.1-r1 to 4.1.1-r2) I also reemerge silo afterwards and ran silo -f to enusre that it would work correctly. But after rebooting SILO just refused to load at all, doesn't even get to the stage1 loader from what I can see. The prom just outputs "Boot device: mybootdevice File and args: " And then nothing at all, just syas like that forever. I first suspected the gcc upgrade as the sme thing had happend on another machine also there during an upgrade including gcc so I booted from a livecd tried downgrading gcc and reemerged SILO with all its dependencys but without success. I stil think it is related to the upgrade of the system in some way (as quite a few other packages was updated aswell) but I'm completely cluless about what exactly the cause can be, if anyone has any suggesting i would be happy.
Here is a link to an emerge.log containing only the packages that was upgraded prior to the failur on one of the computers if it can help.
emerge.log
Any help would be much appreciated.
Edit: Oh, one thing more, I can use SILO from the gentoo livecd to boot into my system.
//Joakim Tosteberg |
|
Back to top |
|
|
Risk n00b
Joined: 05 Mar 2004 Posts: 38 Location: Zoo York City
|
Posted: Wed Nov 29, 2006 3:28 pm Post subject: |
|
|
I'm having the same problem on an E4500 I stupidly decided to migrate to Gentoo.
Using the latest SILO 1.4.13 nothing happens at all.
Upon boot time the disk designation is displayed and it just hangs - no "SIL", no "SI", no "S", nothing
Have installed boot code fourteen or fifteen different times using:
Code: | silo -a
silo -f -a -C /etc/silo.conf
silo -f -C /etc/silo.conf
silo -f
etcetera ad nauseum |
and several different variations of silo.conf information (all of which known working on other SPARC machines)
As we all know, the Enterprise machines take between two and five minutes to POST and finally boot (we're talking twelve processors) so how many waking hours did I waste on this?
Would have tried earlier versions of SILO, but note that earlier versions have trouble with large quantities of memory (this E4500 system has 12G). The ridiculous kludge of removing CPU/mem boards to get the system to boot was not an option before which is why I waited until 2006.1 was available and in use with a SILO that works. Or so I thought.
Sorry that this may not be helpful to you other than the fact you're not the only one with SPARC/SILO problems. |
|
Back to top |
|
|
joakim_tosteberg n00b
Joined: 17 Sep 2005 Posts: 6 Location: Linkoping, Sweden
|
Posted: Wed Nov 29, 2006 3:40 pm Post subject: |
|
|
I have now successfully been able to get SILO to work once again by using nfs to mount /usr/portage from another of our UltaSparc10 computers which I haven't synced in a while and which boots correctly. So now I had an older version of the portage tree so I run an emerge -uD world to _downgrade_ all packages that I had updated (I knew that the portage tree on the two machines was at approximatly the same version before I begun uppgrading the one which failed). After donwgrading some pacakges I reemerge SILO and rebooted, and guess what? It worked. So I'm now inb the proccess of fidning out exactly which packages it was that caused the problem initially, I think I'm down to about 10 possible causes right now so Iäm going to update the one by one, remerge SILO and reboot after each update until I find the problem. |
|
Back to top |
|
|
Risk n00b
Joined: 05 Mar 2004 Posts: 38 Location: Zoo York City
|
Posted: Wed Nov 29, 2006 8:18 pm Post subject: |
|
|
Perhaps you would be kind enough to mention the version number of SILO that you were successful with.
Other version information may also be helpful... what sizes/dates do you show for your:
-rw-r--r-- 1 root root 1024 Jul 7 21:18 fd.b
-rw-r--r-- 1 root root 512 Jul 7 21:18 first.b
-rw-r--r-- 1 root root 1024 Jul 7 21:18 generic.b
-rw-r--r-- 1 root root 724 Jul 7 21:18 ieee32.b
-rw-r--r-- 1 root root 6928 Jul 7 21:18 isofs.b
-rw-r--r-- 1 root root 63488 Nov 15 12:54 second.b
-rw-r--r-- 1 root root 512 Jul 7 21:18 ultra.b
what version of the kernel and udev did you use to successfully get your machine to boot? |
|
Back to top |
|
|
joakim_tosteberg n00b
Joined: 17 Sep 2005 Posts: 6 Location: Linkoping, Sweden
|
Posted: Thu Nov 30, 2006 12:03 pm Post subject: |
|
|
I'm using SILO 1.4.13 and all of the files are the same size except my second.b which is 64512bytes. Dates are irrelevant as they onöy whow the time you emerged silo
I'm currently in the process of tracking down the packages that made SILO stop working, will report back as quick as I have any results. |
|
Back to top |
|
|
joakim_tosteberg n00b
Joined: 17 Sep 2005 Posts: 6 Location: Linkoping, Sweden
|
Posted: Thu Nov 30, 2006 12:52 pm Post subject: |
|
|
There, now do I know whick package that caused SILO to fail =)
Seems like it was the upgrade of bash from 3.1_p17 to 3.2_p5 that was responsible, so I'm just going to add the newer version of bash to my package.mask for now. |
|
Back to top |
|
|
Risk n00b
Joined: 05 Mar 2004 Posts: 38 Location: Zoo York City
|
Posted: Sun Dec 03, 2006 5:59 pm Post subject: |
|
|
How incredibly odd.
I'm wondering if BASH is the only package that affects SILO on SPARC.
I'll double check the versions on 2006.1 "livecd" and on the system I've been trying to install.
I'm thinking it may also be helpful to compare strace output on both versions to see what, if anything, is being done differently between the two bash versions when executing SILO. |
|
Back to top |
|
|
Risk n00b
Joined: 05 Mar 2004 Posts: 38 Location: Zoo York City
|
Posted: Wed Dec 13, 2006 6:18 pm Post subject: |
|
|
VERIFIED
Something about the new Bash version 3.2_p5 is indeed causing SILO to behave improperly
On a Sun E4500 with 12G main mem, using silo 1.4.13 under bash 3.1_p17 results in a system that boots.
Using silo under Bash 3.2_p5 results in a system that just hangs at boot time. |
|
Back to top |
|
|
Weeve Retired Dev
Joined: 30 Oct 2002 Posts: 641
|
Posted: Wed Dec 13, 2006 9:03 pm Post subject: |
|
|
Does the E4500 also run gcc-4.x? This would be something good to know before it becomes unmasked before general consumption. |
|
Back to top |
|
|
Risk n00b
Joined: 05 Mar 2004 Posts: 38 Location: Zoo York City
|
Posted: Thu Dec 14, 2006 2:46 pm Post subject: |
|
|
Oddly enough, this phenomena was evident compiling SILO and Bash with both GCC 4.1.1 and GCC 3.4.6-r2
(albeit the 3.4.6-r2 version was chosen through an already upgraded gcc-config system) |
|
Back to top |
|
|
Attila Tux's lil' helper
Joined: 28 Feb 2003 Posts: 93
|
Posted: Tue Jun 05, 2007 6:41 am Post subject: |
|
|
Hiho,
Very strange! - Last WE i updated my Sparc Gentoo to GCC4 (from 3.3.x) and i run into the same problem. After running silo -f -C /boot/silo.conf the system did not boot anymore (*nothing* happens).
After downgrading bash to 3.1p_17 it works again! Very strange and good to know!
Atti |
|
Back to top |
|
|
Risk n00b
Joined: 05 Mar 2004 Posts: 38 Location: Zoo York City
|
Posted: Sat Jun 23, 2007 1:27 am Post subject: |
|
|
Well, well, well. I just spent my entire 12 hour (so far) day trying to get our Ultra-80 running Gentoo to boot again.
Seems I just never learn to never, EVER use gentoo for anything depended upon. So now I'm stuck again.
Neither GCC 4.1.2 nor 3.4.6 will successfully re-compile silo 1.4.13 whether I use bash 3.1.x or 3.2p11.
Dozens of errors in main.c referring to "no such xxx in union" and other such.
No kernel, previously working or not, will boot. 2.6.21-r3, 2.6.20-r9, 2.6.19-r7, 2.6.18-r7, 2.6.17-r9 (the last working kernel) none of these have gotten past "Booting Linux..."
Please, any suggestion to just get this piece-of-crap system up and running one more time and I will migrate EVERY system off of this junk and onto a better supported hardware/software config. Gentoo has just been a continuous stream of unmitigated disasters for us. This one last Ultra80 running gentoo is crucial and now everyone is just waiting for me to get it running again.
/etc/fstab looks good
installed the boot block using "silo -C /etc/silo.conf -f " and various combinations of that command at least ONE HUNDRED TIMES
used this command after booting from the latest 2007.0 SPARC64 universal CD at least a thirty or so times
e2fsck on the target boot disk six times, no problems
ran SILO using both BASH 3.2_p17 and 3.1_p17 |
|
Back to top |
|
|
|