I still register several crashes and reboots
This is not my first post about kernel crashing (see old posts as: spontaneous reboots). However now I have some more information to provide you. So I hope it could help to determinate the problem that many people have. First: the hardware cpu : 740/750 temperature : 39 C (uncalibrated) clock : 300MHz revision : 2.2 (pvr 0008 0202) bogomips : 598.01 machine : PowerMac1,1 motherboard : PowerMac1,1 MacRISC Power Macintosh L2 cache: 512K unified memory : 96MB pmac-generation : NewWorld Second: the kernel Linux serverg3 2.4.18-6.1mdk #1 lun giu 24 07:53:06 CEST 2002 ppc Third: the facts I upgraded from 8.0 to 8.2 PPC hoping that the crash/reboot problem will solve: well, it still persists. One or 2 times I could see the error on the monitor (it was months I was trying to see it, because when the crash happens the machine reboots by itself and no messages in the logs). The error said something like: Machine check in kernel mode Caused by (from SRR1=41000): Transfer error ... could not read ... Altivec error could not read... So it seems to be something about altivec. Yesterday I recompiled the kernel *without* altivec support (my G3 does not have altivec). Today the new crash with the same error. My next step is to leave mdk kernel and recompile the unpatched one from kernel.org Has someone an idea about this ? Note that the crash is rare: it can crash in 24 hours or in 25 days. Hoping my informations can be helpful, Miguel Beccari
OT: (but I do not know who ask for): buildin an RPM from scratch
I builded a RPM from scratch of an application of mine. The SRC rpm works: it builds 2 i586 architecture packages, it has got good controls, it check dependecies and so on. My trouble is this: how to force the user to answer some questions before install the package ? I wrote down theese lines: %pre echo Bla bla bla.; echo -n Answer (yes | no | later); read chiave while ( test -z $chiave ); do read chiave done; that produce a /var/tmp/rpm-tmp.19101 as follow: echo Bla bla bla.; echo -n Answer (yes | no | later); read chiave while ( test -z $chiave ); do read chiave done; The problem is that $chiave is NEVER != zero. Some one can help me ? Realy tnx, Miguel Beccari
Re: OT: (but I do not know who ask for): buildin an RPM from scratch
Miguel Beccari wrote: %pre echo Bla bla bla.; echo -n Answer (yes | no | later); read chiave while ( test -z $chiave ); do read chiave done; OK. In the docs I read: There are certain caveats about these scripts which you should take into account. Number one, each must fit inside a 8192 buffer, and number two, they should be non-interactive. Anything which requires manual input from the user is wrong, as this will break non-interactive RPM installation procedures. Regards, Miguel Beccari
Re: OT: (but I do not know who ask for): buildin an RPM from scratch
Stew Benedict wrote:
Are you saying the script behave differently in the RPM than if you just
run it? I would do something a little more bulletproof, which would force
them to answer one of your choices, unless they ctrl-C out of it:
Yes (I am saying the script behave differently).
I used your suggestions, but the result is the same.
If I run sh /var/tmp/rpm-tmp.54377 I have got the promt Answer (yes |
no | later)
If I answer the script goes on as well. It is OK.
But when I run rpm -Uhv I have a loop: Answer (yes | no | later) Answer
(yes | no | later) Answer (yes | no | later) Answer (yes | no | later)
Answer (yes | no | later) Answer (yes | no | later) [etc...]
This because of $chiave is empty
GetInput () {
echo -n Answer (yes | no | later);
read chiave
while [ -z $chiave ]; do
GetInput
done;
}
Because of $chiave is empry GetInput is looped.
Miguel Beccari
Re: OT: (but I do not know who ask for): buildin an RPM from scratch
Rick Thomas wrote: When the script is run by rpm, it probably has input redirected from /dev/null. As the docs say, it should be non-interactive. Anything else will break scripted installations. Yes. It is so. In fact I used something like %pre $bashscript = /tmp/myfile echo #!/bin/sh $bashscript echo Stew Benedict suggestion $bashscript chmod 700 $bashscript sh $bashscript So i was REALLY sure it was a bash script to be executed (sorry but I am new to rpms) and the problem was the same. Then I run /tmp/myfile and it worked... so I think the input is redirected from /dev/null. OK. I will write a mini setup program as Microsft does ;) Up to now I just echo out the instructions and the license (with a line that sounds like 'using the software presumes you have accepted the license terms at all Regards, Miguel Beccari
kernel.src.rpm: when a ppc compilable release?
Hi everybody, Just want to know when kernel source packages will compile under ppc. My trouble is always spontaneus reboot, and someone said new kernels are rock solid. I am using kernel-2.4.18-7 (the last compilable under ppc) and spontaneus reboots happen ratherly (once a day). With old kernels (2.4.17 and prior) I had 5, 6 reboots per day. It is not a good thing: my G3 runs as fileserver and has to be always up. I've already posted the console errors I saw just few seconds earlier the reboot (in the logs nothing useful). Regards, Mi[G]hi
Re: kernel.src.rpm: when a ppc compilable release?
My trouble is always spontaneus reboot, and someone said new kernels are rock solid. No one said that about the kenrels under ppc. Most of the time they won't compile without some patching and some fixing. I wanted to say (sorry for my english): My trouble 'are spontaneus reboots'. Someone said that with kernels 2.4.18-6 no reboots happen. So I want to upgrade the kernel (but I can't because new rpms do not compile under ppc): the problem is spontaneus reboot. I am using kernel-2.4.18-7 (the last compilable under ppc) and spontaneus reboots happen ratherly (once a day). We're using kernel-2.4.18-6.2 under PPC for 8.2. I don't think Stew had done any work on 7 to try and make it work right under PPC. Maybe mine is 2.4.18-6mdk too. With old kernels (2.4.17 and prior) I had 5, 6 reboots per day. It is not a good thing: my G3 runs as fileserver and has to be always up. I've already posted the console errors I saw just few seconds earlier the reboot (in the logs nothing useful). So use the right kernel. Thanks for your answer. Regards, Mi[G]hi
Re: Cannot rebuild kernel-2.4.18.10mdk.src.rpm
Miguel Beccari wrote: Patch #1300 (linux-2.4.12-riva-ppc.patch.bz2): + /usr/bin/bzip2 -d + patch -p1 -s 1 out of 5 hunks FAILED -- saving rejects to file drivers/video/riva/fbdev.c.rejerror: Bad exit status from /var/tmp/rpm-tmp.23339 (%prep) I cannot compile kernel-2.4.18.8mdk.src.rpm, kernel-2.4.18.9mdk.src.rpm, and now kernel-2.4.18.10mdk.src.rpm. If I try to comment Patch #1300 other errors come out in compilation step. Many other erros as: init/do_mounts.c: At top level: init/do_mounts.c:956: parse error before `crd_load' init/do_mounts.c:957: warning: return-type defaults to `int' make: *** [init/do_mounts.o] Error 1 error: Bad exit status from /var/tmp/rpm-tmp.67757 (%build) So the question is: Is kernel-2.4.18.10mdk.src.rpm compilable under ppc ? regards, Miguel Beccari
Re: Cannot rebuild kernel-2.4.18.10mdk.src.rpm
Ben Reser wrote: On Fri, Apr 12, 2002 at 10:10:14AM +0200, Miguel Beccari wrote: So the question is: Is kernel-2.4.18.10mdk.src.rpm compilable under ppc ? Probably not. kernel-2.4.18.6mdk.src.rpm should be the right one. But since x86 cooker has marched on I doubt the SRPM is still on the mirrors. I have got kernel-2.4.18.7mdk.src.rpm working on G3 (so the last working is 7mdk). Ok. Tnx, Miguel Beccari
We have spontaneus reboots
I read some mails (in this ml) about spontaneus reboots. I have 2 blue G3. The first has got MDK8.0 (with 2.4.17 kernel), the secons has got MDK8.2beta1 (with kernel 2.4.18). I registered on eitheir machines spontaneus reboots. In the log I cannot find anything useful, but today morning I saw the monitor: I read something like cua state=idle on the first line. I couldn't read the second line because in few seconds the machine rebooted. It was not a crash, but a regular reboot: something like going in init 6. Any suggestion, now ? Regards, Miguel Beccari
Re: We have spontaneus reboots
Luis M wrote: it could be a kernel panic. when that happens to my ibook ( when I was using Mandrake's kernels as you are now), my ibook kept panic'ing for no reason. I could not see anything in the logs. It usually happened after being away from it for a long time -- Yes. My G3s are servers (file servers) and they reboots on nights or on Saturday and Sunday (I see this with 'uptime') and usually whenever I mounted something, like a NFS drive (I compile my kernel with NFS support builtin, not as a module, just in case -- I use it all the time with the ibook). I thought it was some 'modprobe -r' (cleaning the modules) trying to unload a module or something, and that causing the panic. I want to try this. My solution for it was to get the source for the kernel ( got mine from tuxppc.org because I could not get the kernel-source...rpm to compile in my ibook... well, now I know that I should've used 'make mrpropper', but anyway... ) after compiling the kernel using 2.4.19-pre4 my ibook has been rock solid. OK. So I need just a new kernel. tnx, Miguel Beccari
Re: failed to read kmap magic, PowerMac 7200 this time
Stew Benedict wrote: On Thu, 28 Feb 2002, Janne Karjanlahti wrote: Hello! Thu, 21 Feb 2002 Miguel Beccari wrote about errors with his PowerMac 8500 I have identical problems with my old 7200 / 90 with 32 mb memory I want to use it as a router and firewall, so only 32 mb is sufficient I don't remember how much RAM Miguel had, but maybe that's the issue 32mb may just be too little Definitely if you only have 32MB, don't try to allocate all of it for ramdisk, as the kernel needs some too Try some smaller figure like 24MB I may have to pull some RAM from my 7600 and see if I can duplicate this behavior Stew Benedict I had (and have) 96Mb Miguel Beccari
Re: failed to read kmap magic (long reading)
Is is the same :( checksum is ok, burned 3 cds (1 more from linux, 1 from Mac, 1 from PC). Maybe I am using text installation instead of graphical ? In this case, can you tell me what video driver to select ? That's possible I suppose, although others have done a text install. For my 7600, I don't specify anything in the kernel arguments, and the graphical install works. I'll rerun a text install through that point to see if I can duplicate your error, tonight. Stew Benedict Ok. The new burned cd worked. (this is a live description ;) ) The MacOS see the cdrom, ok: it mounted it on the Desktop Folder. I try to unsit the sit file... no stuffit installed: too minime installtion (180 Mb). Scompatted the sit file from the network and copied to the Desktop the files. Double clicked on the star-iconed file... applescript error (maybe too minime installtion). Let's try directly from BootX app passing to kernel theese arguments: ramdisk_size=32000 automatic=method:cdrom text (pls tell me what to write for using a graphical installation: machine is a standard 8600 250Mhz). Linux is booting... no cdrom device found. Selected mesh scsi driver. Ok the installtion goes on (loading program into memory). Language choice: I select italian... failed to read kmap magic. I select all languages one by one... French works, Ungarian works... and so on. So the problem IS IN the cooker; not in the Mac, not in the burned cd, not in the ISO. Now... I do not know 1 word of french or Hungarian and all yoe told me graphical installation works. Can you tell me the kernel parameters to set in BootX app ??? Tnx Miguel Beccari
Re: failed to read kmap magic (long reading)
Stew Benedict wrote: Linux is booting... no cdrom device found. Selected mesh scsi driver. Ok the installtion goes on (loading program into memory). Language choice: I select italian... failed to read kmap magic. I select all languages one by one... French works, Ungarian works... and so on. I said French and Hungarian worked out but I wanted to say the theese worked, the rest no (english did now work, US englush did not work, Spanish did not work... and so on). I first tried known languages (US, Italian, Spanish), then others: French and Hungarian worked. Up to now I am following the Franch installation (I have a little illusion to understand questions instead of Hungarian;) ) and I can say that Diskdrake fails too. I have repeated the installation 5 times and - after partitioned the disk - the error is: error :(segmentswitch to console F2 for a shell is missed as the install crashed. The system freezes and non logs are avaiables. The last time I switched very quickly in the logs and I saw: segmentation fault: seems like memory is missin as the install crashes Regards, Miguel Beccari
Re: failed to read kmap magic (long reading)
Stew Benedict wrote: On Fri, 22 Feb 2002, Miguel Beccari wrote: Language choice: I select italian... failed to read kmap magic. I select all languages one by one... French works, Ungarian works... and so on. So the problem IS IN the cooker; not in the Mac, not in the burned cd, not in the ISO. Just kicked off a text insatll, using the same ISO as published, on the 7600. Selected Italian|Italy, and the language changed, install proceeded. Switching to VT3, there was a warning on kmap magic for the 1st keymap it attempted to load, then it fell back to another one - no error on the main install screen. I'm unable to duplicate your reported error =(. Stew Benedict OK. Let trash my mails: maybe the problem is my old 8500. I changed the cdrom with a cd-r SCSI and used it. The errors are the same... so maybe it is the hardware in some way fails. I want to try for the last time: can you tell me how to install in graphical mode? I tried and tried but I had got an error like: cannot find /usr/X11R6/bin/RGB.pm. It stays on the screen less then a second, the the monitor turn black and the system freezes. Regrads, Miguel
Installation fails (failed to read kmap magic)
Ok, now I can run installation (I posted Trying to use 8600)... but at the first question (select your language) it stops. Every language I choose, the answer is failed to read kmap magic. Grrr The keyboard is a... how can I define it ? is a QZERTY with no F1, F2, F3... It is supposed italian. Regards, Miguel Beccari
failed to read kmap magic
I changed the keyboard (I used tha same of another linux-mac-g3). So now I have F1, F2, F3... to see logs. The error persist and is failed to read kmap magic at /usr/bin/perl-install keyboard.pm : line 305 That's all. Miguel Beccari
Re: failed to read kmap magic
1) A bad ISO burn 2) Hardware problem with your CD drive 3) Kernel/module functionality for that drive is not quite there Since you can't see the stuff in MacOS either, I'm leaning toward 1 or 2. Stew Benedict Maybe. Let's burn another cd ;) Thank you for help, Miguel Beccari
Re: failed to read kmap magic
Stew Benedict wrote: On Thu, 21 Feb 2002, Miguel Beccari wrote: 1) A bad ISO burn 2) Hardware problem with your CD drive 3) Kernel/module functionality for that drive is not quite there Since you can't see the stuff in MacOS either, I'm leaning toward 1 or 2. Stew Benedict Maybe. Let's burn another cd ;) You checked the md5sum on the ISO, correct? Stew Benedict no md5 was present when I downloaded it (or maybe - and it could be so - I do not consider to check md5). I do now... ok I do later: the server is busy. Miguel Beccari
Re: failed to read kmap magic
Ben Reser wrote: On Thu, Feb 21, 2002 at 04:17:33PM +0100, Miguel Beccari wrote: no md5 was present when I downloaded it (or maybe - and it could be so - I did not consider to check md5). I do now... ok I do later: the server is busy. http://mirror.brain.org/linux/mandrake/Mandrake-iso/ppc/md5sums.8.2.ppc It shouldn't be busy. TNX a lot :) I verified the iso and they are good. bash-2.05$ md5sum -c md5sums.8.2.ppc MandrakeLinux-8.2beta1-CD1.ppc.iso: OK MandrakeLinux-8.2beta1-CD2.ppc.iso: OK So it should be the cdrom (in 10 minutes I will have a new hybrid burned cdrom) Miguel Beccari
Re: BootX???
erik wrote: Help.. I've managed to start the Linux installer with bootX but when I see the install screen, it asks me for a CDROM drive device driver. Gives me a huge list and I can't guess wich driver is the right one for my PPC PPC 8200/100 I also have tried to install via http or ftp, but the same happens with the network interface. It asks me for a driver!!! Please, help... Regards, Erik Reidler Messaggio Originale Oggetto: Re: Trying to use a 8500 Data: Wed, 20 Feb 2002 17:07:44 -0500 (EST) Da: Stew Benedict [EMAIL PROTECTED] Rispondi-A: [EMAIL PROTECTED] A: [EMAIL PROTECTED] On Wed, 20 Feb 2002, Miguel Beccari wrote: I have a 250Mhz 8500 series powerPC. It as got very old 2x cdrom (with caddy) It has got no floppy. It has got a build in ehternet, and a 3com (I think). I am unable to boot from cdrom. No way, no way, no way to boot :( So I copied the BootX on the desktop and launched the BootX app. The installer begins and I have to select from a cdrom installation or a ftp, hard disk etc etc. no modules are right for my old cdrom, my built in ethernet and so on... Not booting from the CD is normal on OldWorld - that's why BootX is provided. I don't have access to any 8500's but I can tell you what modules are used on the 7600: SCSI: mesh Ethernet: mace mac53c94 is also used for SCSI on some older machines. I doubt the integrated NIC is 3com, but I suppose anything is possible. Stew Benedict
Trying to use a 8500
I have a 250Mhz 8500 series powerPC. It as got very old 2x cdrom (with caddy) It has got no floppy. It has got a build in ehternet, and a 3com (I think). I am unable to boot from cdrom. No way, no way, no way to boot :( So I copied the BootX on the desktop and launched the BootX app. The installer begins and I have to select from a cdrom installation or a ftp, hard disk etc etc. no modules are right for my old cdrom, my built in ethernet and so on... How can I procede ? Miguel Beccari
Re: Trying to use a 8500
Not booting from the CD is normal on OldWorld - that's why BootX is provided. I don't have access to any 8500's but I can tell you what modules are used on the 7600: I am old with linux but on x86 only (lilo, grub, award bios etc), so I have to understand some things... for example: should I leave a Mac (with the MacOS) partition to have BottX working ? SCSI: mesh Ethernet: mace mac53c94 is also used for SCSI on some older machines. I doubt the integrated NIC is 3com, but I suppose anything is possible. Theese are useful informations. The last question is: if I put the hybrid burned ppc-8.2 cdrom on my linux box (x86) i can mount and read with no problems all; but if I put the cdrom in the old sony (on the Mac) the system say the cd have to be formatted. Why ? Miguel Beccari
Re: empty (was: kernel upgrade failures (long reading))
Juan Quintela wrote:
miguel Infact I updated from lastest cooker src.rpms the folling:
rpm, libbinutils,
miguel fileutils, findutils, mkinitrd, sgml-tools, ghostscript.
miguel Then I downloaded kernel-2.4.17.13mdk-1-1mdk.src.rpm
miguel I did some few hacks to make it works on mdk ppc and now -
ladies and
miguel gentlemen -...
miguel [root@serverg3 SRPMS]# uname -a
miguel Linux serverg3 2.4.17-MACINTOSH #1 ven feb 1 17:54:50 CET
2002 ppc unknown
Wow!!! Juan Quintela aswers a mail of mine... I really appreciate your
work ok kernel rpms.
If you can send me what things did you changed, I can integrate them
in the Mandrake kernel. Notice that I don't have an easy access to
one ppc machine, I will put here the srpm for people to test (I don't
have an easy access to one ppc machine right now).
Ok.
I experienced theese errors (2) with macros:
1) rpm --rebuild kernel-2.4.17.13mdk-1-1mdk.src.rpm
Installing kernel-2.4.17.13mdk-1-1mdk.src.rpm
error: parse error in expression
error: /usr/src/RPM/SPECS/kernel-2.4.spec:1229: parseExpressionBoolean
returns -1
I commented the spec as follow:
1229 #%if %build_BOOT
1230 #CreateKernel BOOT
1231 #%endif
and then the error was:
rpm -bb ../SPECS/kernel-2.4.spec
error: parse error in expression
error: /usr/src/RPM/SPECS/kernel-2.4.spec:1632: parseExpressionBoolean
returns -1
and I commented the following lines:
1632 %if %build_BOOT
1633 %files -n kernel-BOOT-%{mdkversion}
1634 %defattr(-,root,root)
1635 %{_bootdir}/config-%{KVERREL}BOOT
1636 %{_bootdir}/vmlinuz-%{KVERREL}BOOT
1637 %{_bootdir}/System.map-%{KVERREL}BOOT
1638 %dir %{_modulesdir}/%{KVERREL}BOOT/
1639 %{_modulesdir}/%{KVERREL}BOOT/kernel
1640 %{_modulesdir}/%{KVERREL}BOOT/build
1641 %{_modulesdir}/%{KVERREL}BOOT/pcmcia
1642 %{_modulesdir}/%{KVERREL}BOOT/modules.*
1643 %endif
This happens with
build BOOT 0
and
build BOOT 1
2) Then I experienced a dependency error.
This happens with
build doc 1
The compilation starts and goes on without errors
but when it is near to finish (and the macro begin to tar the *.ps
files) an error occurs and the compilation is aborted.
I cannot remember what program was missed, but it was from
sgml-tools.
The *ps files were not created (because sgml-tools were not installed),
and so the macro failed.
I suppose it is necessary to include a build require: sgml-tools.
Theese are my 'few hacks' to make the src.rpm working.
Best regards,
Miguel Beccari
Re: kernel upgrade failures (long reading)
Anne et Bertrand wrote: Le 2002.02.02 14:24, Miguel Beccari a écrit : Anne et Bertrand wrote: I last recompiled 2.4.17 without xmon and the debugging features. No kernel oops message till now (of course !), but I'm running gnome and it doesn't seem to crash. Wonderful. It is going to be a great day ;) ... Play 2 - 4 - 17 on lotto :) I don't play lotto (or loto, in France), but now I've got a kernel that doesn't use more than 60 Mb of swap when I've got 96Mb of ram, as 2.4.4. did ! Really great ! Thanks, B.D. You said you are using old 603 ppc. So today I went in my old pc repository and I found 3 ppc 603. Yes, they work, and each one has got 96Mb, 500Mb HD, Floppy, but no CDROM. I tried linux on old Pentium 90 with 64Mb RAM without specials pleasures. I tried openBSD too on old PCs... but the reality is that todays applications (as apache+php+mysql) need high CPU performance, high disk access, etc etc. I never tried with ppc, so I would like to put a 6Gb HardDrive (I have to buy it because of old PPC were SCSI only), and test all. In your opinion : is it possible expecting great performance then my tests on Pentiums 90-133Mhz ? Best regards, Miguel Beccari
kernel upgrade failures (long reading)
Hi everybody! I installed Mandrake8.0PPC on a G3 with success. The goal was to change OS9 + IPShare with Linux Mandrake + samba + netatalk. The machine has got a 80Gb SCSI hard drive controlled by 2940U2W Adaptec. The problems: 1) installation do NOT let me choose reiserfs as default filesystem (I choosed the old ext2, which is that i do not want) 2) the kernel 2.4.4 has got no drivers for 2940U2W Adaptec So the goal is so far today. The questions: 1) First of all I tried to install from updates a new kernel (2.4.8). I am experienced with rpms, so I preferred to manually upgrade packages. No error was found in the upgrade, except a /usr/local/boot//mkinitrd not found. No problems, I run mkinitrd manually, take a look at /etc/yaboot.conf, /boot, and /lib/modules, typed /sbin/ybin -v (that is - I suppose - the same I do with lilo -v). At the reboot the error was: Ooop: kernel access of bad area. Sig 11. Kernel Panic (rebooting in 180 secs). 2) Then I downloaded the lastest kernel binaries for ppc from Cooker (2.4.17-8). I upgraded a lot of packages (as glibc, iniscripts, bash, setup, perl, rpm) and then - with no errors - I installed the kernel packages. At the reboot the problem was the same: Ooop: kernel access of bad area. Sig 11. Kernel Panic (rebooting in 180 secs). 3) I am not sure about the use of yaboot. Maybe I fail something. I am experienced with x86 machines and lilo or grub, but PPC is a new world. So I installed from the 8.0PPC the 2.2.19 kernel. No errors of course, but at the reboot a new problem: the screen stops at the word booting... 4) So, up to now I am thinking to upgrade the gcc too, and recompile from src.rpms the very last kernel. but the big question is: What is going wrong with my upgrades ? yaboot ? kernels ? Best Regards, Miguel Beccari
Re: kernel upgrade failures (long reading)
Ben Reser wrote: On Fri, Feb 01, 2002 at 10:04:30AM +0100, Miguel Beccari wrote: 1) installation do NOT let me choose reiserfs as default filesystem (I choosed the old ext2, which is that i do not want) Known issue. Will be fixed in 8.2. Very well. At this point my obvious questio is: when 8.2 ppc is planned to come out ? ;) 2) the kernel 2.4.4 has got no drivers for 2940U2W Adaptec I believe this was an error on Stew's part. I'm not sure if it's corrected in the security update kernel but I'm positive he's fixed the support for 8.2 I see that in 2.4.8-31.3mdk (the last update avaiable) have not got it. But I can see it in 2.4.17-8mdk (so I am planning to compile this one) 1) First of all I tried to install from updates a new kernel (2.4.8). I am experienced with rpms, so I preferred to manually upgrade packages. No error was found in the upgrade, except a /usr/local/boot//mkinitrd not found. This is normal for this update. No issue here. OK No problems, I run mkinitrd manually, take a look at /etc/yaboot.conf, /boot, and /lib/modules, typed /sbin/ybin -v (that is - I suppose - the same I do with lilo -v). Sorta ybin also updates the bootstrap as well as the MBR. So maybe OK. My yaboot.conf is: [root@serverg3 root]# cat /etc/yaboot.conf #yaboot.conf - generated by DrakX init-message=\nBenvenuto in Linux Mandrake PPC!\n boot=/dev/hda7 ofboot=hd:7 delay=30 timeout=50 install=/usr/local/lib/yaboot/yaboot magicboot=/usr/local/lib/yaboot/ofboot enablecdboot enableofboot defaultos=linux default=linux nonvram image=hd:8,/boot/vmlinuz label=linux root=/dev/hda8 read-only image=hd:8,/boot/vmlinuz-2.4.8-31.3mdk label=linux2.4.8 root=/dev/hda8 read-only image=hd:8,/boot/vmlinuz-2.4.17-8mdk label=linux2.4.17 root=/dev/hda8 initrd=hd:8,/boot/initrd-2.4.17-8mdk.img read-only image=hd:8,/boot/vmlinuz-2.2 label=linux22 root=/dev/hda8 initrd=hd:8,/boot/initrd-2.2.img read-only image=hd:8,/boot/vmlinuz label=failsafe root=/dev/hda8 append= failsafe read-only macos=hd:6 My /boot is: [root@serverg3 root]# ls /boot/ config@System.map-2.2.19-14mdksecure config-2.4.17-8mdk System.map-2.2.20-6mdk config-2.4.4-6.2mdkSystem.map-2.4.17-8mdk config-2.4.8-31.3mdk System.map-2.4.4-6.2mdk initrd-2.2.19-14mdksecure.img System.map-2.4.8-31.3mdk initrd-2.2.20-6mdk.img vmlinuz@ initrd-2.2.img@vmlinuz-2.2@ initrd-2.4.17-8mdk.img vmlinuz-2.2.19-14mdksecure* initrd.img@vmlinuz-2.2.20-6mdk* kernel.h@ vmlinuz-2.4.17-8mdk* kernel.h-2.4.4-6.2mdk vmlinuz-2.4.4-6.2mdk* messagevmlinuz-2.4.8-31.3mdk* System.map@ and my modules are: [root@serverg3 root]# ls /lib/modules/ 2.2.19-14mdksecure/ 2.4.17-8mdk/ 2.4.4-6.2mdk.tar.bz2 2.2.20-6mdk/ 2.4.4-6.2mdk/ 2.4.8-31.3mdk/ At the reboot the error was: Ooop: kernel access of bad area. Sig 11. Kernel Panic (rebooting in 180 secs). I've never see this maybe Stew has some thoughts on this. This error appens with 2.4.17-8mdk, 2.4.8-31.3mdk. 3) I am not sure about the use of yaboot. Maybe I fail something. I am experienced with x86 machines and lilo or grub, but PPC is a new world. So I installed from the 8.0PPC the 2.2.19 kernel. No errors of course, but at the reboot a new problem: the screen stops at the word booting... Some machines work with 2.2 and refuse to work 2.4 or vice versa. PPC hardware is well very proprietary and support for it is often broken by proprietary changes. Apple is so great in not giving us info on the hardware. OK. This information is useful: I will not try to use 2.2.19-14mdksecure, 2.2.20-6mdk, or other 2.2 4) So, up to now I am thinking to upgrade the gcc too, and recompile from src.rpms the very last kernel. but the big question is: What is going wrong with my upgrades ? yaboot ? kernels ? I don't see anything that gives me the impression your yaboot is wrong. Thought it would be helpful if you posted your yaboot.conf file here. But it really shounds like a kernel issue. And for that I'll defer to Stew our expert. :) Maybe is a kernel issue, so news from me will come after the 2.4.17 compilation. Best Regards, Miguel Beccari PS: TNX for quick answers ;)
