On Monday 29 October 2001 12:43 am, you wrote:

On Sun, Oct 28, PCHintz wrote:

...

Just installed 7.3 and thought I would share my thoughts. SuSE in my opinion continues to be the "Mercedes Benz of Linux" as I saw them referred to in an article once. Each version is more polished and 7.3 seems to be no exception. However when it comes to ease of installation, I feel SuSE has taken a step back. Having seen other listings lately about trouble with the installation suggests I am not alone.

My install went as normal until the first reboot when the system locked up.

Resuming the install, the system again locked up when came time to configure

...

X. I then postponed configuring X until I had the install complete. The system crashed again at the end of the install. Attempting to use SaX2 to configure X afterwards crashed the system repeatedly. Replaced the Voodoo5 5500 with an ATI RADEON with the same result...repeated crashes. I

This sounds like the usual problems with a broken BIOS. Please add "disableapic" as boot option. Then the failsafe option should not longer be necessary.

Thorsten

That worked. Thanks PCHintz

On Wed, Nov 07, 2001 at 11:26:57AM +0000, Fergus Wilde wrote:

I've just started installing 7.3 on new hardware I was obliged to use when old hardware broke - the ordinary install hangs, making me use the failsafe one. I've never had any probs with the SuSE install before, and I'm sure they will be overcome, but since I've yet to commit the installation and while I'm mulling on this can I ask:

a) What's this apic, and why does in need disabling?

APIC, mmm. It is a special form of interrupt handling for SMP boards. In new kernels there is an option to use it on single processor boards as well, although I am ignorant of what it's purpose is there. I believe it has problems with older motherboards, and is sensitive to the physical placement of PCI cards in a system. On most modern boards you should be able to disable it in the BIOS. Disabling it in the kernel configuration seems not to stop it from being recognised by Linux, at least from my reading of the boot messages. Quite what is going on ? God alone knows. SuSE will doubtless blame it on a broken bios. The existance of the failsafe boot option, without much (any) explanation is pathetic.

b) Was I being hopelessly optimistic getting a p4 on the new, fancy intel Winnipeg chipset? (I started worrying when I noticed the chipset name contained the syllable 'win' ... It's just that a p3 with half the spec on an older motherboard was about the same price.)

c) If the bios is 'broken', is there a bios setting I could alter rather than having to give the 'disableapic' option?

Well, mine is an ASUS m/board with a PIII processor, and I can disable it, so I am sure you have that option as well. It is more likely that the implementation is the problem, not the hardware..but you may like to check the web for BIOS updates.

d)When I get the install done, will I need to keep this disableapic line somewhere in lilo.conf, and if so where?

You need an "append" line in the definition for the kernel image, something like: image = /boot/vmlinuz label = Linux root = /dev/sda5 initrd = /boot/initrd append = "disableapic" Note that the failsafe option in the Suse setup has the following line: append = "disableapic ide=nodma apm=off" Apm is also sometimes problematic. Turning off IDE DMA will drastically reduce your hard disk performance. Only ancient disks/controllers should really need this. -- Regards Cliff

Thank you to Michael for this answer - in the case of the board / processor combination I'm using, p4 with intel 845 chipset (mboard model D845WN), presumably it's the newness, not the out-of-dateness of the kit that is making 7.3 install choke w/out the failsafe methods. I certainly don't fancy running without DMA, that would seem to lose too much in terms of speed. Because the 845 chipset is liable to be fairly common (the idea apparently being to provide a p4 chipset that runs well with 'ordinary' pc133 SDRAM), is it perhaps worth SuSE having a word with Intel to see if the issue can be fixed. I do understand the point that time and resources are limited, and not every hardware permutation can be allowed for, but as this is a new product maybe Intel will be interested in providing / documenting a fix? Best to all Fergus Michael Hasenstein wrote:

Cliff Sarginson wrote:

...

... blame it on a broken bios. The existance of the failsafe boot option, without much (any) explanation is pathetic.

It's not, it's a very reasonable thing to do. We can't be blamed for all the things going on in the force field between bioses, boards, buses, and kernel development, for the most part we have to live with what there is and this is a good way to do just that.

...

...

d)When I get the install done, will I need to keep this disableapic line somewhere in lilo.conf, and if so where?

No, the suse installer makes sure these options are there if this is what you used for installation (I'm 98% sure - I also just use it).

-- To unsubscribe send e-mail to suse-linux-e-unsubscribe@suse.com For additional commands send e-mail to suse-linux-e-help@suse.com Also check the FAQ at http://www.suse.com/support/faq and the archives at http://lists.suse.com

On Thu, Nov 08, 2001 at 08:14:23AM -0500, zentara wrote:

Hi, I don't quite follow what the problem with APIC is; but I can tell you my experience with turning it off.

I compiled a kernel without it, and experienced all sorts of random lockups and error messages, such as "spurious interrupt on irq7".

I put it back into the kernel, and the problems stopped.

...

From the help for APIC in make menuconfig, " the local APIC supports CPU-generated self-interrupts, (such as timers and performance counters) and the NMI watchdog which detects hard lockups.

I have a duron processor.

Mmm. I think the summary of all of this is: - Some people have problems with APIC - Some people have problems without it - Some people find APM doesn't work with it - Some people found soft-halts didn't work with APIC and APM - Some people found soft-halts didn't work in any combination - Some people found it all worked fine in more recent kernels So that's clear isn't it. And it's all the fault of the BIOS of course. Strange that the default kernel on installation comes with it on... Still if you don't mind your disks being reduced to a snail's pace you can always choose the fail-safe option. Anyway, it is all explained in the SuSE Reference Manual on page..errm.. page...strange must be missing from my manual copy. -- Regards Cliff

On Thursday 08 November 2001 10:47, Malte Christensen wrote:

Cliff Sarginson wrote:

...

On Thu, Nov 08, 2001 at 08:14:23AM -0500, zentara wrote:

...

Hi, I don't quite follow what the problem with APIC is; but I can tell you my experience with turning it off.

I compiled a kernel without it, and experienced all sorts of random lockups and error messages, such as "spurious interrupt on irq7".

I put it back into the kernel, and the problems stopped.

...

From the help for APIC in make menuconfig, " the local

APIC supports CPU-generated self-interrupts, (such as timers and performance counters) and the NMI watchdog which detects hard lockups.

I have a duron processor.

Mmm. I think the summary of all of this is:

- Some people have problems with APIC - Some people have problems without it - Some people find APM doesn't work with it - Some people found soft-halts didn't work with APIC and APM - Some people found soft-halts didn't work in any combination - Some people found it all worked fine in more recent kernels

So that's clear isn't it. And it's all the fault of the BIOS of course. Strange that the default kernel on installation comes with it on... Still if you don't mind your disks being reduced to a snail's pace you can always choose the fail-safe option.

Anyway, it is all explained in the SuSE Reference Manual on page..errm.. page...strange must be missing from my manual copy.

FWIW, I use nosmp disableapic noapic as my LILO boot args on a Dell Dimension 8100. Without all that garbage, SuSE won't even install (P4 1.4Ghz processor). At least 7.2 won't. I UPGRADED to 7.3, but am going to fdisk and start over on two machines tonight. SuSE 7.3 is a nightmare of instability and useless stuff that doesn't seem to work (KDE and WMware are good examples), and I'm going to rip it off of my machines and go back to 7.2. I do realize, of course, that some of my problems stem from the fact that I use a Compaq Armada laptop (I'll accept no flames. I have to use it). Alas, my machines are now as unstable as they were with Windows ME. At least with Windows ME I got to use all the fancy buttons on the keyboards ;-) Try to get BIOS updates for Linux from most major laptop suppliers. They usually go: 'Linux? Yeah, I heard of it, I thought Bill Gates owned the brand.'

Next time I won't be the first to install a new distribution.

Cheers,

Malte

I heard that folks were having problems with the 'upgrade' option. I knew that I would do a clean install later so I tried the upgrade option ("about everyting") and when it booted I went to root and tested out KDE. It seemed perfect. Expecting the best I logged out of root and into my user account. Total disaster. No menu options would work, save the logout option. I couldn't create a new desktop icon... none of the mimes worked. It was useless. With fear and trepidation I took the clean install route, reformatted my HDs and put ReiserFS on them, and then took the 'about everyting' options. The install was perfect! Everthing was recognized perfectly: the BJC-620 connected to the back of the Iomega Zip250 which was plugged into the LPT1:, the 4832 PlexStor CDRW, the monitor, the wheel mouse, the SBLive! card... For the past week this 1GHz Athlon with 512MB RAM and two 60GB drives has been humming along sweetly. :) My only disappointment so far is that Sword 1.5.2 won't compile, and neither will 1.5.1a, which compiled on this box when it was running 7.2. Something to play with :) I couldn't be more pleased. Some folks have done an upgrade and got excellent results. Some have done clean installs and failed. Too many variables. I'm amazed that any OS can install such a wide variety of boxes with so many different peripherals and BIOS chips. Jerry

Hi, Just a quicky... I grabbed SuSE 7.3 Pro on Saturday and still havent got around to upgrading my 7.2 box. I dont want to 'til I know whats happening. I'm going to shove all data on tape and fresh install it as I have too much cruft on this box at the mo. Anyone found anything REALLY dodgy with the 2.4.10 kernel yet as I heard that there were huge problems with it (Mr Torvalds dodgy branch syndrome). I assume SuSE's normally excellent Quality Control / testing has ironed most of that out of the custom kernel??? ... I just read the power down thing you see! Also, anyone got any stability and performance information comparing ReiserFS, JFS and ext3? Thanks, - Chris. ----------------------------------------------------------- This mail sent through the IMP demo site: demo.horde.org Find out more at http://www.horde.org/imp/

Hi,

...

I'm going to shove all data on tape and fresh install it as I have too much cruft on this box at the mo.

Probably wise :)

I've fallen down that hole once ... that was one time too many ;)

I wouldn't say there's anything really dodgy. I've had problems with APM (trivial), with 3D graphics (hardly critical - and probably just because my card isn't up to it!) and sound (which is annoying, but again not system-threatening, though I have had lock-ups with some alsa mixers under Gnome).

Can you detail the 3D graphics problems at all, and do they involve the NVIDIA 3rd party rpm patches? I'm using a Nvidia GF2u and would like it to work fine (although I cant remember the last time I touched Unreal Tournament / Tribes 2 :P ). Do you know which version of ALSA ships with 7.3? 0.5.? was shipped with 7.2 whereas 0.9.x is current version. Seemed a bit strange to me. Could be the root of the lockups as 2.4.10 is way ahead version-wise and time-wise compared to ALSA. My sound card locks up all the time, but it does that on win2k and NetBSD/OSS drivers too. It appears to be VIA KT266 + Envy24 = PCI latency problem.

...

I chose ext3 for the root fs, and haven't seen any problems so far. When I had to reset to get out of the mixer lockup, the partition fsck'd fine. Performance seems normal - the hdparm test reports the same sort of figures (though that might be filsystem independent?)

Ok. as long as it "feels solid" to a few people I'll probably use it. I'll stress the box for a few days just to see. I was using ReiserFS on 7.2 without a hitch. Thanks for the info, - Chris. ----------------------------------------------------------- This mail sent through the IMP demo site: demo.horde.org Find out more at http://www.horde.org/imp/

...

Also, anyone got any stability and performance information comparing ReiserFS, JFS and ext3?

I chose ext3 for the root fs, and haven't seen any problems so far. When I had to reset to get out of the mixer lockup, the partition fsck'd fine. Performance seems normal - the hdparm test reports the same sort of figures (though that might be filsystem independent?)

If you want to test filesystems you have to use bonnie, not hdparm. It comes with the distribution (at least in 7.1, I dont have 7.3 yet). Praise

On Mon, Oct 29, 2001 at 09:38:36AM +0000, chris@dejection.org.uk wrote:

Anyone found anything REALLY dodgy with the 2.4.10 kernel yet as I heard that there were huge problems with it (Mr Torvalds dodgy branch syndrome). I assume

Where did you hear this ? The only problem I have come across in 2.4.10 and .12 is that APM seems to be broken. Perhaps you are referring to 2.4.11, which had a serious bug that particularly affected SuSE distros (or would have done). LT released 2.4.12 in less than 2 days after 2.4.11 because of it and all 2.4.11 kernels are marked as "DONTUSE" at kernel.org. -- Regards Cliff