Hi,

I have been encountering a problem with the installation of Debian 4.0. In brief, after I install from the CD and reboot, the system appears unable to read the sata hard-drive and the boot process stalls, displaying onscreen a series of lines saying "hda: lost interrupt". I have googled the problem but I have not found anything particularly relevant.

Here are more details...
I have a home-made x86 computer with an intel core 2 duo chip and a Gigabyte GA 965P DQ6 motherboard. The harddrive is a Seagate SATA 80GB.
Using the Debian 4.0r0 full-installation CD 1, I partition the drive and install. However, once the installation is complete and we reboot the boot process stalls and onscreen, the above mentioned "hda: lost interrupt" is repeatedly shown (I would like to give a verbatim output, but don't know how to do that given that it never completes the boot process to allow me to do a dmesg or something).

That's the main story, but there are some extra pieces of information that are relevant. For example, I can successfully install Ubuntu 7.04 from CD on this computer. Also, I can make the debian newly installed system boot up properly if I go into my bios and, in the integrated peripheral section, disable the "Onboard SATA/IDE Device" setting. This, however, has the consequence that my cd/dvd drive is now not usable, and not even seen by the system. Other relevant info: I have tried with the Debian 4.0r1 installation CDs (the netinst and the full install versions) and the cd itself does not even boot up to the point where I can begin the install process. The errors are identical to the above, i.e. "lost interrupt" lines.

So I am a bit puzzled, but I presume it has a lot to do with the SATA drivers that debian uses (ata_piix?) and some issues particular to this motherboard.

Does anybody have any ideas about how to deal with this? I am going to try out lenny to see if I can get anywhere.

-mark

just as an update to my previous post, I tried installing debian with a "lenny" cd, and there were no problems. It booted both the installation cd itself (something that was not working for me with debian 4.0r1 cd) and installed the system from the cd. Upon reboot, it booted the newly installed system with kernel 2.6.18-4-686. So that is good, and I appear to have a working debian system.

Incidentally, something that I might also mention is that during all my recent attempts with debian installations, the installation cd gets its base system from the debs on the cd itself. The process, I presume, is that it first copies the debs from the cd, and then installs them onto the newly partitioned drive. However, what has always happened is that after the debs are copied over the process stops. I am in a curses window saying "Select and Install software" with a progress-bar that indicates a 5% completion of the task at the point that it stops. It reports "retrieving file 557 of 557" indicating that it has got all the debs from the cd (the cd drive goes quiet too). What I do at this point is ctrl+alt+f2 into another console and explicitly kill the "dpkg-preconfigure" process. This does the trick and the installation process back at console 1 proceeds. Does anyone else experience this weirdness?

-mark

Not sure how much I can help, but:

The 5% part of the message seems to be quite specious. I get that and as far as I can tell its 5% irrespective of how many debs it has decided to install. (And if you decide to use the tasksel mechanism to add extra packages, it can force the installer to install fewer debs, which initially seems a bit bizarre. But that still shows up as 5%.)

At this point the system should seem to stop and my initial rather feeble suggestion is that you haven't been waiting long enough and the apps are installed, but not configured.

You mention the potential issue of it being a problem with that particular board - my first reaction was that would be unlikely with a mainstream chipset. However, what I was overlooking was that some of the SATA ports are provided by the Intel Southbridge and some by Gigabyte's own chip (actually it is probably not Gigabyte's own, but something like a Jmicron chip, marked with a Gigabyte part number and Logo and it provides a couple of SATA ports and a PATA one). This would lead me to suspect that the Intel-supplied ports are coped with natively/automagically but that you might need a separate driver for the additional ones.

-first suggestion: only use the ports supplied by the Intel chip (at least for any drive that you need in order to boot)
-second suggestion: see if you can get a driver for the Gigabyte chip/whatever the chip actually is before Gigabyte have it re-marked

The manual ought to be clear which ports come from the ICH8R Southbridge and which come from the add-on chip, although maybe not in the terms you expect. It is possible that the SATA ports are in two groups for raid arrays and that's the only way you can deduce what is going on. Equally, the ports from one source are probably grouped physically together and you need to use the group without the PATA port.

Unfortunately, it is also possible that if you need the PATA port (even today there are few SATA Optical Drives, so what Intel were doing, way back then in dropping all PATA ports from the basic chipset is beyond me), you have to follow the more complicated path of getting the add-on chip working.

Good luck!

thanks for the help salasi
with respect to the sata problems, I think you are right that the separate intel and gigabyte sata ports act differently. I played around with using either one for my drive. Using Debian 4.0r1 install cd, I could boot and install if I use the gigabyte port, but after installation the new system would not boot. On the other hand, with the regular sata ports on the mobo, I could not even use the installation cd to install the system. It just couldn't read the drive, it seemed. From this, it appears that there is definitely something funky happening with the drivers. Perhaps the drivers are conflicting or something. The real trouble is that with a non-booting system, it is (at least for me) very difficult to do proper analysis of the problem and to overcome it.

luckily - very luckily indeed - all of these problems did not occur when I went to the lenny debian distro. everything went just fine, and the system is able to deal without problems with the sata drives. thank linux-god for that.

with respect the other problem I mentioned, that of the install pausing after reporting 5% completion of the "select and install software" step, I think that that pause will just stayed paused indefinitely unless intervention is taken. In one of my attempts, when it hit the 5% wall, I decided to be patient and just see what happened. I went out to a two-hour meeting and came back to see no progress whatsover. Going in to the second console and terminating the dpkg-preconfigure job immediately does the trick, however, with no apparent adverse consequences. It is a strange one though - and it has happened to me with every Debian install cd since the etch was released as the stable version.

thanks again
mark

Two hours should definitely do it; 10-15 minutes seems like the outer limit of a plausible wait time, so you've waited more than long enough.

I suspect that the two sets of ports do something like share an interrupt (or the interrupt from the Gig one is cascaded through the Intel Southbridge). And that means you can use one or the other but not both, until the drivers are in a better state (originally I suspected that you just were not loading the 'supplementary' driver for the Gig one, but your most recent comments make that sound unlikely). I've got to admit that at this point I'm into more guesswork and deduction than actual knowledge.

(Note that using 'one or the other, but not both' probably implies that you don't use the PATA port if you are using any of the Intel SATA ports. Is your optical drive PATA or SATA?)

The base 965 chipset is so widely used that there would have been howls of complaint if that couldn't be made to work somehow, so I've got to feel that it is something to do with the add-on Gigabyte chip. Maybe its prescence is fooling the hardware detection routines into doing something wrong in loading the '965 drivers?

Wish I could help more, but you'll need someone who knows more of the internals if you are to progess this further (or luck, conceivably).

And finally,

I just wanted to add that these problems I am experiencing with the drivers for the cdrom appear to come and go. I have a few machines of identical specs, and sometimes the system will boot just fine, and other times it stalls during the boot up phase with the "hda: lost interrupt" message being repeatedly shown. If I either unplug the ide cable to the cdrom drive, or disable the ide drives in the bios, the system will always boot.

In other words, debian and this motherboard appear to have intractable relationship problems. I will have to get to the bottom of it eventually because even though I don't use the cd/dvd drive that often I will eventually need it. Hopefully, if I figure out a solution I will post back, just in case it is of help to anyone in a similar situation.

-mark

You might take a look on my notes for this problem, first I had a ASUS P5W Deluxe motherboard, and then a Gigabyte 965P-DQ6 (Rev: 1.0) .. Both are notated in my Google Notebook, which you can view publicly here:

http://www.google.com/notebook/publi...BmQgoQse_vuZYi

I hope you find a solution, I am going to install Linux to my computer in the next weeks or two, then I will come back with a solution if possible. Looks like it's fixed with a kernel update like it says here:

http://www.fedoraforum.org/forum/sho...9&postcount=15

Apparently, some guy called Andi Kleen from SuSE is working on a solution, sounds really great!

Just had an "aaaaaaah!" moment. It sounds similar to what i am dealing with here.
I have an msi board with 3 pata ports and 4 sata ports. I dont use the sata ports, but the pata ports have always been a pain as i find that if i boot the system up cold, i get my drives assigned one way, yet if i do a reboot, they are assigned another way. and i suffer mostly after a kernel upgrade.

I have made changes to fstab (generally gets me running again) and to /boot/grub/menu.lst (which generally helps me get things permanent after a kernel upgrade.).

Im guessing this is very annoying for you as your not sure if the system is going to boot each time.

I find that once i can get past the bad fsck message, i can mod fstab and then mount my boot directory and the modify /boot/grub/menu.lst, reboot and im away again.

The thing i notice most of all is that my boot drive is actual hd(0.0) and grub will always try and modify things so that its hd(2,0).
Once this is changed and groot is set to hd(0,0), things stablise and start working mostly hassle free.

Hope this helps you track down the problem.

DJ

thank you djcs and dezza for the advice and notes.

I will continue continue playing with the problem to try to work it out. Currently, my sata drives are working fine, but I can only guarantee a successful system boot if I disable pata from the bios. That means no cd/dvd drive and no pata hard disks. Obviously, that it unacceptable, so I have to find a remedy.

thanks
mark