strange SCSI error while trying to mount a CDrom

[pascaltaf]

---

Well, the bug is not in libata. I recompiled my kernel with libata as a module, loaded ide-core, ide-generic and ide-cd instead of libata and sr_mod, and reproduced the bug.
Then plugging the adapter on another machine with different devices, I didn't reproduce it. So it seems the problem is really bound to the hardware configuration of the machine.
One suggestion is that it comes fropm the fact the adapter is PCI-X and was plugged on a simple PCi bus, while it is now on a PCI-X bus. But PCI-X is ought to work on simple PCI buses, so this explanation is not really satisfying.
I forgot to tell about the BIOS question: the BIOS of the Adaptec card is up to date since there hasn't been new releases since 2004, and the motherboard BIOS was not the very last one but updating it didn't fix the problem.
So... waiting for Adaptec support to give me a new answer, don't think I will bother the lkml with this, and if no solution, my company might requalify the problem and try to use another SCSI adapter, but this is not so easy as Adaptec seems to have a quasi monopol on SCSI adapters.

[Rubberman]

Well, the bug is not in libata. I recompiled my kernel with libata as a module, loaded ide-core, ide-generic and ide-cd instead of libata and sr_mod, and reproduced the bug.
Then plugging the adapter on another machine with different devices, I didn't reproduce it. So it seems the problem is really bound to the hardware configuration of the machine.
One suggestion is that it comes fropm the fact the adapter is PCI-X and was plugged on a simple PCi bus, while it is now on a PCI-X bus. But PCI-X is ought to work on simple PCI buses, so this explanation is not really satisfying.
I forgot to tell about the BIOS question: the BIOS of the Adaptec card is up to date since there hasn't been new releases since 2004, and the motherboard BIOS was not the very last one but updating it didn't fix the problem.
So... waiting for Adaptec support to give me a new answer, don't think I will bother the lkml with this, and if no solution, my company might requalify the problem and try to use another SCSI adapter, but this is not so easy as Adaptec seems to have a quasi monopol on SCSI adapters.

Well, you've done all the logical things to resolve this. I assume that even in a PCI-X slot it still has this problem? What about this machine is different from the others? Is it a different manufacturer, motherboard, ...? The fact that this only happens on this one machine, even with a rebuilt kernel and the board in the correct (PCI-X) slot, tells me that there is some hardware / BIOS thing going on. Sometimes, you just have to change horses!

[Rubberman]

FWIW, besides Adaptec, other manufacturers include HP, Tyan, and Quantum - possibly more, but that's what I came up with in a hurry.

[pascaltaf]

You might have completely forgotten this problem I posted a few weeks ago, I just come back to tell you the last word of the story, at least what will be for me the last word, even if investigation could probably go deeper.
In fact, the problem I had, mount failure on CDrom when booting with an Adatpec 29320ALP SCSI card plugged in, was not caused by the Adaptec card, but by the Cdrom drive.
Plugging the Adaptec card with a completely different hardware around did not reproduce the bug, but plugging it with an identical hardware except the CDrom drive did not reproduce the bug too, and plugging it with the same CDrom drive but a different motherboard reproduced it.
And then, downgrading the firmware of the Cdrom drive made the bug disappear, so this seems clear.
So the last word is : there is an incompatibility somewhere between the Adaptec 29320ALP SCSI card and the Optiarc 5170A CDrom drive with a 1.12 firmware.
Well, now my company will be able to deal with it, thanks for all people who helped me.

[Rubberman]

Ain't that just peachy! Glad you got to the root cause of this. Reminds me of a CD/DVD recorder I had once that could record on a DVD+R just fine, and could play back normal DVD's, but not the DVD+R that it (or any other drive) had recorded. A firmware update fixed that. In your case, you had to backdate the firmware... Go figure!

The nice thing about standards, is that there are so many!