Red Hat Bugzilla – Bug 42070
Promise Ultra66 IDE Controller causes install to hang
Last modified: 2007-04-18 12:33:23 EDT
From Bugzilla Helper:
User-Agent: Mozilla/4.77 [en] (Win98; U)
Description of problem:
Installing off the bootable RH7.1 CD, when my Promise Ultra66 (PDC20262) is installed in the computer, the install hangs during
initial hardware detection.
Steps to Reproduce:
1. Boot off Red Hat 7.1 install CD with Promise Ultra66 card installed in computer with Quantum Fireball Plus KA (18.2 GB) drive in hde
position (IDE2 master)
2. Choose normal install (hit enter), choose text install, choose expert install- it doesn't seem to matter!
Actual Results: Install consistently hangs after minimal initial hardware detection. It does correctly identify the installed hard drives and CD
drives on ide0 and ide1; it even correctly identifies the Promise Ultra66 (PDC20262) on ide2 and ide3- with the correct memory addresses and
IRQ, as well as the Quantum Fireball drive installed as master on ide2. After identifying these components, it gives a full description
of hda (cylinders, heads, sectors) and then hangs.
Expected Results: Install should not hang.
I can get the install to proceed correctly if I remove the Promise Ultra66 card from its PCI slot. The install proceeds perfectly onto my hda hard
drive; I can then re-insert the Promise Ultra66 card and boot into the freshly installed (on hda) OS. Logging in as root I can then mount the
/dev/hde hard drive (on the Ultra66 controller) and everything works fine- it just seems to be a problem with the
boot CD or installation process.
The install also hangs at the same place if I create a boot floppy to start the
install from (and boot
This is strange, because we have a Promise Ultra66 in one of our test machines
and it works fine...
Hi!! I also am having this problem except I have two hard drives one is Maxtor
9 2049H2 (1st hard drive c,d in Windows) and Quantum Fireballp KX IDE (2nd hd
or e, also have Promise Ultra 66 IDE Controller. It detects the hard drives as
hde and hdf then hangs, and I don't even get to installation screen, also tried
expert text mode and some problem. Should I try what you did by taking out the
controller card?? or should I wait for a fix????
I have a Promise Ultra 100 card on a AZZA 633/AS Motherboard with the VIA
Chipset. The Drive was detected correctly identified by the install program as
hde (The IDE controllers on the motherboard are hda, hdb, hdc, hdd) therefore
hde is the 1st drive on the promise card. The install itself went without a
hitch, but upon bootup the system hung after identifying the controllers on the
motherboard (correctly. It never got to the "OK" checks and there were no
error messages -- it just stopped.
What if you try booting with 'linux ide=nodma'? Does that help at all?
Tried it, it didn't work. Still stops after hardware detection. I was
thinking of trying to copy cd's to 2nd hard drive in windows partition then use
swgipson, you can use expert mode without having to try a hard drive install.
That shouldn't change anything.
starsmann, I would recommend filing a bug against the kernel for the issue that
you are seeing. If the install completes, but then the kernel hangs on reboot,
then it's a kernel problem.
Question! When you took out the controller card did you change the BIOS?? Then
When you put back the Controller card did you reset the BIOS?? I'm thinking of
trying this and just want to make sure I do it right. I Have the first part of
my 2nd hard drive partitioned and formatted for Linux so it would be /dev/hdf??
All I did was unplug the IDE cable from the Ultra66 card leading to my hard
drive. At boot, this caused the Ultra66 card to say "no drives detected-
Ultra66 BIOS not installed", and then I could install 7.1 without a single
hitch. The Ultra66 card was still in the computer, but there were no hard
drives physically attached to it, so its BIOS was not installed. I then
proceeded to install onto /dev/hdb (the slave device on ide0, which is handled
by the motherboard). I suspect if you have any drives attached to your Ultra66
(i.e. hdf) that you will continue to have problems. But I did not alter my
normal BIOS in any way- just unplugged the IDE cable from the Ultra66 card to my
fifth hard drive (hde).
I was able to install after unplugging the controller card, I followed
kfrender's instructions, when I went to plug back in the controller card it
hangs before I can get to ok's, If I leave the controller card unpluged I have
no problem booting into Linux. Let me know if there is a fix in future plans,
without the controller card I am limited to only 2 hard drives which is ok for
I didn't alter BIOS either.
This is sounds like a driver problem with the Ultra66 card. Like I said before,
though, this works in our test machine.
My windows drivers are up to date. Is there different drivers for linux?? If
so can you point me in the right direction to find out more?
Sharon the linux newbie
We released a new kernel 2.4.3-12 last friday, with a new Promise IDE driver.
Also, I'm writing a driver for the fasttrak raid, but that isn't 100% stable
Could you please try the 2.4.3-12 kernel and see if that tolerates the
I have the same problem with an onboard Promise Ultra66. On install (7.1-2.4.2
kernel) it hung at hardware detection:
ide1 at somememoryaddr on someirq
ide2 at somememoryaddr on someirq
it hangs here...
I was able to complete the install with the appended parameter 'ide2=noautotune'
I then upgraded to kernel 2.4.3-12 and it now hangs even with the kernel
Interestingly, it now lists ide0, ide1,hang instead of ide1, ide2.
Tyan mainboard Tiger 200 S2505 - Dual P-III with a PROMISE Fast Trak 100
builtin. BIOS V1.02 020801 (The latest available).
Machine has 4 40Gig Maxtor drives on the array controller and 1 80 Gig Maxtor
on the standard IDE controller.
The machine was using all drives including the RAID array when NT 4.0 or W2K
was installed. Array was set up as a RAID 1 with 80 Gig on the array. The
hardware works, and the BIOS settings don't allow for configuring the Promise
controller in any way. During boot, the controller allows for array creation
Install of RH7.1 produces messages ending in
PDC20267 Neither IDE port enabled (BIOS)
Install occurs on the 80Gig with no problems.
Boot of machine shows that the PDC20267 is being detected but the same message
indicates that the system won't use the array.
Fasttrak raid is currently not supported because Promise so far has refused
to give information on how to write drivers for the array. I'm writing a
reverse engineered driver but that takes a bit more time...
If you get to the point where you want to test the fasttrak driver on the
version of their controller that gets imbedded on a main board, let me know.
Since I've got an SMP setup, I can also run tests for that environment.
Thanks for the update
For those of you with the Promise Ultra66 problem, there is a way to install
without having to open your machine. Boot using the RH install disk. When
the "Choose type of install" screen comes up (the first screen asking if you
want graphic, text, expert installation) press CTRL-ALT-F2 to get to a BASH
prompt. At the prompt, type: cat proc/pci. Use SHIFT+PAGE UP to scroll up to
where the information is about the Promise card. There should be a list of I/O
that looks something like: I/O 0x10c0 [0x10c1]
I/O 0x10b4 [0x10b5]
etc. Copy down the first 2 numbers EXACTLY as they appear (including spaces,
brackets, etc). Reboot your installation again. At the boot: prompt at the
first screen, type: linux ide2=0x10c0 [0x10c1],0x10b6 [0x10b7] and then press
enter. The installation will start and it should recognize your hard disks.
There is a HowTo about this at http://www.europe.redhat.com/documentation/mini-
HOWTO/Ultra-DMA-5.php3. A couple notes: You must add two to the second number
(s). If you notice the nubers I had from the I/O printout, b4 and b5 become b6
and b7 when you enter it at the boot: prompt. I don't know why, but that is
the way it has to be. Also, you have to boot up in text mode when you finally
get RH installed and enter in that line again before booting into RH. I have
not figured out how to get around that. There is a patch for the kernel, but I
haven't figured out how to use it yet. I think there might be a way to write
that line into lilo or the kernel itself, but I don't know how to do that.
Anyone that knows, please let me know. I am new to Linux, so any help would be
appreciated. If you have any questions, please feel free to contact me by
email. Hope this helps!
Should be resolved with modern errata kernels. If not please file a new bug
giving version, IDE information and an lspci -v