[linux] Problem with big big hard drive
lists at dansanderson.com
Sun May 25 14:36:17 PDT 2003
And to close the thread: Upgrading to 2.4.20 worked. There's a kernel
module for the Promise cards, and everything else is automatic. I now
have an ide2, an hde, and a nice big 177GB partition called hde1.
Thanks for bearing with the messages. It helps to type these out
Dan Sanderson wrote:
> Just to preserve my reputation for answering my own damned questions,
> here's confirmation that Promise 20269 is in 2.4.20, and not in 2.4.18:
> So I guess I'll try that.
> Comments still welcome...
> -- Dan
> Dan Sanderson wrote:
>> Addendum (no I didn't fix it 5 minutes after I posted this time :) :
>> The Matrox ATA/133 PCI card (a Promise card) is interesting: on boot
>> up, after the BIOS found all the drives for the on-board IDE
>> controller, the new card displayed its own screen where it detected
>> the new drive as a primary on its IDE1 (where I put it). Everything
>> else started up fine. However, Linux didn't automatically set up hde
>> as I was hoping it would.
>> /proc/ide does not contain anything new, only ide0, ide1 (the
>> on-board controllers), hda (ide0/hda) (my boot drive), hdc (ide1/hdc)
>> (an ATAPI Zip drive), hdd (ide1/hdd) (my DVD-ROM drive), and the
>> drivers file ("ide-disk version 1.10"). If everything worked
>> automatically, I would have expected an ide2 and ide3, as well as hde
>> (ide2/hde). No such luck.
>> /proc/pci shows the card just fine:
>> Bus 0, device 13, function 0:
>> Unknown mass storage controller: Promise Technology, Inc. 20269
>> (rev 2).
>> IRQ 9.
>> Master Capable. Latency=64. Min Gnt=4.Max Lat=18.
>> I/O at 0xd800 [0xd807].
>> I/O at 0xdc00 [0xdc03].
>> I/O at 0xe000 [0xe007].
>> I/O at 0xe400 [0xe403].
>> I/O at 0xe800 [0xe80f].
>> Non-prefetchable 32 bit memory at 0xe0000000 [0xe0003fff].
>> More searches about Linux and the Promise U133 show posts from people
>> that have had success using kernel 2.4.19 (I'm at 2.4.18), though no
>> explicit mention that 2.4.18 is required for this (just that everyone
>> seems to be using it). The kernel changelog mentions Promise 20276
>> was added with 2.4.19, which, given the model number difference
>> (20269 mentioned in /proc/pci), isn't enough to convince me to break
>> from Debian stable to upgrade to 2.4.20 (though I suppose it should
>> be considering 2.4.20 is considered the "production" kernel).
>> Someone seems to think that if the kernel is compiled with ide-mod as
>> a module, then /etc/modules.conf could have options ide-mod
>> options="ide2=ata66 ide3=ata66" (or to test, modprobe ide-mod
>> options="ide2=ata66 ide3=ata66"). Someone correct me if I'm wrong,
>> but I gather that the built-into-the-kernel equivalent is putting
>> append="ide2=ata66 ide3=ata66" in lilo.conf (and running lilo to
>> install). No change.
>> Other mentions imply that Promise 20269 support is in the 2.5
>> kernel. Nothing conclusive, but I'm not willing to go 2.5 just for
>> this anyway.
>> Anyhoo, I'd rather settle for EIDE speed, or get a new motherboard,
>> if either of those would give me access to the 200GB drive. Anyone
>> have any similar experiences they can relate?
>> -- Dan
>> Dan Sanderson wrote:
>>> I recently got a brand new Maxtor 7200 RPM 200GB hard drive (Maxtor
>>> 6Y200PO). This is the Maxtor Ultra Series 8MB cache, Ultra
>>> ATA/133. My BIOS detects it fine as the secondary slave, and Linux
>>> assigns it /dev/hdd as appropriate. But any attempt to fdisk
>>> /dev/hdd results in "Unable to open /dev/hdd," and /var/log/messages
>>> to log "hdd: driver not present." I'm using kernel 2.4.18-bf2.4
>>> (Debian 3.0). Does anyone have any ideas on what "driver not
>>> present" means in this context? Anyone know what I might be missing?
>>> The drive came with an Ultra ATA/133 PCI card, which I haven't tried
>>> yet and originally assumed was optional. This has given me the
>>> impression that UDMA is rather new, at least relative to my
>>> motherboard (please correct me if I'm wrong). So I'm learning about
>>> Ultra DMA a little after the fact. The Linux UDMA Mini-Howto ()
>>> says UDMA should be backwards compatible with EIDE (without the UDMA
>>> bus speed), and I shouldn't have to do anything special to get it to
>>> work. I don't know much about my motherboard except that it's at
>>> least a few years old and reports itself as "Via Apollo Pro PCISet"
>>> on start-up (which I assume close to this:
>>> http://www.via.com.tw/en/apollo/PRO.jsp). The mini-howto mentions a
>>> possibly old problem with "TX Pro and other 'Pro' boards," but I
>>> don't know enough to know if that applies here, and by now I'm
>>> probably way off track.
>>> I haven't tried using the PCI card yet, but the mini-howto implies
>>> there's a chance that it might just work, assigning drives to
>>> hde-hdh. I'm not optimistic; the card comes with "Windows-only"
>>> drivers, which reeks of being too new to work with Linux. I've been
>>> tempted to get a new motherboard and processor for a while now, so
>>> maybe this is a good excuse?
>>> Any other thoughts? Anyone got a favorite ATX motherboard they want
>>> to plug? (I'm terrible with motherboards. Lately it seems like the
>>> motherboard/CPU purchasing decision and compatibility issues have
>>> gotten even more complicated, but maybe it's just me.)
>>> Thanks, all!
>>> -- Dan
>>>  UDMA mini-howto:
More information about the Linux