PROBLEM: Highpoint 374 first port device early blocks nonfunctional

I Am Falling I Am Fading (skuld@anime.net)
Fri, 25 Apr 2003 02:02:24 -0600 (MDT)


(This is my first bug report on the Linux kernel ever, so please bear with
me, it will also be a bit sparse on information as I am not following the
bug report template exactly due to laziness on my part. I am sending this
to the LKML because I can't find a particular maintainer for the hpt3xx
driver)

[1.] Highpoint 374 first port device early blocks nonfunctional

[2.] I have a Epox EP-K9A3+ motherboard with integrated HPT374 controller.
Due to issues with the HPT BIOS a bootloader started from the southbridge
IDE controller cannot see the hpt374-attached devices, therefore I am
placing all of my drives on the HPT controller. I am not using them in any
RAID configuration for purposes of just getting the system working
properly to begin with. Basically I am just using it as another IDE
controller.

Using LILO (as GRUB does not function properly with this controller) I can
boot into Linux successfully provided that I do not utilize the first
port.

While experimenting with the controller I discovered that under Linux,
using the native hpt7xx drivers as of kernel 2.4.21-pre5, I could not
successfully mount the first partition on whichever drive was attached to
port 1 of the controller. (in this case /dev/hde). It reports that the
partition is "not a block device". Partitions AFTER this do work properly,
therefore I can mount /dev/hde3, for instance, but not /dev/hde1.

Moving the drive to a different port (e.g. making it /dev/hdg, /dev/hdi,
etc.) fixes the problem, and I can access the first partition properly,
therefore it is not a problem with the drive, but rather with the driver
(as Win2k can address the first partition on the first port without any
problems).

This leads me to believe that there is either a problem addressing the
early blocks of the device on the first port, or a problem with how the
driver handles its partitioning internally.

[4.] (from memory) 2.4.21-pre5, compiled with 2.95.3, but also observed in
2.4.21-pre4.

[5.] There was no Oops:

[6.]

# Take a drive that is properly partitioned and formatted on another
# port/controller with your filesystem of choice, and place it on
# port 1 of the onboard HPT374 controller, in the case of my
# EP-8K9A3+ it is the bottom port of the four.

mount /dev/hde1 /mnt/myfavoritemountpoint

[7.] This is where I'm getting lazy. As the machine is in pieces at the
moment I can't generate the necessary output here. I wish I could give
more info on this. :-( I can only hope that someone with more clue than I
has similar hardware and can try to replecate the problem. I will briefly
describe the system though.

Epox EP-8K9A3+ mb (KT400-based)
Athlon XP 1800+ Palomino core running at stock rates
2 X Crucial PC2700 512MB
ATI Radeon 7500 AIW
3Com 3c905-series NIC (driver not loaded, using onboard NIC right now)
RealMagic Hollywood Plus (no driver loaded)

The error seems to occur regardless of which modules are loaded and what
state the system is in.

I'm sorry I'm not giving better info. :-( In any case I hope someone can
make use of this information. If anyone is interested in taking up the
problem, please write me directly as I do not subscribe to the LKML
(though I do read through the archives from time to time). Thanks!

-----
James Sellman -- ISU CoE-CS/ISLUG Linux Lab Admin |"Lum, did you just see
----------------------------------------------------| a hentai rabbit flying
skuld@inconnu.isu.edu | // A4000/604e/60 128M| through the air?"
skuld@anime.net | \X/ A500/20 3M | - Miyake Shinobu

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/