My disk support UDMA up to level 5, but my current config shows only udma2 active (with hdparm -i /dev/hda which shows udma modes up to 5 but has the asterix next to 2, which I'm interpreting level 2 is active).

I wonder if this even a red hat specific question, but came up - I'm running redhat 9.0 with 2.4.20-8 kernel on a a7n8x motherboard with nvidia's nforce drivers set up (the nvidia onboard nic is up, too)...

Hi,

First a warning: This can be dangerous, it may course some drives to mess up!

Use the -X option in hdparm
Usage:
-X33 -> Multiword DMA
-X66 -> ATA-33 (UDMA 2)
-X68 -> ATA-66 (UDMA 4)
-X69 -> ATA-100 (UDMA 5)

Go single-user (init 1) and run 'hdparm -X69 /dev/hdx'
If it works (!!!) go multi-user again (init 2).

Peace!
Carsten B. Jensen

You can also use hdparm -Xudma5 /dev/hdx for udma5.Just mentioning it because I just found out a while ago that you can set the udma modes direcretly with that command.

hdparm didn't do the trick. the feedback was "setting drive XX to mode 69" or something else, signifying udma5. But it didn't do anything, really. hdparm -t still gave the same results, and the -i option reports udma2 as active.

may be my nForce2 chipset... but if I remember correctly, AMD Viper was enabled in kernel config (nforce uses amd viper for IDE). Perhaps changing the speed of the IDE bus is needed? again, anyove have an idea how this is done w/o recompiling the whole kernel.

I got the same chipset with AMD Viper as driver,too and that works w/o any trouble.What do get for hdparm -T /dev/hdx ?
Also what kernel version do you use? If it shows udma5 as available and the drive info shows it either in hdparm it should work,though.

First of all, I did install the nForce drivers, and I don't know for sure if they are used...

hdparm -t gives time 2.30 seconds -> 27,36 Mbps

I'm running a 2.4.20-8 kernel and red hat 9.0 (downloaded from redhat about 2 months ago - I've only now begun to get to the inner workings of the system.

oh, and if you got an idea how to get an asus P-IN-100-ST-D PCI ISDNLink to work under the system, great!

You'll need to check with hdparm -T - that gives you the speed of the buffer reads.If you get something in the neighborhood of 800 Mbsec I wouldn't worry about the udma thing because the drive won't be faster than the bus anyway.It's probably a kernel thing - those are all patched to some extend.With a gentoo kernel I can get up to udma6 - not that it would change drive perfomance because the bus is much faster than the drive can read and write anyway.
What kind of drive do you got? I get about 30 Mbsec with a Samsung 5200 rpm.
Can't help with ISDN - never used it;sorry.

Edit:My statement about the udma modes is BS.Checked with a drive and it does 28 Mbsec in udma2 mode and 45 Mbsec in udma5 mode.

It could be something in the BIOS... Have you checked it?

I'm getting 28 Mbps from my 80G Samsung drive at udma2. The drive supports udma6.

While on the subject, how about bus frequency... nForce2 chipset should have a fast ide bus on it... I'm running the mainboard at 166 MHz, though, because of my memory. That does mean a processor fsb reading of 266 MHz, which is s'posed to be just right for my XP 2400+

I'm looking at BIOS upgrade, downloading it about now... that could help, at least some have reported better ide performance with updated bios.

thought i might bump this thread as i am currently having exactly the same problem, and have yet to find a working solution...

i am on a 3200+ Barton core in an Abit NF7-S v2.0

I have dl the nforce drivers, and they seem to be working. however, no matter what i have tried, i cannot seem to get my drives to work in anything other than udma2

-----SNIPPED FROM MY DMESG-----

Uniform Multi-Platform E-IDE driver Revision: 7.00beta-2.4
ide: Assuming 33MHz system bus speed for PIO modes; override with idebus=xx
NFORCE2: IDE controller at PCI slot 00:09.0
NFORCE2: chipset revision 162
NFORCE2: not 100% native mode: will probe irqs later
ide0: BM-DMA at 0xf000-0xf007, BIOS settings: hda:DMA, hdb:DMA
ide1: BM-DMA at 0xf008-0xf00f, BIOS settings: hdc:DMA, hdd:DMA
hda: WDC WD400BB-00DEA0, ATA DISK drive
hdb: MAXTOR 6L060J3, ATA DISK drive
blk: queue c03be900, I/O limit 4095Mb (mask 0xffffffff)
blk: queue c03bea44, I/O limit 4095Mb (mask 0xffffffff)
hdc: CRW6206A, ATAPI CD/DVD-ROM drive
hdd: LTN382, ATAPI CD/DVD-ROM drive
ide0 at 0x1f0-0x1f7,0x3f6 on irq 14
ide1 at 0x170-0x177,0x376 on irq 15
hda: host protected area => 1
hda: 78165360 sectors (40021 MB) w/2048KiB Cache, CHS=4865/255/63, UDMA(33)
hdb: host protected area => 1
hdb: 117266688 sectors (60041 MB) w/1819KiB Cache, CHS=7299/255/63, UDMA(33)

-----SNIPPED FROM MY DMESG------

The last couple of lines are cause for concern for me. I cannot get my drives recognized at anything faster than this.

hdparm -X udma5 /dev/hdb results in a warning that speed 3/4/5 are not supported (when in fact they are on the disks themselves).

In fact, in an earlier post I had made to an unrelated problem, i noticed that my same drives were being recognized at UDMA(100).

ANYONE with any thoughts on fixing this problem would earn my infinite thanks and praise. I have gone so far as to try to recompile my kernel, ensuring that CONFIG_BLK_AMD74XX was configured yes, as well as NFORCE. Still no luck :(

I dont know if this may help, but I was poking around /proc/ide and i found this:

[pld@localhost ide]$ ls
drivers hda hdb hdc hdd ide0 ide1 nforce
[pld@localhost ide]$ cat nforce

nVidia 0065 Chipset.
--------------- Primary Channel ---------------- Secondary Channel -------------
enabled enabled
--------------- drive0 --------- drive1 -------- drive0 ---------- drive1 ------
DMA enabled: yes yes yes yes
UDMA
DMA
PIO


I am wondering if perhaps UDMA should be labelled yes?

It should look somewhat like this:

--------------- Primary Channel ---------------- Secondary Channel -------------
enabled enabled
--------------- drive0 --------- drive1 -------- drive0 ---------- drive1 ------
DMA enabled: yes no yes no
UDMA enabled: yes no yes no
UDMA enabled: 5 X 4 X
UDMA
DMA
PIO

But this isn't from a nforce board - I'll have to fire up the old desktop and check it out there.

Do you have an older drive on the bus, if so it will run at the fastest speed that they both support. I just redid my bus assinments so that my fast hdd is the only ony on the bus. Before I did this, I was geting ~30MB/s now I'm geting:
164 MB in 3.02 seconds = 54.37 MB/sec

I have checked all my drives and rechecked them. Both of them are at least UDMA(100), and actually reported correctly on my old system. This is definitely an nforce driver issue on the mobo...

Ok now this is what it's supposed to look like:

bash-2.05b$ cat /proc/ide/amd74xx
----------AMD BusMastering IDE Configuration----------------
Driver Version: 2.9
South Bridge: PCI device 10de:0065 (nVidia Corporation)
Revision: IDE 0xa2
Highest DMA rate: UDMA100
BM-DMA base: 0xf000
PCI clock: 33.3MHz
-----------------------Primary IDE-------Secondary IDE------
Prefetch Buffer: yes yes
Post Write Buffer: yes yes
Enabled: yes yes
Simplex only: no no
Cable Type: 80w 40w
-------------------drive0----drive1----drive2----drive3-----
Transfer Mode: UDMA UDMA DMA DMA
Address Setup: 30ns 30ns 30ns 90ns
Cmd Active: 90ns 90ns 90ns 90ns
Cmd Recovery: 30ns 30ns 30ns 30ns
Data Active: 90ns 90ns 90ns 330ns
Data Recovery: 30ns 30ns 30ns 270ns
Cycle Time: 20ns 20ns 120ns 600ns
Transfer Rate: 99.9MB/s 99.9MB/s 16.6MB/s 3.3MB/s

Check with /var/log/kern.log for something like this:

Feb 9 08:55:13 localhost kernel: NFORCE2: IDE controller on PCI bus 00 dev 48
Feb 9 08:55:13 localhost kernel: NFORCE2: chipset revision 162
Feb 9 08:55:13 localhost kernel: NFORCE2: not 100%% native mode: will probe irqs later
Feb 9 08:55:13 localhost kernel: AMD_IDE: Bios didn't set cable bits corectly. Enabling workaround.
Feb 9 08:55:13 localhost kernel: AMD_IDE: PCI device 10de:0065 (nVidia Corporation) (rev a2) UDMA100 controller on pci00:09.0
Feb 9 08:55:13 localhost kernel: ide0: BM-DMA at 0xf000-0xf007, BIOS settings: hdaMA, hdbMA
Feb 9 08:55:13 localhost kernel: ide1: BM-DMA at 0xf008-0xf00f, BIOS settings: hdcMA, hddMA

Thanks Crash, i'll check it out when i am at home for lunch today and get back to you. from what i can tell so far researching this, though, the problem is that the cable bits are not getting set correctly (hence your line saying:
Feb 9 08:55:13 localhost kernel: AMD_IDE: Bios didn't set cable bits corectly. Enabling workaround.

I do not have this message that I can remember, but i'll check to be sure later.

attempted to compile and install 2.6.2 kernel to see if i could get my drives up to speed. had no problems, auto-detected just fine (just an entire host of other things decided they didnt want to work with the new kernel...