FC5->FC6: hard drives VERY slow!?

Thoddy · 11-09-2006, 10:50 AM

I have 3 hard drives in my system, 1 directly on the mainborad (nForce2 chip) and 2 on an IDE raid controller. The one on the mainboard works fine and the other 2 work, but are damn slow! I am 99% sure this was not in FC5, but after upgrading to FC6, this is somehow the case...

My two candidates are:

SAMSUNG HD300LD
Maxtor 6L300R0

Both are 300GB drives and rather new. They should support UDMA33 / UDMA 66 - also depending on the cables, I know. But this is not the prblem!

It seems that UDMA is deactivated at boot time. Why is this?

Quote:

$ dmesg | grep DMA

DMA zone: 4096 pages, LIFO batch:0
NFORCE2: 0000:00:09.0 (rev a2) UDMA133 controller
ide0: BM-DMA at 0xf000-0xf007, BIOS settings: hda

MA, hdb

MA
ide1: BM-DMA at 0xf008-0xf00f, BIOS settings: hdc

MA, hdd

MA
ide2: BM-DMA at 0xa000-0xa007, BIOS settings: hde

MA, hdf

io
ide3: BM-DMA at 0xa008-0xa00f, BIOS settings: hdg

MA, hdh

io
hda: 80418240 sectors (41174 MB) w/1821KiB Cache, CHS=16383/255/63, UDMA(100)
hde: 586072368 sectors (300069 MB) w/8192KiB Cache, CHS=36481/255/63, BUG DMA OFF
hdg: 586072368 sectors (300069 MB) w/16384KiB Cache, CHS=36481/255/63, BUG DMA OFF
hdc: ATAPI 40X DVD-ROM DVD-R-RAM CD-R/RW drive, 2048kB Cache, UDMA(33)

$ hdparm -v /dev/hde

/dev/hde:
multcount = 0 (off)
IO_support = 0 (default 16-bit)
unmaskirq = 0 (off)
using_dma = 0 (off)
keepsettings = 0 (off)
readonly = 0 (off)
readahead = 256 (on)
geometry = 36481/255/63, sectors = 586072368, start = 0

$ hdparm -v /dev/hdg

/dev/hdg:
multcount = 0 (off)
IO_support = 0 (default 16-bit)
unmaskirq = 0 (off)
using_dma = 0 (off)
keepsettings = 0 (off)
readonly = 0 (off)
readahead = 256 (on)
geometry = 36481/255/63, sectors = 586072368, start = 0

$ hdparm -i /dev/hde

/dev/hde:

Model=SAMSUNG HD300LD, FwRev=WK100-12, SerialNo=S0A4J10L385695
Config={ Fixed }
RawCHS=16383/16/63, TrkSize=34902, SectSize=554, ECCbytes=4
BuffType=DualPortCache, BuffSize=8192kB, MaxMultSect=16, MultSect=off
CurCHS=16383/16/63, CurSects=16514064, LBA=yes, LBAsects=268435455
IORDY=on/off
PIO modes: pio0 pio1 pio2
DMA modes: mdma0 mdma1 mdma2
AdvancedPM=no
Drive conforms to: unknown: ATA/ATAPI-1 ATA/ATAPI-2 ATA/ATAPI-3 ATA/ATAPI-4 ATA/ATAPI-5 ATA/ATAPI-6 ATA/ATAPI-7

$ hdparm -i /dev/hdg

/dev/hdg:

Model=Maxtor 6L300R0, FwRev=BAH41E00, SerialNo=L60NGV7H
Config={ Fixed }
RawCHS=16383/16/63, TrkSize=0, SectSize=0, ECCbytes=57
BuffType=DualPortCache, BuffSize=16384kB, MaxMultSect=16, MultSect=off
CurCHS=16383/16/63, CurSects=16514064, LBA=yes, LBAsects=268435455
IORDY=on/off
PIO modes: pio0 pio1 pio2
DMA modes: mdma0 mdma1 mdma2
AdvancedPM=no
Drive conforms to: ATA/ATAPI-7 T13 1532D revision 0: ATA/ATAPI-1 ATA/ATAPI-2 ATA/ATAPI-3 ATA/ATAPI-4 ATA/ATAPI-5 ATA/ATAPI-6 ATA/ATAPI-7

Any ideas??

glussier · 11-09-2006, 12:47 PM

Have you tried manually setting the dma modes?

For example (udma100 mode 5)

hdparm -X69 /dev/hde
hdparm -X69 /dev/hdg

Thoddy · 11-09-2006, 01:15 PM

Yes - but is doesn't help...

Quote:

# hdparm -X69 /dev/hdg

/dev/hdg:
setting xfermode to 69 (UltraDMA mode5)

# hdparm -t /dev/hdg

/dev/hdg:
Timing buffered disk reads: 6 MB in 3.28 seconds = 1.83 MB/sec

...the 'hdparm -i /dev/hdg' also still shows the old (from my first post) reply.

glussier · 11-09-2006, 08:16 PM

What is the brand and model number of the raid controller? Are you talking about a pci card raid controller?

Thoddy · 11-10-2006, 01:59 AM

Quote:

Originally Posted by glussier

What is the brand and model number of the raid controller? Are you talking about a pci card raid controller?

Yes, it's a PCI card. Actually I forgot the brand and name of the card (I'll try to recall somehow)... but this is what my system tells me:

Vendor: Integrated Technology Express, Inc.
Product: IT/ITE8212 Dual channel ATA RAID controller
Bus: PCI
Driver: ITE821x IDE

Just found the manual

No brand name, just a "ATA-133 IDE RAID Card ITE8212F (Version 1.0)"

Thoddy · 11-11-2006, 08:04 AM

OK, it seems that the kernel versions newer than 2.6.10 do not directly support the IT8212 chip ide raid controllers!?

There also seems to be a patch for the kernel sources by Alan Cox. But does anyone know a way to only compile this one module? That is: is there a linux driver for the ITE8212 ide card that is supported by the kernel version 2.6.18 (as shipped with FC6)?

What I still do not understand is - why did it work in FC5? FC5 had a kernel newer than 2.6.10, didn't it?

Thoddy · 11-13-2006, 12:44 PM

Hey, I have the module compiled and now it seems the controller is detected correctly!

But still: my drives run in slow PIO mode. And I still receive the "BUG DMA OFF" message (in dmesg).

So, I know I can set specific attributes with hdparm, but what do I have to set?

Quote:

hdparm -d1 /dev/hdx
hdparm -X69 /dev/hdx

This seems to be what I want - UDMA 5 (100)!?

But the settings are gone after booting. Where do I have to put them (rc.local ?) to be set at boot time?

If I set them after booting - do I have to umount the drives before setting (and mount again afterwards)?

A lot of questions, I know...

glussier · 11-13-2006, 01:03 PM

hdparm -X69 /dev/hdx should be the only thing you have to do, and yes, you can insert this line into your /etc/rc.local file

Thoddy · 11-13-2006, 03:22 PM

Quote:

Originally Posted by glussier

hdparm -X69 /dev/hdx should be the only thing you have to do, and yes, you can insert this line into your /etc/rc.local file

Thanks, glussier!

BUT:
Loading the module does not work!? I load it with 'modprobe it821x' and do not get an error but it is not loaded?

Quote:

# modprobe -v it821x
insmod /lib/modules/2.6.18-1.2798.fc6/kernel/drivers/ide/pci/it821x.ko
# lsmod | grep it821x
#

As with the 'modprobe -v' you can see the module is there... but it does not get loaded! What I did is compile the module with the 2.6.18 sources and copy it into the directory of my kernel, and afterwards did a 'depmod -a'. Did I miss something?

Also, I still get this, which is not good:

Quote:

# hdparm -v /dev/hde

/dev/hde:
multcount = 0 (off)
IO_support = 0 (default 16-bit)
unmaskirq = 0 (off)
using_dma = 0 (off)
keepsettings = 0 (off)
readonly = 0 (off)
readahead = 256 (on)
geometry = 36481/255/63, sectors = 586072368, start = 0

OH!
I just wanted to post the outcome of my 'dmesg', there I saw that it seems to be a problem with my modules:

Quote:

# dmesg | grep it8
it821x: controller in smart mode.
kobject_add failed for it821x with -EEXIST, don't try to register things with the same name in the same directory.

I am starting to think about putting a different ide controller into my machine!!

glussier · 11-13-2006, 03:46 PM

I think that your problem now, is that the it821x is compile statically in the kernel.

You could try:

/sbin/modprobe -r it821x
/sbin/modprobe it821x

but I doubt that this would work. You might have to recompile the kernel without the it821x support.

Thoddy · 11-13-2006, 04:20 PM

Quote:

Originally Posted by glussier

I think that your problem now, is that the it821x is compile statically in the kernel.

Thanks for your patience!!

Hm, when it is statically in the kernel - then I wouldn't need the module, right? Then the system should recognize the controller right away...

But:

Quote:

# dmesg | grep 821
IT8212: IDE controller at PCI slot 0000:01:08.0
IT8212: chipset revision 19
it821x: controller in smart mode.
IT8212: 100% native mode on irq 177

And:

Quote:

# dmesg | grep DMA
DMA zone: 4096 pages, LIFO batch:0
NFORCE2: 0000:00:09.0 (rev a2) UDMA133 controller
ide0: BM-DMA at 0xf000-0xf007, BIOS settings: hda

MA, hdb

MA
ide1: BM-DMA at 0xf008-0xf00f, BIOS settings: hdc

MA, hdd

MA
ide2: BM-DMA at 0xa000-0xa007, BIOS settings: hde

MA, hdf

io
ide3: BM-DMA at 0xa008-0xa00f, BIOS settings: hdg

MA, hdh

io
hda: 80418240 sectors (41174 MB) w/1821KiB Cache, CHS=16383/255/63, UDMA(100)
hde: 586072368 sectors (300069 MB) w/8192KiB Cache, CHS=36481/255/63, BUG DMA OFF
hdg: 586072368 sectors (300069 MB) w/16384KiB Cache, CHS=36481/255/63, BUG DMA OFF
hdc: ATAPI 40X DVD-ROM DVD-R-RAM CD-R/RW drive, 2048kB Cache, UDMA(33)

What does this "BUG DMA OFF" mean?

So - I am more than ever before questioning myself if I want to continue this way or "simply" plug in another controller...

glussier · 11-13-2006, 06:47 PM

The BUG DMA BUG could be a problem with the controller card bios. I would buy another controller card, if I was you.

Thoddy · 11-14-2006, 02:22 AM

Quote:

Originally Posted by glussier

The BUG DMA BUG could be a problem with the controller card bios. I would buy another controller card, if I was you.

Thnks a lot!!

I think that's what I'm going to do...

Thoddy · 11-15-2006, 09:41 AM

What ever the problem with the ITE8212F is - I installed on old and also rather cheap Promise controller and that one works right away!

Quote:

# dmesg | grep DMA
<...>
hda: 80418240 sectors (41174 MB) w/1821KiB Cache, CHS=16383/255/63, UDMA(100)
hde: 586072368 sectors (300069 MB) w/8192KiB Cache, CHS=36481/255/63, UDMA(100)
hdg: 586072368 sectors (300069 MB) w/16384KiB Cache, CHS=36481/255/63, UDMA(100)
hdc: ATAPI 40X DVD-ROM DVD-R-RAM CD-R/RW drive, 2048kB Cache, UDMA(33)

# hdparm -v /dev/hdg

/dev/hdg:
multcount = 16 (on)
IO_support = 0 (default 16-bit)
unmaskirq = 0 (off)
using_dma = 1 (on)
keepsettings = 0 (off)
readonly = 0 (off)
readahead = 256 (on)
geometry = 36481/255/63, sectors = 586072368, start = 0

hdparm -t /dev/hdg

/dev/hdg:
Timing buffered disk reads: 184 MB in 3.01 seconds = 61.20 MB/sec

That's better than 1.87 MB/sec, isn't it?

glussier · 11-15-2006, 10:22 AM

You must be loading apps and files a little faster using the promise controller.