Elusive SATA speed issue [solved]

duozmo · Last edited by duozmo on Fri Jul 09, 2004 12:26 am; edited 1 time in total

My computer can't recognize that my SATA disk can go faster than UDMA2 (33)

I've tried endlessly to eliminate variables:

Problem happens in Linux, not in Windows so I know it's not a motherboard issue

Occurs both in my Gentoo 2.6 kernel and Knoppix 3.4, so I'm pretty sure it's not kernel config

Also happened in my old 2.4 kernel I don't use anymore, more evidence it's not kernel config

I have been loaned a second SATA drive and it happens there too. Original was Seagate 160GB, new is Maxtor 250GB.

What else could it possibly be?

This sucks, because Linux takes 1.5 minutes to load and only 30 seconds for Windows XP. It's not supposed to be that way!

Monkeywrench · Posted: Sat Jul 03, 2004 5:48 am Post subject:

have you checked out the man page for hdparm?

Sorry I can't help much. But try playing around with hdparm.
_________________
Folding@home -Join the Linux team (163)!

duozmo · Posted: Sat Jul 03, 2004 6:02 am Post subject:

Yeah, hdparm is how I know:

hdparm -i lists:

Model=Maxtor 6Y250M0, FwRev=YAR51EW0, SerialNo=Y63BS1LE
Config={ Fixed }
RawCHS=16383/16/63, TrkSize=0, SectSize=0, ECCbytes=4
BuffType=DualPortCache, BuffSize=7936kB, MaxMultSect=16, MultSect=16
CurCHS=16383/16/63, CurSects=16514064, LBA=yes, LBAsects=268435455
IORDY=on/off, tPIO={min:120,w/IORDY:120}, tDMA={min:120,rec:120}
PIO modes: pio0 pio1 pio2 pio3 pio4
DMA modes: mdma0 mdma1 mdma2
UDMA modes: udma0 udma1 udma2
AdvancedPM=yes: disabled (255) WriteCache=enabled
Drive conforms to: (null):

* signifies the current active mode

If I did this on my IDE drive, the udma modes would go up to 5 or 6

jimbojetset · n00b Joined: 01 Feb 2004 Posts: 67 Location: UK

can you show the results for:

() · l33t Joined: 25 Nov 2002 Posts: 610

You're leaving out a crucial variable, the SATA controller. For instance, the Silicon Image controller can be very slow I think; because of a certain quirk the driver has to throttle througput. Are you using the libata driver (under SCSI in the kernel config)?
_________________
to be concerned is good

duozmo · Posted: Sat Jul 03, 2004 2:36 pm Post subject:

Here's hdparm -tT:

joki · n00b Joined: 30 Sep 2003 Posts: 53 Location: germany

your error sounds crazy.
i also have a maxtor sata, so look at my outputs:

() · l33t Joined: 25 Nov 2002 Posts: 610

pojo: Does the chipset support both PATA and SATA? If the drive is indeed connected to a SATA port, it should show up as /dev/sda, not /dev/hda I think. This is the convention followed by libata, which offers SATA support for ICH. Is it perhaps connected to a PATA port? It looks like the regular IDE driver claims the drive, perhaps the support for ICH is better with libata.

I see you've turned on support for ICH under the IDE section btw, maybe this is why you're getting these IDE port busy messages (the IDE driver has already claimed the ports before libata).
_________________
to be concerned is good

duozmo · Posted: Thu Jul 08, 2004 5:58 am Post subject:

(), you were right about my kernel config. It was picking up the SATA drive as an IDE drive, and my bios was set to allow it to do that. I took the necessary parts out of my kernel and recompiled, but now i'm faced with a problem with lilo: it won't accept "root=/dev/sda3" as a parameter when my root is currently on /dev/hda3. Specifically the error I get is

Fatal: Not a number: "/dev/sda3"

If I leave it the way it is, with root="/dev/hda3", I get a kernel panic on startup, after libata grabs the drive, asking for a valid root=. I've done some searching but can't figure out how to get around this...

Monkeywrench · Posted: Thu Jul 08, 2004 8:53 am Post subject:

pojo, I think you can just have "root=/dev/sda" in your lilo.conf. Mine currently reads:

() · l33t Joined: 25 Nov 2002 Posts: 610

Hmm .. I dunno about lilo, but why should the root parameter matter to the bootloader? Isn't the root parameter simply passed to the kernel? At least with grub, it has it's own numeric root (currently hd0,0), and root=/dev/sdb1 is among the parameters passed to the kernel. Is the bios still set up to masquerade the drive as IDE?
_________________
to be concerned is good

duozmo · Posted: Fri Jul 09, 2004 12:26 am Post subject: problem solved

holy crap, it works!

i realized all i had to do was pass a new 'root' option on the boot prompt, not too hard. i was foiled for a while until i completely ripped ide out of my kernel and updated fstab to hit up the sdas instead of the hdas. this will disable my shiny dvd burner, so i'll figure out how to get it back in without it picking up my disks, but the point is it works. finally, my computer doesn't spin the disk endlessly, it is so much faster now. woot

here's the new dmesg:

() · l33t Joined: 25 Nov 2002 Posts: 610

Great :]
_________________
to be concerned is good

FirechilD · Posted: Fri Jul 09, 2004 8:16 pm Post subject:

may you can just post a little how to?

im trying to change from ide siimage to the scsi one and cant come forward since lilo wont accept an update for the sdaX hdd´s because there arent loaded :/

my system looks like:

- silicon image chip on my a7n8x deluxe
- hde8 = root -> sda8 in the future?

the scsi drivers ive compiled in the kernel:

- scsi device support
- legacy /proc/scsi suport
- scsi disk support

scsi low level drivers ->

- sata support
- silicon image support

edit:

ive found the solution myselv

- compiled the kernel with ide and scsi drivers
- lilo update & reboot
- compiled the kernel with only the scsi drivers & reboot
- at lilo boot-prompt ive done this:
Gentoo root=/dev/sda8

booted fine

- needed to adjust the fstab and lilo config & rebooted

everything runs fine now,
bit ive no speed improvements :/

k, im using love-sources - maybe it were the siimage update wich boosted my hdd´s to that power they have

_________________
The worst foe lies within the self...