Very Computer

Does Linux support the "Hot Rod 66" IDE UDMA PCI controller (ABIT
HA66 chipset)?

I tried with kernel 2.2.13. /proc/pci reports:

  Bus  0, device   8, function  0:
    Unknown mass storage controller: Triones Technologies, Inc. Unknown device (rev 1).
      Vendor id=1103. Device id=4.
      Medium devsel.  IRQ 11.  Master Capable.  Latency=248.  Min Gnt=8.Max Lat=8.
      I/O at 0x6200 [0x6201].
      I/O at 0x6300 [0x6301].
      I/O at 0x6400 [0x6401].
  Bus  0, device   8, function  1:
    Unknown mass storage controller: Triones Technologies, Inc. Unknown device (rev 1).
      Vendor id=1103. Device id=4.
      Medium devsel.  IRQ 11.  Master Capable.  Latency=248.  Min Gnt=8.Max Lat=8.
      I/O at 0x6500 [0x6501].
      I/O at 0x6600 [0x6601].
      I/O at 0x6700 [0x6701].

The harddrives on the controller are recognized by the BIOS, and I
can boot the kernel from them, but Linux doesn't seem to be able to
access them.

Is there any config option (or a kernel patch or something) to use
this controller?

Frank

--

PGP and GPG keys: http://fjf.gnu.de/plan

I run mine in compatibilty mode, linux just thinks its an ide. I needed to
specify the port and irqs as kernel parameters IDE2= etc

Look here:

  http://www.csie.ntu.edu.tw/~b6506063/hpt366/

-- Chuan-kai Lin

A bad idea.  I have tried that once, but I soon realized that HPT366
only runs in PIO mode without native driver, and the performance sucks.

-- Chuan-kai Lin

Thanks. With this infromation, I now got Linux to recognize the
contoller and the harddrive (an IBM-DJNA-351520), but it still
doesn't work as I'd like it to...

To explain my problem: I've got a motherboard with an onboard IDE
controller, but it has the VIA VP2 chipset -- the only chipset the
kernel config help warns about using CONFIG_IDEDMA_AUTO on. And
indeed, when I enable DMA on that controller, heavy disk access
makes the system hang.

OTOH, I want to use my sound blaster, and heavy disk operation
without DMA enabled interrupts the sound card's DMA. This is
annoying when playing sound, and unacceptable when recording.

First, I thought that unmasking the IRQ (hdparm -u1) would help, but
it didn't change anything.

So, I thought that an add-on UDMA controller would be the
solution...

Well, without using hdparm, it seems to work mostly the same as the
onboard controller, interrupting the sound (also with -u1).

I tried -X66 -d1 and -X68 -d1 (UDMA modes 2 and 4 -- in the
controller's BIOS, UDMA 4 is selected). Both cause the DMA flag to
be reset after a short time. So I added -k1, but this made the
process that accesses the HD hang (at least not the whole system
this time ;-), while the syslog was showing messages like the
following:

Dec 14 03:51:47 goedel kernel: hde: status error: status=0x58 { DriveReady SeekComplete DataRequest }
Dec 14 03:51:47 goedel kernel: hde: drive not ready for command
Dec 14 03:51:47 goedel kernel: ide2: reset: success
Dec 14 03:51:57 goedel kernel: hde: timeout waiting for DMA
Dec 14 03:51:57 goedel kernel: hde: irq timeout: status=0x58 { DriveReady SeekComplete DataRequest }
Dec 14 03:51:57 goedel kernel: ide2: reset: success

Am I doing something wrong, or is there anything else I could try?

Or is the problem of the VIA VP2 in the DMA controller rather than
the onboard IDE controller, so I can't get reliable DMA operation
with this motherboard at all? In that case, is there a way to kludge
the kernel to make a short pause after a certain amount of IDE
access, in order to give the sound card time to do its transfers?

Somewhat confused,
Frank

--

PGP and GPG keys:           http://fjf.gnu.de/plan
Pascal code, BP CRT bugfix: http://fjf.gnu.de/programs.html
Free GNU Pascal Compiler:   http://home.pages.de/~GNU-Pascal/

Are you now using the I/O address and IRQ hack, or a patched kernel with
native Linux drivers in it?  HPT366 cannot do UDMA without the native
Linux kernel driver.  If the VIA IDE controller is causing you trouble,
then you might disabling it completely (from BIOS, and remove its driver
from the kernel).

Or have you already done so?  It is not clear in your message.

-- Chuan-kai Lin

The latter.

Quote:> HPT366 cannot do UDMA without the native
> Linux kernel driver.  If the VIA IDE controller is causing you trouble,
> then you might disabling it completely (from BIOS, and remove its driver
> from the kernel).

> Or have you already done so?  It is not clear in your message.

I tried this now (disconnected the other HDs from the onboard
controller, disabled it in the BIOS and removed the VIA driver from
the kernel), but it didn't make any difference.

So, could it be that the VIA DMA controller, rather than the IDE
controller, is at fault?

Frank

--

PGP and GPG keys:           http://fjf.gnu.de/plan
Pascal code, BP CRT bugfix: http://fjf.gnu.de/programs.html
Free GNU Pascal Compiler:   http://home.pages.de/~GNU-Pascal/

I don't know, but that is a possibility.

-- Chuan-kai Lin