MBX IDE

There was a thread back in November about getting IDE going on
an MBX,

http://lists.linuxppc.org/listarcs/linuxppc-embedded/199911/msg00054.html

Did anyone make any progress on this?  Meaning, is there a recent
kernel with working IDE-on-MBX.  I know its an oddball combination,
but I just wanted to check.

With a few tweaks to MontaVista 2.2.13, I get as far as seeing,

W82C105: IDE controller on PCI bus 00 dev 19
W82C105: not 100% native mode: will probe irqs later
SL82C105 command word: 5
IDE timing: 00000909, resetting to PIO0 timing
SL82C105 command word: 5
IDE timing: 000003e4, resetting to PIO0 timing

but the drive never gets detected in ide_scan_devices(), indirectly
because all of ide_hwifs[x].present are set to 0 (I'm in the middle
of tracing this back: ide_hwifs[x].noprobe fields set to 1 seem to be
preventing ide_hwfs[x].present getting set).

If anyone has a console dump of a working IDE-on-MBX, I'd like
to see it just to know what to expect.

And should I continue tweaking 2.2.13 or try 2.4.0-pre? (opinions)

Thanks,
-Jamie

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Jamie Guinan wrote:

There was a thread back in November about getting IDE going on
an MBX,

http://lists.linuxppc.org/listarcs/linuxppc-embedded/199911/msg00054.html

Did anyone make any progress on this?  Meaning, is there a recent
kernel with working IDE-on-MBX.  I know its an oddball combination,
but I just wanted to check.

With a few tweaks to MontaVista 2.2.13, I get as far as seeing,

W82C105: IDE controller on PCI bus 00 dev 19
W82C105: not 100% native mode: will probe irqs later
SL82C105 command word: 5
IDE timing: 00000909, resetting to PIO0 timing
SL82C105 command word: 5
IDE timing: 000003e4, resetting to PIO0 timing

I had a similar problem with a MTX board (not MBX), and the final
thing that deblocked the situation was by changing the file
/usr/src/linux/driver/block/sl82c105.c

I simply enabled the winbond by changing the line

  pci_write_pci_config_dword(dev,0x40,0x10ff08a1) to

  pci_write_pci_config_dword(dev,0x40,0x10ff0011)

I am not sure for your case since this line is enclosed between
"#ifndef CONFIG_MBX", maybe you can try put the line outside the
"ifdef" construct?

I had changed the 0x44 register's value to 0x0260:
  pci_write_pci_config_dword(dev,0x44,0x0260)

hope this helps

---------------------------------------------------------------------
Zirong Wang    Oce Industries  1, Rue J. Lemoine 94015 Creteil France

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

On Wed, 28 Jun 2000, Jamie Guinan wrote:

There was a thread back in November about getting IDE going on
an MBX,

http://lists.linuxppc.org/listarcs/linuxppc-embedded/199911/msg00054.html

Did anyone make any progress on this? ...

<snip>

Well, many hours and printk's later,

  hda: 251350AT Maxtor 2.5" Disk Drive (C) 199, 1287MB w/58kB Cache,
  CHS=2616/16/63

I haven't tried formatting it yet, but I managed to write some
raw data on /dev/hda from a Linux program, then read it back in
EPPCBug.  Looks funny, every other 2bytes are 00's, but perhaps
that's expected for doing a blind write() on /dev/hda.

m8xx_setup.c doesn't seem to want both IDE and PCMCIA ATA devices
to exist at the same time, though.  Not sure what to do about that,
but first I have to study interrupts and PCI/ISA busses some
more (I'm a bit rusty on that subject matter).  If I can hack
them to both work at the same time for my own purposes, I'll
be happy.

-Jamie


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Jamie Guinan wrote:

m8xx_setup.c doesn't seem to want both IDE and PCMCIA ATA devices
to exist at the same time, though.  Not sure what to do about that,

It's an interrupt routing issue.  In fact, all of the support
for SouthBridge/8259 controller parts on an 8xx is an interrupt
routing issue.  Mapping the internal interrupts from something
like the PCMCIA into the same space as the PCI/ISA has never
worked in the 8xx port.  You basically get to configure an
"8xx PC" or an "8xx embedded".  Cort and I are constantly working
on some better interrupt mapping scheme, some of it "sticks" in
the kernel sources.  The entire Linux kernel is either "PC" or
not.

I don't know of any production boards that support PCI with ISA
on an 8xx.  You can program the UPM to provide ISA bus timing,
or for a few pennies put a programmable device on the 8xx bus
if you want an IDE device.

.....  If I can hack
them to both work at the same time for my own purposes, I'll
be happy.

You can.....It just won't run on anything else when you are done.


	-- Dan

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/