2.4 madness!

2.4 madness!

[Tony Mantler]

Ok, so in an effort to exploit cheap-ass IDE HDs, I'm playing around with a
plain vanilla Promise ATA100 card in a Pmac 7600 (which currently runs
2.2.15 just fine).

I first tried compiling a 2.2.17 kernel with the latest ide patch, which
resulted in a machine check on startup. bummer.

Next off I tried a linus-2.4.0-test8 (cross compiled from my x86 box this
time, since the pmac hasn't the HD space - exactly the problem I'm trying
to fix), which blew up too. This was copied down by a friend (I'm trying to
do this all from across the city):

Welcome to Linux..  Yada yada.
Started at: 0x00000000
Linked at:  0xc0000000
Frame Buffer at: 0x94300210 (phys), 0x94c00210 (log)
klimit:     0x00000042
HID0:       0x8000c084

booting...
pmac_init(): exit
id mach(): done
MMU: enter
MMU: hash init
hash: enter
hash: find piece

and that's it.

Next I rsynced over top of this kernel with
rsync://ppc.samba.org/linux-pmac-devel from late-ish last night, and had my
friend hook up a serial cable, and got this:

<4>On node 0 totalpages: 20480
zone(0): 20480 pages.
zone(1): 0 pages.
zone(2): 0 pages.
Kernel command line: root=/dev/sdb7 console=ttyS1,9600 video=ofonly
System has 32 possible interrupts
GMT Delta read from XPRAM: -300 minutes, DST: on
via_calibrate_decr: ticks per jiffy = 100001 (600009 ticks)
Console: colour dummy device 80x25
<4>Total memory = 80MB; using 512kB for hash table (at c0300000)
Linux version 2.4.0-test9 (nicoya@ubb) (gcc version 2.97 20001003
(experimental)) #5 Wed Oct 4 10:41:11 CDT 2000
PCI bus 0 controlled by bandit at f2000000
Cache coherency enabled for bandit/PSX at fdee9000
PCI bus 1 controlled by chaos at f0000000

over and over and over and over and over.

Should I be expecting 2.4 to work, or is it still not quite there yet?


Cheers - Tony 'Nicoya' Mantler :)


--
Tony "Nicoya" Mantler - Renaissance Nerd Extraordinaire - nicoya@apia.dhs.org
Winnipeg, Manitoba, Canada           --           http://nicoya.feline.pp.se/

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

Re: 2.4 madness!

[Michel Lanners]

Hi Tony,

On   4 Oct, this message from Tony Mantler echoed through cyberspace:
> Ok, so in an effort to exploit cheap-ass IDE HDs, I'm playing around with a
> plain vanilla Promise ATA100 card in a Pmac 7600 (which currently runs
> 2.2.15 just fine).

Been there, done that, for a Promise Ultra66 ;-)

> I first tried compiling a 2.2.17 kernel with the latest ide patch, which
> resulted in a machine check on startup. bummer.

That could have been PCI problems....

Anyway, I'd suggest starting with:

1. Andre Hedrick's ATA patches (but I suppose you knew that), available
   somewhere on kernel.org, users/hedrick. AFAIK, the 2.2 patches
   _don't_ contain support for the U/100, though...

2. Get my PCI patches, available here:

   http://www.cpu.lu/~mlan/linux/dev/pci.html

...and cross fingers ;-)

Good luck

Michel

-------------------------------------------------------------------------
Michel Lanners                 |  " Read Philosophy.  Study Art.
23, Rue Paul Henkes            |    Ask Questions.  Make Mistakes.
L-1710 Luxembourg              |
email   mlan@cpu.lu            |
http://www.cpu.lu/~mlan        |                     Learn Always. "

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

Re: 2.4 madness!

[Benjamin Herrenschmidt]

>over and over and over and over and over.
>
>Should I be expecting 2.4 to work, or is it still not quite there yet?

It should work (except for some known 2.4 VM bugs that are not PPC specific).

Can you check if it also crash without the Promise card ?


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

Re: 2.4 madness!

[Eric Lowe]

Hello,

> > I first tried compiling a 2.2.17 kernel with the latest ide patch, which
> > resulted in a machine check on startup. bummer.
>
> That could have been PCI problems....
>
> Anyway, I'd suggest starting with:
>
> 1. Andre Hedrick's ATA patches (but I suppose you knew that), available
>    somewhere on kernel.org, users/hedrick. AFAIK, the 2.2 patches
>    _don't_ contain support for the U/100, though...
>
> 2. Get my PCI patches, available here:
>
>    http://www.cpu.lu/~mlan/linux/dev/pci.html
>
> ...and cross fingers ;-)
>

Hmm.. I'll give your patches a try.  I tried a 2.4 kernel on an MVME 2700
here yesterday and it had some PCI problems where 2.2.17 works..

--
Eric Lowe
FibreChannel Software Engineer, Systran Corporation
elowe@systran.com


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

Re: 2.4 madness!

[Tony Mantler]

At 12:57 AM -0500 10/5/2000, Michel Lanners wrote:
[...]
>> I first tried compiling a 2.2.17 kernel with the latest ide patch, which
>> resulted in a machine check on startup. bummer.
>
>That could have been PCI problems....
>
>Anyway, I'd suggest starting with:
>
>1. Andre Hedrick's ATA patches (but I suppose you knew that), available
>   somewhere on kernel.org, users/hedrick. AFAIK, the 2.2 patches
>   _don't_ contain support for the U/100, though...
>
>2. Get my PCI patches, available here:
>
>   http://www.cpu.lu/~mlan/linux/dev/pci.html
>
>...and cross fingers ;-)

Ok, done that. When I compile it here with gcc 2.97 20001003, it blows up
very early. When I compile it on the pmac itself with gcc 2.95.2 20000220
(debian), it blows up (I quote my friend) like so:

>Got past the first screen and Tux is in the corner now...
>
>Except, it goes...
>Macintosh ADB mouse driver intalled...
>Macintosh non-volatile memory driver v1.0
>Unable to handle kernel NULL pointer dereference at virtual address
>00000530 (error 40000000)
>NIP: C0130784 XER: 2000000 LR: C013072C REGS: c03d3d60 TRAP: 0300
>MSR: 00009032 [EEIRDRME]
>TASK = c03d2000[1] 'swapper' mm->pgd c0194000 Last syscall: 1200
>
>And soforth.. :-)

I had him remove the nvram driver, and it does the same. From a look at the
system map (which I neglected to examine before telling him to remove the
nvram driver), it's dying in ad1848_detect. I'm not sure why he wants that
driver compiled in, but he says he needs it. whatever.

I think I'll go grab the 2.95.2 source and compile myself another cross
compiler [insert picture of me projectile-vomiting on the gcc source], then
switch back to 2.4 again.

Hrm...


Cheers - Tony 'Nicoya' Mantler :)


--
Tony "Nicoya" Mantler - Renaissance Nerd Extraordinaire - nicoya@apia.dhs.org
Winnipeg, Manitoba, Canada           --           http://nicoya.feline.pp.se/


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