linux-apus-devel

A1200 2.4.17 bug - found problem!!!

[Alan B. <al...@ms...>]

hi,

I'm 95% sure that I've found the problem with > 2.4.13 kernels.

its the f'ing GCC compiler.


2.95.4 seems to be the culprit.  last night I compiled
the 2.4.13 source with it....and we all know this kernel
works fine (Its the one i normally run anyway!)...but
after a compile on my current system it freezes at exactly
the same point as the 2.4.17 compile...at the IDE probing/detection
part,

2.95.4 either has a nasty 603 bug...or optimises something away
which shouldnt be touched.  so, either we change things in our
tree to stop this...or advice that everyone leave 2.95.4 well alone
and either backtrack to 2.95.2 or move up to 3.x - anyone playing
with 3.x?

in the meantime...have to figure out how to get my GCC back to a 
nice one with apt-get......


alan - and i thought only 2.96.x was screwed :-(

Re: Booting, CVPPC on board SCSI driver and 1200 IDE problem

[Alan B. <al...@ms...>]

hi,

> I can mount SCSI (A4091) ZIPS with ext2 and IDE fixed disks with either
> AFSS or ext2 on them but not ZIPS with affs - odd.

...interesting. a quick question to those that might know....when the IDE
disks are probed on APUS the RDB is looked at - is there any reference
to the AFFS at this point in time at all?

alan

Re: Booting, CVPPC on board SCSI driver and 1200 IDE problem

[Ken T. <ke...@we...>]

On Mon, 5 Aug 2002, Alan Buxey wrote:

Hello,

> http://www.wirsinds.de/rene/53c770.tar.bz2

Thanks for that.

> welcome to the club - at least your system continues to run.  I guess with a
> Disk in the ZIPdrive, the drive becomes 'active' and issues an interupt. 
> ..one that doesnt get serviced?

I can mount SCSI (A4091) ZIPS with ext2 and IDE fixed disks with either
AFSS or ext2 on them but not ZIPS with affs - odd.

Ken.

Re: CVS Update: 2.3 (branch: trunk)

[Rene B. <re...@we...>]

On 2002.08.02 22:46 Roman Zippel wrote:
> Hi,
> 
> Rene Brothuhn wrote:
> 
> > But there is still another problem with kernel_set_cachemode, if the
> > beginnig address of the region is also not a multiply of PAGE_SIZE,
> then
> > in some cases the last page will also not remaped. I'll fix it if I
> have
> > the time.
> 
> I'd prefer you avoid that function completely. It's an ugly hack and
> absolutely unportable.


Hi!

I have it always used enclosed in #ifdef CONFIG_AMIGA. But is there any 
other or standard method to get some mem uncached?


Ciao, Renè

Re: Booting, CVPPC on board SCSI driver and 1200 IDE problem

[Alan B. <al...@ms...>]

hi,

> On the UW SCSI driver :
> 
> I lost Rene's info on where the tar of his patched driver is and the three
> tests he wanted running, can you repost please ?

http://www.wirsinds.de/rene/53c770.tar.bz2

 
> Possibly related to the 1200 boot failure, I can't mount ZIP disks with
> affs filesystems anymore, pretty sure I could awhile back.
> 
> Booting with a ZIP disk in the drive prouduces 
> 
> last_ipl[2] already set to 2f, now 2d!
> 8278: 2 -2 2 -2 3 -3 2 -2 2 -2 2 -2 2 3 -3 2
> last_ipl[2] already set to 2f, now 2d!
> 8281: -2 3 -3 2 -2 2 -2 2 -2 2 3 -3 2 3 -3 2

welcome to the club - at least your system continues to run.  I guess with a
Disk in the ZIPdrive, the drive becomes 'active' and issues an interupt. 
..one that doesnt get serviced?
 
alan

Booting, CVPPC on board SCSI driver and 1200 IDE problem

[Ken T. <ke...@we...>]

Hello,

Linux is booting again, all is well.  ElfLoadSeg in S:Startup-Sequence
helped enormously in loading the ppc.library.  I never had it there
before, why ?  Because the cgxsystem.library appears to load the
ppc.library and I wasn't running cgx on my emergency boot partition.

On the UW SCSI driver :

I lost Rene's info on where the tar of his patched driver is and the three
tests he wanted running, can you repost please ?

Possibly related to the 1200 boot failure, I can't mount ZIP disks with
affs filesystems anymore, pretty sure I could awhile back.

Booting with a ZIP disk in the drive prouduces 

last_ipl[2] already set to 2f, now 2d!
8278: 2 -2 2 -2 3 -3 2 -2 2 -2 2 -2 2 3 -3 2
last_ipl[2] already set to 2f, now 2d!
8281: -2 3 -3 2 -2 2 -2 2 -2 2 3 -3 2 3 -3 2

I don't normally see any of these messages.

And finally afdisk segfaults and causes a reboot when exiting - might be
related.

Ken. 

Re: G-Rex 4000 and VooDoo 3 3000 support?

[Andreas <an...@po...>]

Hello AmiShop

> Dear Linux Apus developers,
> 
> only one question:
> 
> G-Rex 4000 and VooDoo 3 3000 support for LinuxAPUS?

The last thing I heard of was:

GRex (aka Predator) (PCI board for BlizzardPPC A1200)

No developers have this board
No work has been done

I guess this also includes G-REx 4000. But not for sure! You'd better bought a
Prometheus, AFAIK someones working on this one (hello Adam?)..

-- 
Regards,
Andi

Re: CVS Update: 2.3 (branch: trunk)

[Roman Z. <zi...@li...>]

Hi,

Rene Brothuhn wrote:

> But there is still another problem with kernel_set_cachemode, if the
> beginnig address of the region is also not a multiply of PAGE_SIZE, then
> in some cases the last page will also not remaped. I'll fix it if I have
> the time.

I'd prefer you avoid that function completely. It's an ugly hack and
absolutely unportable.

bye, Roman

[ linux-apus-Feature Requests-590338 ] G-Rex 4000 and VooDoo 3 3000 support

[<no...@so...>]

Feature Requests item #590338, was opened at 2002-08-02 12:40
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=355907&aid=590338&group_id=5907

Category: Kernel
Group: None
Status: Open
Resolution: None
Priority: 5
Submitted By: Nobody/Anonymous (nobody)
Assigned to: Nobody/Anonymous (nobody)
Summary: G-Rex 4000 and VooDoo 3 3000 support

Initial Comment:
Please support G-Rex 4000 and VooDoo 3 3000 PCI graphics card.

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=355907&aid=590338&group_id=5907

G-Rex 4000 and VooDoo 3 3000 support?

[AmiShop On-L. <am...@li...>]

Dear Linux Apus developers,

only one question:

G-Rex 4000 and VooDoo 3 3000 support for LinuxAPUS?

Thanx for all and good work.

WWW AmigaONE WWW

WWW AmigaOS WWW

Emanuele.

Re: The CVPPC on board SCSI driver.

[Rene B. <re...@we...>]

On 2002.08.01 14:35 Alan Buxey wrote:
> hi,
> 
> > http://www.wirsinds.de/rene/53c770.tar.bz2
> >
> > Simply replace the files /linux/drivers/scsi/53c770.x with the files
> from
> > the archive. You probably have to disable module support in your kernel
> 
> > and pass the nobats option to the kernel. Just ignore the compiler
> > warnings you will get.
> 
> maybe make a branch in the APUS tree for this? CSPPCSCSI perhaps?

There is already. The files 53c770.c and 53c770.h in linux/drivers/scsi 
are from Fred Heitkamp which are for the "Cyberstorm Mk III SCSI support", 
which is already configurable.

Maybe we could rename it, because I expect that no one other needs a 
53c770 driver.


Ciao, Renè

Re: CVS Update: 2.3 (branch: trunk)

[Rene B. <re...@we...>]

On 2002.08.01 14:33 Alan Buxey wrote:
> hi,
> 
> > CVSROOT:	/cvsroot/linux-apus
> > Module name:	2.3
> > Repository:	2.3/arch/ppc/kernel/
> > Changes by:	brothuhn@usw-pr-cvs1.	02/07/31 08:10:25
> >
> > Log message:
> >   improved loop counter in kernel_set_cachemode
> >
> > Modified files:
> >       2.3/arch/ppc/kernel/:
> >         apus_setup.c
> >
> >   Revision      Changes    Path
> >   1.34          +4 -2      2.3/arch/ppc/kernel/apus_setup.c
> 
> can you explain this one a little more? Does it fix a current known bug
> etc?


Hello!


O.K. If the size of the region which should remapped are not a multiply of 
PAGE_SIZE, then always the last pages were not remapped. I would call it a 
bug and maybe this has caused some other problems.

But there is still another problem with kernel_set_cachemode, if the 
beginnig address of the region is also not a multiply of PAGE_SIZE, then 
in some cases the last page will also not remaped. I'll fix it if I have 
the time.


Ciao, Renè

Re: The CVPPC on board SCSI driver.

[Alan B. <al...@ms...>]

hi,

> http://www.wirsinds.de/rene/53c770.tar.bz2
> 
> Simply replace the files /linux/drivers/scsi/53c770.x with the files from 
> the archive. You probably have to disable module support in your kernel 
> and pass the nobats option to the kernel. Just ignore the compiler 
> warnings you will get.

maybe make a branch in the APUS tree for this? CSPPCSCSI perhaps?

I dont have CSPPC_SCSI (or CSPPC for that matter) but understand the
interest in this particular development... great work for getting it
along so far!

alan

Re: CVS Update: 2.3 (branch: trunk)

[Alan B. <al...@ms...>]

hi,

> CVSROOT:	/cvsroot/linux-apus
> Module name:	2.3
> Repository:	2.3/arch/ppc/kernel/
> Changes by:	brothuhn@usw-pr-cvs1.	02/07/31 08:10:25
> 
> Log message:
>   improved loop counter in kernel_set_cachemode
> 
> Modified files:
>       2.3/arch/ppc/kernel/:
>         apus_setup.c 
>   
>   Revision      Changes    Path
>   1.34          +4 -2      2.3/arch/ppc/kernel/apus_setup.c

can you explain this one a little more? Does it fix a current known bug etc?

thanks

alan

Re: The CVPPC on board SCSI driver.

[Rene B. <re...@we...>]

On 2002.07.31 23:16 kent wrote:
> 
> 
> On Wed, 31 Jul 2002, Rene Brothuhn wrote:
> 
> Hello,
> 
> > Because someone wants to see my last changes to the UW-SCSI driver I
> have
> > added a link where you can download it. But the driver is still
> unworkable.
> >
> > http://www.wirsinds.de/rene/53c770.tar.bz2
> 
> I was one of the people wanting to see your changes but I've had a SCSI
> HD
> disaster here and can't do anything until I sort things out.

Hi, Ken!

I guess you have bought the UW drive. I know that trouble...
Maybe this can help: don't create partitions greater than 4GB or outside 
4GB unless you have some experience with that. And be realy realy careful 
if you connect any drive to SCSI or voltage when your Amiga is on. I have 
destroyed my 53c770 chip by being a little bit uncareful... (thanks to DCE 
for repair)


Ciao, Renè

Re: The CVPPC on board SCSI driver.

[kent <kt...@we...>]

On Wed, 31 Jul 2002, Rene Brothuhn wrote:

Hello,

> Because someone wants to see my last changes to the UW-SCSI driver I have 
> added a link where you can download it. But the driver is still unworkable.
> 
> http://www.wirsinds.de/rene/53c770.tar.bz2

I was one of the people wanting to see your changes but I've had a SCSI HD
disaster here and can't do anything until I sort things out.

Ken.

CVS Update: 2.3 (branch: trunk)

[Rene B. <bro...@us...>]

CVSROOT:	/cvsroot/linux-apus
Module name:	2.3
Repository:	2.3/arch/ppc/kernel/
Changes by:	brothuhn@usw-pr-cvs1.	02/07/31 08:10:25

Log message:
  improved loop counter in kernel_set_cachemode

Modified files:
      2.3/arch/ppc/kernel/:
        apus_setup.c 
  
  Revision      Changes    Path
  1.34          +4 -2      2.3/arch/ppc/kernel/apus_setup.c

Re: The CVPPC on board SCSI driver.

[Rene B. <re...@we...>]

Hello!


Because someone wants to see my last changes to the UW-SCSI driver I have 
added a link where you can download it. But the driver is still unworkable.

http://www.wirsinds.de/rene/53c770.tar.bz2

Simply replace the files /linux/drivers/scsi/53c770.x with the files from 
the archive. You probably have to disable module support in your kernel 
and pass the nobats option to the kernel. Just ignore the compiler 
warnings you will get.
If you have devices on the SCSI-bus you should at least see that the 
driver detects these devices. On some SCSI-bus configurations (only one 
device...) you may get your system booting (not from UW-SCSI!) and see 
something in /proc/scsi. You may also use dd to try access to a connected 
device and see the LED is working correctly, but beware of your data if 
you do that!

If someone has the time to make a little test for me, just try to compile 
the kernel with module support and test either the driver is a module and 
the driver is not a module. And send me the dmesg in all three cases 
(monolithic kernel; modular kernel, driver not as module; modular kernel, 
driver as module).


Ciao, Renè

Re: IDE issue, no booting and that last_ipl

[Geert U. <ge...@li...>]

On Fri, 26 Jul 2002, Alan Buxey wrote:
> just to confirm, taking IDE support out of the kernel, the
> system boots to mounting of root (ie everything works) so
> its in the IDE stage that we have issues.  dont think
> its the driver code, think its the interupts.  one thing
> that does buzz into my viewpoint is that an interupt call/service
> (cant recall), was REMOVED for when the IDE is initiated.  
> 
> if the A1200 hardware generates an interupt when its first probed
> then this may be it?

Are interrupts disabled during probing?

I remember an issue with some WD drives, when the probing code didn't disable
the real IDE interrupt, but asked the drive not to send an interrupt instead.
Some driver don't obey that command, and still send an interrupt, causing a
lock up.

Gr{oetje,eeting}s,

						Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- ge...@li...

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
							    -- Linus Torvalds

Re: IDE issue, no booting and that last_ipl

[Alan B. <al...@ms...>]

hi,

just to confirm, taking IDE support out of the kernel, the
system boots to mounting of root (ie everything works) so
its in the IDE stage that we have issues.  dont think
its the driver code, think its the interupts.  one thing
that does buzz into my viewpoint is that an interupt call/service
(cant recall), was REMOVED for when the IDE is initiated.  

if the A1200 hardware generates an interupt when its first probed
then this may be it?

alan

Re: IDE issue, no booting and that last_ipl

[Rene B. <re...@we...>]

On 2002.07.25 10:58 Alan Buxey wrote:
> hi,
> 
> > The interrupt controller in the ppc board.
> 
> ..which already has some extra lines in the apus_setup.c file
> (The blizzardPPC was a right pain to get booting into APUS
> at the very start of the project...) . strange how this only manifests
> itself at the part when IDE is being setup.
> 

Hmm, so have you tried a kernel without the IDE stuff within and booted 
from SCSI or the test-ramdisk?


Ciao, Renè

Re: IDE issue, no booting and that last_ipl

[Andreas <an...@po...>]

Hi

> Alan Buxey wrote:
> 
>> I'n not too concerned about the last_ipl printk's right now...as
>> some other people see them and their system still boots (and operates)
>> some info about what it means might be nice...i know its some debugging.
> 
> It's some interrupt event history (positive number for interrupt entry,
> negative for exit). It's printed when it sees unexpected interrupts. The
> basic problem is a buggy interrupt controller, which generates interrupt
> when it shouldn't.

I wonder if it may have something to do with the different ppc firmware
revisions. May it be that the newer kernels somehow dislike newer versions
of the flashrom on cyberstorm and blizzard cards, and therefore crash on those
machines, but work on older ones (like mine or yours) or vice versa?

IIRC there have lately (year 2001) been some new updates floating around to
support pci bridging boards. Maybe we should have asked about the firmware
revision in the poll.

> The new interrupt code ignores these interrupts. It's
> not completely clear why it crashes and without hardware I cannot debug
> this. The problem is that this requires some careful adding of prints
> and lots of reboots, since too much prints will modify the behaviour and
> magically "fix" the problem.

-- 
Kind regards,
Andi

Re: IDE issue, no booting and that last_ipl

[Alan B. <al...@ms...>]

hi,

> The interrupt controller in the ppc board.

..which already has some extra lines in the apus_setup.c file
(The blizzardPPC was a right pain to get booting into APUS
at the very start of the project...) . strange how this only manifests
itself at the part when IDE is being setup.


alan

Re: IDE issue, no booting and that last_ipl

[Roman Z. <zi...@li...>]

Hi,

On Wed, 24 Jul 2002, Rene Brothuhn wrote:

> What do you mean with "buggy interrupt controller"? Is it on the Amiga
> board or on the PPC-board? Or is it simply a peace of software?

The interrupt controller in the ppc board.

bye, Roman

Re: IDE issue, no booting and that last_ipl

[Rene B. <re...@we...>]

On 2002.07.24 22:22 Roman Zippel wrote:
> Hi,
>=20
> Alan Buxey wrote:
>=20
> > I'n not too concerned about the last_ipl printk's right now...as
> > some other people see them and their system still boots (and operates=
)
> > some info about what it means might be nice...i know its some
> debugging.
>=20
> It's some interrupt event history (positive number for interrupt entry,
> negative for exit). It's printed when it sees unexpected interrupts. Th=
e
> basic problem is a buggy interrupt controller, which generates interrup=
t
> when it shouldn't.

Hi!

What do you mean with "buggy interrupt controller"? Is it on the Amiga=20
board or on the PPC-board? Or is it simply a peace of software?


Ciao, Ren=E8