new G4 dual 1 gig is here - how to install?

new G4 dual 1 gig is here - how to install?

[Kevin B. Hendricks]

Hi,

Okay my new G4 1 gig box has finally arrived after sitting in Canadian
customs for close to 5 days for no real good reason.

So I made the kernel changes Ben indicated and built a kernel (up not
smp to start) and since everything is installed on a external scsi drive
I tried the following:

I simply unplugged the scsi chain and installed it on the new machine.
I then held down the option key to get the menu and sure enough the nice
linux disk icon showed in the of menu.

When I select it I immediately get the prompt for linux, mac, or cdrom
boot but as soon as I hit l for linux it recycles.

Obviously it can find the script partion and show it to me but it can
not seem to find yaboot or the root linux partition (on that same drive)
so perhaps some internal of path is set somehow that differs from my old
machine to my new machine.

So what is the best way to get linux working on this thing.  Since
kernel changes are needed to recognize the new 7455 chip, none of the
canned installers from SuSE, or YellowDog will work, will they?

Do I have to roll my own installer with the fixed kernel just to get
linux going?  If so, how  do I go about rolling my own installer image
from the kernel I have built with the 7455 code changes in place?

Ideas anyone.   I really thought since my current installation in on
external scsi that the option startup trick would work (and it does run
the script but yaboot does not seem to get properly launched.

Hope someone can help since I have to give my old machine to my wife
soon (or she will lynch me) so I really want to get linux up on my new
machine before my old machine goes to her.

Any helpful hints would be greatly appreciated.

Thanks,

Kevin


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

Re: new G4 dual 1 gig is here - how to install?

[Tony 'Nicoya' Mantler]

At 7:08 PM -0600 2/12/02, Kevin B. Hendricks wrote:
>So what is the best way to get linux working on this thing.  Since
>kernel changes are needed to recognize the new 7455 chip, none of the
>canned installers from SuSE, or YellowDog will work, will they?

I recall that the debian installer can use essentially any kernel. I think
the only thing it needs to work 100% is a tgz with a pack of modules that
matches the uname, which is easy enough to make/fake.


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: new G4 dual 1 gig is here - how to install?

[Kevin B. Hendricks]

Hi,

BTW: Are there any nice utilities to let me snoop through the OF device
tree under Mac OSX.  I know I can get DisplayName Registry to work under
MacOS 9.2 but would rather do my machine snooping under OSX if possible.

Thanks,

Kevin


On Tuesday, February 12, 2002, at 08:08 PM, Kevin B. Hendricks wrote:

>
> Hi,
>
> Okay my new G4 1 gig box has finally arrived after sitting in Canadian
> customs for close to 5 days for no real good reason.
>
> So I made the kernel changes Ben indicated and built a kernel (up not
> smp to start) and since everything is installed on a external scsi drive
> I tried the following:
>
> I simply unplugged the scsi chain and installed it on the new machine.
> I then held down the option key to get the menu and sure enough the nice
> linux disk icon showed in the of menu.
>
> When I select it I immediately get the prompt for linux, mac, or cdrom
> boot but as soon as I hit l for linux it recycles.
>
> Obviously it can find the script partion and show it to me but it can
> not seem to find yaboot or the root linux partition (on that same drive)
> so perhaps some internal of path is set somehow that differs from my old
> machine to my new machine.
>
> So what is the best way to get linux working on this thing.  Since
> kernel changes are needed to recognize the new 7455 chip, none of the
> canned installers from SuSE, or YellowDog will work, will they?
>
> Do I have to roll my own installer with the fixed kernel just to get
> linux going?  If so, how  do I go about rolling my own installer image
> from the kernel I have built with the 7455 code changes in place?
>
> Ideas anyone.   I really thought since my current installation in on
> external scsi that the option startup trick would work (and it does run
> the script but yaboot does not seem to get properly launched.
>
> Hope someone can help since I have to give my old machine to my wife
> soon (or she will lynch me) so I really want to get linux up on my new
> machine before my old machine goes to her.
>
> Any helpful hints would be greatly appreciated.
>
> Thanks,
>
> Kevin
>
>
>


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

Re: new G4 dual 1 gig is here - how to install?

[Olaf Hering]

On Tue, Feb 12, Kevin B. Hendricks wrote:

> BTW: Are there any nice utilities to let me snoop through the OF device
> tree under Mac OSX.  I know I can get DisplayName Registry to work under
> MacOS 9.2 but would rather do my machine snooping under OSX if possible.

Copy the boot file in the root of a hfs partition, select that one as
boot volume. Boot to OF prompt, type "printenv boot-device", that will
give you a clue about that path name. Then change the ,\\:tbxi part to
,filename.


Gruss Olaf

--
 $ man clone

BUGS
       Main feature not yet implemented...

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

Re: new G4 dual 1 gig is here - how to install?

[Kevin B. Hendricks]

Hi,

Thanks everyone for you helpful hints.

I have now done a make zImage and built  it so I will try that option
tonight.

I was also able to boot into Mac OS 9.2 and run DisplayNameRegistry and
dumped it to a file so I should be able to figure out the correct path to
the scsi boot device from that info (I hope).

If not, I will try the idea of putting a small Mac OS boot script on a
scsi partion and try to use the "print-env" boot-device info to figure out
the of boot path.  And then I will try either changing the ofboot.b script
or trying to boot yaboot directly.

I will give all of these things a try tonight and report back.

Thanks again for all of the ideas.

Kevin

On February 13, 2002 03:00, Olaf Hering wrote:
> On Tue, Feb 12, Kevin B. Hendricks wrote:
> > BTW: Are there any nice utilities to let me snoop through the OF
> > device tree under Mac OSX.  I know I can get DisplayName Registry to
> > work under MacOS 9.2 but would rather do my machine snooping under OSX
> > if possible.
>
> Copy the boot file in the root of a hfs partition, select that one as
> boot volume. Boot to OF prompt, type "printenv boot-device", that will
> give you a clue about that path name. Then change the ,\\:tbxi part to
> ,filename.
>
>
> Gruss Olaf

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

Re: new G4 dual 1 gig is here - how to install?

[Kevin B. Hendricks]

Hi,

I found you can easily find OF paths under MacOSX by simply issuing the
following commands from Terminal.app under MacOSX

ioreg -l

It lists the class hierarchy for the io devices and includes the of
device tree info.

So using that I was able to determine the proper OF boot path for my
scsi card:

/pci@f2000000/ADPT,2930CU@13/@1:6,\\yaboot

So I booted into OF and simply tried it.

There is good news and bad news:

1. The good news is that it seemed to boot

2.  The bad news is that my ATI 7500 with my flat panel display simply
goes black so I can't tell exactly what is detected or not.

But I was able to login blindly and issue a proper shutdown command so
the kernel did boot properly at least to some extent.

I will try with an old VGA monitor I have to see if I can figure out
what is up.

So it looks like once we get past the DFP ATI 7500 issues, it should be
pretty easy for distributions to add support for the new PowerMacs quite
easily.

I will let people know when I actually get something working.

Thanks,

Kevin

On Tuesday, February 12, 2002, at 08:55 PM, Kevin B. Hendricks wrote:

> Hi,
>
> BTW: Are there any nice utilities to let me snoop through the OF device
> tree under Mac OSX.  I know I can get DisplayName Registry to work
> under MacOS 9.2 but would rather do my machine snooping under OSX if
> possible.
>
> Thanks,
>
> Kevin
>
>
> On Tuesday, February 12, 2002, at 08:08 PM, Kevin B. Hendricks wrote:
>
>>
>> Hi,
>>
>> Okay my new G4 1 gig box has finally arrived after sitting in Canadian
>> customs for close to 5 days for no real good reason.
>>
>> So I made the kernel changes Ben indicated and built a kernel (up not
>> smp to start) and since everything is installed on a external scsi
>> drive
>> I tried the following:
>>
>> I simply unplugged the scsi chain and installed it on the new machine.
>> I then held down the option key to get the menu and sure enough the
>> nice
>> linux disk icon showed in the of menu.
>>
>> When I select it I immediately get the prompt for linux, mac, or cdrom
>> boot but as soon as I hit l for linux it recycles.
>>
>> Obviously it can find the script partion and show it to me but it can
>> not seem to find yaboot or the root linux partition (on that same
>> drive)
>> so perhaps some internal of path is set somehow that differs from my
>> old
>> machine to my new machine.
>>
>> So what is the best way to get linux working on this thing.  Since
>> kernel changes are needed to recognize the new 7455 chip, none of the
>> canned installers from SuSE, or YellowDog will work, will they?
>>
>> Do I have to roll my own installer with the fixed kernel just to get
>> linux going?  If so, how  do I go about rolling my own installer image
>> from the kernel I have built with the 7455 code changes in place?
>>
>> Ideas anyone.   I really thought since my current installation in on
>> external scsi that the option startup trick would work (and it does run
>> the script but yaboot does not seem to get properly launched.
>>
>> Hope someone can help since I have to give my old machine to my wife
>> soon (or she will lynch me) so I really want to get linux up on my new
>> machine before my old machine goes to her.
>>
>> Any helpful hints would be greatly appreciated.
>>
>> Thanks,
>>
>> Kevin
>>
>>
>>
>


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

Re: new G4 dual 1 gig is here - how to install?

[Ani Joshi]

On Wed, 13 Feb 2002, Kevin B. Hendricks wrote:

> 2.  The bad news is that my ATI 7500 with my flat panel display simply
> goes black so I can't tell exactly what is detected or not.


You will need to check if the OpenFirmware pll probing code I use in
radeonfb works on 7500.  Look in the device-tree node and see what the
reference clock is.  If you look in radeon_get_pllinfo() you can see the
OF probe code there.  IIRC on x86 7500 boards, the xclk is ~ 22000, the
ref div is 12 and reference clock is 2700 i belive.  The 7500 is supported
on x86 btw.

If you can log in and get a bootlog that would be helpful.


ani


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

Re: new G4 dual 1 gig is here - how to install?

[Kevin B. Hendricks]

Hi Ani,

A couple of problems.

1. Checking the BIOS scratch register

 tmp = INREG(RADEON_BIOS_4_SCRATCH);

        if (rinfo->hasCRTC2) {
                /* primary DVI port */
                if (tmp & 0x08)
                        rinfo->dviDisp_type = MT_DFP;

To determine which port on a 7500 is actualy being used does not work on
PPC, so we have to use the OF names and look in the tree to dtermine which
port is working somehow.  Question is how?

2. For some reason I keep getting the following failure message:

                         default:
                                printk("radeonfb: Failed to detect DFP
panel size\n");
                                return 0;

which means:

        if (radeon_dfp_parse_EDID(rinfo))
                radeon_update_default_var(rinfo);

        if (!rinfo->got_dfpinfo) {

Somehow rinfo->got_dfpinfo is not being set (yet a received no warning
messages about EDID not being detected).

The OF device tree does have the "EDID" block but for some reason it is
not being found properly (the dual head issue of the 7500 leading to two
different device tree nodes?) or not being parsed properly.

But I never received the following message:

        if (!radeon_get_EDID_OF(rinfo))
                RTRACE("radeonfb: could not retrieve EDID from OF\n");


And I never received this message either:

        printk("radeonfb: detected DFP panel size from EDID: %dx%d\n",
                rinfo->panel_xres, rinfo->panel_yres);

        rinfo->got_dfpinfo = 1;


And instead I always get just The "Failed to detect DFP panel size"

I can't figure it out.

I will add some more debug print statements so that I can actually see
what is messing up.

Thanks,

Kevin

On February 13, 2002 10:53, Ani Joshi wrote:
> On Wed, 13 Feb 2002, Kevin B. Hendricks wrote:
> > 2.  The bad news is that my ATI 7500 with my flat panel display simply
> > goes black so I can't tell exactly what is detected or not.
>
> You will need to check if the OpenFirmware pll probing code I use in
> radeonfb works on 7500.  Look in the device-tree node and see what the
> reference clock is.  If you look in radeon_get_pllinfo() you can see the
> OF probe code there.  IIRC on x86 7500 boards, the xclk is ~ 22000, the
> ref div is 12 and reference clock is 2700 i belive.  The 7500 is
> supported on x86 btw.
>
> If you can log in and get a bootlog that would be helpful.
>
>
> ani

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

Re: new G4 dual 1 gig is here - how to install?

[Kevin B. Hendricks]

Hi,

BTW I just finished building XF4.2.0 in under 45 min from scratch with
just one CPU on my new box and fired it up and it works just fine out of
the box.

We really need to write a USB daemon to allow the "Brightness" button to
work on the DFP monitors from Apple.

The only way I could seem to reduce the brightness was to use the Gamma
0.50 setting in the XF86Config-4 file to reduce the gamma

Ani/Ben I will try to send you whatever patches are needed for radeonfb.c
for the 7500 (once I clean up my hacks) so that you can see what I had to
do.

Now on to SMP kernel testing!

Thanks for everyone's HELP!

Thanks,

Kevin

On February 15, 2002 06:08, Kevin B. Hendricks wrote:
> Hi Ani,
>
> A couple of problems.
>
> 1. Checking the BIOS scratch register
>
>  tmp = INREG(RADEON_BIOS_4_SCRATCH);
>
>         if (rinfo->hasCRTC2) {
>                 /* primary DVI port */
>                 if (tmp & 0x08)
>                         rinfo->dviDisp_type = MT_DFP;
>
> To determine which port on a 7500 is actualy being used does not work on
> PPC, so we have to use the OF names and look in the tree to dtermine
> which port is working somehow.  Question is how?
>
> 2. For some reason I keep getting the following failure message:
>
>                          default:
>                                 printk("radeonfb: Failed to detect DFP
> panel size\n");
>                                 return 0;
>
> which means:
>
>         if (radeon_dfp_parse_EDID(rinfo))
>                 radeon_update_default_var(rinfo);
>
>         if (!rinfo->got_dfpinfo) {
>
> Somehow rinfo->got_dfpinfo is not being set (yet a received no warning
> messages about EDID not being detected).
>
> The OF device tree does have the "EDID" block but for some reason it is
> not being found properly (the dual head issue of the 7500 leading to two
> different device tree nodes?) or not being parsed properly.
>
> But I never received the following message:
>
>         if (!radeon_get_EDID_OF(rinfo))
>                 RTRACE("radeonfb: could not retrieve EDID from OF\n");
>
>
> And I never received this message either:
>
>         printk("radeonfb: detected DFP panel size from EDID: %dx%d\n",
>                 rinfo->panel_xres, rinfo->panel_yres);
>
>         rinfo->got_dfpinfo = 1;
>
>
> And instead I always get just The "Failed to detect DFP panel size"
>
> I can't figure it out.
>
> I will add some more debug print statements so that I can actually see
> what is messing up.
>
> Thanks,
>
> Kevin
>
> On February 13, 2002 10:53, Ani Joshi wrote:
> > On Wed, 13 Feb 2002, Kevin B. Hendricks wrote:
> > > 2.  The bad news is that my ATI 7500 with my flat panel display
> > > simply goes black so I can't tell exactly what is detected or not.
> >
> > You will need to check if the OpenFirmware pll probing code I use in
> > radeonfb works on 7500.  Look in the device-tree node and see what the
> > reference clock is.  If you look in radeon_get_pllinfo() you can see
> > the OF probe code there.  IIRC on x86 7500 boards, the xclk is ~
> > 22000, the ref div is 12 and reference clock is 2700 i belive.  The
> > 7500 is supported on x86 btw.
> >
> > If you can log in and get a bootlog that would be helpful.
> >
> >
> > ani

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

SMP kernel configuration question?

[Kevin B. Hendricks]

Hi,

I have never configured a kernel for SMP before.

I turned on the obvious CONFIG_SMP=y in the .config but was prompted if I
would like to distribute interrupts (IRQs) across all CPUs.

What is the recommended setting for:

CONFIG_IRQ_ALL_CPUS

Any guidance here would be greatly appreciated.

Thanks,

Kevin

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

Re: SMP kernel configuration question?

[Benjamin Herrenschmidt]

>
>Hi,
>
>I have never configured a kernel for SMP before.
>
>I turned on the obvious CONFIG_SMP=y in the .config but was prompted if I
>would like to distribute interrupts (IRQs) across all CPUs.
>
>What is the recommended setting for:
>
>CONFIG_IRQ_ALL_CPUS
>
>Any guidance here would be greatly appreciated.

Leave it to N. ideally, you should enable it, but I have experienced
lockups when this feature is enabled, I haven't yet figured out what's
going on (could be related to the lockup some users are experiencing
with SMP + HIGHMEM on some dual 800, let me know if you see the same).

Ben.


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

SMP kernel seems to work fine

[Kevin B. Hendricks]

Hi,

The SMP kernel I just built seems to work fine (no stress testing has been
done yet but the UP seemed quite stable.).

So the bottom line is with only a few patches to help the kernel recognize
the 7455 processor (which are probably already in Ben's tree already) and
some small changes to the  kernel radeon frame buffer code (to help it
find the OF EDID, determine which ports are being used, stop playing with
the TMDS_transmitter_cntl, etc

We can pretty much almost fully support the new PowerMac G4's Apple just
announced (as long as they opt for the Radeon 7500 card in the BTO).

BTW: the sound volume problem in the kernel can be worked around with
external speakers.

Also, with this combination, XF 4.2.0 seems to work out of the box.

So basically any distribution that wanted could simply provide a new
installer built with one of Ben's latest kernels (with some added radeonfb
patches), a recent kernel, and XF420 and you could easily claim support
for the latest G4 machines.

Hope this helps!

Kevin


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

Re: new G4 dual 1 gig is here - how to install?

[Benjamin Herrenschmidt]

>
>BTW I just finished building XF4.2.0 in under 45 min from scratch with
>just one CPU on my new box and fired it up and it works just fine out of
>the box.
>
>We really need to write a USB daemon to allow the "Brightness" button to
>work on the DFP monitors from Apple.

There are at least 2 nice daemons beeing worked on that currently
only handle recent laptop keys but those will be improved over time.
We also need a better API that what the kernel currently provides for
the brightness, especially when dealing with X as it is not safe to
let the kernel tap the card registers the way it does currently while
X is doing the same below.

>The only way I could seem to reduce the brightness was to use the Gamma
>0.50 setting in the XF86Config-4 file to reduce the gamma
>
>Ani/Ben I will try to send you whatever patches are needed for radeonfb.c
>for the 7500 (once I clean up my hacks) so that you can see what I had to
>do.
>
>Now on to SMP kernel testing!

I'm still working on proper L3 support, I'll push when it's ready.

Ben.


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

Re: new G4 dual 1 gig is here - how to install?

[Kevin B. Hendricks]

Hi,

I hooked up an 17inch CRT and the DFP flat panel to the ATI 7500.

Then I tried booting linux, here are some excerpts from the dmesg log
(hand typed)

radeonfb: ref_clk = 2700, ref_div = 12, xclk = 20025 from OF

ATI Radeon 7500 QW DDR SGRAM 32MB

radeonfb:  DVI port no monitor connected
                    CRT port no monitor connected

So it looks like it can't detect which port to use and defaults to CRT?

I will just hack the driver to force use of DVI and try to work around
this issue so that I don't need 2 monitors connected.

But Here is the interesting line that I can't figure out

no framebuffer address found for
/pci@f0000000/ATY,BlueStoneParent@10,ATY,BlueStone_B

So it looks like there is some other ATI device in the system that I
don't understand where or what it is.  I see both entries for this
BlueStone thing and entries for the Radeon 7500 in the device tree.

Also, interestingly enough, SOUND is out of control on this new
machine.  A beep comes out at ear-shattering levels no matter what I
do.  I do believe something has changed when it comes ot sound in these
new systems.

I will try to dump the OF device tree and get the dmesg log posted so
that anyone interested can take a look at things to look for interesting
items.

So I am progressing ....

Anything anyone wants me to dump, just let me know and I will try to
provide the info they need.

Thanks,

Kevin


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

Re: new G4 dual 1 gig is here - how to install?

[benh]

>
>Hi,
>
>I hooked up an 17inch CRT and the DFP flat panel to the ATI 7500.
>
>Then I tried booting linux, here are some excerpts from the dmesg log
>(hand typed)
>
>radeonfb: ref_clk = 2700, ref_div = 12, xclk = 20025 from OF
>
>ATI Radeon 7500 QW DDR SGRAM 32MB
>
>radeonfb:  DVI port no monitor connected
>                    CRT port no monitor connected
>
>So it looks like it can't detect which port to use and defaults to CRT?
>
>I will just hack the driver to force use of DVI and try to work around
>this issue so that I don't need 2 monitors connected.
>
>But Here is the interesting line that I can't figure out
>
>no framebuffer address found for
>/pci@f0000000/ATY,BlueStoneParent@10,ATY,BlueStone_B

That's just the second head, your card is dual head message
is normal.

>So it looks like there is some other ATI device in the system that I
>don't understand where or what it is.  I see both entries for this
>BlueStone thing and entries for the Radeon 7500 in the device tree.
>
>Also, interestingly enough, SOUND is out of control on this new
>machine.  A beep comes out at ear-shattering levels no matter what I
>do.  I do believe something has changed when it comes ot sound in these
>new systems.

Maybe yet another crappy tumbler setup ? Make sure you use my
latest kernel, I also started adding some 7455 support & cleaned
up the NAP/DOZE/idle thing.

Ben.


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

Re: new G4 dual 1 gig is here - how to install?

[Christopher C. Chimelis]

On Thu, 14 Feb 2002 benh@kernel.crashing.org wrote:

> >Also, interestingly enough, SOUND is out of control on this new
> >machine.  A beep comes out at ear-shattering levels no matter what I
> >do.  I do believe something has changed when it comes ot sound in these
> >new systems.
>
> Maybe yet another crappy tumbler setup ? Make sure you use my
> latest kernel, I also started adding some 7455 support & cleaned
> up the NAP/DOZE/idle thing.

If it is tumbler, let me know.  I've become annoyed at the default volume
levels (too loud) in the driver and am planning on changing them.  I'd be
happy to have a few extra people to bounce settings off of.

C


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

Re: new G4 dual 1 gig is here - how to install?

[Tom Rini]

On Tue, Feb 12, 2002 at 08:08:57PM -0500, Kevin B. Hendricks wrote:

> When I select it I immediately get the prompt for linux, mac, or cdrom
> boot but as soon as I hit l for linux it recycles.

Have you tried dropping to OF by hand and booting yaboot directly?  Or
even doing a 'make zImage' and trying to boot
arch/ppc/boot/images/vmlinux.elf-pmac right out of OF?  There's always a
chance the script is 'wrong' in the new box and is trying to boot
something that isn't there.

--
Tom Rini (TR1265)
http://gate.crashing.org/~trini/

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

Re: new G4 dual 1 gig is here - how to install?

[Olaf Hering]

On Tue, Feb 12, Kevin B. Hendricks wrote:

> So what is the best way to get linux working on this thing.  Since
> kernel changes are needed to recognize the new 7455 chip, none of the
> canned installers from SuSE, or YellowDog will work, will they?

Boot a arch/ppc/boot/images/vmlinux.initrd.elf-pmac. A ramdisk.image.gz
can be found on CD1/suse/images/



Gruss Olaf

--
 $ man clone

BUGS
       Main feature not yet implemented...

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

Re: new G4 dual 1 gig is here - how to install?

[Kevin B. Hendricks]

Hi Ani and Ben,

I now have the radeonfb driver working.  I had to make the following
changes:

1. pass "dfp" as the kernel parameters to force use of the DVI port to
get around the BIOS Scratch register issue (see my last mail message)

2. Change the code that looks for OF EDID to use the following:

dp =
find_device_path("/pci@f0000000/ATY,BlueStoneParent@10/ATY,BlueStone_A");
pedid = get_property(dp,"EDID",0);

This is because the pci_device_to_OF_node had dp pointing to
/pci@f0000000/ATY,BlueStoneParent@10 node where the ATY,RefClk property
was and not into ATY,BlueStone_A which is where the OF "EDID" info was
(i.e. changed node structure with a dual head card).

This is obviously a hack.  For dual head cards we need to search the
node we read ATY,RefClk from and all of its children looking for the
EDID info (but what if both BlueStone_A and BlueStone_B have EDID blocks
that differ?  Is BlueStone_A always the DVI port?  If so, if no monitor
is connected to it no EDID is present.  So we could simply look in both
and use that info to see which ports have EDID blocks which will tell us
which ports have monitors attached or not (fixing issue 1)

3.  I had to stop writing out TMDS_TRANSMITTER_CNTL values which I
needed with the old G3 B+W and the Radeom Mac 32 card. Writing anything
to this register value under the 7500 just seems to make the screen go
blank.

I have no idea how to workaround this.  Perhaps you can make it a
compile time option to either include or not include the
TMDS_TRANSMITTER_CNTL setting under radeon_write_mode.  Right now I have
simply commented them out.

With those changes in place we seem to get a very nice console on the
7500 that can be ordered as an option on the new PowerMac G4's

I will now build XF 420 and get X11 going and once there I will move to
SMP to start testing those features.

Hope this info helps,

Kevin


On Tuesday, February 12, 2002, at 08:08 PM, Kevin B. Hendricks wrote:

>
> Hi,
>
> Okay my new G4 1 gig box has finally arrived after sitting in Canadian
> customs for close to 5 days for no real good reason.
>
> So I made the kernel changes Ben indicated and built a kernel (up not
> smp to start) and since everything is installed on a external scsi drive
> I tried the following:
>
> I simply unplugged the scsi chain and installed it on the new machine.
> I then held down the option key to get the menu and sure enough the nice
> linux disk icon showed in the of menu.
>
> When I select it I immediately get the prompt for linux, mac, or cdrom
> boot but as soon as I hit l for linux it recycles.
>
> Obviously it can find the script partion and show it to me but it can
> not seem to find yaboot or the root linux partition (on that same drive)
> so perhaps some internal of path is set somehow that differs from my old
> machine to my new machine.
>
> So what is the best way to get linux working on this thing.  Since
> kernel changes are needed to recognize the new 7455 chip, none of the
> canned installers from SuSE, or YellowDog will work, will they?
>
> Do I have to roll my own installer with the fixed kernel just to get
> linux going?  If so, how  do I go about rolling my own installer image
> from the kernel I have built with the 7455 code changes in place?
>
> Ideas anyone.   I really thought since my current installation in on
> external scsi that the option startup trick would work (and it does run
> the script but yaboot does not seem to get properly launched.
>
> Hope someone can help since I have to give my old machine to my wife
> soon (or she will lynch me) so I really want to get linux up on my new
> machine before my old machine goes to her.
>
> Any helpful hints would be greatly appreciated.
>
> Thanks,
>
> Kevin
>
>
>


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

Re: new G4 dual 1 gig is here - how to install?

[Benjamin Herrenschmidt]

>
>Hi Ani and Ben,
>
>I now have the radeonfb driver working.  I had to make the following
>changes:
>
>1. pass "dfp" as the kernel parameters to force use of the DVI port to
>get around the BIOS Scratch register issue (see my last mail message)

I suppose there must be another property in the device tree telling
us about where the monitor is connected to. Anyway, we should sooner
or later have real working i2c code, Ani ?

>2. Change the code that looks for OF EDID to use the following:
>
>dp =
>find_device_path("/pci@f0000000/ATY,BlueStoneParent@10/ATY,BlueStone_A");
>pedid = get_property(dp,"EDID",0);
>
>This is because the pci_device_to_OF_node had dp pointing to
>/pci@f0000000/ATY,BlueStoneParent@10 node where the ATY,RefClk property
>was and not into ATY,BlueStone_A which is where the OF "EDID" info was
>(i.e. changed node structure with a dual head card).

Yes, that's common to all of Apple/ATI dual head drivers like my pismo's,
however, they don't seem to have a clear way of storing the EDID. On my
tipb, it is in an "EDID1" property in the root ATI node.

I beleive until we have working i2c, you keep the old pci_device_to_OF_node
code, then iterate childs of that node (node->child & node->sibling
pointers). The problem here is that you may well end up with 2 EDIDs if
2 monitors are connected at startup... (well, that depends much on what
the OF driver would do).

>This is obviously a hack.  For dual head cards we need to search the
>node we read ATY,RefClk from and all of its children looking for the
>EDID info (but what if both BlueStone_A and BlueStone_B have EDID blocks
>that differ?  Is BlueStone_A always the DVI port?  If so, if no monitor
>is connected to it no EDID is present.  So we could simply look in both
>and use that info to see which ports have EDID blocks which will tell us
>which ports have monitors attached or not (fixing issue 1)

On my tipb, I have a "display-type" property in the child nodes, aybe
that can help figuring out what is what. Though again, real i2c probing
would probably fix that once for all as I bet not 2 ATI cards will layout
their device tree the same way.
>
>3.  I had to stop writing out TMDS_TRANSMITTER_CNTL values which I
>needed with the old G3 B+W and the Radeom Mac 32 card. Writing anything
>to this register value under the 7500 just seems to make the screen go
>blank.

I'll let Ani look at this one.

>I have no idea how to workaround this.  Perhaps you can make it a
>compile time option to either include or not include the
>TMDS_TRANSMITTER_CNTL setting under radeon_write_mode.  Right now I have
>simply commented them out.
>
>With those changes in place we seem to get a very nice console on the
>7500 that can be ordered as an option on the new PowerMac G4's
>
>I will now build XF 420 and get X11 going and once there I will move to
>SMP to start testing those features.

Well, there are other issues what Ani and I will have to deal with, for
example, I think the DAA_xxx registers (or are they DDA_xxx ?) don't exist
on radeon's, at least not on all of them. Also, I want to change the memory
layout of the card (MC_AGP_LOCATION/MC_FB_LOCATION) in some ways I already
discussed with Ani (basically keeping both AGP and FB out of the region
where system memory is seen to allow clean PCI bus master to main memory),
but I have some trouble when I move them around, sometimes depending on
what OF did to the chip just before. I beleive changing those values requires
some sort of reset of part of the chip, along with adequate change of
other pointers like DISPLAY_BASEADDR, SRC_OFFSET, etc... so that those
end up to the new FB location within card's space.

I haven't fixed my power management issue yet neither.

Ben.


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

radeonfb.c flakiness

[Kevin B. Hendricks]

Hi,

Even with all of my changes, it seems the DFP screen will randomly stay
blank under PPC Linux.

Unfortunately, nothing I do can seem to bring it back to life (but all
of the radeon debug info is absolutely fine).

So some how we are not properly resetting the 7500 card or not properly
initializing it.

I can't understnad why the screen simply goes blank immediately after
the kernel starts to boot.

Ben, you mentioned something about playig with the memory mapping of the
card.  I see code that does something along those lines in the
radeonfb.c for screens that have dual heads.

Do you have any ideas I should test to see if they impact the random
startup blank screen issue (btw the kernel boots fine and I can blindly
login and shutdown).  Changing VTs from 1 to 2 or 3 has no impact.

Ideas welcome.

Thanks,

Kevin

On Saturday, February 16, 2002, at 02:19 PM, Benjamin Herrenschmidt
wrote:

>>
>> Hi Ani and Ben,
>>
>> I now have the radeonfb driver working.  I had to make the following
>> changes:
>>
>> 1. pass "dfp" as the kernel parameters to force use of the DVI port to
>> get around the BIOS Scratch register issue (see my last mail message)
>
> I suppose there must be another property in the device tree telling
> us about where the monitor is connected to. Anyway, we should sooner
> or later have real working i2c code, Ani ?
>
>> 2. Change the code that looks for OF EDID to use the following:
>>
>> dp =
>> find_device_path("/pci@f0000000/ATY,BlueStoneParent@10/ATY,BlueStone_A")
>> ;
>> pedid = get_property(dp,"EDID",0);
>>
>> This is because the pci_device_to_OF_node had dp pointing to
>> /pci@f0000000/ATY,BlueStoneParent@10 node where the ATY,RefClk property
>> was and not into ATY,BlueStone_A which is where the OF "EDID" info was
>> (i.e. changed node structure with a dual head card).
>
> Yes, that's common to all of Apple/ATI dual head drivers like my
> pismo's,
> however, they don't seem to have a clear way of storing the EDID. On my
> tipb, it is in an "EDID1" property in the root ATI node.
>
> I beleive until we have working i2c, you keep the old
> pci_device_to_OF_node
> code, then iterate childs of that node (node->child & node->sibling
> pointers). The problem here is that you may well end up with 2 EDIDs if
> 2 monitors are connected at startup... (well, that depends much on what
> the OF driver would do).
>
>> This is obviously a hack.  For dual head cards we need to search the
>> node we read ATY,RefClk from and all of its children looking for the
>> EDID info (but what if both BlueStone_A and BlueStone_B have EDID
>> blocks
>> that differ?  Is BlueStone_A always the DVI port?  If so, if no monitor
>> is connected to it no EDID is present.  So we could simply look in both
>> and use that info to see which ports have EDID blocks which will tell
>> us
>> which ports have monitors attached or not (fixing issue 1)
>
> On my tipb, I have a "display-type" property in the child nodes, aybe
> that can help figuring out what is what. Though again, real i2c probing
> would probably fix that once for all as I bet not 2 ATI cards will
> layout
> their device tree the same way.
>>
>> 3.  I had to stop writing out TMDS_TRANSMITTER_CNTL values which I
>> needed with the old G3 B+W and the Radeom Mac 32 card. Writing anything
>> to this register value under the 7500 just seems to make the screen go
>> blank.
>
> I'll let Ani look at this one.
>
>> I have no idea how to workaround this.  Perhaps you can make it a
>> compile time option to either include or not include the
>> TMDS_TRANSMITTER_CNTL setting under radeon_write_mode.  Right now I
>> have
>> simply commented them out.
>>
>> With those changes in place we seem to get a very nice console on the
>> 7500 that can be ordered as an option on the new PowerMac G4's
>>
>> I will now build XF 420 and get X11 going and once there I will move to
>> SMP to start testing those features.
>
> Well, there are other issues what Ani and I will have to deal with, for
> example, I think the DAA_xxx registers (or are they DDA_xxx ?) don't
> exist
> on radeon's, at least not on all of them. Also, I want to change the
> memory
> layout of the card (MC_AGP_LOCATION/MC_FB_LOCATION) in some ways I
> already
> discussed with Ani (basically keeping both AGP and FB out of the region
> where system memory is seen to allow clean PCI bus master to main
> memory),
> but I have some trouble when I move them around, sometimes depending on
> what OF did to the chip just before. I beleive changing those values
> requires
> some sort of reset of part of the chip, along with adequate change of
> other pointers like DISPLAY_BASEADDR, SRC_OFFSET, etc... so that those
> end up to the new FB location within card's space.
>
> I haven't fixed my power management issue yet neither.
>
> Ben.
>


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

Re: radeonfb.c flakiness

[Benjamin Herrenschmidt]

>Even with all of my changes, it seems the DFP screen will randomly stay
>blank under PPC Linux.
>
>Unfortunately, nothing I do can seem to bring it back to life (but all
>of the radeon debug info is absolutely fine).
>
>So some how we are not properly resetting the 7500 card or not properly
>initializing it.

Try removing the bit of code in my radeonfb.c that blast various
registers like MC_FB_LOCATION within a #ifdef CONFIG_ALL_PPC
in radeon_pci_register().

I suspect the sequence for changing those isn't good, the chip may
need some additional reset or delays.

>I can't understnad why the screen simply goes blank immediately after
>the kernel starts to boot.
>
>Ben, you mentioned something about playig with the memory mapping of the
>card.  I see code that does something along those lines in the
>radeonfb.c for screens that have dual heads.
>
>Do you have any ideas I should test to see if they impact the random
>startup blank screen issue (btw the kernel boots fine and I can blindly
>login and shutdown).  Changing VTs from 1 to 2 or 3 has no impact.
>
>Ideas welcome.


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

Re: radeonfb.c flakiness

[Kevin B. Hendricks]

Hi Ben and Ani,
>
> Try removing the bit of code in my radeonfb.c that blast various
> registers like MC_FB_LOCATION within a #ifdef CONFIG_ALL_PPC
> in radeon_pci_register().
>
> I suspect the sequence for changing those isn't good, the chip may
> need some additional reset or delays.

Tried that no help.  I have now figured out that the only way to make the
DFP startup properly is to zap the parameter ram (one chime's worth) and
then it will actually startup.  If I don't do that and I have been running
MacOSX then no matter what I do or try it does right after the yaboot
prompt I see a flash and then everything stays blank.

Perhaps Mac OSX driver is writing something to nvram or is moving the
memory mapping for the card's interfaces.  Either way without that single
pram zap, the screen simply goes dead and stays dead.

I am willing to try and debug this further by dumping addesses and things
from the device tree under both Linux and MacOSX to try and figure out
what has changed, if either of you think it would help.

Just let me know what you want me to try.

Thanks,

Kevin

> >I can't understnad why the screen simply goes blank immediately after
> >the kernel starts to boot.
> >
> >Ben, you mentioned something about playig with the memory mapping of
> > the card.  I see code that does something along those lines in the
> > radeonfb.c for screens that have dual heads.
> >
> >Do you have any ideas I should test to see if they impact the random
> >startup blank screen issue (btw the kernel boots fine and I can blindly
> >login and shutdown).  Changing VTs from 1 to 2 or 3 has no impact.
> >
> >Ideas welcome.

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

Re: radeonfb.c flakiness

[Benjamin Herrenschmidt]

>Tried that no help.  I have now figured out that the only way to make the
>DFP startup properly is to zap the parameter ram (one chime's worth) and
>then it will actually startup.  If I don't do that and I have been running
>MacOSX then no matter what I do or try it does right after the yaboot
>prompt I see a flash and then everything stays blank.

that's strange, since a friend of mine is experiencing exactly the same
issue with a Rage M6 (radeon mobility) in a titanium powerbook, while I
never see it here. The only difference is that his has a little bit slower
CPU.

>Perhaps Mac OSX driver is writing something to nvram or is moving the
>memory mapping for the card's interfaces.  Either way without that single
>pram zap, the screen simply goes dead and stays dead.

I think it doesn't write anything but mode & backlight infos. Make sure
your problem isn't caused by the powerbook backlight code ;) I think the
way we currently fetch the backlight setting from pram is kinda broken.

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

question on best way to install other ide drives?

[Kevin B. Hendricks]

Hi,

My dual 1 Gig machine has room to easily mount a second (slave) ide
drive.  I also have an Sonnet ATA-100 pci card and another IDE drive.

I would like to add both.  The Sonnet card can support up to 4 IDE
drives.

Would it be better (performancewize) to:

A.  Hook both drives up the the ATA-100 pci card using two ribbon cables
with both being masters?

B.  Hook only 1 to the ATA-100 (a master) and use the second as slave
off the the included IDE card?

C. Hook both to one ATA-100 card using one as master and one as slave.

Any guidance here would be greatly appreciated.

Thanks,

Kevin


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

Re: question on best way to install other ide drives?

[Tom Rini]

On Sat, Feb 16, 2002 at 06:59:53PM -0500, Kevin B. Hendricks wrote:

> Would it be better (performancewize) to:
>
> A.  Hook both drives up the the ATA-100 pci card using two ribbon cables
> with both being masters?

No difference probably, If you can boot from the sonnet card.

> B.  Hook only 1 to the ATA-100 (a master) and use the second as slave
> off the the included IDE card?
>
> C. Hook both to one ATA-100 card using one as master and one as slave.

Erm, unless you've got some horribly fast IDE drives (and use Andres IDE
patches which means you can't use onboard IDE on pmacs just yet, iirc),
your best bet is either:
a) Throw your other drive in as a slave under the CDRW/DVD, if that bay
is empty.
b) Hook up the drive as a slave to the primary.
c) Throw in the IDE card, and put another drive in as master.

I've done 'a' on the 3 G3s/G4s I've owned, and never really had a
problem (one drive _might_ have gotten pissy after being there for a few
years, but I never could verify what was causing the wierd issues).

--
Tom Rini (TR1265)
http://gate.crashing.org/~trini/

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

Re: question on best way to install other ide drives?

[Timothy A. Seufert]

At 6:59 PM -0500 2/16/02, Kevin B. Hendricks wrote:
>Hi,
>
>My dual 1 Gig machine has room to easily mount a second (slave) ide
>drive.  I also have an Sonnet ATA-100 pci card and another IDE drive.
>
>I would like to add both.  The Sonnet card can support up to 4 IDE
>drives.
>
>Would it be better (performancewize) to:
>
>A.  Hook both drives up the the ATA-100 pci card using two ribbon cables
>with both being masters?

(A).  IDE does not do a terribly good job of sharing bus bandwidth
between two active devices, so the ideal case for performance is one
IDE bus per device.

That said, you'll never be impacted if your workload never attempts
to exercise both the slave and master at the same time.
--
Tim Seufert

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

question: "disabling IRQ 54 defensively?"

[Kevin B. Hendricks]

Hi,

I just finished installing the ATA-100 card with two master IDE drives
(each on its own ribbon cable).

Now when I boot I see 7 to 10 lines that all say:

"disabling IRQ 54 defensively"

Later on I see the following:

ide0 "blah" on irq 19
ide1 "blah" on irq 20
ide2 0x440-0x447, 0x432 on irq 54
ide3 0x420-0x427, 0x412 on irq 54

Should I be worried here?  What does the disabling IRQ 54 defensively mean?

Is it wrong to have two ide devices with the ssame irq?  Are they just
sharing it?

Any info would be greatly appreciated.

Thanks,

Kevin

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

Re: question: "disabling IRQ 54 defensively?"

[Benjamin Herrenschmidt]

>
>Later on I see the following:
>
>ide0 "blah" on irq 19
>ide1 "blah" on irq 20
>ide2 0x440-0x447, 0x432 on irq 54
>ide3 0x420-0x427, 0x412 on irq 54
>
>Should I be worried here?  What does the disabling IRQ 54 defensively mean?
>
>Is it wrong to have two ide devices with the ssame irq?  Are they just
>sharing it?

They are just charing it, nothing wrong

>Any info would be greatly appreciated.

The message means that the IRQ was detected as beeing shared and so disabled
during IDE probe. This is normal behaviour and is done to avoid the machine
locking up under some cicumstances, for example with PCMCIA ATA cards that
share their interrupt with the cardbus controller.

Ben.


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

Final question: best wasy to move /, /usr, /home, etc?

[Kevin B. Hendricks]

Hi,

I am finally at the point I want to give up my old external scsi hard
drives.  Unfrotunately, my current /, /usr, /home, etc all exist on scsi
drives.

I recently purchased a 60 gig hard drive (ide) and have now installed it.

What is the best way to move all of the files and directories, and special
device files, and etc from the scsi drive to new partitions on the IDE
drives?

Can I simply use a recursive "cp" with -a?  Should I use "tar"?  What
about "parted" and its partition copies?

Or is there some way I can run an linux distribution installer from within
linux and make it install to the ide drive instead of the scsi?

Or is there some way to make "rpm" nicely redirect where it installs
things?


In fact, with som much space, I would love to have Debian, SuSE, and
YellowDog distribution sets so I can run and test OpenOffice.org against
those distributions easily (shared build drives) with different
distribution installations.

This is my last step to get a working system so any hints on the best way
to move so much data (and special device files) from one partition to
another preserving the owner,group,word, permissions, file ownership,
setuid root bits, etc. would be greatly appreciated.

Thanks,

Kevin


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

Re: Final question: best wasy to move /, /usr, /home, etc?

[Bastien Nocera]

On Mon, 2002-02-18 at 02:17, Kevin B. Hendricks wrote:
>
> I am finally at the point I want to give up my old external scsi hard
> drives.  Unfrotunately, my current /, /usr, /home, etc all exist on scsi
> drives.
>
> I recently purchased a 60 gig hard drive (ide) and have now installed it.
>
> What is the best way to move all of the files and directories, and special
> device files, and etc from the scsi drive to new partitions on the IDE
> drives?
>
> Can I simply use a recursive "cp" with -a?  Should I use "tar"?  What
> about "parted" and its partition copies?
>
> Or is there some way I can run an linux distribution installer from within
> linux and make it install to the ide drive instead of the scsi?
>
> Or is there some way to make "rpm" nicely redirect where it installs
> things?
>
>
> In fact, with som much space, I would love to have Debian, SuSE, and
> YellowDog distribution sets so I can run and test OpenOffice.org against
> those distributions easily (shared build drives) with different
> distribution installations.
>
> This is my last step to get a working system so any hints on the best way
> to move so much data (and special device files) from one partition to
> another preserving the owner,group,word, permissions, file ownership,
> setuid root bits, etc. would be greatly appreciated.

I've done this several times, to install xfs support on live machines
(ie. on my desktop and laptop that were already full of crap junk and
other kind of files).

rsync -av --exclude /proc --exclude /home/ / /newroot/

Exclude every mount point from the rsync and do this for every
partition. Best is to do this from a separate root, or a rescue/boot
disk, in which case rsync -av /oldroot/ /newroot/ is easy enough without
having to care.

Cheers

PS: sick people will tell you to use tar with loads of pipes, or cpio,
don't listen to them ;)
--
/Bastien Nocera
http://hadess.net

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

Re: Final question: best wasy to move /, /usr, /home, etc?

[Tom Rini]

On Mon, Feb 18, 2002 at 02:36:01AM +0000, Bastien Nocera wrote:
>
> On Mon, 2002-02-18 at 02:17, Kevin B. Hendricks wrote:
> >
> > I am finally at the point I want to give up my old external scsi hard
> > drives.  Unfrotunately, my current /, /usr, /home, etc all exist on scsi
> > drives.
> >
> > I recently purchased a 60 gig hard drive (ide) and have now installed it.
> >
> > What is the best way to move all of the files and directories, and special
> > device files, and etc from the scsi drive to new partitions on the IDE
> > drives?
> >
> > Can I simply use a recursive "cp" with -a?  Should I use "tar"?  What
> > about "parted" and its partition copies?
> >
> > Or is there some way I can run an linux distribution installer from within
> > linux and make it install to the ide drive instead of the scsi?
> >
> > Or is there some way to make "rpm" nicely redirect where it installs
> > things?
> >
> >
> > In fact, with som much space, I would love to have Debian, SuSE, and
> > YellowDog distribution sets so I can run and test OpenOffice.org against
> > those distributions easily (shared build drives) with different
> > distribution installations.
> >
> > This is my last step to get a working system so any hints on the best way
> > to move so much data (and special device files) from one partition to
> > another preserving the owner,group,word, permissions, file ownership,
> > setuid root bits, etc. would be greatly appreciated.
>
> I've done this several times, to install xfs support on live machines
> (ie. on my desktop and laptop that were already full of crap junk and
> other kind of files).
>
> rsync -av --exclude /proc --exclude /home/ / /newroot/
>
> Exclude every mount point from the rsync and do this for every
> partition. Best is to do this from a separate root, or a rescue/boot
> disk, in which case rsync -av /oldroot/ /newroot/ is easy enough without
> having to care.
>

I _think:
rsync -av -x / /newroot/
will get just the / partition (-x means one fs, cp and a few others take
this arg).

> PS: sick people will tell you to use tar with loads of pipes, or cpio,
> don't listen to them ;)

tar -c /{this,this,somethingelse,another} | tar -x -C /newroot

Or so, I think.

--
Tom Rini (TR1265)
http://gate.crashing.org/~trini/

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

Re: Final question: best wasy to move /, /usr, /home, etc?

[Bastien Nocera]

On Mon, 2002-02-18 at 16:16, Tom Rini wrote:
> On Mon, Feb 18, 2002 at 02:36:01AM +0000, Bastien Nocera wrote:
> >
> > On Mon, 2002-02-18 at 02:17, Kevin B. Hendricks wrote:
> > >
> > > I am finally at the point I want to give up my old external scsi hard
> > > drives.  Unfrotunately, my current /, /usr, /home, etc all exist on scsi
> > > drives.
> > >
> > > I recently purchased a 60 gig hard drive (ide) and have now installed it.
> > >
> > > What is the best way to move all of the files and directories, and special
> > > device files, and etc from the scsi drive to new partitions on the IDE
> > > drives?
> > >
> > > Can I simply use a recursive "cp" with -a?  Should I use "tar"?  What
> > > about "parted" and its partition copies?
> > >
> > > Or is there some way I can run an linux distribution installer from within
> > > linux and make it install to the ide drive instead of the scsi?
> > >
> > > Or is there some way to make "rpm" nicely redirect where it installs
> > > things?
> > >
> > >
> > > In fact, with som much space, I would love to have Debian, SuSE, and
> > > YellowDog distribution sets so I can run and test OpenOffice.org against
> > > those distributions easily (shared build drives) with different
> > > distribution installations.
> > >
> > > This is my last step to get a working system so any hints on the best way
> > > to move so much data (and special device files) from one partition to
> > > another preserving the owner,group,word, permissions, file ownership,
> > > setuid root bits, etc. would be greatly appreciated.
> >
> > I've done this several times, to install xfs support on live machines
> > (ie. on my desktop and laptop that were already full of crap junk and
> > other kind of files).
> >
> > rsync -av --exclude /proc --exclude /home/ / /newroot/
> >
> > Exclude every mount point from the rsync and do this for every
> > partition. Best is to do this from a separate root, or a rescue/boot
> > disk, in which case rsync -av /oldroot/ /newroot/ is easy enough without
> > having to care.
> >
>
> I _think:
> rsync -av -x / /newroot/
> will get just the / partition (-x means one fs, cp and a few others take
> this arg).

the man page has it indeed. The good thing is that you can run it
multiple times, or stop it in the middle and restart where you left off,
which you can't with the other solutions.

Cheers

--
/Bastien Nocera
http://hadess.net

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

Re: Final question: best wasy to move /, /usr, /home, etc?

[Jeramy B. Smith]

http://www-106.ibm.com/developerworks/library/partplan.html

There is the IBM way.

On Sun, 17 Feb 2002, Kevin B.Hendricks wrote:

> Hi,
>
> I am finally at the point I want to give up my old external scsi hard
> drives.  Unfrotunately, my current /, /usr, /home, etc all exist on scsi
> drives.
>
> I recently purchased a 60 gig hard drive (ide) and have now installed it.
>
> What is the best way to move all of the files and directories, and special
> device files, and etc from the scsi drive to new partitions on the IDE
> drives?
>
> Can I simply use a recursive "cp" with -a?  Should I use "tar"?  What
> about "parted" and its partition copies?
>
> Or is there some way I can run an linux distribution installer from within
> linux and make it install to the ide drive instead of the scsi?
>
> Or is there some way to make "rpm" nicely redirect where it installs
> things?
>
>
> In fact, with som much space, I would love to have Debian, SuSE, and
> YellowDog distribution sets so I can run and test OpenOffice.org against
> those distributions easily (shared build drives) with different
> distribution installations.
>
> This is my last step to get a working system so any hints on the best way
> to move so much data (and special device files) from one partition to
> another preserving the owner,group,word, permissions, file ownership,
> setuid root bits, etc. would be greatly appreciated.
>
> Thanks,
>
> Kevin
>
>


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

RE: Final question: best wasy to move /, /usr, /home, etc?

[Giuliano Pochini]

> I am finally at the point I want to give up my old external scsi hard
> drives.  Unfrotunately, my current /, /usr, /home, etc all exist on scsi
> drives.
>
> I recently purchased a 60 gig hard drive (ide) and have now installed it.

IDE ?!?  Bleah !!

> What is the best way to move all of the files and directories, and special
> device files, and etc from the scsi drive to new partitions on the IDE
> drives?
>
> Can I simply use a recursive "cp" with -a?  Should I use "tar"?  What
> about "parted" and its partition copies?

cp -a works fine.


Bye.
    Giuliano Pochini ->)|(<- Shiny Network {AS6665} ->)|(<-


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

Re: Final question: best wasy to move /, /usr, /home, etc?

[Derrik Pates]

On Mon, Feb 18, 2002 at 09:42:53AM +0100, Giuliano Pochini wrote:
> cp -a works fine.

Yeah, unless you don't care about any hard links (which a simple 'cp -a'
will happily stomp all over). Using tar is a better bet (I've done it
plenty of times, it works).

--
Derrik Pates      |   Sysadmin, Douglas School   |    #linuxOS on EFnet
dpates@dsdk12.net |     District (dsdk12.net)    |    #linuxOS on OPN

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

Re: Final question: best wasy to move /, /usr, /home, etc?

[Gabriel Paubert]

On Mon, 18 Feb 2002, Derrik Pates wrote:

>
> On Mon, Feb 18, 2002 at 09:42:53AM +0100, Giuliano Pochini wrote:
> > cp -a works fine.
>
> Yeah, unless you don't care about any hard links (which a simple 'cp -a'
> will happily stomp all over). Using tar is a better bet (I've done it
> plenty of times, it works).

Strange enough, cp -a does preserve hard links for me. As it should
since -a is equivalent to -dpR and -d means (from info cp):

`-d' `--no-dereference'
     Copy symbolic links as symbolic links rather than copying the
     files that they point to, and preserve hard links between source
     files in the copies.

I've used it often enough to be pretty sure about this feature.

	Regards,
	Gabriel.


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

Re: Final question: best wasy to move /, /usr, /home, etc?

[Michael Heironimus]

Just a note, "cp -a" is sort of a GNU-ish thing. On most other UNIX
systems you must use tar/cpio/rsync/whatever to preserve ownerships
and permissions. Just thought I'd point that out since I've seen
people make that mistake, and it can be a pain to clean up.

At 16:05 +0100 2/18/02, Gabriel Paubert wrote:
>On Mon, 18 Feb 2002, Derrik Pates wrote:
>  > On Mon, Feb 18, 2002 at 09:42:53AM +0100, Giuliano Pochini wrote:
>>  > cp -a works fine.
>>
>>  Yeah, unless you don't care about any hard links (which a simple 'cp -a'
>>  will happily stomp all over). Using tar is a better bet (I've done it
>>  plenty of times, it works).
>
>Strange enough, cp -a does preserve hard links for me. As it should
>since -a is equivalent to -dpR and -d means (from info cp):
>
>`-d' `--no-dereference'
>      Copy symbolic links as symbolic links rather than copying the
>      files that they point to, and preserve hard links between source
>      files in the copies.
>
>I've used it often enough to be pretty sure about this feature.

--
Michael Heironimus

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

RE: Final question: best wasy to move /, /usr, /home, etc?

[David A. Gatwood]

On Mon, 18 Feb 2002, Giuliano Pochini wrote:

> > I am finally at the point I want to give up my old external scsi hard
> > drives.  Unfrotunately, my current /, /usr, /home, etc all exist on scsi
> > drives.
> >
> > I recently purchased a 60 gig hard drive (ide) and have now installed it.
>
> IDE ?!?  Bleah !!

Yeah.  Be forewarned.  Some IDE drives over 34 gigs, particularly when
attached to ATA cards, behave badly under Linux, causing horrible file
system corruption caused by high block numbers being stupidly remapped
over the top of lower block numbers.

After adding an hdg=x,y,z line, it worked until I shut the machine off, at
which point it said that the file systems that were above the 34 gig limit
weren't there.  So it had failed again.  Several reboots later, it
magically started working.

Long story short, in its power-on state, the Maxtor drive gave incorrect
fictitious geometry, indicating that it was a 2 gig drive.  The Linux
kernel code was ignoring the boot arguments and was proceeding to
overwrite them with data from an inquiry on the drive.  The problem was
that since the drive reported itself as having 4092 cylinders instead of
the 16,653 cylinders that large drives are -supposed- to report as their
fictitious BIOS geometry, the Linux kernel didn't realize that the
geometry was fictitious, and refused to allow the LBA size information to
override the drive-returned geometry.

It took me a couple of days to debug the IDE code in the kernel to create
a workaround.  This is probably the result of a firmware bug in my Maxtor
drive, but if people run into problems where partitions above the 34 gig
limit will fail to work correctly for the first several boots but then
magically start working, I have a possible patch against the 2.2.21
prepatch kernels (should work against any of them, as the code hasn't
changed much).


Later,
David

---------------------------------------------------------------------
                    Check out my weekly web comic:
                     http://www.techmagazine.org


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

Re: question on best way to install other ide drives?

[Benjamin Herrenschmidt]

>
>(A).  IDE does not do a terribly good job of sharing bus bandwidth
>between two active devices, so the ideal case for performance is one
>IDE bus per device.
>
>That said, you'll never be impacted if your workload never attempts
>to exercise both the slave and master at the same time.

iirc, the sonet has 2 busses.

Ben.


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

Re: question on best way to install other ide drives?

[Tom Rini]

On Sun, Feb 17, 2002 at 10:52:04PM +0100, Benjamin Herrenschmidt wrote:
>
> >
> >(A).  IDE does not do a terribly good job of sharing bus bandwidth
> >between two active devices, so the ideal case for performance is one
> >IDE bus per device.
> >
> >That said, you'll never be impacted if your workload never attempts
> >to exercise both the slave and master at the same time.
>
> iirc, the sonet has 2 busses.

Yes, but unless you stress the master and slave at the same time, you'll
never notice.  I bash hda and hdd on my G4 all the time and it never
seems to be 'slow'.

--
Tom Rini (TR1265)
http://gate.crashing.org/~trini/

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

Weird bug problems with timing of NIC driver loading?

[Kevin B. Hendricks]

Hi,

I have a weird bug report I need some help tracking down:

With both the G3 using either tulip or  bmac NICs and the new G4 using
Sungmem I can reliably and repeatedly show funky net behavior when those
drivers are compiled in or loaded early in the boot process as modules.

This behavior is funny in that ifconfig shows no errors and that packets
are being sent and received (and the lights on the cards seem to support
that) but none of the received info ever seems to make it back upstream
from the card (a receive buffer alignment issue?)

This is repeatable with both machines and with BMAC, TULIP, and SUNGEM
when compiled in or loaded as a module during the normal eth0
initialization during bootup

If I simply compile them as modules and wait until the machine is up and
simply do an insmod and configure the network, they ALL work absolutely
perfectly.

So whatever the issue is, it seems to be related to when in the boot
process the NIC code is invoked.

Is this due to some change in memory mapping?
Is this due to some  change in IRQ assignment?
Is this due to some alignment issue with DMA buffers and memory / caches?

So what is different when a module is loaded by modprobe during the eth0
initialization during bootup versus waiting until the end and then
running insmod to load the module and configure the network.

Just to check I made sure there were no firewall modules loaded at all
and that the network routing tables and things were set properly
(identical to the hand done case at the end of boot-up).

This problem seems to exist in every 2.4.X kernel I have tried.

This problem does not exist in 2.2.X kernels.

Ideas anyone?

Thanks,

Kevin

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

Re: Weird bug problems with timing of NIC driver loading?

[Benjamin Herrenschmidt]

>
>Hi,
>
>I have a weird bug report I need some help tracking down:
>
>With both the G3 using either tulip or  bmac NICs and the new G4 using
>Sungmem I can reliably and repeatedly show funky net behavior when those
>drivers are compiled in or loaded early in the boot process as modules.
>
>This behavior is funny in that ifconfig shows no errors and that packets
>are being sent and received (and the lights on the cards seem to support
>that) but none of the received info ever seems to make it back upstream
>from the card (a receive buffer alignment issue?)
>
>This is repeatable with both machines and with BMAC, TULIP, and SUNGEM
>when compiled in or loaded as a module during the normal eth0
>initialization during bootup
>
>If I simply compile them as modules and wait until the machine is up and
>simply do an insmod and configure the network, they ALL work absolutely
>perfectly.

That is weird, you are the first person to report such a problem,
and since such a broad range of HW is affected, I'd rather blame some
other kernel routing problem, possibly some setup of your init scripts,
(or some ECN issue ?)

>So whatever the issue is, it seems to be related to when in the boot
>process the NIC code is invoked.
>
>Is this due to some change in memory mapping?

No, nothing here should matter.

>Is this due to some  change in IRQ assignment?

Neither. IRQ assignement isn't changed, it comes from the firmware
and works on all known HW.

>Is this due to some alignment issue with DMA buffers and memory / caches?

I don't think so, especially recent machines have no known cache
coherency problems.

>So what is different when a module is loaded by modprobe during the eth0
>initialization during bootup versus waiting until the end and then
>running insmod to load the module and configure the network.
>
>Just to check I made sure there were no firewall modules loaded at all
>and that the network routing tables and things were set properly
>(identical to the hand done case at the end of boot-up).
>
>This problem seems to exist in every 2.4.X kernel I have tried.
>
>This problem does not exist in 2.2.X kernels.

That's weird, it could well be an ecn problem. Do you have a switch
or a hub ? to what machine are you trying to talk to ? It really look
like a problem above the drivers.

Ben.


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