Power3 problem?

Power3 problem?

[Scott Carroll]

While building for Power3 (p640), ChangeSet@1.263, 2001-08-15:


-- make[1]: Leaving directory `/usr/src/latest_0815/net'
ld -T arch/ppc/vmlinux.lds -Ttext 0xc0000000 -Bstatic
arch/ppc/kernel/head.o init/main.o init/version.o \
        --start-group \
        arch/ppc/kernel/kernel.o arch/ppc/mm/mm.o arch/ppc/lib/lib.o
kernel/kernel.o mm/mm.o fs/fs.o ipc/ipc.o arch/ppc/xmon/x.o \
         drivers/char/char.o drivers/block/block.o drivers/misc/misc.o
drivers/net/net.o drivers/media/media.o drivers/net/tokenring/tr.o
drivers/ide/idedriver.o drivers/scsi/scsidrv.o drivers/cdrom/driver.o
drivers/pci/driver.o drivers/macintosh/macintosh.o drivers/video/video.o
drivers/usb/usbdrv.o drivers/input/inputdrv.o drivers/md/mddev.o \
        net/network.o \
        /usr/src/latest_0815/lib/lib.a \
        --end-group \
        -o vmlinux
arch/ppc/kernel/kernel.o: In function `proc_dol2crvec':
arch/ppc/kernel/kernel.o(.text+0x634c): undefined reference to
`_set_L2CR'
arch/ppc/kernel/kernel.o(.text+0x634c): relocation truncated to fit:
R_PPC_REL24 _set_L2CR
arch/ppc/kernel/kernel.o(.text+0x6378): undefined reference to
`_get_L2CR'
arch/ppc/kernel/kernel.o(.text+0x6378): relocation truncated to fit:
R_PPC_REL24 _get_L2CR
arch/ppc/kernel/kernel.o: In function `pmac_setup_arch':
arch/ppc/kernel/kernel.o(.text.init+0x2cec): undefined reference to
`_set_L2CR'
arch/ppc/kernel/kernel.o(.text.init+0x2cec): relocation truncated to
fit: R_PPC_REL24 _set_L2CR
arch/ppc/kernel/kernel.o(.text.init+0x2cf4): undefined reference to
`_set_L2CR'
arch/ppc/kernel/kernel.o(.text.init+0x2cf4): relocation truncated to
fit: R_PPC_REL24 _set_L2CR
arch/ppc/kernel/kernel.o: In function `core99_init_l2':
arch/ppc/kernel/kernel.o(.text.pmac+0x3e4): undefined reference to
`_get_L2CR'
arch/ppc/kernel/kernel.o(.text.pmac+0x3e4): relocation truncated to fit:
R_PPC_REL24 _get_L2CR
arch/ppc/kernel/kernel.o(.text.pmac+0x404): undefined reference to
`_get_L2CR'
arch/ppc/kernel/kernel.o(.text.pmac+0x404): relocation truncated to fit:
R_PPC_REL24 _get_L2CR
arch/ppc/kernel/kernel.o(.text.pmac+0x424): undefined reference to
`_set_L2CR'
arch/ppc/kernel/kernel.o(.text.pmac+0x424): relocation truncated to fit:
R_PPC_REL24 _set_L2CR
arch/ppc/kernel/kernel.o(.text.pmac+0x42c): undefined reference to
`_set_L2CR'
arch/ppc/kernel/kernel.o(.text.pmac+0x42c): relocation truncated to fit:
R_PPC_REL24 _set_L2CR
make: *** [vmlinux] Error 1
Scott Carroll                                  IBM RS/6000
Division
11400 Burnet Road, M.S. 9541       Notes:       scottcar@us.ibm.com
Austin, TX  78758-3493 U.S.A       phone:       +1 (512) 838-9642

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

Re: Power3 problem?

[Tom Rini]

On Wed, Aug 15, 2001 Scott Carroll wrote:

> While building for Power3 (p640), ChangeSet@1.263, 2001-08-15:

Arg!  I just fixed this, but the problem is this.  I took
arch/ppc/kernel/l2cr.o out of the always-compiled list, and made it
compile only for CONFIG_6xx (which is 6xx/7xx/74xx/82xx) since only
740/750 and 74xx (more or less) can make use of this.  The problem is
that Power3 and Power4 pretend to be CONFIG_ALL_PPC, so that they can
be a chrp machine easily.  The best way to fix this, I think would be to
make the arch/ppc/kernel/chrp_* files compile without needed pmac_*,
so that these machines could just work.  Paul, does this sound like
a sane plan to you?

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

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

Re: Power3 problem?

[Tom Gall]

Tom Rini wrote:
>
> On Wed, Aug 15, 2001 Scott Carroll wrote:
>
> > While building for Power3 (p640), ChangeSet@1.263, 2001-08-15:
>
> Arg!  I just fixed this, but the problem is this.  I took
> arch/ppc/kernel/l2cr.o out of the always-compiled list, and made it
> compile only for CONFIG_6xx (which is 6xx/7xx/74xx/82xx) since only
> 740/750 and 74xx (more or less) can make use of this.  The problem is
> that Power3 and Power4 pretend to be CONFIG_ALL_PPC, so that they can
> be a chrp machine easily.  The best way to fix this, I think would be to
> make the arch/ppc/kernel/chrp_* files compile without needed pmac_*,
> so that these machines could just work.  Paul, does this sound like
> a sane plan to you?

I agree, this sounds like a good idea. I suspect there's also room for
some of the routines in pmac_* to moved moved into more general files in
the cases they are truely pieces of shared code.

--
Regards,

Tom

--------------------------------------------------------------
Linux Guy             "My heart is human, my blood is boiling,
gall@rochcivictheatre.org    my brain IBM" -- Mr Roboto, Styxx

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

Re: Power3 problem?

[Paul Mackerras]

Tom Rini writes:

> Arg!  I just fixed this, but the problem is this.  I took
> arch/ppc/kernel/l2cr.o out of the always-compiled list, and made it
> compile only for CONFIG_6xx (which is 6xx/7xx/74xx/82xx) since only
> 740/750 and 74xx (more or less) can make use of this.  The problem is
> that Power3 and Power4 pretend to be CONFIG_ALL_PPC, so that they can

Well POWER3/4 aren't just pretending to be CONFIG_ALL_PPC, they *are*
CONFIG_ALL_PPC since they *are* CHRP machines.  Not all CHRPs are 6xx
(actually POWER3 is a PPC630 or PPC630+ so it *is* 6xx in a sense).

> be a chrp machine easily.  The best way to fix this, I think would be to
> make the arch/ppc/kernel/chrp_* files compile without needed pmac_*,
> so that these machines could just work.  Paul, does this sound like
> a sane plan to you?

Nope. :)  Not unless we want to start having separate configs for
prep, pmac and chrp again like we used to.  Which is not impossible
but it would be a fair bit of work.

The fix you pushed (to include l2cr.o for CONFIG_ALL_PPC) is fine.

Paul.

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

Re: Power3 problem?

[Tom Rini]

On Thu, Aug 16, 2001 at 09:15:03AM +1000, Paul Mackerras wrote:
> Tom Rini writes:
>
> > Arg!  I just fixed this, but the problem is this.  I took
> > arch/ppc/kernel/l2cr.o out of the always-compiled list, and made it
> > compile only for CONFIG_6xx (which is 6xx/7xx/74xx/82xx) since only
> > 740/750 and 74xx (more or less) can make use of this.  The problem is
> > that Power3 and Power4 pretend to be CONFIG_ALL_PPC, so that they can
>
> Well POWER3/4 aren't just pretending to be CONFIG_ALL_PPC, they *are*
> CONFIG_ALL_PPC since they *are* CHRP machines.  Not all CHRPs are 6xx
> (actually POWER3 is a PPC630 or PPC630+ so it *is* 6xx in a sense).

Well, pretending is a bad choice of words.  They are real chrp machines,
but they aren't 'ALL_PPC', since you can't run on a power3 and a 604 :)

> > be a chrp machine easily.  The best way to fix this, I think would be to
> > make the arch/ppc/kernel/chrp_* files compile without needed pmac_*,
> > so that these machines could just work.  Paul, does this sound like
> > a sane plan to you?
>
> Nope. :)  Not unless we want to start having separate configs for
> prep, pmac and chrp again like we used to.  Which is not impossible
> but it would be a fair bit of work.

Well, I don't think it will be as bad as that.  It _should_ be possible
to split them out, and move the functions it uses from pmac_setup.c into
a common file.  If I have time I'll play with it sometime and post patches.
If it can be done cleanly, I don't think it'd be a bad idea, but...

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

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

Re: Power3 problem?

[Paul Mackerras]

Tom Rini writes:

> Well, pretending is a bad choice of words.  They are real chrp machines,
> but they aren't 'ALL_PPC', since you can't run on a power3 and a 604 :)

No, but if you had a prep or pmac with a power3 cpu, you could run the
same vmlinux as on a chrp power3. :)

> Well, I don't think it will be as bad as that.  It _should_ be possible
> to split them out, and move the functions it uses from pmac_setup.c into
> a common file.  If I have time I'll play with it sometime and post patches.
> If it can be done cleanly, I don't think it'd be a bad idea, but...

There is an argument for splitting out prep but I would rather keep
pmac and chrp together since they are quite similar.

Paul.

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

Re: Power3 problem?

[Olaf Hering]

On Thu, Aug 16, Paul Mackeras wrote:

> Nope. :)  Not unless we want to start having separate configs for
> prep, pmac and chrp again like we used to.  Which is not impossible
> but it would be a fair bit of work.

oh no, we had that mess in 2.2 and I really like the current state, one
binary for all. Well, power3 is different, but...



Gruss Olaf

--
 $ man clone

BUGS
       Main feature not yet implemented...

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

Re: Power3 problem?

[Tom Rini]

On Thu, Aug 16, 2001 at 10:24:54AM +1000, Paul Mackerras wrote:
> Tom Rini writes:
>
> > Well, pretending is a bad choice of words.  They are real chrp machines,
> > but they aren't 'ALL_PPC', since you can't run on a power3 and a 604 :)
>
> No, but if you had a prep or pmac with a power3 cpu, you could run the
> same vmlinux as on a chrp power3. :)

Well, true...

> > Well, I don't think it will be as bad as that.  It _should_ be possible
> > to split them out, and move the functions it uses from pmac_setup.c into
> > a common file.  If I have time I'll play with it sometime and post patches.
> > If it can be done cleanly, I don't think it'd be a bad idea, but...
>
> There is an argument for splitting out prep but I would rather keep
> pmac and chrp together since they are quite similar.

I think there's lots of good reasons to either split or not split some things
out.  I also think we should probably wait until 2.5 to do any of it.
I also think we should discuss multi platform support a bit more.  I'll
start that in a different email in a bit..

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

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