2.2.18-pre17 candidate

2.2.18-pre17 candidate

[Paul Mackerras]

I have been hacking on the 2.2 kernels for the last day or so.  I have
backported the ADB HID stuff so that you can use either the input
layer for ADB keyboards and mice, or the old mac_keyb.c and adbmouse.c
drivers.

To use the input layer, you currently need to say yes to the 'Support
for USB' and 'USB HID' questions and either (or both) of the 'Keyboard
support' and 'Mouse support' questions (CONFIG_INPUT_KEYBDEV and
CONFIG_INPUT_MOUSEDEV).  Then it will ask 'Use input layer for ADB
keyboard and mouse'.

If you say Y you get ADB keyboards and mice handled by the input
layer.  If you say N it will ask 'Support for ADB keyboard (old
driver)' and 'Support for ADB mouse (old driver)'.

The tree is available by anonymous rsync from these two sites:

penguinppc.org::linux-pmac-stable
ppc.linuxcare.com::linux-pmac-stable

I'll be interested to hear how people go with it.  I'll check the
changes into the bitkeeper linuxppc_2_2 tree within the next day or
so.  If it goes OK I'll send the patch to Alan for inclusion in
2.2.18.

Paul.

--
Paul Mackerras, Senior Open Source Researcher, Linuxcare, Inc.
+61 2 6262 8990 tel, +61 2 6262 8991 fax
paulus@linuxcare.com.au, http://www.linuxcare.com.au/
Linuxcare.  Support for the revolution.

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

Re: 2.2.18-pre17 candidate

[Andreas Tobler]

Paul Mackerras wrote:

> penguinppc.org::linux-pmac-stable
> ppc.linuxcare.com::linux-pmac-stable

synced a few minutes ago from penguinppc.org.

>
> I'll be interested to hear how people go with it.  I'll check the
> changes into the bitkeeper linuxppc_2_2 tree within the next day or
> so.  If it goes OK I'll send the patch to Alan for inclusion in
> 2.2.18.

My menuconfig breaks with the msg 'MCmenu36: command not found report it
to ......'. On the quick it's not very helpful I know.
I also wondered why 'Mac device drivers' is the last entry in the
menuconfig, no sound, nor Kernel hacking.

Andreas

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

Re: 2.2.18-pre17 candidate

[Martin Costabel]

Paul Mackerras wrote:
>
> I'll be interested to hear how people go with it.  I'll check the
> changes into the bitkeeper linuxppc_2_2 tree within the next day or
> so.  If it goes OK I'll send the patch to Alan for inclusion in
> 2.2.18.

I have a hard time trying to compile this.

First of all, you don't use CONFIG_MAGIC_SYSRQ, do you? If you do, you
find that there is confusion over whether it is "ppc_md.kbd_sysrq_xlate"
or "ppc_md.ppc_kbd_sysrq_xlate", and then there are a couple of
"SYSRQ_KEY = 0x69" instead of "ppc_md.SYSRQ_KEY = 0x69".

Having fixed the corresponding 6 or 8 files, compilation went through to
the end, but in the final "ld ... -o vmlinux", there were dozens of
undefined symbols, like:

arch/ppc/kernel/head.o: In function `setup_disp_bat':
arch/ppc/kernel/head.o(.text+0x47ea): undefined reference to `disp_BATL'
arch/ppc/kernel/head.o(.text+0x47fa): undefined reference to `disp_BATU'
arch/ppc/kernel/kernel.o: In function `motopenpic_to_irq':
arch/ppc/kernel/kernel.o(.text+0x8a28): undefined reference to
`openpic_to_irq'

The latter is found at many places. Further:

arch/ppc/kernel/kernel.o: In function `pmac_pic_init':
arch/ppc/kernel/kernel.o(.text.init+0x7f96): undefined reference to
`open_pic_do_IRQ'
arch/ppc/kernel/kernel.o(__ksymtab+0x388): undefined reference to
`pci_dev_io_base'
arch/ppc/kernel/kernel.o(__ksymtab+0x390): undefined reference to
`pci_dev_mem_base'
arch/ppc/kernel/kernel.o: In function `sys_call_table':
arch/ppc/kernel/kernel.o(.data+0x3198): undefined reference to
`sys_pciconfig_read'
arch/ppc/kernel/kernel.o(.data+0x319c): undefined reference to
`sys_pciconfig_write'
drivers/macintosh/macintosh.o: In function `pmu_ioctl':
drivers/macintosh/macintosh.o(.text+0x2b14): undefined reference to
`get_backlight_level'
drivers/macintosh/macintosh.o: In function `via_pmu_init':
drivers/macintosh/macintosh.o(.text.openfirmware+0x5ec): undefined
reference to `register_backlight_controller'

I may have some weird config options (in particular, CONFIG_PMAC_PBOOK=y
when I don't have a Pbook), but they worked before in 2.2.17, and they
work in 2.4.0.

--
Martin

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

Re: 2.2.18-pre17 candidate

[Paul Mackerras]

Andreas Tobler writes:

> My menuconfig breaks with the msg 'MCmenu36: command not found report it
> to ......'. On the quick it's not very helpful I know.
> I also wondered why 'Mac device drivers' is the last entry in the
> menuconfig, no sound, nor Kernel hacking.

Oops, missed an endmenu in arch/ppc/config.in.  Try rsyncing again
now.

Paul.

--
Paul Mackerras, Senior Open Source Researcher, Linuxcare, Inc.
+61 2 6262 8990 tel, +61 2 6262 8991 fax
paulus@linuxcare.com.au, http://www.linuxcare.com.au/
Linuxcare.  Support for the revolution.

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

Re: 2.2.18-pre17 candidate

[Paul Mackerras]

Martin Costabel writes:

> First of all, you don't use CONFIG_MAGIC_SYSRQ, do you? If you do, you

Not usually, no.  How did you guess?  :-)

> Having fixed the corresponding 6 or 8 files, compilation went through to
> the end, but in the final "ld ... -o vmlinux", there were dozens of
> undefined symbols, like:

Could you send me a copy of your .config?

Thanks,
Paul.

--
Paul Mackerras, Senior Open Source Researcher, Linuxcare, Inc.
+61 2 6262 8990 tel, +61 2 6262 8991 fax
paulus@linuxcare.com.au, http://www.linuxcare.com.au/
Linuxcare.  Support for the revolution.

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

Re: 2.2.18-pre17 candidate

[Andreas Tobler]

[-- Attachment #1: Type: text/plain, Size: 714 bytes --]

Paul Mackerras wrote:
>
> Andreas Tobler writes:
>
> > My menuconfig breaks with the msg 'MCmenu36: command not found report it
> > to ......'. On the quick it's not very helpful I know.
> > I also wondered why 'Mac device drivers' is the last entry in the
> > menuconfig, no sound, nor Kernel hacking.
>
> Oops, missed an endmenu in arch/ppc/config.in.  Try rsyncing again
> now.

Yup, make menuconfig works so far. But adbhid.c complains about linux/pmu.h,
adb.h and cuda.h missing. They are there but not in linux, they are in asm.


Build ok, no unresolved....
boots also ;-)

uname -ra:
 Linux wallstreet 2.2.18pre17 #5 Tue Oct 24 13:36:24 CEST 2000 ppc unknown



Thanks
Andreas

BTW, attached my dot.config.

[-- Attachment #2: dot.config.gz --]
[-- Type: application/x-gzip, Size: 3013 bytes --]

Is the difference in IRQ maps between 2.4/2.2 intentional?

[Vitaly Luban]

Hi folks !

Maybe anybody will enlighten me, is the difference between
IRQ maps initialization (arch/ppc/kernel/prep_pci.c) in 2.4
and 2.2 intentional or not? The difference is in that for

Mesquite_pci_IRQ_map,
Sitka_pci_IRQ_map,
MTX_pci_IRQ_map,
MTXplus_pci_IRQ_map and
Genesis2_pci_IRQ_map

some fields have been initialised for 2.2.16 and zero for 2.4.

Thanks,
    Vitaly.


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

Re: 2.2.18-pre17 candidate

[Olaf Hering]

[-- Attachment #1: Type: text/plain, Size: 522 bytes --]

On Tue, Oct 24, Paul Mackerras wrote:


> I'll be interested to hear how people go with it.  I'll check the
> changes into the bitkeeper linuxppc_2_2 tree within the next day or
> so.  If it goes OK I'll send the patch to Alan for inclusion in
> 2.2.18.

This is what I get whit a prep only config.
I will send the .config to you in a separate mail.

Is there a way to not compile the pmac_*.c stuff when pmac is not
defined? At least for 2.2?

Gruss Olaf

--
 $ man clone

BUGS
       Main feature not yet implemented...

[-- Attachment #2: build.log --]
[-- Type: text/plain, Size: 6644 bytes --]

arch/ppc/kernel/kernel.o: In function `pmac_set_rtc_time':
arch/ppc/kernel/kernel.o(.text+0x7dc6): undefined reference to `adb_controller'
arch/ppc/kernel/kernel.o(.text+0x7dca): undefined reference to `adb_controller'
arch/ppc/kernel/kernel.o(.text+0x7e18): undefined reference to `cuda_request'
arch/ppc/kernel/kernel.o(.text+0x7e18): relocation truncated to fit: R_PPC_REL24 cuda_request
arch/ppc/kernel/kernel.o(.text+0x7e28): undefined reference to `cuda_poll'
arch/ppc/kernel/kernel.o(.text+0x7e28): relocation truncated to fit: R_PPC_REL24 cuda_poll
arch/ppc/kernel/kernel.o(.text+0x7e64): undefined reference to `pmu_request'
arch/ppc/kernel/kernel.o(.text+0x7e64): relocation truncated to fit: R_PPC_REL24 pmu_request
arch/ppc/kernel/kernel.o(.text+0x7e74): undefined reference to `pmu_poll'
arch/ppc/kernel/kernel.o(.text+0x7e74): relocation truncated to fit: R_PPC_REL24 pmu_poll
arch/ppc/kernel/kernel.o: In function `pmac_restart':
arch/ppc/kernel/kernel.o(.text+0x8306): undefined reference to `adb_hardware'
arch/ppc/kernel/kernel.o(.text+0x830a): undefined reference to `adb_hardware'
arch/ppc/kernel/kernel.o(.text+0x8334): undefined reference to `cuda_request'
arch/ppc/kernel/kernel.o(.text+0x8334): relocation truncated to fit: R_PPC_REL24 cuda_request
arch/ppc/kernel/kernel.o(.text+0x8338): undefined reference to `cuda_poll'
arch/ppc/kernel/kernel.o(.text+0x8338): relocation truncated to fit: R_PPC_REL24 cuda_poll
arch/ppc/kernel/kernel.o(.text+0x8340): undefined reference to `pmu_restart'
arch/ppc/kernel/kernel.o(.text+0x8340): relocation truncated to fit: R_PPC_REL24 pmu_restart
arch/ppc/kernel/kernel.o: In function `pmac_power_off':
arch/ppc/kernel/kernel.o(.text+0x8366): undefined reference to `adb_hardware'
arch/ppc/kernel/kernel.o(.text+0x836a): undefined reference to `adb_hardware'
arch/ppc/kernel/kernel.o(.text+0x8394): undefined reference to `cuda_request'
arch/ppc/kernel/kernel.o(.text+0x8394): relocation truncated to fit: R_PPC_REL24 cuda_request
arch/ppc/kernel/kernel.o(.text+0x8398): undefined reference to `cuda_poll'
arch/ppc/kernel/kernel.o(.text+0x8398): relocation truncated to fit: R_PPC_REL24 cuda_poll
arch/ppc/kernel/kernel.o(.text+0x83a0): undefined reference to `pmu_shutdown'
arch/ppc/kernel/kernel.o(.text+0x83a0): relocation truncated to fit: R_PPC_REL24 pmu_shutdown
arch/ppc/kernel/kernel.o: In function `pmac_get_rtc_time':
arch/ppc/kernel/kernel.o(.text.pmac+0xe): undefined reference to `adb_controller'
arch/ppc/kernel/kernel.o(.text.pmac+0x12): undefined reference to `adb_controller'
arch/ppc/kernel/kernel.o(.text.pmac+0x48): undefined reference to `cuda_request'
arch/ppc/kernel/kernel.o(.text.pmac+0x48): relocation truncated to fit: R_PPC_REL24 cuda_request
arch/ppc/kernel/kernel.o(.text.pmac+0x58): undefined reference to `cuda_poll'
arch/ppc/kernel/kernel.o(.text.pmac+0x58): relocation truncated to fit: R_PPC_REL24 cuda_poll
arch/ppc/kernel/kernel.o(.text.pmac+0xb4): undefined reference to `pmu_request'
arch/ppc/kernel/kernel.o(.text.pmac+0xb4): relocation truncated to fit: R_PPC_REL24 pmu_request
arch/ppc/kernel/kernel.o(.text.pmac+0xd0): undefined reference to `pmu_poll'
arch/ppc/kernel/kernel.o(.text.pmac+0xd0): relocation truncated to fit: R_PPC_REL24 pmu_poll
arch/ppc/kernel/kernel.o: In function `nvram_read_byte':
arch/ppc/kernel/kernel.o(.text.pmac+0xb40): undefined reference to `pmu_request'
arch/ppc/kernel/kernel.o(.text.pmac+0xb40): relocation truncated to fit: R_PPC_REL24 pmu_request
arch/ppc/kernel/kernel.o(.text.pmac+0xb50): undefined reference to `pmu_poll'
arch/ppc/kernel/kernel.o(.text.pmac+0xb50): relocation truncated to fit: R_PPC_REL24 pmu_poll
arch/ppc/kernel/kernel.o: In function `nvram_write_byte':
arch/ppc/kernel/kernel.o(.text.pmac+0xc3c): undefined reference to `pmu_request'
arch/ppc/kernel/kernel.o(.text.pmac+0xc3c): relocation truncated to fit: R_PPC_REL24 pmu_request
arch/ppc/kernel/kernel.o(.text.pmac+0xc54): undefined reference to `pmu_poll'
arch/ppc/kernel/kernel.o(.text.pmac+0xc54): relocation truncated to fit: R_PPC_REL24 pmu_poll
arch/ppc/kernel/kernel.o: In function `register_backlight_controller':
arch/ppc/kernel/kernel.o(.text.pmac+0xf2c): undefined reference to `pmu_request'
arch/ppc/kernel/kernel.o(.text.pmac+0xf2c): relocation truncated to fit: R_PPC_REL24 pmu_request
arch/ppc/kernel/kernel.o(.text.pmac+0xf34): undefined reference to `pmu_poll'
arch/ppc/kernel/kernel.o(.text.pmac+0xf34): relocation truncated to fit: R_PPC_REL24 pmu_poll
arch/ppc/kernel/kernel.o: In function `scrollscreen':
arch/ppc/kernel/kernel.o(.text.pmac+0x1e90): undefined reference to `pmu_suspend'
arch/ppc/kernel/kernel.o(.text.pmac+0x1e90): relocation truncated to fit: R_PPC_REL24 pmu_suspend
arch/ppc/kernel/kernel.o(.text.pmac+0x1f40): undefined reference to `pmu_resume'
arch/ppc/kernel/kernel.o(.text.pmac+0x1f40): relocation truncated to fit: R_PPC_REL24 pmu_resume
arch/ppc/kernel/kernel.o: In function `chrp_init2':
arch/ppc/kernel/kernel.o(.text.init+0x764): undefined reference to `adb_init'
arch/ppc/kernel/kernel.o(.text.init+0x764): relocation truncated to fit: R_PPC_REL24 adb_init
arch/ppc/kernel/kernel.o: In function `pmac_setup_arch':
arch/ppc/kernel/kernel.o(.text.init+0x1d34): undefined reference to `find_via_cuda'
arch/ppc/kernel/kernel.o(.text.init+0x1d34): relocation truncated to fit: R_PPC_REL24 find_via_cuda
arch/ppc/kernel/kernel.o(.text.init+0x1d38): undefined reference to `find_via_pmu'
arch/ppc/kernel/kernel.o(.text.init+0x1d38): relocation truncated to fit: R_PPC_REL24 find_via_pmu
arch/ppc/kernel/kernel.o: In function `pmac_init2':
arch/ppc/kernel/kernel.o(.text.init+0x1ee4): undefined reference to `adb_init'
arch/ppc/kernel/kernel.o(.text.init+0x1ee4): relocation truncated to fit: R_PPC_REL24 adb_init
arch/ppc/kernel/kernel.o(.text.init+0x1ee8): undefined reference to `media_bay_init'
arch/ppc/kernel/kernel.o(.text.init+0x1ee8): relocation truncated to fit: R_PPC_REL24 media_bay_init
arch/ppc/kernel/kernel.o: In function `pmac_nvram_init':
arch/ppc/kernel/kernel.o(.text.init+0x231e): undefined reference to `adb_controller'
arch/ppc/kernel/kernel.o(.text.init+0x2322): undefined reference to `adb_controller'
drivers/char/char.o: In function `misc_init':
drivers/char/char.o(.text.init+0x4a4): undefined reference to `nvram_init'
drivers/char/char.o(.text.init+0x4a4): relocation truncated to fit: R_PPC_REL24 nvram_init
drivers/char/char.o: In function `chr_dev_init':
drivers/char/char.o(.text.init+0x14d4): undefined reference to `adbdev_init'
drivers/char/char.o(.text.init+0x14d4): relocation truncated to fit: R_PPC_REL24 adbdev_init
make: *** [vmlinux] Error 1

Re: Is the difference in IRQ maps between 2.4/2.2 intentional?

[Matt Porter]

On Tue, Oct 24, 2000 at 09:15:44AM -0700, Vitaly Luban wrote:
>
> Hi folks !
>
> Maybe anybody will enlighten me, is the difference between
> IRQ maps initialization (arch/ppc/kernel/prep_pci.c) in 2.4
> and 2.2 intentional or not? The difference is in that for
>
> Mesquite_pci_IRQ_map,
> Sitka_pci_IRQ_map,
> MTX_pci_IRQ_map,
> MTXplus_pci_IRQ_map and
> Genesis2_pci_IRQ_map
>
> some fields have been initialised for 2.2.16 and zero for 2.4.

Look at all of prep_pci.c and you'll see that many board specific
PCI changes were made in the linuxppc_2_2 tree that have not
yet been merged up to linuxppc_2_3.  It is in the plan, however.

The map changes are benign cruft from a set of changes I merged
in a long time ago.  It was a hack to place 0x1e value at the
PIB's IDSEL for some IDE fixups that aren't really necessary.

Now you know the whole story. :)

--
Matt Porter
MontaVista Software, Inc.
mporter@mvista.com

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

Re: 2.2.18-pre17 candidate

[Martin Costabel]

Paul Mackerras wrote:
>
> Martin Costabel writes:
>
> > First of all, you don't use CONFIG_MAGIC_SYSRQ, do you? If you do, you
>
> Not usually, no.  How did you guess?  :-)
>
> > Having fixed the corresponding 6 or 8 files, compilation went through to
> > the end, but in the final "ld ... -o vmlinux", there were dozens of
> > undefined symbols, like:

Sorry for the noise, my fault. I had some files in arch/ppc/kernel left
over from earlier rsync from the (now dead) hq.fsmlabs.com server that
were newer than your corresponding ones. After a careful cleanup (and
the sysrq stuff patch), it compiles now, and boots, and works. Including
new input layer like on 2.4.0 (tested only with ADB keyboard and ADB
keycodes). And MOL (patched for 603). And MAGIC_SYSRQ :-)

--
Martin

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

Re: 2.2.18-pre17 candidate

[Paul Mackerras]

Olaf Hering writes:

> Is there a way to not compile the pmac_*.c stuff when pmac is not
> defined? At least for 2.2?

Sure, we now have a CONFIG_POWERMAC which is defined if either
CONFIG_PMAC or CONFIG_ALL_PPC are defined, so we can make stuff
conditional on that.

Paul.

--
Paul Mackerras, Senior Open Source Researcher, Linuxcare, Inc.
+61 2 6262 8990 tel, +61 2 6262 8991 fax
paulus@linuxcare.com.au, http://www.linuxcare.com.au/
Linuxcare.  Support for the revolution.

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

linux-pmac-benh off (WAS: 2.2.18-pre17 candidate

[Benjamin Herrenschmidt]

Now that linux-pmac-stable (available from ppc.linuxcare.com or
penguinppc.org) contains all my recent changes, there's no point in using
my linux-pmac-benh tree.

So to everybody using it, you'd rather switch back to Paulus tree (and, I
hope, soon to official kernel, if our patches make it into 2.2.18 final).

Ben.


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

Re: 2.2.18-pre17 candidate

[Chris Leishman]

On Tue, Oct 24, 2000 at 05:29:33PM +1100, Paul Mackerras wrote:
<snip>
> I'll be interested to hear how people go with it.  I'll check the
> changes into the bitkeeper linuxppc_2_2 tree within the next day or
> so.  If it goes OK I'll send the patch to Alan for inclusion in
> 2.2.18.
>
> Paul.

If you compile the GMAC driver as a module it complains about an unresolvable
symbol "feature_set_gmac_phy_reset".  This is a symbol in the matching kernel,
so I'm not sure whats going on...

Also, when I tried using the new input stuff my mapping of right/middle mouse
buttons to F9 and F10 (use append="adb_buttons=101,109" in yaboot) stopped
working.

Chris

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

Re: 2.2.18-pre17 candidate

[Michael Schmitz]

> Also, when I tried using the new input stuff my mapping of right/middle mouse
> buttons to F9 and F10 (use append="adb_buttons=101,109" in yaboot) stopped
> working.

Which is expected. Use the proc interface to set the emulation keys (as
described in Franz' input layer docs IIRC).

	Michael


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