[PATCH 0/1] m68k: add missing I/O macros {in,out}{w,l}_p() for !CONFIG_ISA

[PATCH 0/1] m68k: add missing I/O macros {in,out}{w,l}_p() for !CONFIG_ISA

[Thorsten Glaser]

Hi,

I’m currently trying to get stock Debian unstable linux-2.6 to build
on m68k for the TLS transition. Besides running into gcc-4.3 bugs
(e.g. Debian #598893) there is something mis with the code when
trying to build for ArANYm (atari).

The next mail will contain an attempt at a patch which I hope will
follow patch submission guidelines, however I don’t really know how
to send git diffs out so if this was wrong please tell me how to do
better next time (usually I just attach the 0001-* file to the mail).

As m68k, even emulated, is slow I have yet to test whether this is
indeed the fix, or if other patches are needed, but by submitting
eatly you can already review it. (I’m submitting here because Debian
says to only accept patches after upstream has done so.) Anyway, I’ll
write again once I got a kernel or have given up due to failures.

bye,
//mirabilos • tg@d.o
-- 
I believe no one can invent an algorithm. One just happens to hit upon it
when God enlightens him. Or only God invents algorithms, we merely copy them.
If you don't believe in God, just consider God as Nature if you won't deny
existence.		-- Coywolf Qi Hunt

[PATCH 1/1] m68k: add missing I/O macros {in,out}{w,l}_p() for !CONFIG_ISA

[Thorsten Glaser]

On m68k, I/O macros like inb() outw() etc. are only defined to
something useful if CONFIG_ISA is set; dummies are in place if
not, but four macros were missing from the !CONFIG_ISA case.
Adding these makes some drivers, such as speakup, compile again.
The patch was made against Debian’s linux-2.6_2.6.32-24, but
applies cleanly on linux-m68k master.

Signed-off-by: Thorsten Glaser <tg@debian.org>
---
 arch/m68k/include/asm/io_mm.h |    4 ++++
 1 files changed, 4 insertions(+), 0 deletions(-)

diff --git a/arch/m68k/include/asm/io_mm.h b/arch/m68k/include/asm/io_mm.h
index 56a6555..be9d7da 100644
--- a/arch/m68k/include/asm/io_mm.h
+++ b/arch/m68k/include/asm/io_mm.h
@@ -387,9 +387,13 @@ static inline void isa_delay(void)
 #define outb(val,port)     ((void)0)
 #define outb_p(val,port)   ((void)0)
 #define inw(port)          0xffff
+#define inw_p(port)        0xffff
 #define outw(val,port)     ((void)0)
+#define outw_p(val,port)   ((void)0)
 #define inl(port)          0xffffffffUL
+#define inl_p(port)        0xffffffffUL
 #define outl(val,port)     ((void)0)
+#define outl_p(val,port)   ((void)0)
 
 #define insb(port,buf,nr)  ((void)0)
 #define outsb(port,buf,nr) ((void)0)
-- 
1.6.0.4

Re: [PATCH 1/1] m68k: add missing I/O macros {in,out}{w,l}_p() for !CONFIG_ISA

[Finn Thain]

[-- Attachment #1: Type: TEXT/PLAIN, Size: 1434 bytes --]


On Sun, 3 Oct 2010, Thorsten Glaser wrote:

> On m68k, I/O macros like inb() outw() etc. are only defined to something 
> useful if CONFIG_ISA is set; dummies are in place if not, but four 
> macros were missing from the !CONFIG_ISA case. Adding these makes some 
> drivers, such as speakup, compile again. The patch was made against 
> Debian’s linux-2.6_2.6.32-24, but

I was able to compile that source package. Can you please send the .config 
you used?

Finn

> applies cleanly on linux-m68k master.
> 
> Signed-off-by: Thorsten Glaser <tg@debian.org>
> ---
>  arch/m68k/include/asm/io_mm.h |    4 ++++
>  1 files changed, 4 insertions(+), 0 deletions(-)
> 
> diff --git a/arch/m68k/include/asm/io_mm.h b/arch/m68k/include/asm/io_mm.h
> index 56a6555..be9d7da 100644
> --- a/arch/m68k/include/asm/io_mm.h
> +++ b/arch/m68k/include/asm/io_mm.h
> @@ -387,9 +387,13 @@ static inline void isa_delay(void)
>  #define outb(val,port)     ((void)0)
>  #define outb_p(val,port)   ((void)0)
>  #define inw(port)          0xffff
> +#define inw_p(port)        0xffff
>  #define outw(val,port)     ((void)0)
> +#define outw_p(val,port)   ((void)0)
>  #define inl(port)          0xffffffffUL
> +#define inl_p(port)        0xffffffffUL
>  #define outl(val,port)     ((void)0)
> +#define outl_p(val,port)   ((void)0)
>  
>  #define insb(port,buf,nr)  ((void)0)
>  #define outsb(port,buf,nr) ((void)0)
> 

Re: [PATCH 1/1] m68k: add missing I/O macros {in,out}{w,l}_p() for !CONFIG_ISA

[Thorsten Glaser]

Finn Thain dixit:

>On Sun, 3 Oct 2010, Thorsten Glaser wrote:

>> Debian’s linux-2.6_2.6.32-24, but
>
>I was able to compile that source package. Can you please send the .config 
>you used?

Erm, Debian’s? Otherwise I wouldn’t say “trying to build Debian’s
package”. (The atari variant, to be exact, in order to run ARAnyM.)
You might remember I’m sticking to stock unmodified Debian soruce,
native compilation, etc. as much as possible and then some.

Thing is, the macros are defined correctly if CONFIG_ISA is defined,
which appears to be the case for e.g. amiga but not atari.

bye,
//mirabilos
-- 
FWIW, I'm quite impressed with mksh interactively. I thought it was much
*much* more bare bones. But it turns out it beats the living hell out of
ksh93 in that respect. I'd even consider it for my daily use if I hadn't
wasted half my life on my zsh setup. :-) -- Frank Terbeck in #!/bin/mksh

Re: [PATCH 1/1] m68k: add missing I/O macros {in,out}{w,l}_p() for !CONFIG_ISA

[Finn Thain]

[-- Attachment #1: Type: TEXT/PLAIN, Size: 959 bytes --]


On Mon, 4 Oct 2010, Thorsten Glaser wrote:

> Finn Thain dixit:
> 
> >On Sun, 3 Oct 2010, Thorsten Glaser wrote:
> 
> >> Debian’s linux-2.6_2.6.32-24, but
> >
> >I was able to compile that source package. Can you please send the .config 
> >you used?
> 
> Erm, Debian’s? Otherwise I wouldn’t say “trying to build Debian’s
> package”.

My mistake. I compiled Debian's linux-2.6_2.6.32-23, to be precise.

> (The atari variant, to be exact, in order to run ARAnyM.) You might 
> remember I’m sticking to stock unmodified Debian soruce, native 
> compilation, etc. as much as possible and then some.
> 
> Thing is, the macros are defined correctly if CONFIG_ISA is defined, 
> which appears to be the case for e.g. amiga but not atari.

Perhaps I should try atari_defconfig and enable speakup. Or something 
else.

It would be better if we used the same .config, don't you think?

Finn

> 
> bye,
> //mirabilos
> 

Re: [PATCH 1/1] m68k: add missing I/O macros {in,out}{w,l}_p() for !CONFIG_ISA

[Thorsten Glaser]

Finn Thain dixit:

>It would be better if we used the same .config, don't you think?

*sigh*

# apt-get source linux-2.6
# cd linux-2.6-2.6.32

optionally:
# $EDITOR debian/config/m68k/defines
⇒ in [base] comment out everything except “flavours:” and “atari”
  to save time compiling the other images

then:
# debian/rules clean
(this may give you an error message)
# dpkg-buildpackage -B


So, no modifications to stock Debian. Well, except for #598893…

bye,
//mirabilos
-- 
[...] if maybe ext3fs wasn't a better pick, or jfs, or maybe reiserfs, oh but
what about xfs, and if only i had waited until reiser4 was ready... in the be-
ginning, there was ffs, and in the middle, there was ffs, and at the end, there
was still ffs, and the sys admins knew it was good. :)  -- Ted Unangst über *fs

Re: [PATCH 1/1] m68k: add missing I/O macros {in,out}{w,l}_p() for !CONFIG_ISA

[Finn Thain]

[-- Attachment #1: Type: TEXT/PLAIN, Size: 1263 bytes --]


It builds for me if I disable CONFIG_SPEAKUP. Otherwise, I get:

drivers/staging/speakup/speakup_dtlk.c: In function 'synth_probe':
drivers/staging/speakup/speakup_dtlk.c:339: error: implicit declaration of function 'inw_p'
make[3]: *** [drivers/staging/speakup/speakup_dtlk.o] Error 1
make[2]: *** [drivers/staging/speakup] Error 2
make[1]: *** [drivers/staging] Error 2
make: *** [drivers] Error 2

OTOH, I am using my own build of gcc 4.4.4-8 and none of this was built 
with apt, so YMMV...

I suppose another alternative would be to omit or reverse the patch that 
adds speakup in the first place, if that's more convenient.

Finn


On Mon, 4 Oct 2010, Thorsten Glaser wrote:

> Finn Thain dixit:
> 
> >It would be better if we used the same .config, don't you think?
> 
> *sigh*
> 
> # apt-get source linux-2.6
> # cd linux-2.6-2.6.32
> 
> optionally:
> # $EDITOR debian/config/m68k/defines
> ⇒ in [base] comment out everything except “flavours:” and “atari”
>   to save time compiling the other images
> 
> then:
> # debian/rules clean
> (this may give you an error message)
> # dpkg-buildpackage -B
> 
> 
> So, no modifications to stock Debian. Well, except for #598893…
> 
> bye,
> //mirabilos
> 

Re: [PATCH 1/1] m68k: add missing I/O macros {in,out}{w,l}_p() for !CONFIG_ISA

[Thorsten Glaser]

Finn Thain dixit:

>It builds for me if I disable CONFIG_SPEAKUP.

Yes, but I don’t do that because Debian enables it.
(Also, other drivers use inw_p as well.)

>I suppose another alternative would be to omit or reverse the patch that 
>adds speakup in the first place, if that's more convenient.

Definitively not:

>> So, no modifications to stock Debian. Well, except for #598893…

bye,
//mirabilos
-- 
I believe no one can invent an algorithm. One just happens to hit upon it
when God enlightens him. Or only God invents algorithms, we merely copy them.
If you don't believe in God, just consider God as Nature if you won't deny
existence.		-- Coywolf Qi Hunt

Re: [PATCH 1/1] m68k: add missing I/O macros {in,out}{w,l}_p() for !CONFIG_ISA

[Geert Uytterhoeven]

On Mon, Oct 4, 2010 at 12:22, Finn Thain <fthain@telegraphics.com.au> wrote:
> It builds for me if I disable CONFIG_SPEAKUP. Otherwise, I get:
>
> drivers/staging/speakup/speakup_dtlk.c: In function 'synth_probe':
> drivers/staging/speakup/speakup_dtlk.c:339: error: implicit declaration of function 'inw_p'
> make[3]: *** [drivers/staging/speakup/speakup_dtlk.o] Error 1
> make[2]: *** [drivers/staging/speakup] Error 2
> make[1]: *** [drivers/staging] Error 2
> make: *** [drivers] Error 2
>
> OTOH, I am using my own build of gcc 4.4.4-8 and none of this was built
> with apt, so YMMV...
>
> I suppose another alternative would be to omit or reverse the patch that
> adds speakup in the first place, if that's more convenient.

I don't think it makes sense to build staging drivers on Debian/m68k.
So can't you default to CONFIG_STAGING=n?

While I do enable them in my test builds to catch portability issues, most of
them don't matter for m68k at all.

> On Mon, 4 Oct 2010, Thorsten Glaser wrote:
>> Finn Thain dixit:
>>
>> >It would be better if we used the same .config, don't you think?
>>
>> *sigh*
>>
>> # apt-get source linux-2.6
>> # cd linux-2.6-2.6.32
>>
>> optionally:
>> # $EDITOR debian/config/m68k/defines
>> ⇒ in [base] comment out everything except “flavours:” and “atari”
>>   to save time compiling the other images
>>
>> then:
>> # debian/rules clean
>> (this may give you an error message)
>> # dpkg-buildpackage -B
>>
>>
>> So, no modifications to stock Debian. Well, except for #598893…

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

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

Re: [PATCH 1/1] m68k: add missing I/O macros {in,out}{w,l}_p() for !CONFIG_ISA

[Finn Thain]

On Mon, 4 Oct 2010, Geert Uytterhoeven wrote:

> 
> I don't think it makes sense to build staging drivers on Debian/m68k. So 
> can't you default to CONFIG_STAGING=n?

It appears that this could be done by adding
"# CONFIG_STAGING is not set"
to debian/config/m68k/config
But this is probably a discussion for debian-m68k...

Finn

Re: [PATCH 1/1] m68k: add missing I/O macros {in,out}{w,l}_p() for !CONFIG_ISA

[Thorsten Glaser]

Geert Uytterhoeven dixit:

>I don't think it makes sense to build staging drivers on Debian/m68k.

Yes. Tell that to *them*, though.

>So can't you default to CONFIG_STAGING=n?

No *I* can’t.

Besides, this still looks like an obvious omission to me.
Look a few lines above, the macros are defined in the other
case (CONFIG_ISA defined).

>Gr{oetje,eeting}s,

Terug,
//mirabilos
-- 
15:39⎜«mika:#grml» mira|AO: "mit XFree86® wär’ das nicht passiert" - muhaha
15:48⎜<thkoehler:#grml> also warum machen die xorg Jungs eigentlich alles
kaputt? :)    15:49⎜<novoid:#grml> thkoehler: weil sie als Kinder nie den
gebauten Turm selber umschmeissen durften?	-- ~/.Xmodmap wonders…

Re: [PATCH 1/1] m68k: add missing I/O macros {in,out}{w,l}_p() for !CONFIG_ISA

[Geert Uytterhoeven]

On Mon, Oct 4, 2010 at 15:40, Thorsten Glaser <tg@mirbsd.de> wrote:
> Geert Uytterhoeven dixit:
>
>>I don't think it makes sense to build staging drivers on Debian/m68k.
>
> Yes. Tell that to *them*, though.
>
>>So can't you default to CONFIG_STAGING=n?
>
> No *I* can’t.
>
> Besides, this still looks like an obvious omission to me.
> Look a few lines above, the macros are defined in the other
> case (CONFIG_ISA defined).

I'm not questioning they're missing.

I am questioning a distro building staging drivers for m68k.
IMHO it's a waste of CPU cycles (especially for native builds).

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

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

Re: [PATCH 1/1] m68k: add missing I/O macros {in,out}{w,l}_p() for !CONFIG_ISA

[Thorsten Glaser]

Geert Uytterhoeven <geert <at> linux-m68k.org> writes:

> I am questioning a distro building staging drivers for m68k.
> IMHO it's a waste of CPU cycles (especially for native builds).

Yeah. However, I don’t speak for the kernel team. I’m “just”
someone playing buildd, until they can do that automatedly
again (after getting new kernel+gcc+eglibc). It’s annoying
to me, too (just how much longer do I need to wait?).

bye,
//mirabilos
-- 
«MyISAM tables -will- get corrupted eventually. This is a fact of life. »
“mysql is about as much database as ms access” – “MSSQL at least descends
from a database” “it's a rebranded SyBase” “MySQL however was born from a
flatfile and went downhill from there” – “at least jetDB doesn’t claim to
be a database”  (#nosec)    ‣‣‣ Please let MySQL and MariaDB finally die!

Re: [PATCH 1/1] m68k: add missing I/O macros {in,out}{w,l}_p() for !CONFIG_ISA

[Thorsten Glaser]

Geert Uytterhoeven dixit:

>I'm not questioning they're missing.

There are other, non-staging, drivers using them as well.

Anyway.
-rw-r--r-- 1 root root 17251 Oct  4 17:43 linux-2.6_2.6.32-24+m68k_m68k.changes

So, with that one patch, I can successfully build a kernel.
Now, let’s see whether it boots… then, after you accepted that
patch, I can submit that to the Debian kernel people.

bye,
//mirabilos
-- 
I believe no one can invent an algorithm. One just happens to hit upon it
when God enlightens him. Or only God invents algorithms, we merely copy them.
If you don't believe in God, just consider God as Nature if you won't deny
existence.		-- Coywolf Qi Hunt

Re: [PATCH 1/1] m68k: add missing I/O macros {in,out}{w,l}_p() for !CONFIG_ISA

[Thorsten Glaser]

Geert Uytterhoeven dixit:

>> Besides, this still looks like an obvious omission to me.
>> Look a few lines above, the macros are defined in the other
>> case (CONFIG_ISA defined).
>
>I'm not questioning they're missing.

I can’t help but notice the patch hasn’t yet shown up in the
gitweb (but Stephen committed it to Debian already, thanks).

//mirabilos, currently trying to build again
-- 
I believe no one can invent an algorithm. One just happens to hit upon it
when God enlightens him. Or only God invents algorithms, we merely copy them.
If you don't believe in God, just consider God as Nature if you won't deny
existence.		-- Coywolf Qi Hunt

Re: [PATCH 1/1] m68k: add missing I/O macros {in,out}{w,l}_p() for !CONFIG_ISA

[Stephen R. Marenka]

On Mon, October 4, 2010 8:40 am, Thorsten Glaser wrote:
> Geert Uytterhoeven dixit:
>
>>I don't think it makes sense to build staging drivers on Debian/m68k.
>
> Yes. Tell that to *them*, though.
>
>>So can't you default to CONFIG_STAGING=n?
>
> No *I* can’t.

But I can. Feel free to ask to join the debian-kernel team for m68k. After
all, they let me have access. ;-)

> Besides, this still looks like an obvious omission to me.
> Look a few lines above, the macros are defined in the other
> case (CONFIG_ISA defined).

I'll apply the patch too.

I'll try to get it done in the next day or two when I can get to a
computer with appropriate access.

Peace,

Stephen

-- 
Stephen R. Marenka     If life's not fun, you're not doing it right!
<stephen@marenka.net>

debian kernel config, was Re: [PATCH 1/1] m68k: add missing I/O macros {in,out}{w,l}_p() for !CONFIG_ISA

[Finn Thain]

On Tue, 5 Oct 2010, Stephen R. Marenka wrote:

> On Mon, October 4, 2010 8:40 am, Thorsten Glaser wrote:
>
> > Geert Uytterhoeven dixit:
> >
> > > I don't think it makes sense to build staging drivers on Debian/m68k.
> >
> > Yes. Tell that to *them*, though.
> >
> > > So can't you default to CONFIG_STAGING=n?
> >
> > No *I* can't.
> 
> But I can.

I don't know whether this driver works or not, but the mac config should 
probably add:

CONFIG_BLK_DEV_SWIM=m

And while we are on the subject of configs... ;-)

# CONFIG_CAN is not set
# CONFIG_WIRELESS is not set
# CONFIG_WIMAX is not set
# CONFIG_RFKILL is not set
# CONFIG_C2PORT is not set
# CONFIG_EEPROM_93CX6 is not set
# CONFIG_IDE_GD_ATAPI is not set
# CONFIG_SCSI_FC_ATTRS is not set
# CONFIG_SCSI_SAS_LIBSAS is not set
# CONFIG_LIBFC is not set
# CONFIG_MII is not set
# CONFIG_ETHOC is not set
# CONFIG_DNET is not set
# CONFIG_KS8842 is not set
# CONFIG_KS8851_MLL is not set
# CONFIG_NETDEV_1000 is not set
# CONFIG_NETDEV_10000 is not set
# CONFIG_WLAN is not set
# CONFIG_IEEE802154_DRIVERS is not set
# CONFIG_INPUT_FF_MEMLESS is not set
# CONFIG_KEYBOARD_OPENCORES is not set
# CONFIG_KEYBOARD_STOWAWAY is not set
# CONFIG_MOUSE_PS2 is not set
# CONFIG_TCG_TPM is not set
# CONFIG_SPI is not set
# CONFIG_PPS is not set
# CONFIG_HWMON is not set
# CONFIG_SSB is not set
# CONFIG_MFD_SM501 is not set
# CONFIG_HTC_PASIC3 is not set
# CONFIG_REGULATOR is not set
# CONFIG_MEDIA_SUPPORT is not set
# CONFIG_FIRMWARE_EDID is not set
# CONFIG_FB_METRONOME is not set
# CONFIG_FB_MB862XX is not set
# CONFIG_BACKLIGHT_LCD_SUPPORT is not set
# CONFIG_DISPLAY_SUPPORT is not set
# CONFIG_HID_PID is not set
# CONFIG_USB_SUPPORT is not set
# CONFIG_MEMSTICK is not set
# CONFIG_NEW_LEDS is not set
# CONFIG_UIO is not set

The mac config for 2.6.32-23 has all of the above enabled, but I think 
none are applicable.

OTOH, it might be less work for the Debian kernel team if instead we only 
picked the low fruit from this list and then disable for all Debian/m68k 
kernels what is inappropriate for all Debian/m68k kernels...

Finn

Re: [PATCH 1/1] m68k: add missing I/O macros {in,out}{w,l}_p() for !CONFIG_ISA

[Stephen R. Marenka]

On Tue, October 5, 2010 7:56 am, Stephen R. Marenka wrote:
>
> On Mon, October 4, 2010 8:40 am, Thorsten Glaser wrote:
>> Geert Uytterhoeven dixit:
>>
>>>I don't think it makes sense to build staging drivers on Debian/m68k.
>>
>> Yes. Tell that to *them*, though.
>>
>>>So can't you default to CONFIG_STAGING=n?
>>
>> No *I* can’t.
>
> But I can. Feel free to ask to join the debian-kernel team for m68k. After
> all, they let me have access. ;-)
>
>> Besides, this still looks like an obvious omission to me.
>> Look a few lines above, the macros are defined in the other
>> case (CONFIG_ISA defined).
>
> I'll apply the patch too.
>
> I'll try to get it done in the next day or two when I can get to a
> computer with appropriate access.

Done in -25.

Peace,

Stephen

-- 
Stephen R. Marenka     If life's not fun, you're not doing it right!
<stephen@marenka.net>

Re: [PATCH 1/1] m68k: add missing I/O macros {in,out}{w,l}_p() for !CONFIG_ISA

[Thorsten Glaser]

Stephen R. Marenka dixit:

>>>So can't you default to CONFIG_STAGING=n?
>>
>> No *I* can’t.
>
>But I can. Feel free to ask to join the debian-kernel team for m68k. After
>all, they let me have access. ;-)

Mh. But this is a for-fun weekend “job” for me, and I don’t know
enough about Linux. It’s too fast-paced, too. So thanks but nope.
(Besides, building a kernel is only secondary to what I do.)

>I'll apply the patch too.

Thanks, but #599121 also needs love… waldi said that m68k has
been ignored during even not-so-recent cleanup actions; the
ARAnyM RTC driver added after my last “report” probably vanished
too (hwclock still says it can’t access it).

bye,
//mirabilos
-- 
I believe no one can invent an algorithm. One just happens to hit upon it
when God enlightens him. Or only God invents algorithms, we merely copy them.
If you don't believe in God, just consider God as Nature if you won't deny
existence.		-- Coywolf Qi Hunt

Re: [PATCH 1/1] m68k: add missing I/O macros {in,out}{w,l}_p() for !CONFIG_ISA

[Stephen R. Marenka]

On Wed, October 6, 2010 4:26 am, Thorsten Glaser wrote:
> Stephen R. Marenka dixit:
>
>>>>So can't you default to CONFIG_STAGING=n?
>>>
>>> No *I* can’t.
>>
>>But I can. Feel free to ask to join the debian-kernel team for m68k.
>> After
>>all, they let me have access. ;-)
>
> Mh. But this is a for-fun weekend “job” for me, and I don’t know
> enough about Linux. It’s too fast-paced, too. So thanks but nope.

Hey now worries, it's not my day-job either. I don't blame you, but it
would be nice to have a backup person.

> (Besides, building a kernel is only secondary to what I do.)

I only needed the kernels to get d-i working and to support my buildd habit.

>>I'll apply the patch too.
>
> Thanks, but #599121 also needs love… waldi said that m68k has
> been ignored during even not-so-recent cleanup actions; the
> ARAnyM RTC driver added after my last “report” probably vanished
> too (hwclock still says it can’t access it).

Did those patches ever go upstream of linux-m68k? I recall that some of
the aranym patches were sitting there. They were certainly on the list of
patches I was adding to the debian kernel at some point in the past.

Peace,

Stephen

-- 
Stephen R. Marenka     If life's not fun, you're not doing it right!
<stephen@marenka.net>

Re: [PATCH 1/1] m68k: add missing I/O macros {in,out}{w,l}_p() for !CONFIG_ISA

[Finn Thain]

[-- Attachment #1: Type: TEXT/PLAIN, Size: 878 bytes --]


On Wed, 6 Oct 2010, Thorsten Glaser wrote:

> the ARAnyM RTC driver added after my last “report” probably vanished too 
> (hwclock still says it can’t access it).

I gather that atari uses CONFIG_RTC_DRV_CMOS which needs CONFIG_RTC_CLASS. 
The Debian 2.6.32-23 configs lack these:

linux-2.6-2.6.32$ grep -r RTC_CLASS debian/config/
debian/config/config:CONFIG_RTC_CLASS=y
debian/config/m68k/config.atari:# CONFIG_RTC_CLASS is not set
debian/config/sh4/config.sh7785lcr:CONFIG_RTC_CLASS=y
debian/config/sh4/config.sh7751r:CONFIG_RTC_CLASS=y
debian/config/hppa/config:# CONFIG_RTC_CLASS is not set
linux-2.6-2.6.32$ 

That's about all I can tell from looking at the sources. Hopefully someone 
who knows something about atari/aranym can say more. This link might jog 
some memories: 
http://permalink.gmane.org/gmane.linux.debian.ports.68k/9157

Finn

Re: [PATCH 1/1] m68k: add missing I/O macros {in,out}{w,l}_p() for !CONFIG_ISA

[Michael Schmitz]

Hi,

please take care not to disturb these macros for any other case than
the default one (nothing related to ISA emulation configured at all).

Are you certain the speakup driver has a chance to work, once you got
it to compile?

Cheers,

  Michael


On Sun, Oct 3, 2010 at 8:48 PM, Thorsten Glaser <tg@debian.org> wrote:
> On m68k, I/O macros like inb() outw() etc. are only defined to
> something useful if CONFIG_ISA is set; dummies are in place if
> not, but four macros were missing from the !CONFIG_ISA case.
> Adding these makes some drivers, such as speakup, compile again.
> The patch was made against Debian’s linux-2.6_2.6.32-24, but
> applies cleanly on linux-m68k master.
>
> Signed-off-by: Thorsten Glaser <tg@debian.org>
> ---
>  arch/m68k/include/asm/io_mm.h |    4 ++++
>  1 files changed, 4 insertions(+), 0 deletions(-)
>
> diff --git a/arch/m68k/include/asm/io_mm.h b/arch/m68k/include/asm/io_mm.h
> index 56a6555..be9d7da 100644
> --- a/arch/m68k/include/asm/io_mm.h
> +++ b/arch/m68k/include/asm/io_mm.h
> @@ -387,9 +387,13 @@ static inline void isa_delay(void)
>  #define outb(val,port)     ((void)0)
>  #define outb_p(val,port)   ((void)0)
>  #define inw(port)          0xffff
> +#define inw_p(port)        0xffff
>  #define outw(val,port)     ((void)0)
> +#define outw_p(val,port)   ((void)0)
>  #define inl(port)          0xffffffffUL
> +#define inl_p(port)        0xffffffffUL
>  #define outl(val,port)     ((void)0)
> +#define outl_p(val,port)   ((void)0)
>
>  #define insb(port,buf,nr)  ((void)0)
>  #define outsb(port,buf,nr) ((void)0)
> --
> 1.6.0.4
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-m68k" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>

Re: [PATCH 1/1] m68k: add missing I/O macros {in,out}{w,l}_p() for !CONFIG_ISA

[Thorsten Glaser]

Michael Schmitz dixit:

>please take care not to disturb these macros for any other case than
>the default one (nothing related to ISA emulation configured at all).

The chunk is inside the !CONFIG_ISA part.

>Are you certain the speakup driver has a chance to work, once you got
>it to compile?

Not much, except for the software one. But it’s not the only one, only
the first one, to use the *_p() macros, and given they’re defined in
the CONFIG_ISA case and _some_ are defined for !CONFIG_ISA, this looks
like an obvious omission to me.

Besides, it’s easier to get a bunch of modules that don’t do anything
except say the hardware they steer isn’t available to compile than to
change the config (especially as the latter potentially has to be done
for every new kernel version, and I’d rather keep things working, as
they seem to change very quickly and nobody can keep up really).

bye,
//mirabilos
-- 
“It is inappropriate to require that a time represented as
 seconds since the Epoch precisely represent the number of
 seconds between the referenced time and the Epoch.”
	-- IEEE Std 1003.1b-1993 (POSIX) Section B.2.2.2

Re: [PATCH 1/1] m68k: add missing I/O macros {in,out}{w,l}_p() for !CONFIG_ISA

[Finn Thain]

[-- Attachment #1: Type: TEXT/PLAIN, Size: 938 bytes --]


On Mon, 4 Oct 2010, Thorsten Glaser wrote:

> Besides, it’s easier to get a bunch of modules that don’t do anything 
> except say the hardware they steer isn’t available to compile than to 
> change the config

Which modules are we talking about? In 2.6.32-23, debian patches speakup 
into drivers/staging but I don't see it in mainline drivers/accessibility 
or drivers/staging.

If the bug is not in the mainline, then the fix should arguably be 
downstream too. (Policy seems to be that the mainline doesn't carry code 
that doesn't have in-tree users.)

> (especially as the latter potentially has to be done for every new 
> kernel version, and I’d rather keep things working, as they seem to 
> change very quickly and nobody can keep up really).

Is it easier to maintain local patches from one debian release to another 
than it is to maintain .config changes?

Finn

> 
> bye,
> //mirabilos
> 

Re: [PATCH 1/1] m68k: add missing I/O macros {in,out}{w,l}_p() for !CONFIG_ISA

[Thorsten Glaser]

Finn Thain dixit:

>Which modules are we talking about? In 2.6.32-23, debian patches speakup 
>into drivers/staging

Yes. They are, however, compiled. This is a problem for me. At the
moment, the one important thing for me is to get this &*#)*^&*(
stuff to compile so we can go on with migration to TLS. I cannot
upload modified source to unstable, so patches must go into the
official Debian packages. Patches are only accepted if they are
accepted by “someone” – so if they live in linux-m68k for a while
it’s okay.

http://kernel-handbook.alioth.debian.org/ch-source.html#s-acceptance

Besides, this seems like a simple fix or an accidental omission
to me. What is with all this bikeshedding? Some macros are not
available on m68k if CONFIG_ISA is not set, but they are available
“almost” or even really everywhere else, so drivers use them, so
m68k is buggy in not providing them. (Outsider’s PoV.)

bye,
//mirabilos
-- 
16:47⎜«mika:#grml» .oO(mira ist einfach gut....)      23:22⎜«mikap:#grml»
mirabilos: und dein bootloader ist geil :)    23:29⎜«mikap:#grml» und ich
finds saugeil dass ich ein bsd zum booten mit grml hab, das muss ich dann
gleich mal auf usb-stick installieren	-- Michael Prokop über MirOS bsd4grml

Re: [PATCH 1/1] m68k: add missing I/O macros {in,out}{w,l}_p() for !CONFIG_ISA

[Finn Thain]

[-- Attachment #1: Type: TEXT/PLAIN, Size: 1283 bytes --]


On Mon, 4 Oct 2010, Thorsten Glaser wrote:

> Finn Thain dixit:
> 
> Besides, this seems like a simple fix or an accidental omission to me. 
> What is with all this bikeshedding?

It isn't bikeshedding. Recall that I built these sources a few weeks ago 
when you mentioned the speakup issue in private correspondence. At the 
time, I built speakup in 2.6.32-23 to see if it would behave differently 
than the 2.6.3[45] sources that were giving you grief.

It built for me at the time, so I was suprised to find out it didn't build 
for you, and naturally I wanted to know what I did wrong.

It may be that speakup works on amiga and atari (I don't know). It doesn't 
work on mac as there are no audio drivers. It has long frustrated me that 
there is bloat in the debian kernel configs. So perhaps I am guilty of 
being unfairly obstructionist here, in the hope that the config file 
issues could be addressed. My apologies.

> Some macros are not available on m68k if CONFIG_ISA is not set, but they 
> are available “almost” or even really everywhere else, so drivers use 
> them, so m68k is buggy in not providing them. (Outsider’s PoV.)

You may be right. I confess to being ignorant when it comes to ISA.

Finn

> 
> bye,
> //mirabilos
> 

Re: [PATCH 1/1] m68k: add missing I/O macros {in,out}{w,l}_p() for !CONFIG_ISA

[Thorsten Glaser]

Finn Thain dixit:

>It built for me at the time, so I was suprised to find out it didn't build
>for you, and naturally I wanted to know what I did wrong.

Well, we found out it didn’t build for you now too.

>It may be that speakup works on amiga and atari (I don't know). It doesn't
>work on mac as there are no audio drivers. It has long frustrated me that

There’s a software backend. (You could probably use network audio.)

>there is bloat in the debian kernel configs. So perhaps I am guilty of

Yes, there is bloat. However, to remove *that* would be even more
difficult to both do and find reasons for, is *WAY* beyong my
knowledge; remember I’m first and foremost a BSD developer.

So, we have a speakup driver broken without¹ the patch I submitted.

>being unfairly obstructionist here, in the hope that the config file
>issues could be addressed. My apologies.

I don’t even *want* to go there. I do believe I’d fail, and I want
to focus on getting enough packages built to be able to use eglibc
with TLS and cowbuilder, which includes having a newer kernel only
as side effect (and the kernel is said to be the least problematic
part of them all… *sigh*).

>You may be right. I confess to being ignorant when it comes to ISA.

I’m ignorant regarding both the Linux kernel and m68k, which may be
a reason for different views. And bringing in fresh wind/blood…

Anyway, it appears to build with the exact patch I submitted applied:

[…]
  CC [M]  drivers/ssb/driver_chipcommon_pmu.o
  LD [M]  drivers/ssb/ssb.o
  CC      drivers/staging/staging.o
  LD      drivers/staging/built-in.o
  LD      drivers/staging/speakup/built-in.o
  CC [M]  drivers/staging/speakup/speakup_acntsa.o
  CC [M]  drivers/staging/speakup/speakup_acntpc.o
  CC [M]  drivers/staging/speakup/speakup_apollo.o
  CC [M]  drivers/staging/speakup/speakup_audptr.o
  CC [M]  drivers/staging/speakup/speakup_bns.o
  CC [M]  drivers/staging/speakup/speakup_dectlk.o
  CC [M]  drivers/staging/speakup/speakup_decext.o
  CC [M]  drivers/staging/speakup/speakup_dtlk.o
  CC [M]  drivers/staging/speakup/speakup_keypc.o
  CC [M]  drivers/staging/speakup/speakup_ltlk.o
  CC [M]  drivers/staging/speakup/speakup_soft.o
  CC [M]  drivers/staging/speakup/speakup_spkout.o
  CC [M]  drivers/staging/speakup/speakup_txprt.o
  CC [M]  drivers/staging/speakup/speakup_dummy.o
  CC [M]  drivers/staging/speakup/buffers.o
  CC [M]  drivers/staging/speakup/devsynth.o
  CC [M]  drivers/staging/speakup/i18n.o
  CC [M]  drivers/staging/speakup/fakekey.o
  CC [M]  drivers/staging/speakup/main.o
  CC [M]  drivers/staging/speakup/keyhelp.o
  CC [M]  drivers/staging/speakup/kobjects.o
  CC [M]  drivers/staging/speakup/selection.o
  CC [M]  drivers/staging/speakup/serialio.o
/root/linux-2.6-2.6.32/debian/build/source_m68k_none/drivers/staging/speakup/serialio.c: In function ‘spk_serial_init’:                                                                                                 /root/linux-2.6-2.6.32/debian/build/source_m68k_none/drivers/staging/speakup/serialio.c:44: warning: format ‘%x’ expects type ‘unsigned int’, but argument 2 has type ‘long unsigned int’
/root/linux-2.6-2.6.32/debian/build/source_m68k_none/drivers/staging/speakup/serialio.c: In function ‘start_serial_interrupt’:                                                                                          /root/linux-2.6-2.6.32/debian/build/source_m68k_none/drivers/staging/speakup/serialio.c:112: warning: statement with no effect
/root/linux-2.6-2.6.32/debian/build/source_m68k_none/drivers/staging/speakup/serialio.c:113: warning: statement with no effect
/root/linux-2.6-2.6.32/debian/build/source_m68k_none/drivers/staging/speakup/serialio.c:114: warning: statement with no effect
/root/linux-2.6-2.6.32/debian/build/source_m68k_none/drivers/staging/speakup/serialio.c:115: warning: statement with no effect
  CC [M]  drivers/staging/speakup/synth.o
  CC [M]  drivers/staging/speakup/thread.o
  CC [M]  drivers/staging/speakup/varhandlers.o
  LD [M]  drivers/staging/speakup/speakup.o
  LD      drivers/uio/built-in.o
  CC [M]  drivers/uio/uio.o
[…]

This lets me think I’m doing something right here. Still building…

[…]
  CC [M]  net/ipv6/xfrm6_input.o

This is where I am at right now, and I wonder how much longer it will
take. However, I somewhat feel proud to get so far even with Debian’s
admitted bloat, which on the other hand is something I’d *expect* from
a general-purpose distribution.

bye,
//mirabilos
-- 
> Hi, does anyone sell openbsd stickers by themselves and not packaged
> with other products?
No, the only way I've seen them sold is for $40 with a free OpenBSD CD.
	-- Haroon Khalid and Steve Shockley in gmane.os.openbsd.misc

Re: [PATCH 1/1] m68k: add missing I/O macros {in,out}{w,l}_p() for !CONFIG_ISA

[Finn Thain]

On Mon, 4 Oct 2010, Thorsten Glaser wrote:

> Finn Thain wrote:
> 
> >It may be that speakup works on amiga and atari (I don't know). It 
> >doesn't work on mac as there are no audio drivers.
> 
> There's a software backend.

Fair enough. The docs tell me that it can also use serial port connected 
synthesisers.

Finn

Re: [PATCH 1/1] m68k: add missing I/O macros {in,out}{w,l}_p() for !CONFIG_ISA

[Geert Uytterhoeven]

On Sun, Oct 3, 2010 at 20:48, Thorsten Glaser <tg@debian.org> wrote:
> On m68k, I/O macros like inb() outw() etc. are only defined to
> something useful if CONFIG_ISA is set; dummies are in place if
> not, but four macros were missing from the !CONFIG_ISA case.
> Adding these makes some drivers, such as speakup, compile again.
> The patch was made against Debian’s linux-2.6_2.6.32-24, but
> applies cleanly on linux-m68k master.

Thanks, applied!

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

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