At first I thought my build was incorrect, but I reproduced this error
with a clean build and ALSA CVS from today:
CC [M] /home/rlrevell/cvs/alsa/alsa-driver/kbuild/../acore/pcm_native.o
/home/rlrevell/cvs/alsa/alsa-driver/acore/pcm_native.c:3202:57: macro "io_remap_page_range" requires 5 arguments, but only 4 given
/home/rlrevell/cvs/alsa/alsa-driver/acore/pcm_native.c: In function `snd_pcm_lib_mmap_iomem':
/home/rlrevell/cvs/alsa/alsa-driver/acore/pcm_native.c:3200: error: `io_remap_page_range' undeclared (first use in this function)
/home/rlrevell/cvs/alsa/alsa-driver/acore/pcm_native.c:3200: error: (Each undeclared identifier is reported only once
/home/rlrevell/cvs/alsa/alsa-driver/acore/pcm_native.c:3200: error: for each function it appears in.)
make[3]: *** [/home/rlrevell/cvs/alsa/alsa-driver/kbuild/../acore/pcm_native.o] Error 1
make[2]: *** [/home/rlrevell/cvs/alsa/alsa-driver/kbuild/../acore] Error 2
make[1]: *** [_module_/home/rlrevell/cvs/alsa/alsa-driver/kbuild] Error 2
make[1]: Leaving directory `/home/rlrevell/kernel-source/linux-2.6.9-rc2-mm4-S7'
make: *** [compile] Error 2
I am not sure if this is an ALSA issue or -mm4. I suspect -mm4 because
-mm3-S6 worked. The VP patch does not seem to be involved.
Lee
Lee Revell wrote:
> At first I thought my build was incorrect, but I reproduced this error
> with a clean build and ALSA CVS from today:
>
> CC [M] /home/rlrevell/cvs/alsa/alsa-driver/kbuild/../acore/pcm_native.o
> /home/rlrevell/cvs/alsa/alsa-driver/acore/pcm_native.c:3202:57: macro
> "io_remap_page_range" requires 5 arguments, but only 4 given
> /home/rlrevell/cvs/alsa/alsa-driver/acore/pcm_native.c: In function
> `snd_pcm_lib_mmap_iomem':
> /home/rlrevell/cvs/alsa/alsa-driver/acore/pcm_native.c:3200: error:
> `io_remap_page_range' undeclared (first use in this function)
> /home/rlrevell/cvs/alsa/alsa-driver/acore/pcm_native.c:3200: error: (Each
> undeclared identifier is reported only once
> /home/rlrevell/cvs/alsa/alsa-driver/acore/pcm_native.c:3200: error: for
> each function it appears in.)
> make[3]: ***
> [/home/rlrevell/cvs/alsa/alsa-driver/kbuild/../acore/pcm_native.o] Error 1
> make[2]: *** [/home/rlrevell/cvs/alsa/alsa-driver/kbuild/../acore] Error 2
> make[1]: *** [_module_/home/rlrevell/cvs/alsa/alsa-driver/kbuild] Error 2
> make[1]: Leaving directory
> `/home/rlrevell/kernel-source/linux-2.6.9-rc2-mm4-S7'
> make: *** [compile] Error 2
>
> I am not sure if this is an ALSA issue or -mm4. I suspect -mm4 because
> -mm3-S6 worked. The VP patch does not seem to be involved.
>
> Lee
>
Good grief! I'm having this too, and I was desperate thinking I was the
only one, and ultimately offering the blame to gcc 3.4.1 which is what I'm
test-driving now on my laptop (Mdk 10.1c).
Now I remember that -mm4 has some issue about remap_page_range kernel
symbol being renamed to something else, which is breaking the build of
outsider modules (i.e. not the ones bundled under the kernel source tree).
Or so it seems.
Maybe someone on the lkml may have a clue and help here?
--
rncbc aka Rui Nuno Capela
[email protected]
Salut,
On Sat, Oct 02, 2004 at 01:34:29AM +0100, Rui Nuno Capela wrote:
> Maybe someone on the lkml may have a clue and help here?
http://marc.theaimsgroup.com/?l=linux-kernel&m=109526630905797&w=2
Tonnerre
On Fri, 2004-10-01 at 20:34, Rui Nuno Capela wrote:
> Lee Revell wrote:
> Good grief! I'm having this too, and I was desperate thinking I was the
> only one, and ultimately offering the blame to gcc 3.4.1 which is what I'm
> test-driving now on my laptop (Mdk 10.1c).
>
> Now I remember that -mm4 has some issue about remap_page_range kernel
> symbol being renamed to something else, which is breaking the build of
> outsider modules (i.e. not the ones bundled under the kernel source tree).
> Or so it seems.
Looking through my archives I cannot find a report of this exact issue,
but you are probably right. Looks like ALSA drivers need to be updated.
Lee
On Fri, 2004-10-01 at 20:48, Tonnerre wrote:
> Salut,
>
> On Sat, Oct 02, 2004 at 01:34:29AM +0100, Rui Nuno Capela wrote:
> > Maybe someone on the lkml may have a clue and help here?
>
> http://marc.theaimsgroup.com/?l=linux-kernel&m=109526630905797&w=2
Thanks, but it's already obvious that this is related to the iomem
changes. This does not address this issue of what specifically broke
ALSA between -mm3 and -mm4.
If it is clear to you what the issue is then a patch would be much more
helpful ;-)
Lee
On Fri, 2004-10-01 at 20:53, Lee Revell wrote:
> On Fri, 2004-10-01 at 20:48, Tonnerre wrote:
> > Salut,
> >
> > On Sat, Oct 02, 2004 at 01:34:29AM +0100, Rui Nuno Capela wrote:
> > > Maybe someone on the lkml may have a clue and help here?
> >
> > http://marc.theaimsgroup.com/?l=linux-kernel&m=109526630905797&w=2
>
> Thanks, but it's already obvious that this is related to the iomem
> changes. This does not address this issue of what specifically broke
> ALSA between -mm3 and -mm4.
>
> If it is clear to you what the issue is then a patch would be much more
> helpful ;-)
>
OK found the problem. Looks like the ALSA folks have the hooks in there
for the iomem changes but it is currently a NOOP. In
alsa-driver/configure:
6632 #define CONFIG_OLD_REMAP_PAGE_RANGE 1
Then:
6569 echo "$as_me:$LINENO: checking for new remap_page_range" >&5
6570 echo $ECHO_N "checking for new remap_page_range... $ECHO_C" >&6
For now unset CONFIG_OLD_REMAP_PAGE_RANGE and it should work. Not yet
tested.
Lee
On Fri, 2004-10-01 at 21:13, Lee Revell wrote:
> For now unset CONFIG_OLD_REMAP_PAGE_RANGE and it should work. Not yet
> tested.
Yup, this works. Trivial patch:
--- alsa-driver/configure~ 2004-09-29 16:05:19.000000000 -0400
+++ alsa-driver/configure 2004-10-01 21:14:20.000000000 -0400
@@ -6629,7 +6629,7 @@
CFLAGS=$ac_save_CFLAGS
if test "$new_remap" != "1"; then
cat >>confdefs.h <<\_ACEOF
-#define CONFIG_OLD_REMAP_PAGE_RANGE 1
+#undef CONFIG_OLD_REMAP_PAGE_RANGE
_ACEOF
fi
Lee
On Fri, 01 Oct 2004 20:12:10 -0400
Lee Revell <[email protected]> wrote:
> At first I thought my build was incorrect, but I reproduced this error
> with a clean build and ALSA CVS from today:
>
> CC [M]
> /home/rlrevell/cvs/alsa/alsa-driver/kbuild/../acore/pcm_native.o
> /home/rlrevell/cvs/alsa/alsa-driver/acore/pcm_native.c:3202:57: macro
> "io_remap_page_range" requires 5 arguments, but only 4
> given/home/rlrevell/cvs/alsa/alsa-driver/acore/pcm_native.c: In
> function
> `snd_pcm_lib_mmap_iomem':/home/rlrevell/cvs/alsa/alsa-driver/acore/pc
> m_native.c:3200: error: `io_remap_page_range' undeclared (first use in
> this
> function)/home/rlrevell/cvs/alsa/alsa-driver/acore/pcm_native.c:3200:
> error: (Each undeclared identifier is reported only
> once/home/rlrevell/cvs/alsa/alsa-driver/acore/pcm_native.c:3200:
> error: for each function it appears in.) make[3]: ***
> [/home/rlrevell/cvs/alsa/alsa-driver/kbuild/../acore/pcm_native.o]
> Error 1 make[2]: ***
> [/home/rlrevell/cvs/alsa/alsa-driver/kbuild/../acore] Error 2 make[1]:
> *** [_module_/home/rlrevell/cvs/alsa/alsa-driver/kbuild] Error 2
> make[1]: Leaving directory
> `/home/rlrevell/kernel-source/linux-2.6.9-rc2-mm4-S7' make: ***
> [compile] Error 2
>
> I am not sure if this is an ALSA issue or -mm4. I suspect -mm4
> because-mm3-S6 worked. The VP patch does not seem to be involved.
Hmm, i have no problems compiling the alsa drivers included with the
kernel [yes, i'm on mm4-S7]..
tapas@mango:~$ gcc --version
gcc (GCC) 3.3.4 (Debian 1:3.3.4-13)
flo
On Sat, 2 Oct 2004 13:47:32 +0200
Florian Schmidt <[email protected]> wrote:
> > I am not sure if this is an ALSA issue or -mm4. I suspect -mm4
> > because-mm3-S6 worked. The VP patch does not seem to be involved.
>
>
> Hmm, i have no problems compiling the alsa drivers included with the
> kernel [yes, i'm on mm4-S7]..
>
> tapas@mango:~$ gcc --version
> gcc (GCC) 3.3.4 (Debian 1:3.3.4-13)
oops misread. sorry..
flo
At Fri, 01 Oct 2004 20:51:08 -0400,
Lee Revell wrote:
>
> On Fri, 2004-10-01 at 20:34, Rui Nuno Capela wrote:
> > Lee Revell wrote:
> > Good grief! I'm having this too, and I was desperate thinking I was the
> > only one, and ultimately offering the blame to gcc 3.4.1 which is what I'm
> > test-driving now on my laptop (Mdk 10.1c).
> >
> > Now I remember that -mm4 has some issue about remap_page_range kernel
> > symbol being renamed to something else, which is breaking the build of
> > outsider modules (i.e. not the ones bundled under the kernel source tree).
> > Or so it seems.
>
> Looking through my archives I cannot find a report of this exact issue,
> but you are probably right. Looks like ALSA drivers need to be updated.
The alsa-kernel code there (pcm_native.c) is ok but the patch in
alsa-driver looks broken for the recent change of remap_pfn_range().
Also, there was another API brekage about pci_save/restore_state().
Fixed both on CVS now.
Takahsi
On Mon, 2004-10-04 at 11:26, Takashi Iwai wrote:
> At Fri, 01 Oct 2004 20:51:08 -0400,
> Lee Revell wrote:
> >
> > On Fri, 2004-10-01 at 20:34, Rui Nuno Capela wrote:
> > > Lee Revell wrote:
> > > Good grief! I'm having this too, and I was desperate thinking I was the
> > > only one, and ultimately offering the blame to gcc 3.4.1 which is what I'm
> > > test-driving now on my laptop (Mdk 10.1c).
> > >
> > > Now I remember that -mm4 has some issue about remap_page_range kernel
> > > symbol being renamed to something else, which is breaking the build of
> > > outsider modules (i.e. not the ones bundled under the kernel source tree).
> > > Or so it seems.
> >
> > Looking through my archives I cannot find a report of this exact issue,
> > but you are probably right. Looks like ALSA drivers need to be updated.
>
> The alsa-kernel code there (pcm_native.c) is ok but the patch in
> alsa-driver looks broken for the recent change of remap_pfn_range().
>
> Also, there was another API brekage about pci_save/restore_state().
>
> Fixed both on CVS now.
I think this is still broken. Same problem when I went to configure
ALSA for -mm3-T3. The configure script gets
CONFIG_HAVE_OLD_REMAP_PAGE_RANGE wrong; I have the new version but
configure fails to detect it.
AFAICT the test in configure is a NOOP and
CONFIG_HAVE_OLD_REMAP_PAGE_RANGE always gets set to 1.
Lee
On Fri, 2004-10-08 at 15:23, Lee Revell wrote:
> On Mon, 2004-10-04 at 11:26, Takashi Iwai wrote:
> > At Fri, 01 Oct 2004 20:51:08 -0400,
> > Lee Revell wrote:
> > >
> > > On Fri, 2004-10-01 at 20:34, Rui Nuno Capela wrote:
> > > > Lee Revell wrote:
> > > > Good grief! I'm having this too, and I was desperate thinking I was the
> > > > only one, and ultimately offering the blame to gcc 3.4.1 which is what I'm
> > > > test-driving now on my laptop (Mdk 10.1c).
> > > >
> > > > Now I remember that -mm4 has some issue about remap_page_range kernel
> > > > symbol being renamed to something else, which is breaking the build of
> > > > outsider modules (i.e. not the ones bundled under the kernel source tree).
> > > > Or so it seems.
> > >
> > > Looking through my archives I cannot find a report of this exact issue,
> > > but you are probably right. Looks like ALSA drivers need to be updated.
> >
> > The alsa-kernel code there (pcm_native.c) is ok but the patch in
> > alsa-driver looks broken for the recent change of remap_pfn_range().
> >
> > Also, there was another API brekage about pci_save/restore_state().
> >
> > Fixed both on CVS now.
>
> I think this is still broken. Same problem when I went to configure
> ALSA for -mm3-T3. The configure script gets
> CONFIG_HAVE_OLD_REMAP_PAGE_RANGE wrong; I have the new version but
> configure fails to detect it.
>
> AFAICT the test in configure is a NOOP and
> CONFIG_HAVE_OLD_REMAP_PAGE_RANGE always gets set to 1.
Nope, sorry, it works now. But, I did cvs update, make clean, ./build
prep then configure and I was still hitting the bug. I was only able to
fix it by deleting my working directory and checking out again. I don't
think "make clean" cleans everything up.
Lee