Skip to contentSkip to conversation

modprobe fails for via82xx in alsa-0.9.3a

DISCUSSION John Covici 3 participants 19 messages May 21, 2003 – May 28, 2003 Kbuild

lore
Share & tools
lore
#0 lore raw

Hi. I tried to install alsa-0.9.3a and when it did the modprobe for the via82xx module, I got unresolved symbols -- all of which had to do with memory (I don't have them exactly to report now), but they seemed to come from another module snd-page-alloc .

Any assistance would be appreciated.

Signature · 9 lines
         John Covici
         [email protected]


-------------------------------------------------------
This SF.net email is sponsored by: ObjectStore.
If flattening out C++ or Java code to make your application fit in a
relational database is painful, don't do it! Check out ObjectStore.
Now part of Progress Software. http://www.objectstore.net/sourceforge
#1 lore raw

At Wed, 21 May 2003 06:44:36 -0400,

Quoted 7 lines · John Covici

Hi. I tried to install alsa-0.9.3a and when it did the modprobe for the via82xx module, I got unresolved symbols -- all of which had to do with memory (I don't have them exactly to report now), but they seemed to come from another module snd-page-alloc .

Any assistance would be appreciated.

please try to unload snd-page-alloc module manually if it still persists. alsasound init script won't unload it to keep the bulk memory areas.

Takashi

-------------------------------------------------------
This SF.net email is sponsored by: ObjectStore.
If flattening out C++ or Java code to make your application fit in a relational database is painful, don't do it! Check out ObjectStore. Now part of Progress Software. http://www.objectstore.net/sourceforge

#2 lore raw

Should I delete it from the module tree, or if not, when should I do an rmmod -- seems that the script would just reload it again as it loads everything with snd if I remember correctly.

on Wednesday 05/21/2003 Takashi Iwai([email protected]) wrote

Quoted 16 lines

At Wed, 21 May 2003 06:44:36 -0400,

Quoted 7 lines · John Covici

Hi. I tried to install alsa-0.9.3a and when it did the modprobe for the via82xx module, I got unresolved symbols -- all of which had to do with memory (I don't have them exactly to report now), but they seemed to come from another module snd-page-alloc .

Any assistance would be appreciated.

please try to unload snd-page-alloc module manually if it still persists. alsasound init script won't unload it to keep the bulk memory areas.

Takashi

Signature · 9 lines
         John Covici
         [email protected]


-------------------------------------------------------
This SF.net email is sponsored by: ObjectStore.
If flattening out C++ or Java code to make your application fit in a
relational database is painful, don't do it! Check out ObjectStore.
Now part of Progress Software. http://www.objectstore.net/sourceforge
#3 lore raw

At Wed, 21 May 2003 07:34:26 -0400,

Quoted 4 lines · John covici

Should I delete it from the module tree, or if not, when should I do an rmmod -- seems that the script would just reload it again as it loads everything with snd if I remember correctly.

just make sure that you really reloaded snd-page-alloc. as said, the script was changed not to reload snd-page-alloc module.

Takashi

-------------------------------------------------------
This SF.net email is sponsored by: ObjectStore.
If flattening out C++ or Java code to make your application fit in a relational database is painful, don't do it! Check out ObjectStore. Now part of Progress Software. http://www.objectstore.net/sourceforge

#4 lore raw

Here is what I got when I tried to install-- note the page-alloc module is there:

Starting sound driver: snd-via82xx /lib/modules/2.4.20-xfs-newspeech/kernel/sound/acore/snd-pcm.o: unresolved symbol snd_dma_free_reserved_Rd027fa8d /lib/modules/2.4.20-xfs-newspeech/kernel/sound/acore/snd-pcm.o: unresolved symbol snd_dma_free_pages_Rce5f2dfe /lib/modules/2.4.20-xfs-newspeech/kernel/sound/acore/snd-pcm.o: unresolved symbol snd_dma_get_reserved_R3761189e /lib/modules/2.4.20-xfs-newspeech/kernel/sound/acore/snd-pcm.o: unresolved symbol snd_dma_set_reserved_Rd569b169 /lib/modules/2.4.20-xfs-newspeech/kernel/sound/acore/snd-pcm.o: unresolved symbol snd_dma_alloc_pages_Rfc708157 /lib/modules/2.4.20-xfs-newspeech/kernel/sound/acore/snd-pcm.o: insmod /lib/modules/2.4.20-xfs-newspeech/kernel/sound/acore/snd-pcm.o failed /lib/modules/2.4.20-xfs-newspeech/kernel/sound/acore/snd-pcm.o: insmod snd-via82xx failed
done
Module Size Used by Not tainted

snd-page-alloc          5932   0 
w83781d                20464   0  (unused)
eeprom                  3636   0  (unused)
i2c-proc                7472   0  [w83781d eeprom]
i2c-isa                 1192   0  (unused)
i2c-viapro              3984   0  (unused)
i2c-core               15848   0  [w83781d eeprom i2c-proc i2c-isa i2c-viapro]
3c59x                  27152   2  (autoclean)
nls_iso8859-1           2876   4  (autoclean)
advansys               86400   0  (unused)
scsi_mod               95476   1  [advansys]
ipt_MARK                 792   4  (autoclean)
iptable_mangle          2168   1  (autoclean)
iptable_filter          1740   1 
ipt_LOG                 3416   1 
ipt_limit                920   1 
ip_nat_ftp              3088   0  (unused)
iptable_nat            15576   2  [ip_nat_ftp]
ip_conntrack_ftp        4080   1 
ip_conntrack           18976   2  [ip_nat_ftp iptable_nat ip_conntrack_ftp]
ip_tables              11896   8  [ipt_MARK iptable_mangle iptable_filter ipt_LOG ipt_limit iptable_nat]
8139too                15208   1  (autoclean)
mii                     2512   0  (autoclean) [8139too]

on Wednesday 05/21/2003 Takashi Iwai([email protected]) wrote

Quoted 12 lines

At Wed, 21 May 2003 07:34:26 -0400,

Quoted 4 lines · John covici

Should I delete it from the module tree, or if not, when should I do an rmmod -- seems that the script would just reload it again as it loads everything with snd if I remember correctly.

just make sure that you really reloaded snd-page-alloc. as said, the script was changed not to reload snd-page-alloc module.

Takashi

Signature · 9 lines
         John Covici
         [email protected]


-------------------------------------------------------
This SF.net email is sponsored by: ObjectStore.
If flattening out C++ or Java code to make your application fit in a
relational database is painful, don't do it! Check out ObjectStore.
Now part of Progress Software. http://www.objectstore.net/sourceforge
#5 lore raw

At Wed, 21 May 2003 09:29:13 -0400,

Quoted 3 lines · John covici

Here is what I got when I tried to install-- note the page-alloc module is there:

(snip)

Quoted 3 lines

done
Module Size Used by Not tainted
snd-page-alloc 5932 0

and what happens if you run "modprobe -r snd-page-alloc" and "/etc/init.d/alsasound start"?

Takashi

-------------------------------------------------------
This SF.net email is sponsored by: ObjectStore.
If flattening out C++ or Java code to make your application fit in a relational database is painful, don't do it! Check out ObjectStore. Now part of Progress Software. http://www.objectstore.net/sourceforge

#6 lore raw

I am not sure what you mean -- before I run /etc/init.d/alsasound there are no snd modules at all, so what would modprobe -r snd-page-alloc do for the situation?

Quoted 15 lines · on Wed, 21 May 2003 17:14:04 +0200 Takashi Iwai <[email protected]> wrote:

At Wed, 21 May 2003 09:29:13 -0400,

Quoted 3 lines · John covici

Here is what I got when I tried to install-- note the page-alloc module is there:

(snip)

Quoted 3 lines

done
Module Size Used by Not tainted
snd-page-alloc 5932 0

and what happens if you run "modprobe -r snd-page-alloc" and "/etc/init.d/alsasound start"?

Takashi

Signature · 9 lines
         John Covici
         [email protected]


-------------------------------------------------------
This SF.net email is sponsored by: ObjectStore.
If flattening out C++ or Java code to make your application fit in a
relational database is painful, don't do it! Check out ObjectStore.
Now part of Progress Software. http://www.objectstore.net/sourceforge
#7 lore raw

At Thu, 22 May 2003 07:29:27 -0400,

Quoted 4 lines · John Covici

I am not sure what you mean -- before I run /etc/init.d/alsasound there are no snd modules at all, so what would modprobe -r snd-page-alloc do for the situation?

well i was not sure whether it's really not there.

then something wrong in the compilation of the modules. please make sure that you did make clean and the configure refers to the correct linux kernel tree.

Takashi

-------------------------------------------------------
This SF.net email is sponsored by: ObjectStore.
If flattening out C++ or Java code to make your application fit in a relational database is painful, don't do it! Check out ObjectStore. Now part of Progress Software. http://www.objectstore.net/sourceforge

#8 lore raw

0.9.3c build failure + partial patch

Hi,
attached is a patch to attempt to fix some problems with the current release:
Hunk 1 contains a minor fix that will create directories during installation for some files (especially useful for building packages) Hunk 2 fixes a dependency that appears to be no longer required. Hunk 3 is a partial fix to the pnp.c file in support/pnp not being built. It is a partial fix as it still does not get compiled at all, even though issuing a make will compile it...

tom.

Makefile renamed from Makefile.org +3 -3
@@ -153,11 +153,11 @@
 .PHONY: install-scripts
 install-scripts:
 	if [ -d /sbin/init.d ]; then \
-	  install -m 755 -g $(IGROUP) -o $(IUSER) utils/alsasound $(DESTDIR)/sbin/init.d/alsasound; \
+	  install -D -m 755 -g $(IGROUP) -o $(IUSER) utils/alsasound $(DESTDIR)/sbin/init.d/alsasound; \
 	elif [ -d /etc/rc.d/init.d ]; then \
-	  install -m 755 -g $(IGROUP) -o $(IUSER) utils/alsasound $(DESTDIR)/etc/rc.d/init.d/alsasound; \
+	  install -D -m 755 -g $(IGROUP) -o $(IUSER) utils/alsasound $(DESTDIR)/etc/rc.d/init.d/alsasound; \
 	elif [ -d /etc/init.d ]; then \
-	  install -m 755 -g $(IGROUP) -o $(IUSER) utils/alsasound $(DESTDIR)/etc/init.d/alsasound; \
+	  install -D -m 755 -g $(IGROUP) -o $(IUSER) utils/alsasound $(DESTDIR)/etc/init.d/alsasound; \
 	fi
 
 .PHONY: check-snd-prefix
support/pnp/pnp.c renamed from support/pnp/pnp.c.org +0 -1
@@ -33,7 +33,6 @@
 #if defined(CONFIG_MODVERSIONS) && !defined(__GENKSYMS__) && !defined(__DEPEND__)
 #define MODVERSIONS
 #include <linux/modversions.h>
-#include "sndversions.h"
 #endif
 #endif
 
support/pnp/Makefile renamed from support/pnp/Makefile.org +1 -1
@@ -17,7 +17,7 @@
 # Toplevel Module Dependency
 ifndef CONFIG_PNP_KERNEL
 ifeq (y,$(CONFIG_ISAPNP))
-obj-m += snd-pnp.o
+obj-m += pnp.o
 endif
 endif
 
#9 lore raw

Re: 0.9.3c build failure + partial patch

At Thu, 22 May 2003 22:33:13 +1000 (EST),

Quoted 7 lines · tom burkart

[1 <text/plain; US-ASCII (7bit)>]
Hi,
attached is a patch to attempt to fix some problems with the current release:
Hunk 1 contains a minor fix that will create directories during installation for some files (especially useful for building packages)

thanks for pointing this out.
i think checking $(DESTDIR)/xxx in the if is better than forcing to install scripts onto non-existing directories. fixed on cvs now.

Quoted 1 line

Hunk 2 fixes a dependency that appears to be no longer required.

the file is still required if you use a kernel with MODVERSIONS.

Quoted 3 lines

Hunk 3 is a partial fix to the pnp.c file in support/pnp not being built. It is a partial fix as it still does not get compiled at all, even though issuing a make will compile it...

this is strange. the current version works for me.

Takashi

-------------------------------------------------------
This SF.net email is sponsored by: ObjectStore.
If flattening out C++ or Java code to make your application fit in a relational database is painful, don't do it! Check out ObjectStore. Now part of Progress Software. http://www.objectstore.net/sourceforge

#10 lore raw

Re: 0.9.3c build failure + partial patch

Quoted 2 lines · Takashi Iwai
Quoted 1 line

Hunk 2 fixes a dependency that appears to be no longer required.

the file is still required if you use a kernel with MODVERSIONS.

It is not created and the only place it is used is here...

Quoted 4 lines
Quoted 3 lines

Hunk 3 is a partial fix to the pnp.c file in support/pnp not being built. It is a partial fix as it still does not get compiled at all, even though issuing a make will compile it...

this is strange. the current version works for me.

Basically the install complains about a missing file which is not there because it does not seem to get compiled.

tom.

-------------------------------------------------------
This SF.net email is sponsored by: ObjectStore.
If flattening out C++ or Java code to make your application fit in a relational database is painful, don't do it! Check out ObjectStore. Now part of Progress Software. http://www.objectstore.net/sourceforge

#11 lore raw

Re: 0.9.3c build failure + partial patch

At Mon, 26 May 2003 13:54:01 +1000 (EST),

Quoted 6 lines · tom burkart
Quoted 2 lines · Takashi Iwai
Quoted 1 line

Hunk 2 fixes a dependency that appears to be no longer required.

the file is still required if you use a kernel with MODVERSIONS.

It is not created and the only place it is used is here...

?? i see the file in fact.

Quoted 6 lines
Quoted 4 lines
Quoted 3 lines

Hunk 3 is a partial fix to the pnp.c file in support/pnp not being built. It is a partial fix as it still does not get compiled at all, even though issuing a make will compile it...

this is strange. the current version works for me.

Basically the install complains about a missing file which is not there because it does not seem to get compiled.

then this might be the problem of make.
the older make cannot build some modules correctly.

Takashi

-------------------------------------------------------
This SF.net email is sponsored by: ObjectStore.
If flattening out C++ or Java code to make your application fit in a relational database is painful, don't do it! Check out ObjectStore. Now part of Progress Software. http://www.objectstore.net/sourceforge

#12 lore raw

Re: 0.9.3c build failure + partial patch

Quoted 3 lines · Takashi Iwai

?? i see the file in fact.
then this might be the problem of make.
the older make cannot build some modules correctly.

Ok, the problem is that "make all" (or just "make") is not a build prerequisite to make install.

This obsoletes hunk 2 and 3 of my patch. Please discard.

tom.
Consultant

AUSSEC Phone: 61 4 1768 2202
339 Blaxland Rd., Ryde NSW 2112
Email: [email protected]

-------------------------------------------------------
This SF.net email is sponsored by: ObjectStore.
If flattening out C++ or Java code to make your application fit in a relational database is painful, don't do it! Check out ObjectStore. Now part of Progress Software. http://www.objectstore.net/sourceforge

#13 lore raw

Re: 0.9.3c build failure + partial patch

In reply to Takashi Iwai
Quoted 4 lines · Takashi Iwai
Quoted 2 lines

Hunk 1 contains a minor fix that will create directories during installation for some files (especially useful for building packages)

i think checking $(DESTDIR)/xxx in the if is better than forcing to install scripts onto non-existing directories. fixed on cvs now.

That may be incorrect (I haven't checked the CVS), as those if statements check for existence of the particular directories to set it up correctly for the given distribution. It may be better to get install to create them if they don't exist.

tom.
Consultant

AUSSEC Phone: 61 4 1768 2202
339 Blaxland Rd., Ryde NSW 2112
Email: [email protected]

-------------------------------------------------------
This SF.net email is sponsored by: ObjectStore.
If flattening out C++ or Java code to make your application fit in a relational database is painful, don't do it! Check out ObjectStore. Now part of Progress Software. http://www.objectstore.net/sourceforge

#14 lore raw

Re: 0.9.3c build failure + partial patch

At Mon, 26 May 2003 14:27:59 +1000 (EST),

Quoted 11 lines · tom burkart
Quoted 4 lines · Takashi Iwai
Quoted 2 lines

Hunk 1 contains a minor fix that will create directories during installation for some files (especially useful for building packages)

i think checking $(DESTDIR)/xxx in the if is better than forcing to install scripts onto non-existing directories. fixed on cvs now.

That may be incorrect (I haven't checked the CVS), as those if statements check for existence of the particular directories to set it up correctly for the given distribution. It may be better to get install to create them if they don't exist.

hmm, then how can you check which directory to be installed? each distribution has different file location (although LSB exists). creating the file always is not preferred way in the case of init script.

perhaps it's better not to install it as default but via "make install-debian" or something like that.

Takashi

-------------------------------------------------------
This SF.net email is sponsored by: ObjectStore.
If flattening out C++ or Java code to make your application fit in a relational database is painful, don't do it! Check out ObjectStore. Now part of Progress Software. http://www.objectstore.net/sourceforge

#15 lore raw

Re: 0.9.3c build failure + partial patch

Quoted 4 lines · Takashi Iwai

hmm, then how can you check which directory to be installed? each distribution has different file location (although LSB exists). creating the file always is not preferred way in the case of init script.

This is correct. That is why make install checks which one of the various directories actually exists and then does the appropriate install in DESTDIR/whatever, which is the correct way of doing it. All I was adding was for install to actually create the directories in case they did not exist (for the case of package creation installs).

Quoted 2 lines

perhaps it's better not to install it as default but via "make install-debian" or something like that.

This is more painful, I would have thought.

Basically, I am suggesting to leave the install alone as it is correct, I am only adding the option of creating the directories if they don't already exist.

tom.
Consultant

AUSSEC Phone: 61 4 1768 2202
339 Blaxland Rd., Ryde NSW 2112
Email: [email protected]

-------------------------------------------------------
This SF.net email is sponsored by: ObjectStore.
If flattening out C++ or Java code to make your application fit in a relational database is painful, don't do it! Check out ObjectStore. Now part of Progress Software. http://www.objectstore.net/sourceforge

#16 lore raw

Re: 0.9.3c build failure + partial patch

In reply to Takashi Iwai
Quoted 4 lines · Takashi Iwai

hmm, then how can you check which directory to be installed? each distribution has different file location (although LSB exists). creating the file always is not preferred way in the case of init script.

Correct. The existing checks are fine as they check /etc/... for the most common distributions and so it installs it in the correct place. What I am adding is that install checks whether the destination directory exists and creates it if it doesn't. If the "if" statement evaluates to true then the directory exists and is the correct one. If you have a $DESTDIR set, the actual installation dir will most likely not exist and install fails. Hence the reason for getting install to check that the directories are there and to install them as required.

Quoted 2 lines

perhaps it's better not to install it as default but via "make install-debian" or something like that.

Absolutely not necessary as the existing code is nearly correct.

tom.

-------------------------------------------------------
This SF.net email is sponsored by: ObjectStore.
If flattening out C++ or Java code to make your application fit in a relational database is painful, don't do it! Check out ObjectStore. Now part of Progress Software. http://www.objectstore.net/sourceforge

#17 lore raw
In reply to Takashi Iwai

Yup, created the directory right from scratch the first time, did configure, etc. and it seemed to put things in the correct places -- if you want I can send you the config output, and the output of make and make install.

on Thursday 05/22/2003 Takashi Iwai([email protected]) wrote

Quoted 15 lines

At Thu, 22 May 2003 07:29:27 -0400,

Quoted 4 lines · John Covici

I am not sure what you mean -- before I run /etc/init.d/alsasound there are no snd modules at all, so what would modprobe -r snd-page-alloc do for the situation?

well i was not sure whether it's really not there.

then something wrong in the compilation of the modules. please make sure that you did make clean and the configure refers to the correct linux kernel tree.

Takashi

Signature · 9 lines
         John Covici
         [email protected]


-------------------------------------------------------
This SF.net email is sponsored by: ObjectStore.
If flattening out C++ or Java code to make your application fit in a
relational database is painful, don't do it! Check out ObjectStore.
Now part of Progress Software. http://www.objectstore.net/sourceforge
#18 lore raw

Well 3c isn't having the problem, so maybe this is moot.

on Thursday 05/22/2003 John covici([email protected]) wrote

Quoted 36 lines

Yup, created the directory right from scratch the first time, did configure, etc. and it seemed to put things in the correct places -- if you want I can send you the config output, and the output of make and make install.

on Thursday 05/22/2003 Takashi Iwai([email protected]) wrote

Quoted 15 lines

At Thu, 22 May 2003 07:29:27 -0400,

Quoted 4 lines · John Covici

I am not sure what you mean -- before I run /etc/init.d/alsasound there are no snd modules at all, so what would modprobe -r snd-page-alloc do for the situation?

well i was not sure whether it's really not there.

then something wrong in the compilation of the modules. please make sure that you did make clean and the configure refers to the correct linux kernel tree.

Takashi

Signature · 13 lines
         John Covici
         [email protected]


-------------------------------------------------------
This SF.net email is sponsored by: ObjectStore.
If flattening out C++ or Java code to make your application fit in a
relational database is painful, don't do it! Check out ObjectStore.
Now part of Progress Software. http://www.objectstore.net/sourceforge
_______________________________________________
Alsa-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/alsa-devel
Signature · 9 lines
         John Covici
         [email protected]


-------------------------------------------------------
This SF.net email is sponsored by: ObjectStore.
If flattening out C++ or Java code to make your application fit in a
relational database is painful, don't do it! Check out ObjectStore.
Now part of Progress Software. http://www.objectstore.net/sourceforge