CONFIG_KMOD breakage in next kernel

CONFIG_KMOD breakage in next kernel

[Takashi Iwai]

Hi,

I found that CONFIG_KMOD was already removed in linux-next tree.
This seems to result in many breakages.  For example, I cannot mount a
FAT device automatically because of failure of automatic loading of
nls modules.

My particular problem would be fixed by the patch below, but I found
CONFIG_KMOD is still referred in many places.

% find * -name '*.[ch]' | xargs grep -l CONFIG_KMOD
drivers/media/video/w9968cf.c
drivers/media/video/v4l2-common.c
drivers/media/video/v4l1-compat.c
drivers/media/video/usbvision/usbvision-video.c
drivers/media/video/usbvision/usbvision-core.c
drivers/media/video/vino.c
drivers/media/video/cpia.c
drivers/video/fbmem.c
drivers/md/md.c
drivers/net/irda/sir_dongle.c
drivers/net/pppox.c
drivers/net/ppp_generic.c
drivers/mtd/mtdpart.c
fs/char_dev.c
fs/exec.c
kernel/kmod.c
net/dccp/ccid.c
net/sunrpc/auth.c
net/ipv4/devinet.c
net/ipv4/inet_diag.c
net/ipv4/tcp_cong.c
net/bluetooth/af_bluetooth.c
net/sched/cls_api.c
net/sched/act_api.c
net/sched/ematch.c
net/sched/sch_api.c
net/can/af_can.c
net/core/rtnetlink.c
net/core/dev.c
net/decnet/dn_dev.c
net/netlink/af_netlink.c
net/socket.c
net/netfilter/nfnetlink.c
net/bridge/netfilter/ebtables.c


Anyone working on this?  It must be trivial.

Or, revert the following for the time being until all these are fixed:
commit ba90ad7a589abdf59553a622c35cc149a348034f
Author: Johannes Berg <johannes@sipsolutions.net>
Date:   Tue Jul 8 19:00:26 2008 +0200

    remove CONFIG_KMOD
    
    Now that nothing depends on it any more, remove CONFIG_KMOD.
    
    Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
    Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>


thanks,

Takashi

---
diff --git a/fs/nls/nls_base.c b/fs/nls/nls_base.c
index 64965e1..9f1a54c 100644
--- a/fs/nls/nls_base.c
+++ b/fs/nls/nls_base.c
@@ -13,9 +13,6 @@
 #include <linux/nls.h>
 #include <linux/kernel.h>
 #include <linux/errno.h>
-#ifdef CONFIG_KMOD
-#include <linux/kmod.h>
-#endif
 #include <linux/spinlock.h>
 
 static struct nls_table default_table;
@@ -216,7 +213,7 @@ static struct nls_table *find_nls(char *charset)
 struct nls_table *load_nls(char *charset)
 {
 	struct nls_table *nls;
-#ifdef CONFIG_KMOD
+#ifdef CONFIG_MODULES
 	int ret;
 #endif
 
@@ -224,7 +221,7 @@ struct nls_table *load_nls(char *charset)
 	if (nls)
 		return nls;
 
-#ifdef CONFIG_KMOD
+#ifdef CONFIG_MODULES
 	ret = request_module("nls_%s", charset);
 	if (ret != 0) {
 		printk("Unable to load NLS charset %s\n", charset);

Re: CONFIG_KMOD breakage in next kernel

[Marcel Holtmann]

Hi guys,

> I found that CONFIG_KMOD was already removed in linux-next tree.
> This seems to result in many breakages.  For example, I cannot mount a
> FAT device automatically because of failure of automatic loading of
> nls modules.
> 
> My particular problem would be fixed by the patch below, but I found
> CONFIG_KMOD is still referred in many places.

I am happy to fix the net/bluetooth/ part, you someone has to put out a
guideline how this should be fixed.

Regards

Marcel

Re: CONFIG_KMOD breakage in next kernel

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 776 bytes --]

On Wed, 2008-08-20 at 19:16 +0200, Marcel Holtmann wrote:
> Hi guys,
> 
> > I found that CONFIG_KMOD was already removed in linux-next tree.
> > This seems to result in many breakages.  For example, I cannot mount a
> > FAT device automatically because of failure of automatic loading of
> > nls modules.
> > 
> > My particular problem would be fixed by the patch below, but I found
> > CONFIG_KMOD is still referred in many places.
> 
> I am happy to fix the net/bluetooth/ part, you someone has to put out a
> guideline how this should be fixed.

Bluetooth was already part of http://lkml.org/lkml/2008/7/9/144

Basically, I got rid of CONFIG_KMOD, always enabling it when
CONFIG_MODULES is defined and making it a dummy no-op inline when not.

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: CONFIG_KMOD breakage in next kernel

[Marcel Holtmann]

Hi Johannes,

> > > I found that CONFIG_KMOD was already removed in linux-next tree.
> > > This seems to result in many breakages.  For example, I cannot mount a
> > > FAT device automatically because of failure of automatic loading of
> > > nls modules.
> > > 
> > > My particular problem would be fixed by the patch below, but I found
> > > CONFIG_KMOD is still referred in many places.
> > 
> > I am happy to fix the net/bluetooth/ part, you someone has to put out a
> > guideline how this should be fixed.
> 
> Bluetooth was already part of http://lkml.org/lkml/2008/7/9/144
> 
> Basically, I got rid of CONFIG_KMOD, always enabling it when
> CONFIG_MODULES is defined and making it a dummy no-op inline when not.

I missed the patches, but that happens from time to time. Just looked at
them and they look pretty nice. I like it this way.

Acked-by: Marcel Holtmann <marcel@holtmann.org>

Regards

Marcel

Re: CONFIG_KMOD breakage in next kernel

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 747 bytes --]

Hi,

> I found that CONFIG_KMOD was already removed in linux-next tree.
> This seems to result in many breakages.  For example, I cannot mount a
> FAT device automatically because of failure of automatic loading of
> nls modules.

This is odd. I had posted a long series of patches removing all of the
users I had found (similar grep like the one you used), see
http://lkml.org/lkml/2008/7/9/150 for the whole thread.


For some reason, only seven of those eleven patches ended up in the
kernel, you had taken the alsa one and Rusty posted only six of them.

I'd hope that the patches posted at the URL above still apply, and I
sure intended the series to go in as posted with the removal of
CONFIG_KMOD coming last.

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: CONFIG_KMOD breakage in next kernel

[Rusty Russell]

On Thursday 21 August 2008 04:16:05 Johannes Berg wrote:
> Hi,
>
> > I found that CONFIG_KMOD was already removed in linux-next tree.
> > This seems to result in many breakages.  For example, I cannot mount a
> > FAT device automatically because of failure of automatic loading of
> > nls modules.
>
> This is odd. I had posted a long series of patches removing all of the
> users I had found (similar grep like the one you used), see
> http://lkml.org/lkml/2008/7/9/150 for the whole thread.
>
>
> For some reason, only seven of those eleven patches ended up in the
> kernel, you had taken the alsa one and Rusty posted only six of them.

OK, not quite sure what happened.  I'll grab the rest and push them into 
tomorrow's linux-next.

For today, I'll pull out the CONFIG_KMOD removal.

Thanks,
Rusty.

Re: CONFIG_KMOD breakage in next kernel

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 366 bytes --]

Hi Rusty,

On Mon, 25 Aug 2008 10:13:58 +1000 Rusty Russell <rusty@rustcorp.com.au> wrote:
>
> For today, I'll pull out the CONFIG_KMOD removal.

Does that mean you want me to revert "remove CONFIG_KMOD" from your tree
for today's linux-next build?

-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

[-- Attachment #2: Type: application/pgp-signature, Size: 197 bytes --]