[-mm patch] fix fs/nfs/nfsroot.c compile error

[-mm patch] fix fs/nfs/nfsroot.c compile error

[Adrian Bunk]

On Sat, May 05, 2007 at 01:49:55AM -0700, Andrew Morton wrote:
>...
> Changes since 2.6.21-rc7-mm2:
>...
>  git-net.patch
>...
>  git trees
>...

match_table_t was made "const" and gcc doesn't like "const __initdata":

<--  snip  -->

...
  CC      fs/nfs/nfsroot.o
/home/bunk/linux/kernel-2.6/linux-2.6.21-mm1/fs/nfs/nfsroot.c:131: error: tokens causes a section type conflict
make[3]: *** [fs/nfs/nfsroot.o] Error 1

<--  snip  -->

Signed-off-by: Adrian Bunk <bunk@stusta.de>

---

BTW: Why is this change to include/linux/parser.h in the net tree?

--- linux-2.6.21-mm1/fs/nfs/nfsroot.c.old	2007-05-05 18:36:26.000000000 +0200
+++ linux-2.6.21-mm1/fs/nfs/nfsroot.c	2007-05-05 18:36:44.000000000 +0200
@@ -128,7 +128,7 @@
 	Opt_err
 };
 
-static match_table_t __initdata tokens = {
+static match_table_t tokens = {
 	{Opt_port, "port=%u"},
 	{Opt_rsize, "rsize=%u"},
 	{Opt_wsize, "wsize=%u"},

Re: [-mm patch] fix fs/nfs/nfsroot.c compile error

[Trond Myklebust]

On Sat, 2007-05-05 at 18:44 +0200, Adrian Bunk wrote:
> On Sat, May 05, 2007 at 01:49:55AM -0700, Andrew Morton wrote:
> >...
> > Changes since 2.6.21-rc7-mm2:
> >...
> >  git-net.patch
> >...
> >  git trees
> >...
> 
> match_table_t was made "const" and gcc doesn't like "const __initdata":

Then please revert the "fix" to match_table_t...

There is no reason for something like the nfsroot parser to be kept in
memory after the system has booted. That would be code bloat.

Trond

Re: [-mm patch] fix fs/nfs/nfsroot.c compile error

[Trond Myklebust]

On Sat, 2007-05-05 at 13:20 -0400, Trond Myklebust wrote:
> On Sat, 2007-05-05 at 18:44 +0200, Adrian Bunk wrote:
> > On Sat, May 05, 2007 at 01:49:55AM -0700, Andrew Morton wrote:
> > >...
> > > Changes since 2.6.21-rc7-mm2:
> > >...
> > >  git-net.patch
> > >...
> > >  git trees
> > >...
> > 
> > match_table_t was made "const" and gcc doesn't like "const __initdata":
> 
> Then please revert the "fix" to match_table_t...
> 
> There is no reason for something like the nfsroot parser to be kept in
> memory after the system has booted. That would be code bloat.
> 
> Trond

Alternatively, please change the nfsroot parser to use

        static struct match_token tokens[] __initdata = {

That is in any case cleaner than using a typedef.

Trond

Re: [-mm patch] fix fs/nfs/nfsroot.c compile error

[David Howells]

Adrian Bunk <bunk@stusta.de> wrote:

> match_table_t was made "const" and gcc doesn't like "const __initdata":

Hmmm...  My gcc seems quite happy with it.  What gcc are you using?

Shouldn't __initdata override const?

> BTW: Why is this change to include/linux/parser.h in the net tree?

Because DaveM picked it up as part of my AFS patches, I guess.

David

Re: [-mm patch] fix fs/nfs/nfsroot.c compile error

[Adrian Bunk]

On Mon, May 07, 2007 at 11:47:51AM +0100, David Howells wrote:
> Adrian Bunk <bunk@stusta.de> wrote:
> 
> > match_table_t was made "const" and gcc doesn't like "const __initdata":
> 
> Hmmm...  My gcc seems quite happy with it.  What gcc are you using?

4.1

> Shouldn't __initdata override const?
> 
> > BTW: Why is this change to include/linux/parser.h in the net tree?
> 
> Because DaveM picked it up as part of my AFS patches, I guess.

It might make sense, but it doesn't seem to be in any way AFS specific.

> David

cu
Adrian

-- 

       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed

Re: [-mm patch] fix fs/nfs/nfsroot.c compile error

[Trond Myklebust]

On Mon, 2007-05-07 at 12:56 +0200, Adrian Bunk wrote:
> On Mon, May 07, 2007 at 11:47:51AM +0100, David Howells wrote:
> > Adrian Bunk <bunk@stusta.de> wrote:
> > 
> > > match_table_t was made "const" and gcc doesn't like "const __initdata":
> > 
> > Hmmm...  My gcc seems quite happy with it.  What gcc are you using?
> 
> 4.1
> 
> > Shouldn't __initdata override const?
> > 
> > > BTW: Why is this change to include/linux/parser.h in the net tree?
> > 
> > Because DaveM picked it up as part of my AFS patches, I guess.
> 
> It might make sense, but it doesn't seem to be in any way AFS specific.

It makes little or no sense to be forcing attributes like 'const' in a
typedef. That is basically stating that it is inconceivable to use
anything else. Clearly not the case here...

Trond

Re: [-mm patch] fix fs/nfs/nfsroot.c compile error

[David Howells]

Trond Myklebust <trond.myklebust@fys.uio.no> wrote:

> It makes little or no sense to be forcing attributes like 'const' in a
> typedef. That is basically stating that it is inconceivable to use
> anything else. Clearly not the case here...

Yeah.  I think that there shouldn't be a typedef.  There's no real need for
it.

David