ext4: don't work without procfs

Regression from commit dd68314ccf3fb918c1fb6471817edbc60ece4b52
---
 fs/ext4/super.c |    8 ++++++++
 1 files changed, 8 insertions(+), 0 deletions(-)

On Mon, 3 Oct 2011, Yargil wrote:

> Regression from commit dd68314ccf3fb918c1fb6471817edbc60ece4b52

Hi,

the commit you've mentioned does not have anything to do with procfs.
Maybe you meant commit 

c9de560ded61faa5b754137b7753da252391c55a
	ext4: Add multi block allocator for ext4

which adds multi block allocator for ext4 (mballoc.c) and has been added
in Jan 29 2008.

> ---
>  fs/ext4/super.c |    8 ++++++++
>  1 files changed, 8 insertions(+), 0 deletions(-)
> 
> diff --git a/fs/ext4/super.c b/fs/ext4/super.c
> index 44d0c8d..8e7298d 100644
> --- a/fs/ext4/super.c
> +++ b/fs/ext4/super.c
> @@ -53,7 +53,9 @@
>  #define CREATE_TRACE_POINTS
>  #include <trace/events/ext4.h>
>  
> +#ifdef CONFIG_PROC_FS
>  static struct proc_dir_entry *ext4_proc_root;
> +#endif

This is not needed I think, since proc_dir_entry is defined in
linux/proc_fs.h even if CONFIG_PROC_FS is not defined.

>  static struct kset *ext4_kset;
>  static struct ext4_lazy_init *ext4_li_info;
>  static struct mutex ext4_li_mtx;
> @@ -812,9 +814,11 @@ static void ext4_put_super(struct super_block *sb)
>  		es->s_state = cpu_to_le16(sbi->s_mount_state);
>  		ext4_commit_super(sb, 1);
>  	}
> +#ifdef CONFIG_PROC_FS
>  	if (sbi->s_proc) {
>  		remove_proc_entry(sb->s_id, ext4_proc_root);

This function exists even if CONFIG_PROC_FS is not defined.

#define remove_proc_entry(name, parent) do {} while (0)


>  	}
> +#endif
>  	kobject_del(&sbi->s_kobj);
>  
>  	for (i = 0; i < sbi->s_gdb_count; i++)
> @@ -4984,9 +4988,11 @@ static int __init ext4_init_fs(void)
>  	ext4_kset = kset_create_and_add("ext4", NULL, fs_kobj);
>  	if (!ext4_kset)
>  		goto out6;
> +#ifdef CONFIG_PROC_FS
>  	ext4_proc_root = proc_mkdir("fs/ext4", NULL);

The same here

static inline struct proc_dir_entry *proc_mkdir(const char *name,
	struct proc_dir_entry *parent) {return NULL;}


>  	if (!ext4_proc_root)
>  		goto out5;
However this is wrong, because in case that procfs is not compiled in
ext4_proc_root will be NULL, but there is no reason to error out in this
case.

The question is whether we should error out in case that it fails even
if we have procfs compiled in. I am slightly in favour of just printing
a warning in !CONFIG_PROC_FS case.

Also this is probably why you blame commit
dd68314ccf3fb918c1fb6471817edbc60ece4b52 for the regression, since this
check was added with this commit. But I think that there is no reasong
to #ifdef everything procfs related.


> +#endif
>  
>  	err = ext4_init_feat_adverts();
>  	if (err)
> @@ -5022,8 +5028,10 @@ out2:
>  out3:
>  	ext4_exit_feat_adverts();
>  out4:
> +#ifdef CONFIG_PROC_FS
>  	remove_proc_entry("fs/ext4", NULL);

same here.

>  out5:
> +#endif
>  	kset_unregister(ext4_kset);
>  out6:
>  	ext4_exit_system_zone();
> 

-Lukas
--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Selon Lukas Czerner <lczerner@redhat.com>:

> On Mon, 3 Oct 2011, Yargil wrote:
>
> > Regression from commit dd68314ccf3fb918c1fb6471817edbc60ece4b52
>
> Hi,
>
> the commit you've mentioned does not have anything to do with procfs.
> Maybe you meant commit
>
> c9de560ded61faa5b754137b7753da252391c55a
> 	ext4: Add multi block allocator for ext4
>
> which adds multi block allocator for ext4 (mballoc.c) and has been added
> in Jan 29 2008.
>
> > ---
> >  fs/ext4/super.c |    8 ++++++++
> >  1 files changed, 8 insertions(+), 0 deletions(-)
> >
> > diff --git a/fs/ext4/super.c b/fs/ext4/super.c
> > index 44d0c8d..8e7298d 100644
> > --- a/fs/ext4/super.c
> > +++ b/fs/ext4/super.c
> > @@ -53,7 +53,9 @@
> >  #define CREATE_TRACE_POINTS
> >  #include <trace/events/ext4.h>
> >
> > +#ifdef CONFIG_PROC_FS
> >  static struct proc_dir_entry *ext4_proc_root;
> > +#endif
>
> This is not needed I think, since proc_dir_entry is defined in
> linux/proc_fs.h even if CONFIG_PROC_FS is not defined.
>
> >  static struct kset *ext4_kset;
> >  static struct ext4_lazy_init *ext4_li_info;
> >  static struct mutex ext4_li_mtx;
> > @@ -812,9 +814,11 @@ static void ext4_put_super(struct super_block *sb)
> >  		es->s_state = cpu_to_le16(sbi->s_mount_state);
> >  		ext4_commit_super(sb, 1);
> >  	}
> > +#ifdef CONFIG_PROC_FS
> >  	if (sbi->s_proc) {
> >  		remove_proc_entry(sb->s_id, ext4_proc_root);
>
> This function exists even if CONFIG_PROC_FS is not defined.
>
> #define remove_proc_entry(name, parent) do {} while (0)
>
>
> >  	}
> > +#endif
> >  	kobject_del(&sbi->s_kobj);
> >
> >  	for (i = 0; i < sbi->s_gdb_count; i++)
> > @@ -4984,9 +4988,11 @@ static int __init ext4_init_fs(void)
> >  	ext4_kset = kset_create_and_add("ext4", NULL, fs_kobj);
> >  	if (!ext4_kset)
> >  		goto out6;
> > +#ifdef CONFIG_PROC_FS
> >  	ext4_proc_root = proc_mkdir("fs/ext4", NULL);
>
> The same here
>
> static inline struct proc_dir_entry *proc_mkdir(const char *name,
> 	struct proc_dir_entry *parent) {return NULL;}
>
>
> >  	if (!ext4_proc_root)
> >  		goto out5;
> However this is wrong, because in case that procfs is not compiled in
> ext4_proc_root will be NULL, but there is no reason to error out in this
> case.
>
> The question is whether we should error out in case that it fails even
> if we have procfs compiled in. I am slightly in favour of just printing
> a warning in !CONFIG_PROC_FS case.
>
> Also this is probably why you blame commit
> dd68314ccf3fb918c1fb6471817edbc60ece4b52 for the regression, since this
> check was added with this commit. But I think that there is no reasong
> to #ifdef everything procfs related.

I did that to avoid warning. But if printing warning is not a problem then the
only part of code to be #ifdef is the test of the return value of proc_mkdir in
the ext4_init_fs function.

>
>
> > +#endif
> >
> >  	err = ext4_init_feat_adverts();
> >  	if (err)
> > @@ -5022,8 +5028,10 @@ out2:
> >  out3:
> >  	ext4_exit_feat_adverts();
> >  out4:
> > +#ifdef CONFIG_PROC_FS
> >  	remove_proc_entry("fs/ext4", NULL);
>
> same here.
>
> >  out5:
> > +#endif
> >  	kset_unregister(ext4_kset);
> >  out6:
> >  	ext4_exit_system_zone();
> >
>
> -Lukas
>


--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

On Tue, 4 Oct 2011, yargil@free.fr wrote:

> Selon Lukas Czerner <lczerner@redhat.com>:
> 
> > On Mon, 3 Oct 2011, Yargil wrote:
> >
> > > Regression from commit dd68314ccf3fb918c1fb6471817edbc60ece4b52
> >
> > Hi,
> >
> > the commit you've mentioned does not have anything to do with procfs.
> > Maybe you meant commit
> >
> > c9de560ded61faa5b754137b7753da252391c55a
> > 	ext4: Add multi block allocator for ext4
> >
> > which adds multi block allocator for ext4 (mballoc.c) and has been added
> > in Jan 29 2008.
> >
> > > ---
> > >  fs/ext4/super.c |    8 ++++++++
> > >  1 files changed, 8 insertions(+), 0 deletions(-)
> > >
> > > diff --git a/fs/ext4/super.c b/fs/ext4/super.c
> > > index 44d0c8d..8e7298d 100644
> > > --- a/fs/ext4/super.c
> > > +++ b/fs/ext4/super.c
> > > @@ -53,7 +53,9 @@
> > >  #define CREATE_TRACE_POINTS
> > >  #include <trace/events/ext4.h>
> > >
> > > +#ifdef CONFIG_PROC_FS
> > >  static struct proc_dir_entry *ext4_proc_root;
> > > +#endif
> >
> > This is not needed I think, since proc_dir_entry is defined in
> > linux/proc_fs.h even if CONFIG_PROC_FS is not defined.
> >
> > >  static struct kset *ext4_kset;
> > >  static struct ext4_lazy_init *ext4_li_info;
> > >  static struct mutex ext4_li_mtx;
> > > @@ -812,9 +814,11 @@ static void ext4_put_super(struct super_block *sb)
> > >  		es->s_state = cpu_to_le16(sbi->s_mount_state);
> > >  		ext4_commit_super(sb, 1);
> > >  	}
> > > +#ifdef CONFIG_PROC_FS
> > >  	if (sbi->s_proc) {
> > >  		remove_proc_entry(sb->s_id, ext4_proc_root);
> >
> > This function exists even if CONFIG_PROC_FS is not defined.
> >
> > #define remove_proc_entry(name, parent) do {} while (0)
> >
> >
> > >  	}
> > > +#endif
> > >  	kobject_del(&sbi->s_kobj);
> > >
> > >  	for (i = 0; i < sbi->s_gdb_count; i++)
> > > @@ -4984,9 +4988,11 @@ static int __init ext4_init_fs(void)
> > >  	ext4_kset = kset_create_and_add("ext4", NULL, fs_kobj);
> > >  	if (!ext4_kset)
> > >  		goto out6;
> > > +#ifdef CONFIG_PROC_FS
> > >  	ext4_proc_root = proc_mkdir("fs/ext4", NULL);
> >
> > The same here
> >
> > static inline struct proc_dir_entry *proc_mkdir(const char *name,
> > 	struct proc_dir_entry *parent) {return NULL;}
> >
> >
> > >  	if (!ext4_proc_root)
> > >  		goto out5;
> > However this is wrong, because in case that procfs is not compiled in
> > ext4_proc_root will be NULL, but there is no reason to error out in this
> > case.
> >
> > The question is whether we should error out in case that it fails even
> > if we have procfs compiled in. I am slightly in favour of just printing
> > a warning in !CONFIG_PROC_FS case.
> >
> > Also this is probably why you blame commit
> > dd68314ccf3fb918c1fb6471817edbc60ece4b52 for the regression, since this
> > check was added with this commit. But I think that there is no reasong
> > to #ifdef everything procfs related.
> 
> I did that to avoid warning. But if printing warning is not a problem then the
> only part of code to be #ifdef is the test of the return value of proc_mkdir in
> the ext4_init_fs function.

Exactly, it has the advantage of not spawning #ifdefs in super.c and the
advantage of not preventing the mount in case that the proc_mkdir() fails
for some reason, because we actually do not need that for file system to
work properly.

Thanks!
-Lukas

> 
> >
> >
> > > +#endif
> > >
> > >  	err = ext4_init_feat_adverts();
> > >  	if (err)
> > > @@ -5022,8 +5028,10 @@ out2:
> > >  out3:
> > >  	ext4_exit_feat_adverts();
> > >  out4:
> > > +#ifdef CONFIG_PROC_FS
> > >  	remove_proc_entry("fs/ext4", NULL);
> >
> > same here.
> >
> > >  out5:
> > > +#endif
> > >  	kset_unregister(ext4_kset);
> > >  out6:
> > >  	ext4_exit_system_zone();
> > >
> >
> > -Lukas
> >
> 
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>