Re: [PATCH v4 2/5] fs: Check if utf8 encoding is loaded before calling utf8_unload()

[Eric Biggers]

On Thu, Mar 25, 2021 at 03:31:42PM -0400, Gabriel Krisman Bertazi wrote:
> Eric Biggers <ebiggers@xxxxxxxxxx> writes:
> 
> > On Thu, Mar 25, 2021 at 05:38:08AM +0530, Shreeya Patel wrote:
> >> utf8_unload is being called if CONFIG_UNICODE is enabled.
> >> The ifdef block doesn't check if utf8 encoding has been loaded
> >> or not before calling the utf8_unload() function.
> >> This is not the expected behavior since it would sometimes lead
> >> to unloading utf8 even before loading it.
> >> Hence, add a condition which will check if sb->encoding is NOT NULL
> >> before calling the utf8_unload().
> >> 
> >> Reviewed-by: Gabriel Krisman Bertazi <krisman@xxxxxxxxxxxxx>
> >> Signed-off-by: Shreeya Patel <shreeya.patel@xxxxxxxxxxxxx>
> >> ---
> >>  fs/ext4/super.c | 6 ++++--
> >>  fs/f2fs/super.c | 9 ++++++---
> >>  2 files changed, 10 insertions(+), 5 deletions(-)
> >> 
> >> diff --git a/fs/ext4/super.c b/fs/ext4/super.c
> >> index ad34a37278cd..e438d14f9a87 100644
> >> --- a/fs/ext4/super.c
> >> +++ b/fs/ext4/super.c
> >> @@ -1259,7 +1259,8 @@ static void ext4_put_super(struct super_block *sb)
> >>  	fs_put_dax(sbi->s_daxdev);
> >>  	fscrypt_free_dummy_policy(&sbi->s_dummy_enc_policy);
> >>  #ifdef CONFIG_UNICODE
> >> -	utf8_unload(sb->s_encoding);
> >> +	if (sb->s_encoding)
> >> +		utf8_unload(sb->s_encoding);
> >>  #endif
> >>  	kfree(sbi);
> >>  }
> >
> >
> > What's the benefit of this change?  utf8_unload is a no-op when passed a NULL
> > pointer; why not keep it that way?
> 
> For the record, it no longer is a no-op after patch 5 of this series.
> Honestly, I prefer making it explicitly at the caller that we are not
> entering the function, like the patch does, instead of returning from it
> immediately.  Makes it more readable, IMO.
> 

I don't think making all the callers do the NULL check is more readable.  It's
conventional for free-like functions to accept NULL pointers.  See for example
every other function in the code snippet above -- fs_put_dax(),
fscrypt_free_dummy_policy(), and kfree().

This seems more like an issue with patch 5; it shouldn't be dropping the NULL
check from unicode_unload().

- Eric