[PATCH v2 6/6] staging: exfat: replace kmalloc with kmalloc_array

[PATCH v2 6/6] staging: exfat: replace kmalloc with kmalloc_array

[Roi Martin]

Replace expressions of the form:
	kmalloc(count * size, GFP_KERNEL);
With:
	kmalloc_array(count, size, GFP_KERNEL);

Signed-off-by: Roi Martin <jroi.martin@gmail.com>
---
v2 changes:

This patch has been rebased against the branch "staging-testing" of the
tree:

https://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging.git

 drivers/staging/exfat/exfat_core.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/drivers/staging/exfat/exfat_core.c b/drivers/staging/exfat/exfat_core.c
index f71235c6a338..f4f82aecc05d 100644
--- a/drivers/staging/exfat/exfat_core.c
+++ b/drivers/staging/exfat/exfat_core.c
@@ -713,8 +713,8 @@ static s32 __load_upcase_table(struct super_block *sb, sector_t sector,
 
 	u32 checksum = 0;
 
-	upcase_table = p_fs->vol_utbl = kmalloc(UTBL_COL_COUNT * sizeof(u16 *),
-						GFP_KERNEL);
+	upcase_table = kmalloc_array(UTBL_COL_COUNT, sizeof(u16 *), GFP_KERNEL);
+	p_fs->vol_utbl = upcase_table;
 	if (!upcase_table)
 		return -ENOMEM;
 	memset(upcase_table, 0, UTBL_COL_COUNT * sizeof(u16 *));
@@ -793,8 +793,8 @@ static s32 __load_default_upcase_table(struct super_block *sb)
 	u16	uni = 0;
 	u16 **upcase_table;
 
-	upcase_table = p_fs->vol_utbl = kmalloc(UTBL_COL_COUNT * sizeof(u16 *),
-						GFP_KERNEL);
+	upcase_table = kmalloc_array(UTBL_COL_COUNT, sizeof(u16 *), GFP_KERNEL);
+	p_fs->vol_utbl = upcase_table;
 	if (!upcase_table)
 		return -ENOMEM;
 	memset(upcase_table, 0, UTBL_COL_COUNT * sizeof(u16 *));
-- 
2.20.1

Re: [PATCH v2 6/6] staging: exfat: replace kmalloc with kmalloc_array

[Dan Carpenter]

On Thu, Oct 31, 2019 at 01:31:39PM +0100, Roi Martin wrote:
> diff --git a/drivers/staging/exfat/exfat_core.c b/drivers/staging/exfat/exfat_core.c
> index f71235c6a338..f4f82aecc05d 100644
> --- a/drivers/staging/exfat/exfat_core.c
> +++ b/drivers/staging/exfat/exfat_core.c
> @@ -713,8 +713,8 @@ static s32 __load_upcase_table(struct super_block *sb, sector_t sector,
>  
>  	u32 checksum = 0;
>  
> -	upcase_table = p_fs->vol_utbl = kmalloc(UTBL_COL_COUNT * sizeof(u16 *),
> -						GFP_KERNEL);
> +	upcase_table = kmalloc_array(UTBL_COL_COUNT, sizeof(u16 *), GFP_KERNEL);
> +	p_fs->vol_utbl = upcase_table;

This patch is fine, but one idea for future patches is that you could
remove the "upcase_table" variable and use "p_fs->vol_utbl" everywhere
instead.

>  	if (!upcase_table)
>  		return -ENOMEM;

regards,
dan carpenter

Re: [PATCH v2 6/6] staging: exfat: replace kmalloc with kmalloc_array

[Roi Martin]

> > diff --git a/drivers/staging/exfat/exfat_core.c b/drivers/staging/exfat/exfat_core.c
> > index f71235c6a338..f4f82aecc05d 100644
> > --- a/drivers/staging/exfat/exfat_core.c
> > +++ b/drivers/staging/exfat/exfat_core.c
> > @@ -713,8 +713,8 @@ static s32 __load_upcase_table(struct super_block *sb, sector_t sector,
> >  
> >  	u32 checksum = 0;
> >  
> > -	upcase_table = p_fs->vol_utbl = kmalloc(UTBL_COL_COUNT * sizeof(u16 *),
> > -						GFP_KERNEL);
> > +	upcase_table = kmalloc_array(UTBL_COL_COUNT, sizeof(u16 *), GFP_KERNEL);
> > +	p_fs->vol_utbl = upcase_table;
> 
> This patch is fine, but one idea for future patches is that you could
> remove the "upcase_table" variable and use "p_fs->vol_utbl" everywhere
> instead.

Thanks for the suggestion.

This is my first contribution and I tried to introduce the minimum
number of changes necessary to fix the issues reported by checkpatch.pl.
Also, I'm still immersed in getting familiar with the contribution
process and the code.

Do you think it makes sense to include this change in a future patch
series along with other refactoring? Or, should I modify this patch?

By the way, upcase_table is sometimes accessed in quite complex ways.
For instance:

	upcase_table[col_index][get_row_index(index)] = uni;

Where having an intermediate variable instead of using the struct field
directly seems to improve readability a bit. Otherwise:

	p_fs->vol_utbl[col_index][get_row_index(index)] = uni;

I assume, in cases like this, from a coding style perspective, the
following approach is preferred:

	row_index = get_row_index(index);
	p_fs->vol_utbl[col_index][row_index] = uni;

Is that correct?

Regards,

	Roi Martin

Re: [PATCH v2 6/6] staging: exfat: replace kmalloc with kmalloc_array

[Dan Carpenter]

On Thu, Oct 31, 2019 at 05:03:56PM +0100, Roi Martin wrote:
> > > diff --git a/drivers/staging/exfat/exfat_core.c b/drivers/staging/exfat/exfat_core.c
> > > index f71235c6a338..f4f82aecc05d 100644
> > > --- a/drivers/staging/exfat/exfat_core.c
> > > +++ b/drivers/staging/exfat/exfat_core.c
> > > @@ -713,8 +713,8 @@ static s32 __load_upcase_table(struct super_block *sb, sector_t sector,
> > >  
> > >  	u32 checksum = 0;
> > >  
> > > -	upcase_table = p_fs->vol_utbl = kmalloc(UTBL_COL_COUNT * sizeof(u16 *),
> > > -						GFP_KERNEL);
> > > +	upcase_table = kmalloc_array(UTBL_COL_COUNT, sizeof(u16 *), GFP_KERNEL);
> > > +	p_fs->vol_utbl = upcase_table;
> > 
> > This patch is fine, but one idea for future patches is that you could
> > remove the "upcase_table" variable and use "p_fs->vol_utbl" everywhere
> > instead.
> 
> Thanks for the suggestion.
> 
> This is my first contribution and I tried to introduce the minimum
> number of changes necessary to fix the issues reported by checkpatch.pl.
> Also, I'm still immersed in getting familiar with the contribution
> process and the code.
> 
> Do you think it makes sense to include this change in a future patch
> series along with other refactoring? Or, should I modify this patch?

No don't modify the patch.  The patch is fine.

> 
> By the way, upcase_table is sometimes accessed in quite complex ways.
> For instance:
> 
> 	upcase_table[col_index][get_row_index(index)] = uni;
> 
> Where having an intermediate variable instead of using the struct field
> directly seems to improve readability a bit. Otherwise:
> 
> 	p_fs->vol_utbl[col_index][get_row_index(index)] = uni;

This line isn't very complex.  It's fine.


> 
> I assume, in cases like this, from a coding style perspective, the
> following approach is preferred:
> 
> 	row_index = get_row_index(index);
> 	p_fs->vol_utbl[col_index][row_index] = uni;

But this is better, yes.

regards,
dan carpenter