[PATCH] fat: Use pointer to d_name[0] in put_user() for compat case

[PATCH] fat: Use pointer to d_name[0] in put_user() for compat case

[Helge Deller]

The put_user(val,ptr) macro wants a pointer in the second parameter, but in
fat_ioctl_filldir() the d_name field references a whole "array of chars".
Usually the compiler automatically converts it and uses a pointer to that
array, but it's more clean to explicitly give the real pointer to where someting
is put, which is in this case the first character of the d_name[] array.

I noticed that issue while trying to optimize the parisc put_user() macro
and used an intermediate variable to store the pointer. In that case I
got this error:

In file included from include/linux/uaccess.h:11,
                 from include/linux/compat.h:17,
                 from fs/fat/dir.c:18:
fs/fat/dir.c: In function ‘fat_ioctl_filldir’:
fs/fat/dir.c:725:33: error: invalid initializer
  725 |                 if (put_user(0, d2->d_name)                     ||         \
      |                                 ^~
include/asm/uaccess.h:152:33: note: in definition of macro ‘__put_user’
  152 |         __typeof__(ptr) __ptr = ptr;                            \
      |                                 ^~~
fs/fat/dir.c:759:1: note: in expansion of macro ‘FAT_IOCTL_FILLDIR_FUNC’
  759 | FAT_IOCTL_FILLDIR_FUNC(fat_ioctl_filldir, __fat_dirent)

The patch below cleans it up.

Signed-off-by: Helge Deller <deller@gmx.de>

diff --git a/fs/fat/dir.c b/fs/fat/dir.c
index c4a274285858..249825017da7 100644
--- a/fs/fat/dir.c
+++ b/fs/fat/dir.c
@@ -722,7 +722,7 @@ static int func(struct dir_context *ctx, const char *name, int name_len,   \
 		if (name_len >= sizeof(d1->d_name))			   \
 			name_len = sizeof(d1->d_name) - 1;		   \
 									   \
-		if (put_user(0, d2->d_name)			||	   \
+		if (put_user(0, &d2->d_name[0])			||	   \
 		    put_user(0, &d2->d_reclen)			||	   \
 		    copy_to_user(d1->d_name, name, name_len)	||	   \
 		    put_user(0, d1->d_name + name_len)		||	   \

Re: [PATCH] fat: Use pointer to d_name[0] in put_user() for compat case

[OGAWA Hirofumi]

Helge Deller <deller@gmx.de> writes:

> The put_user(val,ptr) macro wants a pointer in the second parameter, but in
> fat_ioctl_filldir() the d_name field references a whole "array of chars".
> Usually the compiler automatically converts it and uses a pointer to that
> array, but it's more clean to explicitly give the real pointer to where someting
> is put, which is in this case the first character of the d_name[] array.
>
> I noticed that issue while trying to optimize the parisc put_user() macro
> and used an intermediate variable to store the pointer. In that case I
> got this error:
>
> In file included from include/linux/uaccess.h:11,
>                  from include/linux/compat.h:17,
>                  from fs/fat/dir.c:18:
> fs/fat/dir.c: In function ‘fat_ioctl_filldir’:
> fs/fat/dir.c:725:33: error: invalid initializer
>   725 |                 if (put_user(0, d2->d_name)                     ||         \
>       |                                 ^~
> include/asm/uaccess.h:152:33: note: in definition of macro ‘__put_user’
>   152 |         __typeof__(ptr) __ptr = ptr;                            \
>       |                                 ^~~
> fs/fat/dir.c:759:1: note: in expansion of macro ‘FAT_IOCTL_FILLDIR_FUNC’
>   759 | FAT_IOCTL_FILLDIR_FUNC(fat_ioctl_filldir, __fat_dirent)
>
> The patch below cleans it up.
>
> Signed-off-by: Helge Deller <deller@gmx.de>

Looks good.

Acked-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>

> diff --git a/fs/fat/dir.c b/fs/fat/dir.c
> index c4a274285858..249825017da7 100644
> --- a/fs/fat/dir.c
> +++ b/fs/fat/dir.c
> @@ -722,7 +722,7 @@ static int func(struct dir_context *ctx, const char *name, int name_len,   \
>  		if (name_len >= sizeof(d1->d_name))			   \
>  			name_len = sizeof(d1->d_name) - 1;		   \
>  									   \
> -		if (put_user(0, d2->d_name)			||	   \
> +		if (put_user(0, &d2->d_name[0])			||	   \
>  		    put_user(0, &d2->d_reclen)			||	   \
>  		    copy_to_user(d1->d_name, name, name_len)	||	   \
>  		    put_user(0, d1->d_name + name_len)		||	   \

-- 
OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>

RE: [PATCH] fat: Use pointer to d_name[0] in put_user() for compat case

[David Laight]

From: Helge Deller
> Sent: 13 February 2022 22:10
> 
> The put_user(val,ptr) macro wants a pointer in the second parameter, but in
> fat_ioctl_filldir() the d_name field references a whole "array of chars".
> Usually the compiler automatically converts it and uses a pointer to that
> array, but it's more clean to explicitly give the real pointer to where someting
> is put, which is in this case the first character of the d_name[] array.

That just isn't true.

In C both x->char_array and &x->char_array[0] have the same type
'char *'.

The 'bug' is caused by put_user() trying to do:
	__typeof__(ptr) __ptr = ptr;
where __typeof__ is returning char[n] not char *.

I've tried a few things but can't get __typeof__ to
generate a suitable type for both a simple type and array.

	David

-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
Registration No: 1397386 (Wales)

RE: [PATCH] fat: Use pointer to d_name[0] in put_user() for compat case

[David Laight]

From: David Laight
> Sent: 14 February 2022 09:12
> 
> From: Helge Deller
> > Sent: 13 February 2022 22:10
> >
> > The put_user(val,ptr) macro wants a pointer in the second parameter, but in
> > fat_ioctl_filldir() the d_name field references a whole "array of chars".
> > Usually the compiler automatically converts it and uses a pointer to that
> > array, but it's more clean to explicitly give the real pointer to where someting
> > is put, which is in this case the first character of the d_name[] array.
> 
> That just isn't true.
> 
> In C both x->char_array and &x->char_array[0] have the same type
> 'char *'.
> 
> The 'bug' is caused by put_user() trying to do:
> 	__typeof__(ptr) __ptr = ptr;
> where __typeof__ is returning char[n] not char *.
> 
> I've tried a few things but can't get __typeof__ to
> generate a suitable type for both a simple type and array.

Actually the issue is that put_user() writes a single variable
and needs a pointer to one.
So changing to:
	put_user(0, &array[0]);
is probably fine.
But the description is all wrong.

	David

-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
Registration No: 1397386 (Wales)

Re: [PATCH] fat: Use pointer to d_name[0] in put_user() for compat case

[Andreas Schwab]

On Feb 14 2022, David Laight wrote:

> The 'bug' is caused by put_user() trying to do:
> 	__typeof__(ptr) __ptr = ptr;
> where __typeof__ is returning char[n] not char *.
>
> I've tried a few things but can't get __typeof__ to
> generate a suitable type for both a simple type and array.

Does it work to use __typeof__(&*(ptr))?

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 7578 EB47 D4E5 4D69 2510  2552 DF73 E780 A9DA AEC1
"And now for something completely different."

Re: [PATCH] fat: Use pointer to d_name[0] in put_user() for compat case

[Helge Deller]

On 2/14/22 10:27, Andreas Schwab wrote:
> On Feb 14 2022, David Laight wrote:
>
>> The 'bug' is caused by put_user() trying to do:
>> 	__typeof__(ptr) __ptr = ptr;
>> where __typeof__ is returning char[n] not char *.
>>
>> I've tried a few things but can't get __typeof__ to
>> generate a suitable type for both a simple type and array.
>
> Does it work to use __typeof__(&*(ptr))?

Yes, this works.

Helge

Re: [PATCH] fat: Use pointer to d_name[0] in put_user() for compat case

[Helge Deller]

On 2/14/22 10:26, David Laight wrote:
> From: David Laight
>> Sent: 14 February 2022 09:12
>>
>> From: Helge Deller
>>> Sent: 13 February 2022 22:10
>>>
>>> The put_user(val,ptr) macro wants a pointer in the second parameter, but in
>>> fat_ioctl_filldir() the d_name field references a whole "array of chars".
>>> Usually the compiler automatically converts it and uses a pointer to that
>>> array, but it's more clean to explicitly give the real pointer to where someting
>>> is put, which is in this case the first character of the d_name[] array.
>>
>> That just isn't true.
>>
>> In C both x->char_array and &x->char_array[0] have the same type
>> 'char *'.
>>
>> The 'bug' is caused by put_user() trying to do:
>> 	__typeof__(ptr) __ptr = ptr;
>> where __typeof__ is returning char[n] not char *.
>>
>> I've tried a few things but can't get __typeof__ to
>> generate a suitable type for both a simple type and array.
>
> Actually the issue is that put_user() writes a single variable
> and needs a pointer to one.
> So changing to:
> 	put_user(0, &array[0]);
> is probably fine.

Ok.

> But the description is all wrong.

I agree it can be improved.
Would you mind proposing a better description?

Helge

RE: [PATCH] fat: Use pointer to d_name[0] in put_user() for compat case

[David Laight]

From: Helge Deller
> Sent: 14 February 2022 11:05
> 
> On 2/14/22 10:26, David Laight wrote:
> > From: David Laight
> >> Sent: 14 February 2022 09:12
> >>
> >> From: Helge Deller
> >>> Sent: 13 February 2022 22:10
> >>>
> >>> The put_user(val,ptr) macro wants a pointer in the second parameter, but in
> >>> fat_ioctl_filldir() the d_name field references a whole "array of chars".
> >>> Usually the compiler automatically converts it and uses a pointer to that
> >>> array, but it's more clean to explicitly give the real pointer to where someting
> >>> is put, which is in this case the first character of the d_name[] array.
> >>
> >> That just isn't true.
> >>
> >> In C both x->char_array and &x->char_array[0] have the same type
> >> 'char *'.
> >>
> >> The 'bug' is caused by put_user() trying to do:
> >> 	__typeof__(ptr) __ptr = ptr;
> >> where __typeof__ is returning char[n] not char *.
> >>
> >> I've tried a few things but can't get __typeof__ to
> >> generate a suitable type for both a simple type and array.
> >
> > Actually the issue is that put_user() writes a single variable
> > and needs a pointer to one.
> > So changing to:
> > 	put_user(0, &array[0]);
> > is probably fine.
> 
> Ok.
> 
> > But the description is all wrong.
> 
> I agree it can be improved.
> Would you mind proposing a better description?

put_user() needs a pointer to a simple type.

	David

-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
Registration No: 1397386 (Wales)