Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1583593 > unrolled thread

[PATCH] asm-generic: Force cast get_user() setting x to 0

Back to article view | Back to linux.kernel

Contents

  [PATCH] asm-generic: Force cast get_user() setting x to 0 Stephen Boyd <stephen.boyd@linaro.org> - 2017-02-17 18:00 +0100
    Re: [PATCH] asm-generic: Force cast get_user() setting x to 0 Stephen Boyd <stephen.boyd@linaro.org> - 2017-02-25 02:30 +0100
      Re: [PATCH] asm-generic: Force cast get_user() setting x to 0 Arnd Bergmann <arnd@arndb.de> - 2017-02-27 17:00 +0100

#1583593 — [PATCH] asm-generic: Force cast get_user() setting x to 0

If we call get_user() with an __le* or __be* type sparse will
complain when we assign the result to 0 on the faulting path.
Let's force cast here so that sparse doesn't complain. This
mirrors what we do in __get_user() as well.

Signed-off-by: Stephen Boyd <stephen.boyd@linaro.org>
---

I made a similar fix to arm64 headers, and it looks like asm-generic
would need the same fix.

 include/asm-generic/uaccess.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/include/asm-generic/uaccess.h b/include/asm-generic/uaccess.h
index cc6bb319e464..6352d61df7eb 100644
--- a/include/asm-generic/uaccess.h
+++ b/include/asm-generic/uaccess.h
@@ -227,7 +227,7 @@ extern int __put_user_bad(void) __attribute__((noreturn));
 	might_fault();						\
 	access_ok(VERIFY_READ, __p, sizeof(*ptr)) ?		\
 		__get_user((x), (__typeof__(*(ptr)) *)__p) :	\
-		((x) = (__typeof__(*(ptr)))0,-EFAULT);		\
+		((x) = (__force __typeof__(*(ptr)))0, -EFAULT);	\
 })
 
 #ifndef __get_user_fn
-- 
2.10.0.297.gf6727b0

[toc] | [next] | [standalone]

#1588024

Quoting Stephen Boyd (2017-02-17 08:52:29)
> If we call get_user() with an __le* or __be* type sparse will
> complain when we assign the result to 0 on the faulting path.
> Let's force cast here so that sparse doesn't complain. This
> mirrors what we do in __get_user() as well.
> 
> Signed-off-by: Stephen Boyd <stephen.boyd@linaro.org>
> ---
> 
> I made a similar fix to arm64 headers, and it looks like asm-generic
> would need the same fix.

I suppose we can ignore this. I was using sparse v0.5.0 but it seems
that later versions of sparse don't complain here anymore as Luc has
pointed out.

-Stephen

> 
>  include/asm-generic/uaccess.h | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/include/asm-generic/uaccess.h b/include/asm-generic/uaccess.h
> index cc6bb319e464..6352d61df7eb 100644
> --- a/include/asm-generic/uaccess.h
> +++ b/include/asm-generic/uaccess.h
> @@ -227,7 +227,7 @@ extern int __put_user_bad(void) __attribute__((noreturn));
>         might_fault();                                          \
>         access_ok(VERIFY_READ, __p, sizeof(*ptr)) ?             \
>                 __get_user((x), (__typeof__(*(ptr)) *)__p) :    \
> -               ((x) = (__typeof__(*(ptr)))0,-EFAULT);          \
> +               ((x) = (__force __typeof__(*(ptr)))0, -EFAULT); \
>  })
>  
>  #ifndef __get_user_fn

[toc] | [prev] | [next] | [standalone]

#1588773

On Sat, Feb 25, 2017 at 2:24 AM, Stephen Boyd <stephen.boyd@linaro.org> wrote:
> Quoting Stephen Boyd (2017-02-17 08:52:29)
>> If we call get_user() with an __le* or __be* type sparse will
>> complain when we assign the result to 0 on the faulting path.
>> Let's force cast here so that sparse doesn't complain. This
>> mirrors what we do in __get_user() as well.
>>
>> Signed-off-by: Stephen Boyd <stephen.boyd@linaro.org>
>> ---
>>
>> I made a similar fix to arm64 headers, and it looks like asm-generic
>> would need the same fix.
>
> I suppose we can ignore this. I was using sparse v0.5.0 but it seems
> that later versions of sparse don't complain here anymore as Luc has
> pointed out.

Ok, thanks for the follow-up.

   Arnd

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web