put_user/get_user warnings for x86

• Next message: Adam J. Richter: "Re: bio_chain: proposed solution for bio_alloc failure and large IO simplification"
• Previous message: Sancho Dauskardt: "Re: /proc/scsi/map"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

It's not *neccessarily wrong* to use put_user(int, long *), but it's
probably worth warning for:

Name: put_user/get_user extra checks (x86 only)
Author: Rusty Russell
Status: Trivial

D: This makes put_user and get_user warn if the type of the userspace ptr
D: and the kernel variable are incompatible, eg:
D: int x;
D: char *uptr;
D: ...
D: put_user(x,uptr); <= Warn here
D: get_user(x,uptr); <= Warn here

diff -urN -I \$.*\$ --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal linux-2.5.21/include/asm-i386/uaccess.h working-2.5.21-putusercheck/include/asm-i386/uaccess.h
--- linux-2.5.21/include/asm-i386/uaccess.h Wed Feb 20 17:56:40 2002
+++ working-2.5.21-putusercheck/include/asm-i386/uaccess.h Sun Jun 16 05:55:27 2002
@@ -133,16 +133,18 @@
 extern void __put_user_bad(void);
 
 #define put_user(x,ptr) \
- __put_user_check((__typeof__(*(ptr)))(x),(ptr),sizeof(*(ptr)))
+ __put_user_check((x),(ptr),sizeof(*(ptr)))
 
 #define __get_user(x,ptr) \
   __get_user_nocheck((x),(ptr),sizeof(*(ptr)))
 #define __put_user(x,ptr) \
- __put_user_nocheck((__typeof__(*(ptr)))(x),(ptr),sizeof(*(ptr)))
+ __put_user_nocheck((x),(ptr),sizeof(*(ptr)))
 
 #define __put_user_nocheck(x,ptr,size) \
 ({ \
         long __pu_err; \
+ /* Elicit warning */ \
+ __typeof__(x) *__pu_check __attribute__((unused)) = ptr; \
         __put_user_size((x),(ptr),(size),__pu_err); \
         __pu_err; \
 })
@@ -151,7 +153,8 @@
 #define __put_user_check(x,ptr,size) \
 ({ \
         long __pu_err = -EFAULT; \
- __typeof__(*(ptr)) *__pu_addr = (ptr); \
+ /* Elicit warning */ \
+ __typeof__(x) *__pu_addr = (ptr); \
         if (access_ok(VERIFY_WRITE,__pu_addr,size)) \
                 __put_user_size((x),__pu_addr,(size),__pu_err); \
         __pu_err; \

--
  Anyone who quotes me in their sig is an idiot. -- Rusty Russell.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

• Next message: Adam J. Richter: "Re: bio_chain: proposed solution for bio_alloc failure and large IO simplification"
• Previous message: Sancho Dauskardt: "Re: /proc/scsi/map"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

This archive was generated by hypermail 2b29 : Sat Jun 15 2002 - 22:00:33 EST