linux-m68k.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH] m68knommu: switch to using asm-generic/uaccess.h
@ 2020-09-09  2:18 Greg Ungerer
  2020-09-09  6:55 ` Geert Uytterhoeven
  2020-09-09  6:56 ` Christoph Hellwig
  0 siblings, 2 replies; 5+ messages in thread
From: Greg Ungerer @ 2020-09-09  2:18 UTC (permalink / raw)
  To: linux-m68k; +Cc: hch, geert, arnd, Greg Ungerer, kernel test robot

Switch to using the asm-generic/uaccess functions for non-MMU builds.
Remove all the m68knommu local specific uaccess defines and macros.

There is nothing so special about the m68knommu targets that they cannot
use all of the asm-generic uaccess support. Using the asm-generic
uaccess definitions also resolves some of the existing problems with
missing __user annotations in the m68knommu specific functions.

The elimination of all of the contents of uaccess_no.h means we can fold
the uaccess_mm.h back into uaccess.h - and just have the single file
now.

The resulting generated code ends up being slightly smaller (by a few
hundred bytes) due to the compilers ability to better optimize load
and stores without forcing its hand with asm statements.

Specifically trivial cases like this contrived example:

    get_user(x, ptr);
    x++;
    put_user(x, ptr);

end up now being optimized to a single instruction on m68k. More
generally the compiler can avoid using a temporary register in many
cases as well.

Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Greg Ungerer <gerg@linux-m68k.org>
---
 arch/m68k/Kconfig                             |   1 +
 .../include/asm/{uaccess_mm.h => uaccess.h}   |   7 +
 arch/m68k/include/asm/uaccess_no.h            | 160 ------------------
 3 files changed, 8 insertions(+), 160 deletions(-)
 rename arch/m68k/include/asm/{uaccess_mm.h => uaccess.h} (98%)
 delete mode 100644 arch/m68k/include/asm/uaccess_no.h

diff --git a/arch/m68k/Kconfig b/arch/m68k/Kconfig
index 6f2f38d05772..aefffebc0afa 100644
--- a/arch/m68k/Kconfig
+++ b/arch/m68k/Kconfig
@@ -24,6 +24,7 @@ config M68K
 	select GENERIC_IOMAP
 	select GENERIC_STRNCPY_FROM_USER if MMU
 	select GENERIC_STRNLEN_USER if MMU
+	select UACCESS_MEMCPY if !MMU
 	select ARCH_WANT_IPC_PARSE_VERSION
 	select HAVE_FUTEX_CMPXCHG if MMU && FUTEX
 	select HAVE_MOD_ARCH_SPECIFIC
diff --git a/arch/m68k/include/asm/uaccess_mm.h b/arch/m68k/include/asm/uaccess.h
similarity index 98%
rename from arch/m68k/include/asm/uaccess_mm.h
rename to arch/m68k/include/asm/uaccess.h
index 9ae9f8d05925..f98208ccbbcd 100644
--- a/arch/m68k/include/asm/uaccess_mm.h
+++ b/arch/m68k/include/asm/uaccess.h
@@ -2,12 +2,15 @@
 #ifndef __M68K_UACCESS_H
 #define __M68K_UACCESS_H
 
+#ifdef CONFIG_MMU
+
 /*
  * User space memory access functions
  */
 #include <linux/compiler.h>
 #include <linux/types.h>
 #include <asm/segment.h>
+#include <asm/extable.h>
 
 /* We let the MMU do all checking */
 static inline int access_ok(const void __user *addr,
@@ -387,4 +390,8 @@ unsigned long __clear_user(void __user *to, unsigned long n);
 
 #define clear_user	__clear_user
 
+#else /* !CONFIG_MMU */
+#include <asm-generic/uaccess.h>
+#endif
+
 #endif /* _M68K_UACCESS_H */
diff --git a/arch/m68k/include/asm/uaccess_no.h b/arch/m68k/include/asm/uaccess_no.h
deleted file mode 100644
index dcfb69361408..000000000000
--- a/arch/m68k/include/asm/uaccess_no.h
+++ /dev/null
@@ -1,160 +0,0 @@
-/* SPDX-License-Identifier: GPL-2.0 */
-#ifndef __M68KNOMMU_UACCESS_H
-#define __M68KNOMMU_UACCESS_H
-
-/*
- * User space memory access functions
- */
-#include <linux/string.h>
-
-#include <asm/segment.h>
-
-#define access_ok(addr,size)	_access_ok((unsigned long)(addr),(size))
-
-/*
- * It is not enough to just have access_ok check for a real RAM address.
- * This would disallow the case of code/ro-data running XIP in flash/rom.
- * Ideally we would check the possible flash ranges too, but that is
- * currently not so easy.
- */
-static inline int _access_ok(unsigned long addr, unsigned long size)
-{
-	return 1;
-}
-
-/*
- * These are the main single-value transfer routines.  They automatically
- * use the right size if we just have the right pointer type.
- */
-
-#define put_user(x, ptr)				\
-({							\
-    int __pu_err = 0;					\
-    typeof(*(ptr)) __pu_val = (x);			\
-    switch (sizeof (*(ptr))) {				\
-    case 1:						\
-	__put_user_asm(__pu_err, __pu_val, ptr, b);	\
-	break;						\
-    case 2:						\
-	__put_user_asm(__pu_err, __pu_val, ptr, w);	\
-	break;						\
-    case 4:						\
-	__put_user_asm(__pu_err, __pu_val, ptr, l);	\
-	break;						\
-    case 8:						\
-	memcpy((void __force *)ptr, &__pu_val, sizeof(*(ptr))); \
-	break;						\
-    default:						\
-	__pu_err = __put_user_bad();			\
-	break;						\
-    }							\
-    __pu_err;						\
-})
-#define __put_user(x, ptr) put_user(x, ptr)
-
-extern int __put_user_bad(void);
-
-/*
- * Tell gcc we read from memory instead of writing: this is because
- * we do not write to any memory gcc knows about, so there are no
- * aliasing issues.
- */
-
-#define __ptr(x) ((unsigned long __user *)(x))
-
-#define __put_user_asm(err,x,ptr,bwl)				\
-	__asm__ ("move" #bwl " %0,%1"				\
-		: /* no outputs */						\
-		:"d" (x),"m" (*__ptr(ptr)) : "memory")
-
-#define get_user(x, ptr)					\
-({								\
-    int __gu_err = 0;						\
-    switch (sizeof(*(ptr))) {					\
-    case 1:							\
-	__get_user_asm(__gu_err, x, ptr, b, "=d");		\
-	break;							\
-    case 2:							\
-	__get_user_asm(__gu_err, x, ptr, w, "=r");		\
-	break;							\
-    case 4:							\
-	__get_user_asm(__gu_err, x, ptr, l, "=r");		\
-	break;							\
-    case 8: {							\
-	union {							\
-	    u64 l;						\
-	    __typeof__(*(ptr)) t;				\
-	} __gu_val;						\
-	memcpy(&__gu_val.l, (const void __force *)ptr, sizeof(__gu_val.l)); \
-	(x) = __gu_val.t;					\
-	break;							\
-    }								\
-    default:							\
-	__gu_err = __get_user_bad();				\
-	break;							\
-    }								\
-    __gu_err;							\
-})
-#define __get_user(x, ptr) get_user(x, ptr)
-
-extern int __get_user_bad(void);
-
-#define __get_user_asm(err,x,ptr,bwl,reg)			\
-	__asm__ ("move" #bwl " %1,%0"				\
-		 : "=d" (x)					\
-		 : "m" (*__ptr(ptr)))
-
-static inline unsigned long
-raw_copy_from_user(void *to, const void __user *from, unsigned long n)
-{
-	memcpy(to, (__force const void *)from, n);
-	return 0;
-}
-
-static inline unsigned long
-raw_copy_to_user(void __user *to, const void *from, unsigned long n)
-{
-	memcpy((__force void *)to, from, n);
-	return 0;
-}
-#define INLINE_COPY_FROM_USER
-#define INLINE_COPY_TO_USER
-
-/*
- * Copy a null terminated string from userspace.
- */
-
-static inline long
-strncpy_from_user(char *dst, const char *src, long count)
-{
-	char *tmp;
-	strncpy(dst, src, count);
-	for (tmp = dst; *tmp && count > 0; tmp++, count--)
-		;
-	return(tmp - dst); /* DAVIDM should we count a NUL ?  check getname */
-}
-
-/*
- * Return the size of a string (including the ending 0)
- *
- * Return 0 on exception, a value greater than N if too long
- */
-static inline long strnlen_user(const char *src, long n)
-{
-	return(strlen(src) + 1); /* DAVIDM make safer */
-}
-
-/*
- * Zero Userspace
- */
-
-static inline unsigned long
-__clear_user(void *to, unsigned long n)
-{
-	memset(to, 0, n);
-	return 0;
-}
-
-#define	clear_user(to,n)	__clear_user(to,n)
-
-#endif /* _M68KNOMMU_UACCESS_H */
-- 
2.25.1


^ permalink raw reply related	[flat|nested] 5+ messages in thread

* Re: [PATCH] m68knommu: switch to using asm-generic/uaccess.h
  2020-09-09  2:18 [PATCH] m68knommu: switch to using asm-generic/uaccess.h Greg Ungerer
@ 2020-09-09  6:55 ` Geert Uytterhoeven
  2020-09-09  7:54   ` Greg Ungerer
  2020-09-09  6:56 ` Christoph Hellwig
  1 sibling, 1 reply; 5+ messages in thread
From: Geert Uytterhoeven @ 2020-09-09  6:55 UTC (permalink / raw)
  To: Greg Ungerer
  Cc: Linux/m68k, Christoph Hellwig, Arnd Bergmann, kernel test robot

Hi Greg,

On Wed, Sep 9, 2020 at 4:18 AM Greg Ungerer <gerg@linux-m68k.org> wrote:
> Switch to using the asm-generic/uaccess functions for non-MMU builds.
> Remove all the m68knommu local specific uaccess defines and macros.
>
> There is nothing so special about the m68knommu targets that they cannot
> use all of the asm-generic uaccess support. Using the asm-generic
> uaccess definitions also resolves some of the existing problems with
> missing __user annotations in the m68knommu specific functions.
>
> The elimination of all of the contents of uaccess_no.h means we can fold
> the uaccess_mm.h back into uaccess.h - and just have the single file
> now.
>
> The resulting generated code ends up being slightly smaller (by a few
> hundred bytes) due to the compilers ability to better optimize load
> and stores without forcing its hand with asm statements.
>
> Specifically trivial cases like this contrived example:
>
>     get_user(x, ptr);
>     x++;
>     put_user(x, ptr);
>
> end up now being optimized to a single instruction on m68k. More
> generally the compiler can avoid using a temporary register in many
> cases as well.
>
> Reported-by: kernel test robot <lkp@intel.com>
> Signed-off-by: Greg Ungerer <gerg@linux-m68k.org>

Thanks for your patch!

> rename from arch/m68k/include/asm/uaccess_mm.h
> rename to arch/m68k/include/asm/uaccess.h
> index 9ae9f8d05925..f98208ccbbcd 100644
> --- a/arch/m68k/include/asm/uaccess_mm.h
> +++ b/arch/m68k/include/asm/uaccess.h
> @@ -2,12 +2,15 @@
>  #ifndef __M68K_UACCESS_H
>  #define __M68K_UACCESS_H
>
> +#ifdef CONFIG_MMU
> +
>  /*
>   * User space memory access functions
>   */
>  #include <linux/compiler.h>
>  #include <linux/types.h>
>  #include <asm/segment.h>
> +#include <asm/extable.h>

For a moment, I wondered where this new include came from...

Seems like git doesn't show what happened to the old
arch/m68k/include/asm/uaccess.h file, which was overwritten by the
rename:

    -/* SPDX-License-Identifier: GPL-2.0 */
    -#ifdef __uClinux__
    -#include <asm/uaccess_no.h>
    -#else
    -#include <asm/uaccess_mm.h>
    -#endif
    -#include <asm/extable.h>

There it is ;-)

Acked-by: Geert Uytterhoeven <geert@linux-m68k.org>

>  /* We let the MMU do all checking */
>  static inline int access_ok(const void __user *addr,
> @@ -387,4 +390,8 @@ unsigned long __clear_user(void __user *to, unsigned long n);
>
>  #define clear_user     __clear_user
>
> +#else /* !CONFIG_MMU */
> +#include <asm-generic/uaccess.h>
> +#endif
> +
>  #endif /* _M68K_UACCESS_H */

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: [PATCH] m68knommu: switch to using asm-generic/uaccess.h
  2020-09-09  2:18 [PATCH] m68knommu: switch to using asm-generic/uaccess.h Greg Ungerer
  2020-09-09  6:55 ` Geert Uytterhoeven
@ 2020-09-09  6:56 ` Christoph Hellwig
  2020-09-09  7:58   ` Greg Ungerer
  1 sibling, 1 reply; 5+ messages in thread
From: Christoph Hellwig @ 2020-09-09  6:56 UTC (permalink / raw)
  To: Greg Ungerer; +Cc: linux-m68k, hch, geert, arnd, kernel test robot

On Wed, Sep 09, 2020 at 12:18:31PM +1000, Greg Ungerer wrote:
> Switch to using the asm-generic/uaccess functions for non-MMU builds.
> Remove all the m68knommu local specific uaccess defines and macros.
> 
> There is nothing so special about the m68knommu targets that they cannot
> use all of the asm-generic uaccess support. Using the asm-generic
> uaccess definitions also resolves some of the existing problems with
> missing __user annotations in the m68knommu specific functions.
> 
> The elimination of all of the contents of uaccess_no.h means we can fold
> the uaccess_mm.h back into uaccess.h - and just have the single file
> now.
> 
> The resulting generated code ends up being slightly smaller (by a few
> hundred bytes) due to the compilers ability to better optimize load
> and stores without forcing its hand with asm statements.
> 
> Specifically trivial cases like this contrived example:
> 
>     get_user(x, ptr);
>     x++;
>     put_user(x, ptr);
> 
> end up now being optimized to a single instruction on m68k. More
> generally the compiler can avoid using a temporary register in many
> cases as well.

This looks great!

Reviewed-by: Christoph Hellwig <hch@lst.de>

^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: [PATCH] m68knommu: switch to using asm-generic/uaccess.h
  2020-09-09  6:55 ` Geert Uytterhoeven
@ 2020-09-09  7:54   ` Greg Ungerer
  0 siblings, 0 replies; 5+ messages in thread
From: Greg Ungerer @ 2020-09-09  7:54 UTC (permalink / raw)
  To: Geert Uytterhoeven
  Cc: Linux/m68k, Christoph Hellwig, Arnd Bergmann, kernel test robot

Hi Geert,

On 9/9/20 4:55 pm, Geert Uytterhoeven wrote:
> Hi Greg,
> 
> On Wed, Sep 9, 2020 at 4:18 AM Greg Ungerer <gerg@linux-m68k.org> wrote:
>> Switch to using the asm-generic/uaccess functions for non-MMU builds.
>> Remove all the m68knommu local specific uaccess defines and macros.
>>
>> There is nothing so special about the m68knommu targets that they cannot
>> use all of the asm-generic uaccess support. Using the asm-generic
>> uaccess definitions also resolves some of the existing problems with
>> missing __user annotations in the m68knommu specific functions.
>>
>> The elimination of all of the contents of uaccess_no.h means we can fold
>> the uaccess_mm.h back into uaccess.h - and just have the single file
>> now.
>>
>> The resulting generated code ends up being slightly smaller (by a few
>> hundred bytes) due to the compilers ability to better optimize load
>> and stores without forcing its hand with asm statements.
>>
>> Specifically trivial cases like this contrived example:
>>
>>      get_user(x, ptr);
>>      x++;
>>      put_user(x, ptr);
>>
>> end up now being optimized to a single instruction on m68k. More
>> generally the compiler can avoid using a temporary register in many
>> cases as well.
>>
>> Reported-by: kernel test robot <lkp@intel.com>
>> Signed-off-by: Greg Ungerer <gerg@linux-m68k.org>
> 
> Thanks for your patch!
> 
>> rename from arch/m68k/include/asm/uaccess_mm.h
>> rename to arch/m68k/include/asm/uaccess.h
>> index 9ae9f8d05925..f98208ccbbcd 100644
>> --- a/arch/m68k/include/asm/uaccess_mm.h
>> +++ b/arch/m68k/include/asm/uaccess.h
>> @@ -2,12 +2,15 @@
>>   #ifndef __M68K_UACCESS_H
>>   #define __M68K_UACCESS_H
>>
>> +#ifdef CONFIG_MMU
>> +
>>   /*
>>    * User space memory access functions
>>    */
>>   #include <linux/compiler.h>
>>   #include <linux/types.h>
>>   #include <asm/segment.h>
>> +#include <asm/extable.h>
> 
> For a moment, I wondered where this new include came from...
> 
> Seems like git doesn't show what happened to the old
> arch/m68k/include/asm/uaccess.h file, which was overwritten by the
> rename:
> 
>      -/* SPDX-License-Identifier: GPL-2.0 */
>      -#ifdef __uClinux__
>      -#include <asm/uaccess_no.h>
>      -#else
>      -#include <asm/uaccess_mm.h>
>      -#endif
>      -#include <asm/extable.h>
> 
> There it is ;-)
> 
> Acked-by: Geert Uytterhoeven <geert@linux-m68k.org>

Thanks Geert.

I noticed that in the diff too. I generated this one with
"git format-patch -1 -M -B" to make it easier to review. Otherwise you end up
with all those identical lines being inserted and deleted.
But it makes that "#include <asm/extable.h>" appear to come out of nowhere.

Regards
Greg



>>   /* We let the MMU do all checking */
>>   static inline int access_ok(const void __user *addr,
>> @@ -387,4 +390,8 @@ unsigned long __clear_user(void __user *to, unsigned long n);
>>
>>   #define clear_user     __clear_user
>>
>> +#else /* !CONFIG_MMU */
>> +#include <asm-generic/uaccess.h>
>> +#endif
>> +
>>   #endif /* _M68K_UACCESS_H */
> 
> Gr{oetje,eeting}s,
> 
>                          Geert
> 

^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: [PATCH] m68knommu: switch to using asm-generic/uaccess.h
  2020-09-09  6:56 ` Christoph Hellwig
@ 2020-09-09  7:58   ` Greg Ungerer
  0 siblings, 0 replies; 5+ messages in thread
From: Greg Ungerer @ 2020-09-09  7:58 UTC (permalink / raw)
  To: Christoph Hellwig; +Cc: linux-m68k, geert, arnd, kernel test robot


On 9/9/20 4:56 pm, Christoph Hellwig wrote:
> On Wed, Sep 09, 2020 at 12:18:31PM +1000, Greg Ungerer wrote:
>> Switch to using the asm-generic/uaccess functions for non-MMU builds.
>> Remove all the m68knommu local specific uaccess defines and macros.
>>
>> There is nothing so special about the m68knommu targets that they cannot
>> use all of the asm-generic uaccess support. Using the asm-generic
>> uaccess definitions also resolves some of the existing problems with
>> missing __user annotations in the m68knommu specific functions.
>>
>> The elimination of all of the contents of uaccess_no.h means we can fold
>> the uaccess_mm.h back into uaccess.h - and just have the single file
>> now.
>>
>> The resulting generated code ends up being slightly smaller (by a few
>> hundred bytes) due to the compilers ability to better optimize load
>> and stores without forcing its hand with asm statements.
>>
>> Specifically trivial cases like this contrived example:
>>
>>      get_user(x, ptr);
>>      x++;
>>      put_user(x, ptr);
>>
>> end up now being optimized to a single instruction on m68k. More
>> generally the compiler can avoid using a temporary register in many
>> cases as well.
> 
> This looks great!
> 
> Reviewed-by: Christoph Hellwig <hch@lst.de>

Thanks Christoph.
I'll push this into the for-next branch of the m68knommu git tree.

Regards
Greg


^ permalink raw reply	[flat|nested] 5+ messages in thread

end of thread, other threads:[~2020-09-09  7:58 UTC | newest]

Thread overview: 5+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-09-09  2:18 [PATCH] m68knommu: switch to using asm-generic/uaccess.h Greg Ungerer
2020-09-09  6:55 ` Geert Uytterhoeven
2020-09-09  7:54   ` Greg Ungerer
2020-09-09  6:56 ` Christoph Hellwig
2020-09-09  7:58   ` Greg Ungerer

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).