[PATCH v3 1/2] kexec: fix wrong types of some local variables

[PATCH v3 1/2] kexec: fix wrong types of some local variables

[Zhang Yanfei]

The types of the following local variables:
  - ubytes/mbytes in kimage_load_crash_segment()/kimage_load_normal_segment()
  - r in vmcoreinfo_append_str()
are wrong, so fix them.

Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Simon Horman <horms@verge.net.au>
Signed-off-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>
---
 kernel/kexec.c |    6 +++---
 1 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/kernel/kexec.c b/kernel/kexec.c
index 2436ffc..3cbfcc7 100644
--- a/kernel/kexec.c
+++ b/kernel/kexec.c
@@ -789,7 +789,7 @@ static int kimage_load_normal_segment(struct kimage *image,
 					 struct kexec_segment *segment)
 {
 	unsigned long maddr;
-	unsigned long ubytes, mbytes;
+	size_t ubytes, mbytes;
 	int result;
 	unsigned char __user *buf;
 
@@ -853,7 +853,7 @@ static int kimage_load_crash_segment(struct kimage *image,
 	 * We do things a page at a time for the sake of kmap.
 	 */
 	unsigned long maddr;
-	unsigned long ubytes, mbytes;
+	size_t ubytes, mbytes;
 	int result;
 	unsigned char __user *buf;
 
@@ -1455,7 +1455,7 @@ void vmcoreinfo_append_str(const char *fmt, ...)
 {
 	va_list args;
 	char buf[0x50];
-	int r;
+	size_t r;
 
 	va_start(args, fmt);
 	r = vsnprintf(buf, sizeof(buf), fmt, args);
-- 
1.7.1

[PATCH v3 2/2] kexec: Use min_t to simplify logic

[Zhang Yanfei]

This is just a tweak: using min_t to simplify logic of variable
assignments.

v3:
- cast type of (PAGE_SIZE - (maddr & ~PAGE_MASK)) into size_t.

v2:
- Rewrite patch description as Simon suggested.
- Fix an inappropriate if test introduced by v1. Thanks Simon.

Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Simon Horman <horms@verge.net.au>
Signed-off-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>
---
 kernel/kexec.c |   24 ++++++++----------------
 1 files changed, 8 insertions(+), 16 deletions(-)

diff --git a/kernel/kexec.c b/kernel/kexec.c
index 3cbfcc7..b6b16f9 100644
--- a/kernel/kexec.c
+++ b/kernel/kexec.c
@@ -822,13 +822,9 @@ static int kimage_load_normal_segment(struct kimage *image,
 		/* Start with a clear page */
 		clear_page(ptr);
 		ptr += maddr & ~PAGE_MASK;
-		mchunk = PAGE_SIZE - (maddr & ~PAGE_MASK);
-		if (mchunk > mbytes)
-			mchunk = mbytes;
-
-		uchunk = mchunk;
-		if (uchunk > ubytes)
-			uchunk = ubytes;
+		mchunk = min_t(size_t, mbytes,
+			       (size_t)(PAGE_SIZE - (maddr & ~PAGE_MASK)));
+		uchunk = min_t(size_t, ubytes, mchunk);
 
 		result = copy_from_user(ptr, buf, uchunk);
 		kunmap(page);
@@ -874,13 +870,10 @@ static int kimage_load_crash_segment(struct kimage *image,
 		}
 		ptr = kmap(page);
 		ptr += maddr & ~PAGE_MASK;
-		mchunk = PAGE_SIZE - (maddr & ~PAGE_MASK);
-		if (mchunk > mbytes)
-			mchunk = mbytes;
-
-		uchunk = mchunk;
-		if (uchunk > ubytes) {
-			uchunk = ubytes;
+		mchunk = min_t(size_t, mbytes,
+			       (size_t)(PAGE_SIZE - (maddr & ~PAGE_MASK)));
+		uchunk = min_t(size_t, ubytes, mchunk);
+		if (mchunk > uchunk) {
 			/* Zero the trailing part of the page */
 			memset(ptr + uchunk, 0, mchunk - uchunk);
 		}
@@ -1461,8 +1454,7 @@ void vmcoreinfo_append_str(const char *fmt, ...)
 	r = vsnprintf(buf, sizeof(buf), fmt, args);
 	va_end(args);
 
-	if (r + vmcoreinfo_size > vmcoreinfo_max_size)
-		r = vmcoreinfo_max_size - vmcoreinfo_size;
+	r = min_t(size_t, r, vmcoreinfo_max_size - vmcoreinfo_size);
 
 	memcpy(&vmcoreinfo_data[vmcoreinfo_size], buf, r);
 
-- 
1.7.1

Re: [PATCH v3 2/2] kexec: Use min_t to simplify logic

[Joe Perches]

On Tue, 2013-02-26 at 13:30 +0800, Zhang Yanfei wrote:
> This is just a tweak: using min_t to simplify logic of variable
> assignments.
> 
> v3:
> - cast type of (PAGE_SIZE - (maddr & ~PAGE_MASK)) into size_t.

Why?  Isn't this just a redundant cast?

> diff --git a/kernel/kexec.c b/kernel/kexec.c
[]
> @@ -822,13 +822,9 @@ static int kimage_load_normal_segment(struct kimage *image,
[]
> +		mchunk = min_t(size_t, mbytes,
> +			       (size_t)(PAGE_SIZE - (maddr & ~PAGE_MASK)));

#define min_t(type, x, y) ({			\
	type __min1 = (x);			\
	type __min2 = (y);			\
	__min1 < __min2 ? __min1: __min2; })

Re: [PATCH v3 2/2] kexec: Use min_t to simplify logic

[Zhang Yanfei]

于 2013年02月26日 16:38, Joe Perches 写道:
> On Tue, 2013-02-26 at 13:30 +0800, Zhang Yanfei wrote:
>> This is just a tweak: using min_t to simplify logic of variable
>> assignments.
>>
>> v3:
>> - cast type of (PAGE_SIZE - (maddr & ~PAGE_MASK)) into size_t.
> 
> Why?  Isn't this just a redundant cast?
> 
>> diff --git a/kernel/kexec.c b/kernel/kexec.c
> []
>> @@ -822,13 +822,9 @@ static int kimage_load_normal_segment(struct kimage *image,
> []
>> +		mchunk = min_t(size_t, mbytes,
>> +			       (size_t)(PAGE_SIZE - (maddr & ~PAGE_MASK)));
> 
> #define min_t(type, x, y) ({			\
> 	type __min1 = (x);			\
> 	type __min2 = (y);			\
> 	__min1 < __min2 ? __min1: __min2; })
> 
> 
> 

Hmm, from the definition, the cast is redundant.

Maybe I misunderstood what Andrew meant in the mail related to v2:

"The types of PAGE_SIZE and PAGE_MASK are vague - iirc they once had
different types on different architectures, so some form of casting is
unavoidable here."

Andrew, could you please explain the casting you meant above?

Thanks
Zhang

Re: [PATCH v3 2/2] kexec: Use min_t to simplify logic

[Andrew Morton]

On Tue, 26 Feb 2013 16:49:02 +0800 Zhang Yanfei <zhangyanfei@cn.fujitsu.com> wrote:

> >> diff --git a/kernel/kexec.c b/kernel/kexec.c
> > []
> >> @@ -822,13 +822,9 @@ static int kimage_load_normal_segment(struct kimage *image,
> > []
> >> +		mchunk = min_t(size_t, mbytes,
> >> +			       (size_t)(PAGE_SIZE - (maddr & ~PAGE_MASK)));
> > 
> > #define min_t(type, x, y) ({			\
> > 	type __min1 = (x);			\
> > 	type __min2 = (y);			\
> > 	__min1 < __min2 ? __min1: __min2; })
> > 
> > 
> > 
> 
> Hmm, from the definition, the cast is redundant.
> 
> Maybe I misunderstood what Andrew meant in the mail related to v2:
> 
> "The types of PAGE_SIZE and PAGE_MASK are vague - iirc they once had
> different types on different architectures, so some form of casting is
> unavoidable here."
> 
> Andrew, could you please explain the casting you meant above?

I mean that a cast (or min_t, which is a cast) will be needed. 
The code you have here casts the same thing two times, which isn't
necessary.

Re: [PATCH v3 2/2] kexec: Use min_t to simplify logic

[Zhang Yanfei]

于 2013年02月26日 16:53, Andrew Morton 写道:
> On Tue, 26 Feb 2013 16:49:02 +0800 Zhang Yanfei <zhangyanfei@cn.fujitsu.com> wrote:
> 
>>>> diff --git a/kernel/kexec.c b/kernel/kexec.c
>>> []
>>>> @@ -822,13 +822,9 @@ static int kimage_load_normal_segment(struct kimage *image,
>>> []
>>>> +		mchunk = min_t(size_t, mbytes,
>>>> +			       (size_t)(PAGE_SIZE - (maddr & ~PAGE_MASK)));
>>>
>>> #define min_t(type, x, y) ({			\
>>> 	type __min1 = (x);			\
>>> 	type __min2 = (y);			\
>>> 	__min1 < __min2 ? __min1: __min2; })
>>>
>>>
>>>
>>
>> Hmm, from the definition, the cast is redundant.
>>
>> Maybe I misunderstood what Andrew meant in the mail related to v2:
>>
>> "The types of PAGE_SIZE and PAGE_MASK are vague - iirc they once had
>> different types on different architectures, so some form of casting is
>> unavoidable here."
>>
>> Andrew, could you please explain the casting you meant above?
> 
> I mean that a cast (or min_t, which is a cast) will be needed. 
> The code you have here casts the same thing two times, which isn't
> necessary.
> 
> 

Thanks.

Should I resend the patch again? After removing this redundant cast,
it is the same with v2.

Zhang

Re: [PATCH v3 2/2] kexec: Use min_t to simplify logic

[Andrew Morton]

On Tue, 26 Feb 2013 16:55:57 +0800
Zhang Yanfei <zhangyanfei@cn.fujitsu.com> wrote:

> ___ 2013___02___26___ 16:53, Andrew Morton ______:
> > On Tue, 26 Feb 2013 16:49:02 +0800 Zhang Yanfei <zhangyanfei@cn.fujitsu.com> wrote:
> > 
> >>>> diff --git a/kernel/kexec.c b/kernel/kexec.c
> >>> []
> >>>> @@ -822,13 +822,9 @@ static int kimage_load_normal_segment(struct kimage *image,
> >>> []
> >>>> +		mchunk = min_t(size_t, mbytes,
> >>>> +			       (size_t)(PAGE_SIZE - (maddr & ~PAGE_MASK)));
> >>>
> >>> #define min_t(type, x, y) ({			\
> >>> 	type __min1 = (x);			\
> >>> 	type __min2 = (y);			\
> >>> 	__min1 < __min2 ? __min1: __min2; })
> >>>
> >>>
> >>>
> >>
> >> Hmm, from the definition, the cast is redundant.
> >>
> >> Maybe I misunderstood what Andrew meant in the mail related to v2:
> >>
> >> "The types of PAGE_SIZE and PAGE_MASK are vague - iirc they once had
> >> different types on different architectures, so some form of casting is
> >> unavoidable here."
> >>
> >> Andrew, could you please explain the casting you meant above?
> > 
> > I mean that a cast (or min_t, which is a cast) will be needed. 
> > The code you have here casts the same thing two times, which isn't
> > necessary.
> > 
> > 
> 
> Thanks.
> 
> Should I resend the patch again? After removing this redundant cast,
> it is the same with v2.

That patch was pretty busted.   This should fix everything up:

--- a/kernel/kexec.c~kexec-use-min_t-to-simplify-logic-fix
+++ a/kernel/kexec.c
@@ -820,8 +820,8 @@ static int kimage_load_normal_segment(st
 		clear_page(ptr);
 		ptr += maddr & ~PAGE_MASK;
 		mchunk = min_t(size_t, mbytes,
-			       (size_t)(PAGE_SIZE - (maddr & ~PAGE_MASK)));
-		uchunk = min_t(size_t, ubytes, mchunk);
+				PAGE_SIZE - (maddr & ~PAGE_MASK));
+		uchunk = min(ubytes, mchunk);
 
 		result = copy_from_user(ptr, buf, uchunk);
 		kunmap(page);
@@ -868,8 +868,8 @@ static int kimage_load_crash_segment(str
 		ptr = kmap(page);
 		ptr += maddr & ~PAGE_MASK;
 		mchunk = min_t(size_t, mbytes,
-			       (size_t)(PAGE_SIZE - (maddr & ~PAGE_MASK)));
-		uchunk = min_t(size_t, ubytes, mchunk);
+				PAGE_SIZE - (maddr & ~PAGE_MASK));
+		uchunk = min(ubytes, mchunk);
 		if (mchunk > uchunk) {
 			/* Zero the trailing part of the page */
 			memset(ptr + uchunk, 0, mchunk - uchunk);
@@ -1451,7 +1451,7 @@ void vmcoreinfo_append_str(const char *f
 	r = vsnprintf(buf, sizeof(buf), fmt, args);
 	va_end(args);
 
-	r = min_t(size_t, r, vmcoreinfo_max_size - vmcoreinfo_size);
+	r = min(r, vmcoreinfo_max_size - vmcoreinfo_size);
 
 	memcpy(&vmcoreinfo_data[vmcoreinfo_size], buf, r);
 
_

Re: [PATCH v3 1/2] kexec: fix wrong types of some local variables

[Simon Horman]

On Tue, Feb 26, 2013 at 01:15:16PM +0800, Zhang Yanfei wrote:
> The types of the following local variables:
>   - ubytes/mbytes in kimage_load_crash_segment()/kimage_load_normal_segment()
>   - r in vmcoreinfo_append_str()
> are wrong, so fix them.
> 
> Cc: "Eric W. Biederman" <ebiederm@xmission.com>
> Cc: Andrew Morton <akpm@linux-foundation.org>
> Cc: Simon Horman <horms@verge.net.au>
> Signed-off-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>

No complaints here.

Reviewed-by: Simon Horman <horms@verge.net.au>

> ---
>  kernel/kexec.c |    6 +++---
>  1 files changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/kernel/kexec.c b/kernel/kexec.c
> index 2436ffc..3cbfcc7 100644
> --- a/kernel/kexec.c
> +++ b/kernel/kexec.c
> @@ -789,7 +789,7 @@ static int kimage_load_normal_segment(struct kimage *image,
>  					 struct kexec_segment *segment)
>  {
>  	unsigned long maddr;
> -	unsigned long ubytes, mbytes;
> +	size_t ubytes, mbytes;
>  	int result;
>  	unsigned char __user *buf;
>  
> @@ -853,7 +853,7 @@ static int kimage_load_crash_segment(struct kimage *image,
>  	 * We do things a page at a time for the sake of kmap.
>  	 */
>  	unsigned long maddr;
> -	unsigned long ubytes, mbytes;
> +	size_t ubytes, mbytes;
>  	int result;
>  	unsigned char __user *buf;
>  
> @@ -1455,7 +1455,7 @@ void vmcoreinfo_append_str(const char *fmt, ...)
>  {
>  	va_list args;
>  	char buf[0x50];
> -	int r;
> +	size_t r;
>  
>  	va_start(args, fmt);
>  	r = vsnprintf(buf, sizeof(buf), fmt, args);
> -- 
> 1.7.1
>