Re: [PATCH] pre7-1 semicolon & nicely readableB

Re: [PATCH] pre7-1 semicolon & nicely readableB

[Roel van der Goot]

Hi Rik,

I want to inform you that there is a subtle difference between
the following two loops:

(i)

   while ((mm->swap_cnt << 2 * (i + 1) < max_cnt)
                   && i++ < 10);

(ii)

   while ((mm->swap_cnt << 2 * (i + 1) < max_cnt)
                   && i < 10)
           i++;

Cheers,
Roel.


--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux.eu.org/Linux-MM/

Re: [PATCH] pre7-1 semicolon & nicely readableB

[Rik van Riel]

On Mon, 1 May 2000, Roel van der Goot wrote:

> I want to inform you that there is a subtle difference between
> the following two loops:
> 
> (i)
> 
>    while ((mm->swap_cnt << 2 * (i + 1) < max_cnt)
>                    && i++ < 10);
> 
> (ii)
> 
>    while ((mm->swap_cnt << 2 * (i + 1) < max_cnt)
>                    && i < 10)
>            i++;

I want to inform you that you're wrong. The only difference is
in readability.

If the first test fails, the clause behind the && won't be run.

Furthermore, i will only reach 10 if the RSS difference between
the current process and the biggest process is more than a factor
2^21 ... which can never happen on 32-bit hardware, unless the
RSS of the current process is 0.

In fact, the <10 test is only there to prevent infinite looping
for when a process with 0 swap_cnt "slips through" the tests above.

regards,

Rik
--
The Internet is not a network of computers. It is a network
of people. That is its real strength.

Wanna talk about the kernel?  irc.openprojects.net / #kernelnewbies
http://www.conectiva.com/		http://www.surriel.com/

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux.eu.org/Linux-MM/

Re: [PATCH] pre7-1 semicolon & nicely readable

[Roel van der Goot]

On Mon, 1 May 2000, Rik van Riel wrote:

> In fact, the <10 test is only there to prevent infinite looping
> for when a process with 0 swap_cnt "slips through" the tests above.

In case of a "slip through" variable i will have a different
value after the loop. But I understand from your reply that you
covered that case.

Cheers,
Roel.


--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux.eu.org/Linux-MM/

Re: [PATCH] pre7-1 semicolon & nicely readableB

[Chris Evans]

On Mon, 1 May 2000, Rik van Riel wrote:

> I want to inform you that you're wrong. The only difference is
> in readability.

[..]

> In fact, the <10 test is only there to prevent infinite looping
> for when a process with 0 swap_cnt "slips through" the tests above.

If such a value should never "slip through", then, for readability, you
want an assert (e.g. BUG() ).

Chris

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux.eu.org/Linux-MM/

Re: [PATCH] pre7-1 semicolon & nicely readableB

[Rik van Riel]

On Tue, 2 May 2000, Chris Evans wrote:
> On Mon, 1 May 2000, Rik van Riel wrote:
> 
> > I want to inform you that you're wrong. The only difference is
> > in readability.
> 
> [..]
> 
> > In fact, the <10 test is only there to prevent infinite looping
> > for when a process with 0 swap_cnt "slips through" the tests above.
> 
> If such a value should never "slip through", then, for
> readability, you want an assert (e.g. BUG() ).

It's ok for them to slip through, it's just not ok for the kernel
to go into an infinite loop here...

Rik
--
The Internet is not a network of computers. It is a network
of people. That is its real strength.

Wanna talk about the kernel?  irc.openprojects.net / #kernelnewbies
http://www.conectiva.com/		http://www.surriel.com/

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux.eu.org/Linux-MM/

[PATCH] pre7-1 semicolon & nicely readable

[Rik van Riel]

Hi Linus,

after getting a number of complaints that the semicolon
thing wasn't nicely readable (they're right) I changed my
code as per the patch below.

Could you please put the more readable version in the
next pre kernel?

thanks,

Rik
--
The Internet is not a network of computers. It is a network
of people. That is its real strength.

Wanna talk about the kernel?  irc.openprojects.net / #kernelnewbies
http://www.conectiva.com/		http://www.surriel.com/



--- mm/vmscan.c.org2	Sat Apr 29 06:53:00 2000
+++ mm/vmscan.c	Mon May  1 11:42:34 2000
@@ -384,11 +384,12 @@
 			for (; p != &init_task; p = p->next_task) {
 				int i = 0;
 				struct mm_struct *mm = p->mm;
-				if (!p->swappable || !mm || mm->rss <= 0)
+				if (!p->swappable || !mm || mm->swap_cnt <= 0)
 					continue;
 				/* small processes are swapped out less */
 				while ((mm->swap_cnt << 2 * (i + 1) < max_cnt)
-						&& i++ < 10)
+						&& i < 10)
+					i++;
 				mm->swap_cnt >>= i;
 				mm->swap_cnt += i; /* if swap_cnt reaches 0 */
 				/* we're big -> hog treatment */

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux.eu.org/Linux-MM/