public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
* tmpfs doesn't update free memory stats?
@ 2001-05-04 10:09 Jacek Kopecky
  2001-05-04 13:13 ` Christoph Rohland
  0 siblings, 1 reply; 2+ messages in thread
From: Jacek Kopecky @ 2001-05-04 10:09 UTC (permalink / raw)
  To: linux-kernel

 Hello. 8-)
 I'm not in the list, please cc your replies to me.
 After upgrading to 2.4.4 I started using tmpfs for /tmp and I
noticed a strange behavior:

 dd if=/dev/zero of=blah bs=1024 count=102400
	# increased my used swap space by approx. 100MiB (correct)
 rm blah
	# did not decrease it back

 Multiple retries showed what looked like a random behavior of
the used swap stats. Is this a correct behavior? Should the swap
stats be dismissed as 'unreliable'? I expected that when creating
a 100MiB file in memory it should increase the swap (or memory)
usage by cca 100MiB and that removing a file from tmpfs means
freeing the memory.
 Best regards

                            Jacek Kopecky
                               Idoox




^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: tmpfs doesn't update free memory stats?
  2001-05-04 10:09 tmpfs doesn't update free memory stats? Jacek Kopecky
@ 2001-05-04 13:13 ` Christoph Rohland
  0 siblings, 0 replies; 2+ messages in thread
From: Christoph Rohland @ 2001-05-04 13:13 UTC (permalink / raw)
  To: Jacek Kopecky; +Cc: linux-kernel

Hi Jacek,

On Fri, 4 May 2001, Jacek Kopecky wrote:
>  I'm not in the list, please cc your replies to me.
>  After upgrading to 2.4.4 I started using tmpfs for /tmp and I
> noticed a strange behavior:
> 
>  dd if=/dev/zero of=blah bs=1024 count=102400
> 	# increased my used swap space by approx. 100MiB (correct)
>  rm blah
> 	# did not decrease it back
> 
>  Multiple retries showed what looked like a random behavior of
> the used swap stats. Is this a correct behavior? Should the swap
> stats be dismissed as 'unreliable'? I expected that when creating
> a 100MiB file in memory it should increase the swap (or memory)
> usage by cca 100MiB and that removing a file from tmpfs means
> freeing the memory.

It will be adjusted under memory pressure. At this time there is no
way to release swap cached pages without the potential of deadlocks.

This is not nice but the only short term solution and should not
affect anything besides stats.

Greetings
		Christoph



^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2001-05-04 13:33 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2001-05-04 10:09 tmpfs doesn't update free memory stats? Jacek Kopecky
2001-05-04 13:13 ` Christoph Rohland

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox