Hi,
This patch was generated against 2.6.12-rc1
Working on some code lately I've been getting huge values
for "Cached". The cause is that get_page_cache_size() is an
approximate value, and for a sufficiently small returned value
of get_page_cache_size() the value underflows.
Signed-off-by: Martin Hicks <[email protected]>
proc_misc.c | 9 ++++++++-
1 files changed, 8 insertions(+), 1 deletion(-)
Index: linux-2.6.11.cached-limit/fs/proc/proc_misc.c
===================================================================
--- linux-2.6.11.cached-limit.orig/fs/proc/proc_misc.c 2005-03-31 11:36:46.000000000 -0800
+++ linux-2.6.11.cached-limit/fs/proc/proc_misc.c 2005-03-31 11:37:12.000000000 -0800
@@ -126,6 +126,7 @@ static int meminfo_read_proc(char *page,
unsigned long committed;
unsigned long allowed;
struct vmalloc_info vmi;
+ unsigned long cached;
get_page_state(&ps);
get_zone_counts(&active, &inactive, &free);
@@ -140,6 +141,12 @@ static int meminfo_read_proc(char *page,
allowed = ((totalram_pages - hugetlb_total_pages())
* sysctl_overcommit_ratio / 100) + total_swap_pages;
+ cached = get_page_cache_size();
+ if (total_swapcache_pages+i.bufferram < cached)
+ cached -= total_swapcache_pages + i.bufferram;
+ else
+ cached = 0;
+
get_vmalloc_info(&vmi);
/*
@@ -172,7 +179,7 @@ static int meminfo_read_proc(char *page,
K(i.totalram),
K(i.freeram),
K(i.bufferram),
- K(get_page_cache_size()-total_swapcache_pages-i.bufferram),
+ K(cached),
K(total_swapcache_pages),
K(active),
K(inactive),
Martin Hicks <[email protected]> wrote:
>
> Working on some code lately I've been getting huge values
> for "Cached". The cause is that get_page_cache_size() is an
> approximate value, and for a sufficiently small returned value
> of get_page_cache_size() the value underflows.
OK..
I think I'd prefer to do it this way - it's simpler and the original patch
had a teeny race wrt changes in total_swapcache_pages.
diff -puN fs/proc/proc_misc.c~meminfo-add-cached-underflow-check fs/proc/proc_misc.c
--- 25/fs/proc/proc_misc.c~meminfo-add-cached-underflow-check Mon Apr 4 15:06:38 2005
+++ 25-akpm/fs/proc/proc_misc.c Mon Apr 4 15:10:24 2005
@@ -126,6 +126,7 @@ static int meminfo_read_proc(char *page,
unsigned long committed;
unsigned long allowed;
struct vmalloc_info vmi;
+ long cached;
get_page_state(&ps);
get_zone_counts(&active, &inactive, &free);
@@ -140,6 +141,10 @@ static int meminfo_read_proc(char *page,
allowed = ((totalram_pages - hugetlb_total_pages())
* sysctl_overcommit_ratio / 100) + total_swap_pages;
+ cached = get_page_cache_size() - total_swapcache_pages - i.bufferram;
+ if (cached < 0)
+ cached = 0;
+
get_vmalloc_info(&vmi);
/*
@@ -172,7 +177,7 @@ static int meminfo_read_proc(char *page,
K(i.totalram),
K(i.freeram),
K(i.bufferram),
- K(get_page_cache_size()-total_swapcache_pages-i.bufferram),
+ K(cached),
K(total_swapcache_pages),
K(active),
K(inactive),
_
On Mon, Apr 04, 2005 at 03:10:49PM -0700, Andrew Morton wrote:
> Martin Hicks <[email protected]> wrote:
> >
> > Working on some code lately I've been getting huge values
> > for "Cached". The cause is that get_page_cache_size() is an
> > approximate value, and for a sufficiently small returned value
> > of get_page_cache_size() the value underflows.
>
> OK..
>
> I think I'd prefer to do it this way - it's simpler and the original patch
> had a teeny race wrt changes in total_swapcache_pages.
Fine by me.
mh
--
Martin Hicks || Silicon Graphics Inc. || [email protected]