Changelog:
v2 -> v3:
* update patch description, spotted by Michal
v1 -> v2:
* update patch description, spotted by Michal
"hugetlb_total_pages is used for overcommit calculations but the
current implementation considers only default hugetlb page size (which
is either the first defined hugepage size or the one specified by
default_hugepagesz kernel boot parameter).
If the system is configured for more than one hugepage size (which is
possible since a137e1cc hugetlbfs: per mount huge page sizes) then
the overcommit estimation done by __vm_enough_memory (resp. shown by
meminfo_proc_show) is not precise - there is an impression of more
available/allowed memory. This can lead to an unexpected ENOMEM/EFAULT
resp. SIGSEGV when memory is accounted."
The patch should also push to 2.6.27 stable tree.
Testcase:
boot: hugepagesz=1G hugepages=1
the default overcommit ratio is 50
before patch:
egrep 'CommitLimit' /proc/meminfo
CommitLimit: 55434168 kB
after patch:
egrep 'CommitLimit' /proc/meminfo
CommitLimit: 54909880 kB
Signed-off-by: Wanpeng Li <[email protected]>
---
mm/hugetlb.c | 7 +++++--
1 file changed, 5 insertions(+), 2 deletions(-)
diff --git a/mm/hugetlb.c b/mm/hugetlb.c
index cdb64e4..9e25040 100644
--- a/mm/hugetlb.c
+++ b/mm/hugetlb.c
@@ -2124,8 +2124,11 @@ int hugetlb_report_node_meminfo(int nid, char *buf)
/* Return the number pages of memory we physically have, in PAGE_SIZE units. */
unsigned long hugetlb_total_pages(void)
{
- struct hstate *h = &default_hstate;
- return h->nr_huge_pages * pages_per_huge_page(h);
+ struct hstate *h;
+ unsigned long nr_total_pages = 0;
+ for_each_hstate(h)
+ nr_total_pages += h->nr_huge_pages * pages_per_huge_page(h);
+ return nr_total_pages;
}
static int hugetlb_acct_memory(struct hstate *h, long delta)
--
1.7.11.7
On Thu 14-03-13 19:30:46, Wanpeng Li wrote:
> Changelog:
> v2 -> v3:
> * update patch description, spotted by Michal
> v1 -> v2:
> * update patch description, spotted by Michal
>
> "hugetlb_total_pages is used for overcommit calculations but the
> current implementation considers only default hugetlb page size (which
> is either the first defined hugepage size or the one specified by
> default_hugepagesz kernel boot parameter).
>
> If the system is configured for more than one hugepage size (which is
> possible since a137e1cc hugetlbfs: per mount huge page sizes) then
> the overcommit estimation done by __vm_enough_memory (resp. shown by
> meminfo_proc_show) is not precise - there is an impression of more
> available/allowed memory. This can lead to an unexpected ENOMEM/EFAULT
> resp. SIGSEGV when memory is accounted."
Any reason for keeping the quotes I put in the previous email just to
show what changelog I had in mind?
> The patch should also push to 2.6.27 stable tree.
There are no such trees, I have mentioned that only to point out which
kernels are affected.
>
> Testcase:
> boot: hugepagesz=1G hugepages=1
> the default overcommit ratio is 50
> before patch:
> egrep 'CommitLimit' /proc/meminfo
> CommitLimit: 55434168 kB
> after patch:
> egrep 'CommitLimit' /proc/meminfo
> CommitLimit: 54909880 kB
>
Is this information useful at all?
Cc: <[email protected]> # 3.0+
please
> Signed-off-by: Wanpeng Li <[email protected]>
Acked-by: Michal Hocko <[email protected]>
> ---
> mm/hugetlb.c | 7 +++++--
> 1 file changed, 5 insertions(+), 2 deletions(-)
>
> diff --git a/mm/hugetlb.c b/mm/hugetlb.c
> index cdb64e4..9e25040 100644
> --- a/mm/hugetlb.c
> +++ b/mm/hugetlb.c
> @@ -2124,8 +2124,11 @@ int hugetlb_report_node_meminfo(int nid, char *buf)
> /* Return the number pages of memory we physically have, in PAGE_SIZE units. */
> unsigned long hugetlb_total_pages(void)
> {
> - struct hstate *h = &default_hstate;
> - return h->nr_huge_pages * pages_per_huge_page(h);
> + struct hstate *h;
> + unsigned long nr_total_pages = 0;
> + for_each_hstate(h)
> + nr_total_pages += h->nr_huge_pages * pages_per_huge_page(h);
> + return nr_total_pages;
> }
>
> static int hugetlb_acct_memory(struct hstate *h, long delta)
> --
> 1.7.11.7
>
> --
> To unsubscribe, send a message with 'unsubscribe linux-mm' in
> the body to [email protected]. For more info on Linux MM,
> see: http://www.linux-mm.org/ .
> Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
--
Michal Hocko
SUSE Labs