On Mon, Oct 21, 2013 at 11:01:07AM -0600, Khalid Aziz wrote:
> Hi Andrea,
>
> Sure, this can happen but wouldn't this case be caught by the last line
> in PageHuge() - "return dtor == free_huge_page;". This conditional
> should return false for THP PageTail, right?
The problem is that get_compo
On 10/18/2013 10:06 AM, Andrea Arcangeli wrote:
Hi,
Sorry for the self-followup but looking further into this, there are
more subtle troubles with this patch that I think are worth discussing
(not just the PageHuge check to skip the __page_cache_release).
- struct page *page_head = compo
On Fri, 2013-10-18 at 18:06 +0200, Andrea Arcangeli wrote:
> Hi,
>
> Sorry for the self-followup but looking further into this, there are
> more subtle troubles with this patch that I think are worth discussing
> (not just the PageHuge check to skip the __page_cache_release).
[...]
Given these ob
On 10/17/2013 05:34 PM, Andrea Arcangeli wrote:
Hi,
On Tue, Oct 15, 2013 at 01:46:08PM -0600, Khalid Aziz wrote:
@@ -68,13 +69,26 @@ static void __put_compound_page(struct page *page)
{
compound_page_dtor *dtor;
- __page_cache_release(page);
+ if (!PageHuge(page))
+
Hi,
Sorry for the self-followup but looking further into this, there are
more subtle troubles with this patch that I think are worth discussing
(not just the PageHuge check to skip the __page_cache_release).
> > - struct page *page_head = compound_trans_head(page);
> > + struct page *page_hea
Hi,
On Tue, Oct 15, 2013 at 01:46:08PM -0600, Khalid Aziz wrote:
> @@ -68,13 +69,26 @@ static void __put_compound_page(struct page *page)
> {
> compound_page_dtor *dtor;
>
> - __page_cache_release(page);
> + if (!PageHuge(page))
> + __page_cache_release(page);
>
On Tue, 2013-10-15 at 13:46 -0600, Khalid Aziz wrote:
> commit 7cb2ef56e6a8b7b368b2e883a0a47d02fed66911 upstream.
> Some backport work was needed to account for changes in upstream mm/swap.c.
> Reduced indentation level from upstream patch following Kirill's
> suggestion.
[...]
> Cc: # 3.0.x
> Cc:
On Tue, Oct 15, 2013 at 02:30:33PM -0600, Khalid Aziz wrote:
> On 10/15/2013 02:10 PM, Greg KH wrote:
> >On Tue, Oct 15, 2013 at 01:46:08PM -0600, Khalid Aziz wrote:
> >>
> >>commit 7cb2ef56e6a8b7b368b2e883a0a47d02fed66911 upstream.
> >>Some backport work was needed to account for changes in upstre
On 10/15/2013 02:10 PM, Greg KH wrote:
On Tue, Oct 15, 2013 at 01:46:08PM -0600, Khalid Aziz wrote:
commit 7cb2ef56e6a8b7b368b2e883a0a47d02fed66911 upstream.
Some backport work was needed to account for changes in upstream mm/swap.c.
Reduced indentation level from upstream patch following Kiril
On Tue, Oct 15, 2013 at 01:46:08PM -0600, Khalid Aziz wrote:
>
> commit 7cb2ef56e6a8b7b368b2e883a0a47d02fed66911 upstream.
> Some backport work was needed to account for changes in upstream mm/swap.c.
> Reduced indentation level from upstream patch following Kirill's
> suggestion.
Which stable tr
commit 7cb2ef56e6a8b7b368b2e883a0a47d02fed66911 upstream.
Some backport work was needed to account for changes in upstream mm/swap.c.
Reduced indentation level from upstream patch following Kirill's
suggestion.
I am working with a tool that simulates oracle database I/O workload.
This tool (orion
11 matches
Mail list logo