Re: [patch v2] thp, memcg: split hugepage for memcg oom on cow

[David Rientjes <rientjes@google.com>]

On Mon, 23 Apr 2012, David Rientjes wrote:

> > Can you instead put a __split_huge_page_pmd(mm, pmd) here?  It has to
> > redo the get-page-ref-through-pagetable dance, but it's more robust
> > and obvious than splitting the COW page before returning OOM in the
> > thp wp handler.
> > 
> 
> I agree it's more robust if do_huge_pmd_wp_page() were modified later and 
> mistakenly returned VM_FAULT_OOM without the page being split, but 
> __split_huge_page_pmd() has the drawback of also requiring to retake 
> mm->page_table_lock to test whether orig_pmd is still legitimate so it 
> will be slower.  Do you feel strongly about the way it's currently written 
> which will be faster at runtime?
> 

Andrew, please merge this patch.  I'd rather not unnecessarily take 
another reference on the cow page and unnecessarily take 
mm->page_table_lock in the page fault handler so the code is cleaner.  
It's faster this way.

--
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-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>