[PATCH 1/5] staging: zsmalloc: zsmalloc memory allocation library

[Dan Magenheimer]

> From: Seth Jennings [mailto:sjenning at linux.vnet.ibm.com]
> Subject: Re: [PATCH 1/5] staging: zsmalloc: zsmalloc memory allocation library
> 
> On 01/11/2012 11:19 AM, Dan Magenheimer wrote:
> >> From: Seth Jennings [mailto:sjenning at linux.vnet.ibm.com]
> >> Subject: [PATCH 1/5] staging: zsmalloc: zsmalloc memory allocation library
> >>
> >> From: Nitin Gupta <ngupta at vflare.org>
> >>
> >> This patch creates a new memory allocation library named
> >> zsmalloc.
> >>
> >> +/*
> >> + * Allocate a zspage for the given size class
> >> + */
> >> +static struct page *alloc_zspage(struct size_class *class, gfp_t flags)
> >> +{
> >> +	int i, error;
> >> +	struct page *first_page = NULL;
> >> +
> >> +	/*
> >> +	 * Allocate individual pages and link them together as:
> >> +	 * 1. first page->private = first sub-page
> >> +	 * 2. all sub-pages are linked together using page->lru
> >> +	 * 3. each sub-page is linked to the first page using page->first_page
> >> +	 *
> >> +	 * For each size class, First/Head pages are linked together using
> >> +	 * page->lru. Also, we set PG_private to identify the first page
> >> +	 * (i.e. no other sub-page has this flag set) and PG_private_2 to
> >> +	 * identify the last page.
> >> +	 */
> >> +	error = -ENOMEM;
> >> +	for (i = 0; i < class->zspage_order; i++) {
> >> +		struct page *page, *prev_page;
> >> +
> >> +		page = alloc_page(flags);
> >
> > Hmmm... I thought we agreed offlist that the new allocator API would
> > provide for either preloads or callbacks (which may differ per pool)
> > instead of directly allocating raw pages from the kernel.  The caller
> > (zcache or ramster or ???) needs to be able to somehow manage maximum
> > memory capacity to avoid OOMs.
> >
> > Or am I missing the code that handles that?
> 
> No, you aren't missing it; it's not there.  And I agree that we
> should add that.
> 
> However, the existing allocator, xvmalloc, doesn't support callback
> functionality either.  Would it be simpler to add the that as
> a separate patch, that way we can keep the changes to zcache/zram
> in this patchset isolated to just changing the xvmalloc calls to
> zsmalloc calls?

OK, I'll buy that.  Since you haven't changed any of the policy
code, and the allocator is (purportedly) more storage-efficient on
any sequence of chunks stored in it, the potential for OOMs shouldn't
get any worse.

Dan