Re: [PATCH 10 of 20] ipath - support for userspace apps using core driver

["Bryan O'Sullivan" <bos@pathscale.com>]

On Thu, 2006-03-16 at 17:23 +0000, Hugh Dickins wrote:

> But your backport driver will
> have to be using PageReserved still, not relying on __GFP_COMP: although
> __GFP_COMP was defined in 2.6.9 and a few earlier, it used to take effect
> only when #ifdef CONFIG_HUGETLB_PAGE - only in 2.6.15 did we make it
> available to all configurations.  You'll have irritating accounting
> differences between the two drivers: it used to be the case that put_page
> on a PageReserved page did nothing, so you had to avoid get_page on it to
> get the page accounting right; we straightened that out in 2.6.15.

OK.  Would it be correct to say that this is what we should do, then?

      * On 2.6.15 and later kernels, use __GFP_COMP at allocation time,
        and get_page in ->nopage.  This is what we're doing as of this
        morning, and it works.
      * For backports to 2.6.14 and earlier, avoid __GFP_COMP, mark each
        page with SetPageReserved at allocation time, and do nothing
        special in ->nopage.  Do we need to ClearPageReserved before
        freeing?

Thanks,

	<b