-----Original Message----- From: Christian König [mailto:ckoenig.leichtzumerken@xxxxxxxxx] Sent: Monday, December 04, 2017 8:13 PM To: dri-devel at lists.freedesktop.org; amd-gfx at lists.freedesktop.org; He, Roger <Hongbo.He at amd.com>; ckoenig.leichtzumerken at gmail.com Subject: [PATCH 1/2] drm/ttm: swap consecutive allocated cached pages v2 When we detect consecutive allocation of pages swap them to avoid accidentally freeing them as huge page. v2: use swap Signed-off-by: Christian König <christian.koenig at amd.com> --- drivers/gpu/drm/ttm/ttm_page_alloc.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/drivers/gpu/drm/ttm/ttm_page_alloc.c b/drivers/gpu/drm/ttm/ttm_page_alloc.c index c82d94cbbabc..b6c5148607e9 100644 --- a/drivers/gpu/drm/ttm/ttm_page_alloc.c +++ b/drivers/gpu/drm/ttm/ttm_page_alloc.c @@ -921,6 +921,10 @@ static int ttm_get_pages(struct page **pages, unsigned npages, int flags, return -ENOMEM; } How about add first index as below? first = i; //init first here while (npages) { p = alloc_page(gfp_flags); if (!p) { pr_debug("Unable to allocate page\n"); return -ENOMEM; } /* Swap the pages if we detect consecutive order */ if (i > first && pages[i - 1] == p - 1) //change to (i > first), because I worry if npages is 513, the first regular page is exactly consecutive with last page of huge page. Then swap them is not what I want here. swap(p, pages[i - 1]); pages[i++] = p; --npages; } pages[i++] = p; --npages; } -- 2.11.0