| Message ID | 20210330114847.GX3697@techsingularity.net (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | mm/page_alloc: Add a bulk page allocator -fix -fix | expand |
On 30/03/2021 12:48, Mel Gorman wrote: > Colin Ian King reported the following problem (slightly edited) > > Author: Mel Gorman <mgorman@techsingularity.net> > Date: Mon Mar 29 11:12:24 2021 +1100 > > mm/page_alloc: add a bulk page allocator > > ... > > Static analysis on linux-next with Coverity has found a potential > uninitialized variable issue in function __alloc_pages_bulk with > the following commit: > > ... > > Uninitialized scalar variable (UNINIT) > 15. uninit_use_in_call: Using uninitialized value alloc_flags when > calling prepare_alloc_pages. > > 5056 if (!prepare_alloc_pages(gfp, 0, preferred_nid, nodemask, > &ac, &alloc_gfp, &alloc_flags)) > > The problem is that prepare_alloc_flags only updates alloc_flags > which must have a valid initial value. The appropriate initial value is > ALLOC_WMARK_LOW to avoid the bulk allocator pushing a zone below the low > watermark without waking kswapd assuming the GFP mask allows kswapd to > be woken. > > This is a second fix to the mmotm patch > mm-page_alloc-add-a-bulk-page-allocator.patch . It will cause a mild conflict > with a later patch due to renaming of an adjacent variable that is trivially > resolved. I can post a full series with the fixes merged if that is preferred. > > Signed-off-by: Mel Gorman <mgorman@techsingularity.net> > --- > mm/page_alloc.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/mm/page_alloc.c b/mm/page_alloc.c > index 92d55f80c289..dabef0b910c9 100644 > --- a/mm/page_alloc.c > +++ b/mm/page_alloc.c > @@ -4990,7 +4990,7 @@ unsigned long __alloc_pages_bulk(gfp_t gfp, int preferred_nid, > struct list_head *pcp_list; > struct alloc_context ac; > gfp_t alloc_gfp; > - unsigned int alloc_flags; > + unsigned int alloc_flags = ALLOC_WMARK_LOW; > int allocated = 0; > > if (WARN_ON_ONCE(nr_pages <= 0)) > Thanks Mel, that definitely fixes the issue. Reviewed-by: Colin Ian King <colin.king@canonical.com>
diff --git a/mm/page_alloc.c b/mm/page_alloc.c index 92d55f80c289..dabef0b910c9 100644 --- a/mm/page_alloc.c +++ b/mm/page_alloc.c @@ -4990,7 +4990,7 @@ unsigned long __alloc_pages_bulk(gfp_t gfp, int preferred_nid, struct list_head *pcp_list; struct alloc_context ac; gfp_t alloc_gfp; - unsigned int alloc_flags; + unsigned int alloc_flags = ALLOC_WMARK_LOW; int allocated = 0; if (WARN_ON_ONCE(nr_pages <= 0))
Colin Ian King reported the following problem (slightly edited) Author: Mel Gorman <mgorman@techsingularity.net> Date: Mon Mar 29 11:12:24 2021 +1100 mm/page_alloc: add a bulk page allocator ... Static analysis on linux-next with Coverity has found a potential uninitialized variable issue in function __alloc_pages_bulk with the following commit: ... Uninitialized scalar variable (UNINIT) 15. uninit_use_in_call: Using uninitialized value alloc_flags when calling prepare_alloc_pages. 5056 if (!prepare_alloc_pages(gfp, 0, preferred_nid, nodemask, &ac, &alloc_gfp, &alloc_flags)) The problem is that prepare_alloc_flags only updates alloc_flags which must have a valid initial value. The appropriate initial value is ALLOC_WMARK_LOW to avoid the bulk allocator pushing a zone below the low watermark without waking kswapd assuming the GFP mask allows kswapd to be woken. This is a second fix to the mmotm patch mm-page_alloc-add-a-bulk-page-allocator.patch . It will cause a mild conflict with a later patch due to renaming of an adjacent variable that is trivially resolved. I can post a full series with the fixes merged if that is preferred. Signed-off-by: Mel Gorman <mgorman@techsingularity.net> --- mm/page_alloc.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)