[PATCH] COMPACTION: bugfix of improper cache flush in MIGRATION code.

From: Leonid Yegoshin
Date: Wed May 08 2013 - 20:19:35 EST


Page 'new' during MIGRATION can't be flushed by flush_cache_page().
Using flush_cache_page(vma, addr, pfn) is justified only if
page is already placed in process page table, and that is done right
after flush_cache_page(). But without it the arch function has
no knowledge of process PTE and does nothing.

Besides that, flush_cache_page() flushes an application cache,
kernel has a different page virtual address and dirtied it.

Replace it with flush_dcache_page(new) which is a proper usage.

Old page is flushed in try_to_unmap_one() before MIGRATION.

This bug takes place in Sead3 board with M14Kc MIPS CPU without
cache aliasing (but Harvard arch - separate I and D cache)
in tight memory environment (128MB) each 1-3days on SOAK test.
It fails in cc1 during kernel build (SIGILL, SIGBUS, SIGSEG) if
CONFIG_COMPACTION is switched ON.

Author: Leonid Yegoshin <yegoshin@xxxxxxxx>
Signed-off-by: Leonid Yegoshin <Leonid.Yegoshin@xxxxxxxxxx>
---
mm/migrate.c | 2 +-
1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/mm/migrate.c b/mm/migrate.c
index 2fd8b4a..4c6250a 100644
--- a/mm/migrate.c
+++ b/mm/migrate.c
@@ -165,7 +165,7 @@ static int remove_migration_pte(struct page *new, struct vm_area_struct *vma,
pte = arch_make_huge_pte(pte, vma, new, 0);
}
#endif
- flush_cache_page(vma, addr, pte_pfn(pte));
+ flush_dcache_page(new);
set_pte_at(mm, addr, ptep, pte);

if (PageHuge(new)) {


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/