Message ID | 20150601082203.GA11946@gondor.apana.org.au (mailing list archive) |
---|---|
State | Accepted |
Delegated to: | Herbert Xu |
Headers | show |
diff --git a/crypto/scatterwalk.c b/crypto/scatterwalk.c index 2ef9cbb..ea5815c 100644 --- a/crypto/scatterwalk.c +++ b/crypto/scatterwalk.c @@ -54,7 +54,11 @@ static void scatterwalk_pagedone(struct scatter_walk *walk, int out, struct page *page; page = sg_page(walk->sg) + ((walk->offset - 1) >> PAGE_SHIFT); - if (!PageSlab(page)) + /* Test ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE first as + * PageSlab cannot be optimised away per se due to + * use of volatile pointer. + */ + if (ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE && !PageSlab(page)) flush_dcache_page(page); }
On architectures where flush_dcache_page is not needed, we will end up generating all the code up to the PageSlab call. This is because PageSlab operates on a volatile pointer and thus cannot be optimised away. This patch works around this by checking whether flush_dcache_page is needed before we call PageSlab which then allows PageSlab to be compiled awy. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>