Patchwork [V3,2/3] mm: dmapool: Align to ARCH_DMA_MINALIGN in non-coherent DMA mode

login
register
mail settings
Submitter Huacai Chen
Date Sept. 13, 2017, 9:20 a.m.
Message ID <1505294451-21312-1-git-send-email-chenhc@lemote.com>
Download mbox | patch
Permalink /patch/9950815/
State New
Headers show

Comments

Huacai Chen - Sept. 13, 2017, 9:20 a.m.
In non-coherent DMA mode, kernel uses cache flushing operations to
maintain I/O coherency, so the dmapool objects should be aligned to
ARCH_DMA_MINALIGN.

Cc: stable@vger.kernel.org
Signed-off-by: Huacai Chen <chenhc@lemote.com>
---
 mm/dmapool.c | 3 +++
 1 file changed, 3 insertions(+)
Andrew Morton - Sept. 13, 2017, 9:52 p.m.
On Wed, 13 Sep 2017 17:20:51 +0800 Huacai Chen <chenhc@lemote.com> wrote:

> In non-coherent DMA mode, kernel uses cache flushing operations to
> maintain I/O coherency, so the dmapool objects should be aligned to
> ARCH_DMA_MINALIGN.

What are the user-visible effects of this bug?
Huacai Chen - Sept. 14, 2017, 1:13 a.m.
Hi, Andrew,

It will cause data corruption, at least on MIPS:
step 1, dma_map_single
step 2, cache_invalidate (no writeback)
step 3, dma_from_device
step 4, dma_unmap_single
If a DMA buffer and a kernel structure share a same cache line, and if the kernel structure has dirty data, cache_invalidate (no writeback) may cause data lost.
 
Huacai
 
------------------ Original ------------------
From:  "Andrew Morton"<akpm@linux-foundation.org>;

Date:  Thu, Sep 14, 2017 05:52 AM
To:  "Huacai Chen"<chenhc@lemote.com>; 
Cc:  "Fuxin Zhang"<zhangfx@lemote.com>; "linux-mm"<linux-mm@kvack.org>; "linux-kernel"<linux-kernel@vger.kernel.org>; "stable"<stable@vger.kernel.org>; 
Subject:  Re: [PATCH V3 2/3] mm: dmapool: Align to ARCH_DMA_MINALIGN innon-coherent DMA mode

 
On Wed, 13 Sep 2017 17:20:51 +0800 Huacai Chen <chenhc@lemote.com> wrote:

> In non-coherent DMA mode, kernel uses cache flushing operations to

> maintain I/O coherency, so the dmapool objects should be aligned to

> ARCH_DMA_MINALIGN.


What are the user-visible effects of this bug?

Patch

diff --git a/mm/dmapool.c b/mm/dmapool.c
index 4d90a64..2ac6f4a 100644
--- a/mm/dmapool.c
+++ b/mm/dmapool.c
@@ -140,6 +140,9 @@  struct dma_pool *dma_pool_create(const char *name, struct device *dev,
 	else if (align & (align - 1))
 		return NULL;
 
+	if (!plat_device_is_coherent(dev))
+		align = max_t(size_t, align, dma_get_cache_alignment());
+
 	if (size == 0)
 		return NULL;
 	else if (size < 4)