Re: [PATCH 4.4 26/28] x86, pmem: fix broken __copy_user_nocache cache-bypass assumptions

[Ben Hutchings <ben.hutchings@codethink.co.uk>]

On Tue, 2017-04-25 at 16:08 +0100, Greg Kroah-Hartman wrote:
> 4.4-stable review patch.  If anyone has any objections, please let me know.
> 
> ------------------
> 
> From: Dan Williams <dan.j.williams@intel.com>
> 
> commit 11e63f6d920d6f2dfd3cd421e939a4aec9a58dcd upstream.
[...]
> +	if (iter_is_iovec(i)) {
> +		unsigned long flushed, dest = (unsigned long) addr;
> +
> +		if (bytes < 8) {
> +			if (!IS_ALIGNED(dest, 4) || (bytes != 4))
> +				__arch_wb_cache_pmem(addr, 1);
[...]

What if the write crosses a cache line boundary?  I think you need the
following fix-up (untested, I don't have this kind of hardware).

Ben.

---
From: Ben Hutchings <ben.hutchings@codethink.co.uk>
Subject: x86, pmem: Fix cache flushing for iovec write < 8 bytes

Commit 11e63f6d920d added cache flushing for unaligned writes from an
iovec, covering the first and last cache line of a >= 8 byte write and
the first cache line of a < 8 byte write.  But an unaligned write of
2-7 bytes can still cover two cache lines, so make sure we flush both
in that case.

Fixes: 11e63f6d920d ("x86, pmem: fix broken __copy_user_nocache ...")
Signed-off-by: Ben Hutchings <ben.hutchings@codethink.co.uk>
---
 arch/x86/include/asm/pmem.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/x86/include/asm/pmem.h b/arch/x86/include/asm/pmem.h
index d5a22bac9988..0ff8fe71b255 100644
--- a/arch/x86/include/asm/pmem.h
+++ b/arch/x86/include/asm/pmem.h
@@ -98,7 +98,7 @@ static inline size_t arch_copy_from_iter_pmem(void *addr, size_t bytes,
 
 		if (bytes < 8) {
 			if (!IS_ALIGNED(dest, 4) || (bytes != 4))
-				arch_wb_cache_pmem(addr, 1);
+				arch_wb_cache_pmem(addr, bytes);
 		} else {
 			if (!IS_ALIGNED(dest, 8)) {
 				dest = ALIGN(dest, boot_cpu_data.x86_clflush_size);

-- 
Ben Hutchings
Software Developer, Codethink Ltd.