Skip to contentSkip to conversation

Patch "mmc: sd: Meet alignment requirements for raw_ssr DMA" has been added to the 4.9-stable tree

PATCH [email protected] 1 participant 1 message Jan 6, 2017

Subsystems (2)
lore
Share & tools
lore

[email protected]

author maintainer Applied
#0 lore raw

This is a note to let you know that I've just added the patch titled

mmc: sd: Meet alignment requirements for raw_ssr DMA

to the 4.9-stable tree which can be found at:
http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=summary

The filename of the patch is:
mmc-sd-meet-alignment-requirements-for-raw_ssr-dma.patch
and it can be found in the queue-4.9 subdirectory.

If you, or anyone else, feels it should not be added to the stable tree, please let <[email protected]> know about it.

Quoted 1 line

From e85baa8868b016513c0f5738362402495b1a66a5 Mon Sep 17 00:00:00 2001

From: Paul Burton <[email protected]>
Date: Fri, 11 Nov 2016 14:22:36 +0000
Subject: mmc: sd: Meet alignment requirements for raw_ssr DMA
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

From: Paul Burton <[email protected]>

commit e85baa8868b016513c0f5738362402495b1a66a5 upstream.

The mmc_read_ssr() function results in DMA to the raw_ssr member of struct mmc_card, which is not guaranteed to be cache line aligned & thus might not meet the requirements set out in Documentation/DMA-API.txt:

Warnings: Memory coherency operates at a granularity called the cache line width. In order for memory mapped by this API to operate correctly, the mapped region must begin exactly on a cache line boundary and end exactly on one (to prevent two separately mapped regions from sharing a single cache line). Since the cache line size may not be known at compile time, the API will not enforce this requirement. Therefore, it is recommended that driver writers who don't take special care to determine the cache line size at run time only map virtual regions that begin and end on page boundaries (which are guaranteed also to be cache line boundaries).

On some systems where DMA is non-coherent this can lead to us losing data that shares cache lines with the raw_ssr array.

Fix this by kmalloc'ing a temporary buffer to perform DMA into. kmalloc will ensure the buffer is suitably aligned, allowing the DMA to be performed without any loss of data.

Fixes 5275a652d296 ("mmc: sd: Export SD Status via “ssr” device attribute")
---
 drivers/mmc/core/sd.c |   12 ++++++++++--
 1 file changed, 10 insertions(+), 2 deletions(-)
drivers/mmc/core/sd.c +10 -2
@@ -223,6 +223,7 @@ static int mmc_decode_scr(struct mmc_car
 static int mmc_read_ssr(struct mmc_card *card)
 {
 	unsigned int au, es, et, eo;
+	u32 *raw_ssr;
 	int i;
 
 	if (!(card->csd.cmdclass & CCC_APP_SPEC)) {
@@ -231,14 +232,21 @@ static int mmc_read_ssr(struct mmc_card
 		return 0;
 	}
 
-	if (mmc_app_sd_status(card, card->raw_ssr)) {
+	raw_ssr = kmalloc(sizeof(card->raw_ssr), GFP_KERNEL);
+	if (!raw_ssr)
+		return -ENOMEM;
+
+	if (mmc_app_sd_status(card, raw_ssr)) {
 		pr_warn("%s: problem reading SD Status register\n",
 			mmc_hostname(card->host));
+		kfree(raw_ssr);
 		return 0;
 	}
 
 	for (i = 0; i < 16; i++)
-		card->raw_ssr[i] = be32_to_cpu(card->raw_ssr[i]);
+		card->raw_ssr[i] = be32_to_cpu(raw_ssr[i]);
+
+	kfree(raw_ssr);
 
 	/*
 	 * UNSTUFF_BITS only works with four u32s so we have to offset the

Patches currently in stable-queue which might be from [email protected] are

queue-4.9/mmc-sd-meet-alignment-requirements-for-raw_ssr-dma.patch