mmc: sdhci: 64-bit DMA actually has 4-byte alignment

The version 3.00 SDHCI spec. was a bit unclear about the
required data alignment for 64-bit DMA, whereas the version
4.10 spec. uses different language and indicates that only
4-byte alignment is required rather than the 8-byte alignment
currently implemented.  That make no difference to SD and EMMC
which invariably transfer data in sector-aligned blocks.
However with SDIO, it results in using more DMA descriptors
than necessary.  Theoretically that slows DMA slightly although
DMA is not the limiting factor for throughput, so there is no
discernable impact on performance.  Nevertheless, the driver
should follw the spec unless there is good reason not to, so
this patch corrects the alignment criterion.

There is a more complicated criterion for the DMA descriptor
table itself.  However the table is allocated by dma_alloc_coherent()
which allocates pages (i.e. aligned to a page boundary).
For simplicity just check it is 8-byte aligned, but add a comment
that some Intel controllers actually require 8-byte alignment
even when using 32-bit DMA.
Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>

drivers/mmc/host/sdhci.c

@@ -492,7 +492,7 @@ static int sdhci_adma_table_pre(struct sdhci_host *host,
 		host->align_buffer, host->align_buffer_sz, direction);
 	if (dma_mapping_error(mmc_dev(host->mmc), host->align_addr))
 		goto fail;
-	BUG_ON(host->align_addr & host->align_mask);
+	BUG_ON(host->align_addr & SDHCI_ADMA2_MASK);
 
 	host->sg_count = sdhci_pre_dma_transfer(host, data);
 	if (host->sg_count < 0)
@@ -514,8 +514,8 @@ static int sdhci_adma_table_pre(struct sdhci_host *host,
 		 * the (up to three) bytes that screw up the
 		 * alignment.
 		 */
-		offset = (host->align_sz - (addr & host->align_mask)) &
-			 host->align_mask;
+		offset = (SDHCI_ADMA2_ALIGN - (addr & SDHCI_ADMA2_MASK)) &
+			 SDHCI_ADMA2_MASK;
 		if (offset) {
 			if (data->flags & MMC_DATA_WRITE) {
 				buffer = sdhci_kmap_atomic(sg, &flags);
@@ -529,8 +529,8 @@ static int sdhci_adma_table_pre(struct sdhci_host *host,
 
 			BUG_ON(offset > 65536);
 
-			align += host->align_sz;
-			align_addr += host->align_sz;
+			align += SDHCI_ADMA2_ALIGN;
+			align_addr += SDHCI_ADMA2_ALIGN;
 
 			desc += host->desc_sz;
 
@@ -611,7 +611,7 @@ static void sdhci_adma_table_post(struct sdhci_host *host,
 	/* Do a quick scan of the SG list for any unaligned mappings */
 	has_unaligned = false;
 	for_each_sg(data->sg, sg, host->sg_count, i)
-		if (sg_dma_address(sg) & host->align_mask) {
+		if (sg_dma_address(sg) & SDHCI_ADMA2_MASK) {
 			has_unaligned = true;
 			break;
 		}
@@ -623,15 +623,15 @@ static void sdhci_adma_table_post(struct sdhci_host *host,
 		align = host->align_buffer;
 
 		for_each_sg(data->sg, sg, host->sg_count, i) {
-			if (sg_dma_address(sg) & host->align_mask) {
-				size = host->align_sz -
-				       (sg_dma_address(sg) & host->align_mask);
+			if (sg_dma_address(sg) & SDHCI_ADMA2_MASK) {
+				size = SDHCI_ADMA2_ALIGN -
+				       (sg_dma_address(sg) & SDHCI_ADMA2_MASK);
 
 				buffer = sdhci_kmap_atomic(sg, &flags);
 				memcpy(buffer, align, size);
 				sdhci_kunmap_atomic(buffer, &flags);
 
-				align += host->align_sz;
+				align += SDHCI_ADMA2_ALIGN;
 			}
 		}
 	}
@@ -2961,24 +2961,17 @@ int sdhci_add_host(struct sdhci_host *host)
 		if (host->flags & SDHCI_USE_64_BIT_DMA) {
 			host->adma_table_sz = (SDHCI_MAX_SEGS * 2 + 1) *
 					      SDHCI_ADMA2_64_DESC_SZ;
-			host->align_buffer_sz = SDHCI_MAX_SEGS *
-						SDHCI_ADMA2_64_ALIGN;
 			host->desc_sz = SDHCI_ADMA2_64_DESC_SZ;
-			host->align_sz = SDHCI_ADMA2_64_ALIGN;
-			host->align_mask = SDHCI_ADMA2_64_ALIGN - 1;
 		} else {
 			host->adma_table_sz = (SDHCI_MAX_SEGS * 2 + 1) *
 					      SDHCI_ADMA2_32_DESC_SZ;
-			host->align_buffer_sz = SDHCI_MAX_SEGS *
-						SDHCI_ADMA2_32_ALIGN;
 			host->desc_sz = SDHCI_ADMA2_32_DESC_SZ;
-			host->align_sz = SDHCI_ADMA2_32_ALIGN;
-			host->align_mask = SDHCI_ADMA2_32_ALIGN - 1;
 		}
 		host->adma_table = dma_alloc_coherent(mmc_dev(mmc),
 						      host->adma_table_sz,
 						      &host->adma_addr,
 						      GFP_KERNEL);
+		host->align_buffer_sz = SDHCI_MAX_SEGS * SDHCI_ADMA2_ALIGN;
 		host->align_buffer = kmalloc(host->align_buffer_sz, GFP_KERNEL);
 		if (!host->adma_table || !host->align_buffer) {
 			if (host->adma_table)
@@ -2992,7 +2985,7 @@ int sdhci_add_host(struct sdhci_host *host)
 			host->flags &= ~SDHCI_USE_ADMA;
 			host->adma_table = NULL;
 			host->align_buffer = NULL;
-		} else if (host->adma_addr & host->align_mask) {
+		} else if (host->adma_addr & (SDHCI_ADMA2_DESC_ALIGN - 1)) {
 			pr_warn("%s: unable to allocate aligned ADMA descriptor\n",
 				mmc_hostname(mmc));
 			host->flags &= ~SDHCI_USE_ADMA;

drivers/mmc/host/sdhci.h

@@ -272,22 +272,27 @@
 /* ADMA2 32-bit DMA descriptor size */
 #define SDHCI_ADMA2_32_DESC_SZ	8
 
-/* ADMA2 32-bit DMA alignment */
-#define SDHCI_ADMA2_32_ALIGN	4
-
 /* ADMA2 32-bit descriptor */
 struct sdhci_adma2_32_desc {
 	__le16	cmd;
 	__le16	len;
 	__le32	addr;
-}  __packed __aligned(SDHCI_ADMA2_32_ALIGN);
+}  __packed __aligned(4);
+
+/* ADMA2 data alignment */
+#define SDHCI_ADMA2_ALIGN	4
+#define SDHCI_ADMA2_MASK	(SDHCI_ADMA2_ALIGN - 1)
+
+/*
+ * ADMA2 descriptor alignment.  Some controllers (e.g. Intel) require 8 byte
+ * alignment for the descriptor table even in 32-bit DMA mode.  Memory
+ * allocation is at least 8 byte aligned anyway, so just stipulate 8 always.
+ */
+#define SDHCI_ADMA2_DESC_ALIGN	8
 
 /* ADMA2 64-bit DMA descriptor size */
 #define SDHCI_ADMA2_64_DESC_SZ	12
 
-/* ADMA2 64-bit DMA alignment */
-#define SDHCI_ADMA2_64_ALIGN	8
-
 /*
  * ADMA2 64-bit descriptor. Note 12-byte descriptor can't always be 8-byte
  * aligned.
@@ -482,8 +487,6 @@ struct sdhci_host {
 	dma_addr_t align_addr;	/* Mapped bounce buffer */
 
 	unsigned int desc_sz;	/* ADMA descriptor size */
-	unsigned int align_sz;	/* ADMA alignment */
-	unsigned int align_mask;	/* ADMA alignment mask */
 
 	struct tasklet_struct finish_tasklet;	/* Tasklet structures */