ICX: dmaengine: ioatdma: add descriptor pre-fetch support for v3.4

commit e0100d40906d5dbe6d09d31083c1a5aaccc947fa upstream.

Adding support for new feature on ioatdma 3.4 hardware that provides
descriptor pre-fetching in order to reduce small DMA latencies.
Signed-off-by: NDave Jiang <dave.jiang@intel.com>
Signed-off-by: NVinod Koul <vkoul@kernel.org>
Signed-off-by: NLin Wang <lin.x.wang@intel.com>
Signed-off-by: NJeffle Xu <jefflexu@linux.alibaba.com>
Acked-by: NJoseph Qi <joseph.qi@linux.alibaba.com>
Acked-by: NCaspar Zhang <caspar@linux.alibaba.com>

drivers/dma/ioat/dma.c

@@ -372,6 +372,7 @@ struct ioat_ring_ent **
 ioat_alloc_ring(struct dma_chan *c, int order, gfp_t flags)
 {
 	struct ioatdma_chan *ioat_chan = to_ioat_chan(c);
+	struct ioatdma_device *ioat_dma = ioat_chan->ioat_dma;
 	struct ioat_ring_ent **ring;
 	int total_descs = 1 << order;
 	int i, chunks;
@@ -437,6 +438,17 @@ ioat_alloc_ring(struct dma_chan *c, int order, gfp_t flags)
 	}
 	ring[i]->hw->next = ring[0]->txd.phys;
 
+	/* setup descriptor pre-fetching for v3.4 */
+	if (ioat_dma->cap & IOAT_CAP_DPS) {
+		u16 drsctl = IOAT_CHAN_DRSZ_2MB | IOAT_CHAN_DRS_EN;
+
+		if (chunks == 1)
+			drsctl |= IOAT_CHAN_DRS_AUTOWRAP;
+
+		writew(drsctl, ioat_chan->reg_base + IOAT_CHAN_DRSCTL_OFFSET);
+
+	}
+
 	return ring;
 }
 

drivers/dma/ioat/init.c

@@ -138,10 +138,10 @@ static int ioat3_dma_self_test(struct ioatdma_device *ioat_dma);
 static int ioat_dca_enabled = 1;
 module_param(ioat_dca_enabled, int, 0644);
 MODULE_PARM_DESC(ioat_dca_enabled, "control support of dca service (default: 1)");
-int ioat_pending_level = 4;
+int ioat_pending_level = 7;
 module_param(ioat_pending_level, int, 0644);
 MODULE_PARM_DESC(ioat_pending_level,
-		 "high-water mark for pushing ioat descriptors (default: 4)");
+		 "high-water mark for pushing ioat descriptors (default: 7)");
 static char ioat_interrupt_style[32] = "msix";
 module_param_string(ioat_interrupt_style, ioat_interrupt_style,
 		    sizeof(ioat_interrupt_style), 0644);
@@ -1189,6 +1189,10 @@ static int ioat3_dma_probe(struct ioatdma_device *ioat_dma, int dca)
 	if (err)
 		return err;
 
+	if (ioat_dma->cap & IOAT_CAP_DPS)
+		writeb(ioat_pending_level + 1,
+		       ioat_dma->reg_base + IOAT_PREFETCH_LIMIT_OFFSET);
+
 	return 0;
 }
 

drivers/dma/ioat/registers.h

@@ -84,6 +84,9 @@
 #define IOAT_CAP_PQ				0x00000200
 #define IOAT_CAP_DWBES				0x00002000
 #define IOAT_CAP_RAID16SS			0x00020000
+#define IOAT_CAP_DPS				0x00800000
+
+#define IOAT_PREFETCH_LIMIT_OFFSET		0x4C	/* CHWPREFLMT */
 
 #define IOAT_CHANNEL_MMIO_SIZE			0x80	/* Each Channel MMIO space is this size */
 
@@ -243,4 +246,11 @@
 
 #define IOAT_CHANERR_MASK_OFFSET		0x2C	/* 32-bit Channel Error Register */
 
+#define IOAT_CHAN_DRSCTL_OFFSET			0xB6
+#define IOAT_CHAN_DRSZ_4KB			0x0000
+#define IOAT_CHAN_DRSZ_8KB			0x0001
+#define IOAT_CHAN_DRSZ_2MB			0x0009
+#define IOAT_CHAN_DRS_EN			0x0100
+#define IOAT_CHAN_DRS_AUTOWRAP			0x0200
+
 #endif /* _IOAT_REGISTERS_H_ */