[PATCH] IB/mthca: add fast memory region implementation

Implement fast memory regions (FMRs), where the driver writes directly into
the HCA's translation tables rather than requiring a firmware command.  For
Tavor, MTTs for FMR are separate from regular MTTs, and are reserved at driver
initialization.  This is done to limit the amount of virtual memory needed to
map the MTTs.  For Arbel, there's no such limitation, and all MTTs and MPTs
may be used for FMR or for regular MR.

Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il>
Signed-off-by: Roland Dreier <roland@topspin.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

drivers/infiniband/hw/mthca/mthca_dev.h

@@ -61,7 +61,8 @@ enum {
 	MTHCA_FLAG_SRQ        = 1 << 2,
 	MTHCA_FLAG_MSI        = 1 << 3,
 	MTHCA_FLAG_MSI_X      = 1 << 4,
-	MTHCA_FLAG_NO_LAM     = 1 << 5
+	MTHCA_FLAG_NO_LAM     = 1 << 5,
+	MTHCA_FLAG_FMR        = 1 << 6
 };
 
 enum {
@@ -134,6 +135,7 @@ struct mthca_limits {
 	int      reserved_eqs;
 	int      num_mpts;
 	int      num_mtt_segs;
+	int      fmr_reserved_mtts;
 	int      reserved_mtts;
 	int      reserved_mrws;
 	int      reserved_uars;
@@ -178,10 +180,17 @@ struct mthca_buddy {
 
 struct mthca_mr_table {
 	struct mthca_alloc      mpt_alloc;
-	struct mthca_buddy	mtt_buddy;
+	struct mthca_buddy      mtt_buddy;
+	struct mthca_buddy     *fmr_mtt_buddy;
 	u64                     mtt_base;
+	u64                     mpt_base;
 	struct mthca_icm_table *mtt_table;
 	struct mthca_icm_table *mpt_table;
+	struct {
+		void __iomem   *mpt_base;
+		void __iomem   *mtt_base;
+		struct mthca_buddy mtt_buddy;
+	} tavor_fmr;
 };
 
 struct mthca_eq_table {
@@ -380,7 +389,17 @@ int mthca_mr_alloc_phys(struct mthca_dev *dev, u32 pd,
 			u64 *buffer_list, int buffer_size_shift,
 			int list_len, u64 iova, u64 total_size,
 			u32 access, struct mthca_mr *mr);
-void mthca_free_mr(struct mthca_dev *dev, struct mthca_mr *mr);
+void mthca_free_mr(struct mthca_dev *dev,  struct mthca_mr *mr);
+
+int mthca_fmr_alloc(struct mthca_dev *dev, u32 pd,
+		    u32 access, struct mthca_fmr *fmr);
+int mthca_tavor_map_phys_fmr(struct ib_fmr *ibfmr, u64 *page_list,
+			     int list_len, u64 iova);
+void mthca_tavor_fmr_unmap(struct mthca_dev *dev, struct mthca_fmr *fmr);
+int mthca_arbel_map_phys_fmr(struct ib_fmr *ibfmr, u64 *page_list,
+			     int list_len, u64 iova);
+void mthca_arbel_fmr_unmap(struct mthca_dev *dev, struct mthca_fmr *fmr);
+int mthca_free_fmr(struct mthca_dev *dev,  struct mthca_fmr *fmr);
 
 int mthca_map_eq_icm(struct mthca_dev *dev, u64 icm_virt);
 void mthca_unmap_eq_icm(struct mthca_dev *dev);

drivers/infiniband/hw/mthca/mthca_main.c

@@ -73,14 +73,15 @@ static const char mthca_version[] __devinitdata =
 	DRV_VERSION " (" DRV_RELDATE ")\n";
 
 static struct mthca_profile default_profile = {
-	.num_qp     = 1 << 16,
-	.rdb_per_qp = 4,
-	.num_cq     = 1 << 16,
-	.num_mcg    = 1 << 13,
-	.num_mpt    = 1 << 17,
-	.num_mtt    = 1 << 20,
-	.num_udav   = 1 << 15,	/* Tavor only */
-	.uarc_size  = 1 << 18,	/* Arbel only */
+	.num_qp		   = 1 << 16,
+	.rdb_per_qp	   = 4,
+	.num_cq		   = 1 << 16,
+	.num_mcg	   = 1 << 13,
+	.num_mpt	   = 1 << 17,
+	.num_mtt	   = 1 << 20,
+	.num_udav	   = 1 << 15,	/* Tavor only */
+	.fmr_reserved_mtts = 1 << 18,	/* Tavor only */
+	.uarc_size	   = 1 << 18,	/* Arbel only */
 };
 
 static int __devinit mthca_tune_pci(struct mthca_dev *mdev)