[ofa-general] [PATCH 9/10] nes: support systems with other than 4k page size

Glenn Grundstrom NetEffect glenn at lists.openfabrics.org
Thu Dec 20 12:37:49 PST 2007


Originally assumed 4k page size.  Get the page size from
the kernel and map the memory accordingly.

Signed-off-by: Glenn Grundstrom <ggrundstrom at neteffect.com>

---

diff --git a/drivers/infiniband/hw/nes/nes_hw.c b/drivers/infiniband/hw/nes/nes_hw.c
index 515133d..5646e5a 100644
--- a/drivers/infiniband/hw/nes/nes_hw.c
+++ b/drivers/infiniband/hw/nes/nes_hw.c
@@ -259,7 +259,7 @@ struct nes_adapter *nes_init_adapter(struct nes_device *nesdev, u8 hw_rev) {
 	}
 
 	/* no adapter found */
-	num_pds = pci_resource_len(nesdev->pcidev, BAR_1) / 4096;
+	num_pds = pci_resource_len(nesdev->pcidev, BAR_1) >> PAGE_SHIFT;
 	if ((hw_rev != NE020_REV) && (hw_rev != NE020_REV1)) {
 		nes_debug(NES_DBG_INIT, "NE020 driver detected unknown hardware revision 0x%x\n",
 				hw_rev);
diff --git a/drivers/infiniband/hw/nes/nes_verbs.c b/drivers/infiniband/hw/nes/nes_verbs.c
index 9751d0d..aa53cbd 100644
--- a/drivers/infiniband/hw/nes/nes_verbs.c
+++ b/drivers/infiniband/hw/nes/nes_verbs.c
@@ -839,7 +839,7 @@ static struct ib_ucontext *nes_alloc_ucontext(struct ib_device *ibdev,
 		return ERR_PTR(-ENOMEM);
 
 	nes_ucontext->nesdev = nesdev;
-	nes_ucontext->mmap_wq_offset = ((uresp.max_pds * 4096) + PAGE_SIZE-1) / PAGE_SIZE;
+	nes_ucontext->mmap_wq_offset = uresp.max_pds;
 	nes_ucontext->mmap_cq_offset = nes_ucontext->mmap_wq_offset +
 			((sizeof(struct nes_hw_qp_wqe) * uresp.max_qps * 2) + PAGE_SIZE-1) /
 			PAGE_SIZE;
@@ -960,7 +960,7 @@ static struct ib_pd *nes_alloc_pd(struct ib_device *ibdev,
 	nes_debug(NES_DBG_PD, "Allocating PD (%p) for ib device %s\n",
 			nespd, nesvnic->nesibdev->ibdev.name);
 
-	nespd->pd_id = pd_num + nesadapter->base_pd;
+	nespd->pd_id = (pd_num << (PAGE_SHIFT-12)) + nesadapter->base_pd;
 
 	if (context) {
 		nesucontext = to_nesucontext(context);
@@ -1018,7 +1018,7 @@ static int nes_dealloc_pd(struct ib_pd *ibpd)
 	nes_debug(NES_DBG_PD, "Deallocating PD%u structure located @%p.\n",
 			nespd->pd_id, nespd);
 	nes_free_resource(nesadapter, nesadapter->allocated_pds,
-			nespd->pd_id-nesadapter->base_pd);
+			(nespd->pd_id-nesadapter->base_pd)>>(PAGE_SHIFT-12));
 	kfree(nespd);
 
 	return 0;
@@ -1089,8 +1089,8 @@ static int nes_setup_virt_qp(struct nes_qp *nesqp, struct nes_pbl *nespbl,
 	pbl = nespbl->pbl_vbase; /* points to first pbl entry */
 	/* now lets set the sq_vbase as well as rq_vbase addrs we will assign */
 	/* the first pbl to be fro the rq_vbase... */
-	rq_pbl_entries = (rq_size * sizeof(struct nes_hw_qp_wqe)) >> PAGE_SHIFT;
-	sq_pbl_entries = (sq_size * sizeof(struct nes_hw_qp_wqe)) >> PAGE_SHIFT;
+	rq_pbl_entries = (rq_size * sizeof(struct nes_hw_qp_wqe)) >> 12;
+	sq_pbl_entries = (sq_size * sizeof(struct nes_hw_qp_wqe)) >> 12;
 	nesqp->hwqp.sq_pbase  = (le32_to_cpu (((u32 *)pbl)[0]) ) | ((u64)((le32_to_cpu (((u32 *)pbl)[1]))) << 32);
 	if (!nespbl->page) {
 		nes_debug(NES_DBG_QP, "QP nespbl->page is NULL \n");
@@ -2433,6 +2433,7 @@ static struct ib_mr *nes_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
 	u32 driver_key;
 	u32 root_pbl_index = 0;
 	u32 cur_pbl_index = 0;
+	u32 skip_pages;
 	u16 pbl_count;
 	u8 single_page = 1;
 	u8 stag_key;
@@ -2442,8 +2443,12 @@ static struct ib_mr *nes_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
 		return (struct ib_mr *)region;
 	}
 
-	nes_debug(NES_DBG_MR, "User base = 0x%lX, Virt base = 0x%lX, length = %u\n",
-			(unsigned long int)start, (unsigned long int)virt, (u32)length);
+	nes_debug(NES_DBG_MR, "User base = 0x%lX, Virt base = 0x%lX, length = %u,"
+			" offset = %u, page size = %u.\n",
+			(unsigned long int)start, (unsigned long int)virt, (u32)length,
+			region->offset, region->page_size);
+
+	skip_pages = ((u32)region->offset) >> 12;
 
 	if (ib_copy_from_udata(&req, udata, sizeof(req)))
 		return ERR_PTR(-EFAULT);
@@ -2460,7 +2465,7 @@ static struct ib_mr *nes_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
 			get_random_bytes(&next_stag_index, sizeof(next_stag_index));
 			stag_key = (u8)next_stag_index;
 
-			driver_key = 0;
+			driver_key = next_stag_index & 0x70000000;
 
 			next_stag_index >>= 8;
 			next_stag_index %= nesadapter->max_mr;
@@ -2484,70 +2489,6 @@ static struct ib_mr *nes_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
 				nes_debug(NES_DBG_MR, "Chunk: nents = %u, nmap = %u .\n",
 						chunk->nents, chunk->nmap);
 				for (nmap_index = 0; nmap_index < chunk->nmap; ++nmap_index) {
-					if ((page_count&0x01FF) == 0) {
-						if (page_count>(1024*512)) {
-							ib_umem_release(region);
-							pci_free_consistent(nesdev->pcidev, 4096, vpbl.pbl_vbase,
-									vpbl.pbl_pbase);
-							nes_free_resource(nesadapter,
-									nesadapter->allocated_mrs, stag_index);
-							kfree(nesmr);
-							return ERR_PTR(-E2BIG);
-						}
-						if (root_pbl_index == 1) {
-							root_vpbl.pbl_vbase = pci_alloc_consistent(nesdev->pcidev,
-									8192, &root_vpbl.pbl_pbase);
-							nes_debug(NES_DBG_MR, "Allocating root PBL, va = %p, pa = 0x%08X\n",
-									root_vpbl.pbl_vbase, (unsigned int)root_vpbl.pbl_pbase);
-							if (!root_vpbl.pbl_vbase) {
-								ib_umem_release(region);
-								pci_free_consistent(nesdev->pcidev, 4096, vpbl.pbl_vbase,
-										vpbl.pbl_pbase);
-								nes_free_resource(nesadapter, nesadapter->allocated_mrs,
-										stag_index);
-								kfree(nesmr);
-								return ERR_PTR(-ENOMEM);
-							}
-							root_vpbl.leaf_vpbl = kzalloc(sizeof(*root_vpbl.leaf_vpbl)*1024,
-									GFP_KERNEL);
-							if (!root_vpbl.leaf_vpbl) {
-								ib_umem_release(region);
-								pci_free_consistent(nesdev->pcidev, 8192, root_vpbl.pbl_vbase,
-										root_vpbl.pbl_pbase);
-								pci_free_consistent(nesdev->pcidev, 4096, vpbl.pbl_vbase,
-										vpbl.pbl_pbase);
-								nes_free_resource(nesadapter, nesadapter->allocated_mrs,
-										stag_index);
-								kfree(nesmr);
-								return ERR_PTR(-ENOMEM);
-							}
-							root_vpbl.pbl_vbase[0].pa_low =
-									cpu_to_le32((u32)vpbl.pbl_pbase);
-							root_vpbl.pbl_vbase[0].pa_high =
-									cpu_to_le32((u32)((((u64)vpbl.pbl_pbase) >> 32)));
-							root_vpbl.leaf_vpbl[0] = vpbl;
-						}
-						vpbl.pbl_vbase = pci_alloc_consistent(nesdev->pcidev, 4096,
-								&vpbl.pbl_pbase);
-						nes_debug(NES_DBG_MR, "Allocating leaf PBL, va = %p, pa = 0x%08X\n",
-								vpbl.pbl_vbase, (unsigned int)vpbl.pbl_pbase);
-						if (!vpbl.pbl_vbase) {
-							ib_umem_release(region);
-							nes_free_resource(nesadapter, nesadapter->allocated_mrs, stag_index);
-							ibmr = ERR_PTR(-ENOMEM);
-							kfree(nesmr);
-							goto reg_user_mr_err;
-						}
-						if (1 <= root_pbl_index) {
-							root_vpbl.pbl_vbase[root_pbl_index].pa_low =
-									cpu_to_le32((u32)vpbl.pbl_pbase);
-							root_vpbl.pbl_vbase[root_pbl_index].pa_high =
-									cpu_to_le32((u32)((((u64)vpbl.pbl_pbase)>>32)));
-							root_vpbl.leaf_vpbl[root_pbl_index] = vpbl;
-						}
-						root_pbl_index++;
-						cur_pbl_index = 0;
-					}
 					if (sg_dma_address(&chunk->page_list[nmap_index]) & ~PAGE_MASK) {
 						ib_umem_release(region);
 						nes_free_resource(nesadapter, nesadapter->allocated_mrs, stag_index);
@@ -2569,35 +2510,106 @@ static struct ib_mr *nes_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
 					}
 
 					region_length += sg_dma_len(&chunk->page_list[nmap_index]);
-					chunk_pages = sg_dma_len(&chunk->page_list[nmap_index]) >> PAGE_SHIFT;
-					for (page_index=0; page_index < chunk_pages; page_index++) {
+					chunk_pages = sg_dma_len(&chunk->page_list[nmap_index]) >> 12;
+					region_length -= skip_pages << 12;
+					for (page_index=skip_pages; page_index < chunk_pages; page_index++) {
+						skip_pages = 0;
+						if ((page_count!=0)&&(page_count<<12)-(region->offset&(4096-1))>=region->length)
+							goto enough_pages;
+						if ((page_count&0x01FF) == 0) {
+							if (page_count>(1024*512)) {
+								ib_umem_release(region);
+								pci_free_consistent(nesdev->pcidev, 4096, vpbl.pbl_vbase,
+										vpbl.pbl_pbase);
+								nes_free_resource(nesadapter,
+										nesadapter->allocated_mrs, stag_index);
+								kfree(nesmr);
+								nesmr = ERR_PTR(-E2BIG);
+								goto reg_user_mr_err;
+							}
+							if (root_pbl_index == 1) {
+								root_vpbl.pbl_vbase = pci_alloc_consistent(nesdev->pcidev,
+										8192, &root_vpbl.pbl_pbase);
+								nes_debug(NES_DBG_MR, "Allocating root PBL, va = %p, pa = 0x%08X\n",
+										root_vpbl.pbl_vbase, (unsigned int)root_vpbl.pbl_pbase);
+								if (!root_vpbl.pbl_vbase) {
+									ib_umem_release(region);
+									pci_free_consistent(nesdev->pcidev, 4096, vpbl.pbl_vbase,
+											vpbl.pbl_pbase);
+									nes_free_resource(nesadapter, nesadapter->allocated_mrs,
+											stag_index);
+									kfree(nesmr);
+									nesmr = ERR_PTR(-ENOMEM);
+									goto reg_user_mr_err;
+								}
+								root_vpbl.leaf_vpbl = kzalloc(sizeof(*root_vpbl.leaf_vpbl)*1024,
+										GFP_KERNEL);
+								if (!root_vpbl.leaf_vpbl) {
+									ib_umem_release(region);
+									pci_free_consistent(nesdev->pcidev, 8192, root_vpbl.pbl_vbase,
+											root_vpbl.pbl_pbase);
+									pci_free_consistent(nesdev->pcidev, 4096, vpbl.pbl_vbase,
+											vpbl.pbl_pbase);
+									nes_free_resource(nesadapter, nesadapter->allocated_mrs,
+											stag_index);
+									kfree(nesmr);
+									nesmr = ERR_PTR(-ENOMEM);
+									goto reg_user_mr_err;
+								}
+								root_vpbl.pbl_vbase[0].pa_low =
+										cpu_to_le32((u32)vpbl.pbl_pbase);
+								root_vpbl.pbl_vbase[0].pa_high =
+										cpu_to_le32((u32)((((u64)vpbl.pbl_pbase) >> 32)));
+								root_vpbl.leaf_vpbl[0] = vpbl;
+							}
+							vpbl.pbl_vbase = pci_alloc_consistent(nesdev->pcidev, 4096,
+									&vpbl.pbl_pbase);
+							nes_debug(NES_DBG_MR, "Allocating leaf PBL, va = %p, pa = 0x%08X\n",
+									vpbl.pbl_vbase, (unsigned int)vpbl.pbl_pbase);
+							if (!vpbl.pbl_vbase) {
+								ib_umem_release(region);
+								nes_free_resource(nesadapter, nesadapter->allocated_mrs, stag_index);
+								ibmr = ERR_PTR(-ENOMEM);
+								kfree(nesmr);
+								goto reg_user_mr_err;
+							}
+							if (1 <= root_pbl_index) {
+								root_vpbl.pbl_vbase[root_pbl_index].pa_low =
+										cpu_to_le32((u32)vpbl.pbl_pbase);
+								root_vpbl.pbl_vbase[root_pbl_index].pa_high =
+										cpu_to_le32((u32)((((u64)vpbl.pbl_pbase)>>32)));
+								root_vpbl.leaf_vpbl[root_pbl_index] = vpbl;
+							}
+							root_pbl_index++;
+							cur_pbl_index = 0;
+						}
 						if (single_page) {
 							if (page_count != 0) {
-								if ((last_dma_addr+PAGE_SIZE) !=
+								if ((last_dma_addr+4096) !=
 										(sg_dma_address(&chunk->page_list[nmap_index])+
-										(page_index*PAGE_SIZE)))
+										(page_index*4096)))
 									single_page = 0;
 								last_dma_addr = sg_dma_address(&chunk->page_list[nmap_index])+
-										(page_index*PAGE_SIZE);
+										(page_index*4096);
 							} else {
 								first_dma_addr = sg_dma_address(&chunk->page_list[nmap_index])+
-										(page_index*PAGE_SIZE);
+										(page_index*4096);
 								last_dma_addr = first_dma_addr;
 							}
 						}
 
 						vpbl.pbl_vbase[cur_pbl_index].pa_low =
 								cpu_to_le32((u32)(sg_dma_address(&chunk->page_list[nmap_index])+
-								(page_index*PAGE_SIZE)));
+								(page_index*4096)));
 						vpbl.pbl_vbase[cur_pbl_index].pa_high =
 								cpu_to_le32((u32)((((u64)(sg_dma_address(&chunk->page_list[nmap_index])+
-								(page_index*PAGE_SIZE))) >> 32)));
+								(page_index*4096))) >> 32)));
 						cur_pbl_index++;
 						page_count++;
 					}
 				}
 			}
-
+			enough_pages:
 			nes_debug(NES_DBG_MR, "calculating stag, stag_index=0x%08x, driver_key=0x%08x,"
 					" stag_key=0x%08x\n",
 					stag_index, driver_key, stag_key);
@@ -2609,12 +2621,6 @@ static struct ib_mr *nes_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
 			}
 
 			iova_start = virt;
-			nes_debug(NES_DBG_MR, "Registering STag 0x%08X, VA = 0x%08X, length = 0x%08X,"
-					" index = 0x%08X, region->length=0x%08llx\n",
-					stag, (unsigned int)iova_start,
-					(unsigned int)region_length, stag_index,
-					(unsigned long long)region->length);
-
 			/* Make the leaf PBL the root if only one PBL */
 			if (root_pbl_index == 1) {
 				root_vpbl.pbl_pbase = vpbl.pbl_pbase;
@@ -2626,6 +2632,11 @@ static struct ib_mr *nes_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
 				pbl_count = root_pbl_index;
 				first_dma_addr = 0;
 			}
+			nes_debug(NES_DBG_MR, "Registering STag 0x%08X, VA = 0x%08X, length = 0x%08X,"
+					" index = 0x%08X, region->length=0x%08llx, pbl_count = %u\n",
+					stag, (unsigned int)iova_start,
+					(unsigned int)region_length, stag_index,
+					(unsigned long long)region->length, pbl_count);
 			ret = nes_reg_mr( nesdev, nespd, stag, region->length, &root_vpbl,
 					first_dma_addr, pbl_count, (u16)cur_pbl_index, acc, &iova_start);
 
@@ -2684,8 +2695,8 @@ static struct ib_mr *nes_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
 			}
 			nesmr->region = region;
 			nes_ucontext = to_nesucontext(pd->uobject->context);
-			pbl_depth = region->length >> PAGE_SHIFT;
-			pbl_depth += (region->length & ~PAGE_MASK) ? 1 : 0;
+			pbl_depth = region->length >> 12;
+			pbl_depth += (region->length & (4096-1)) ? 1 : 0;
 			nespbl->pbl_size = pbl_depth*sizeof(u64);
 			if (req.reg_type == IWNES_MEMREG_TYPE_QP) {
 				nes_debug(NES_DBG_MR, "Attempting to allocate QP PBL memory");
@@ -2714,15 +2725,16 @@ static struct ib_mr *nes_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
 
 			list_for_each_entry(chunk, &region->chunk_list, list) {
 				for (nmap_index = 0; nmap_index < chunk->nmap; ++nmap_index) {
-					chunk_pages = sg_dma_len(&chunk->page_list[nmap_index]) >> PAGE_SHIFT;
+					chunk_pages = sg_dma_len(&chunk->page_list[nmap_index]) >> 12;
+					chunk_pages += (sg_dma_len(&chunk->page_list[nmap_index]) & (4096-1)) ? 1 : 0;
 					nespbl->page = sg_page(&chunk->page_list[0]);
 					for (page_index=0; page_index<chunk_pages; page_index++) {
 						((u32 *)pbl)[0] = cpu_to_le32((u32)
 								(sg_dma_address(&chunk->page_list[nmap_index])+
-								(page_index*PAGE_SIZE)));
+								(page_index*4096)));
 						((u32 *)pbl)[1] = cpu_to_le32(((u64)
 								(sg_dma_address(&chunk->page_list[nmap_index])+
-								(page_index*PAGE_SIZE)))>>32);
+								(page_index*4096)))>>32);
 						nes_debug(NES_DBG_MR, "pbl=%p, *pbl=0x%016llx, 0x%08x%08x\n", pbl,
 								(unsigned long long)*pbl,
 								le32_to_cpu(((u32 *)pbl)[1]), le32_to_cpu(((u32 *)pbl)[0]));



More information about the general mailing list