Numa-aware, growable physical memory allocator (pmalloc)

Change-Id: Ic4c46bc733afae8bf0d8146623ed15633928de30
Signed-off-by: Damjan Marion <damarion@cisco.com>

4 files changed, 68 insertions, 345 deletions

diff --git a/src/vlib/linux/pci.c b/src/vlib/linux/pci.c
index b55fb5042f7..0e2241b0e58 100644
--- a/src/vlib/linux/pci.c
+++ b/src/vlib/linux/pci.c
@@ -951,6 +951,21 @@ add_device_vfio (vlib_main_t * vm, linux_pci_device_t * p,
       linux_pci_vfio_unmask_intx (vm, p);
     }
 
+  if (p->supports_va_dma)
+    {
+      vlib_buffer_pool_t *bp;
+      /* *INDENT-OFF* */
+      vec_foreach (bp, buffer_main.buffer_pools)
+	{
+	  u32 i;
+	  vlib_physmem_map_t *pm;
+	  pm = vlib_physmem_get_map (vm, bp->physmem_map_index);
+	  for (i = 0; i < pm->n_pages; i++)
+	    vfio_map_physmem_page (vm, pm->base + (i << pm->log2_page_size));
+	}
+      /* *INDENT-ON* */
+    }
+
   if (r && r->init_function)
     err = r->init_function (lpm->vlib_main, p->handle);
 
@@ -1092,6 +1107,25 @@ vlib_pci_map_region_fixed (vlib_main_t * vm, vlib_pci_dev_handle_t h,
 }
 
 clib_error_t *
+vlib_pci_map_dma (vlib_main_t * vm, vlib_pci_dev_handle_t h, void *ptr)
+{
+  linux_pci_device_t *p = linux_pci_get_device (h);
+
+  if (!p->supports_va_dma)
+    return 0;
+
+  return vfio_map_physmem_page (vm, ptr);
+}
+
+int
+vlib_pci_supports_virtual_addr_dma (vlib_main_t * vm, vlib_pci_dev_handle_t h)
+{
+  linux_pci_device_t *p = linux_pci_get_device (h);
+
+  return p->supports_va_dma != 0;
+}
+
+clib_error_t *
 vlib_pci_device_open (vlib_main_t * vm, vlib_pci_addr_t * addr,
 		      pci_device_id_t ids[], vlib_pci_dev_handle_t * handle)
 {
diff --git a/src/vlib/linux/physmem.c b/src/vlib/linux/physmem.c
deleted file mode 100755
index 90b0f8cab3d..00000000000
--- a/src/vlib/linux/physmem.c
+++ /dev/null
@@ -1,307 +0,0 @@
-/*
- * Copyright (c) 2015 Cisco and/or its affiliates.
- * Licensed under the Apache License, Version 2.0 (the "License");
- * you may not use this file except in compliance with the License.
- * You may obtain a copy of the License at:
- *
- *     http://www.apache.org/licenses/LICENSE-2.0
- *
- * Unless required by applicable law or agreed to in writing, software
- * distributed under the License is distributed on an "AS IS" BASIS,
- * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
- * See the License for the specific language governing permissions and
- * limitations under the License.
- */
-/*
- * physmem.c: Unix physical memory
- *
- * Copyright (c) 2008 Eliot Dresselhaus
- *
- * Permission is hereby granted, free of charge, to any person obtaining
- * a copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sublicense, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- *
- * The above copyright notice and this permission notice shall be
- * included in all copies or substantial portions of the Software.
- *
- *  THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- *  EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- *  MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
- *  NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
- *  LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- *  OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- *  WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- */
-
-#include <unistd.h>
-#include <sys/types.h>
-#include <sys/mount.h>
-#include <sys/mman.h>
-#include <sys/fcntl.h>
-#include <sys/stat.h>
-#include <unistd.h>
-
-#include <vppinfra/linux/syscall.h>
-#include <vppinfra/linux/sysfs.h>
-#include <vlib/vlib.h>
-#include <vlib/physmem.h>
-#include <vlib/unix/unix.h>
-#include <vlib/pci/pci.h>
-#include <vlib/linux/vfio.h>
-
-static void *
-unix_physmem_alloc_aligned (vlib_main_t * vm, vlib_physmem_region_index_t idx,
-			    uword n_bytes, uword alignment)
-{
-  vlib_physmem_region_t *pr = vlib_physmem_get_region (vm, idx);
-  uword lo_offset, hi_offset;
-  uword *to_free = 0;
-
-  if (pr->heap == 0)
-    return 0;
-
-  /* IO memory is always at least cache aligned. */
-  alignment = clib_max (alignment, CLIB_CACHE_LINE_BYTES);
-
-  while (1)
-    {
-#if USE_DLMALLOC == 0
-
-      mheap_get_aligned (pr->heap, n_bytes,
-			 /* align */ alignment,
-			 /* align offset */ 0,
-			 &lo_offset);
-#else
-      lo_offset = (uword) mspace_get_aligned (pr->heap, n_bytes,
-					      alignment, ~0ULL /* offset */ );
-      if (lo_offset == 0)
-	lo_offset = ~0ULL;
-#endif
-
-      /* Allocation failed? */
-      if (lo_offset == ~0)
-	break;
-
-      /* Make sure allocation does not span DMA physical chunk boundary. */
-      hi_offset = lo_offset + n_bytes - 1;
-
-      if (((pointer_to_uword (pr->heap) + lo_offset) >> pr->log2_page_size) ==
-	  ((pointer_to_uword (pr->heap) + hi_offset) >> pr->log2_page_size))
-	break;
-
-      /* Allocation would span chunk boundary, queue it to be freed as soon as
-         we find suitable chunk. */
-      vec_add1 (to_free, lo_offset);
-    }
-
-  if (to_free != 0)
-    {
-      uword i;
-      for (i = 0; i < vec_len (to_free); i++)
-	{
-#if USE_DLMALLOC == 0
-	  mheap_put (pr->heap, to_free[i]);
-#else
-	  mspace_put_no_offset (pr->heap, (void *) to_free[i]);
-#endif
-	}
-      vec_free (to_free);
-    }
-
-#if USE_DLMALLOC == 0
-  return lo_offset != ~0 ? (void *) (pr->heap + lo_offset) : 0;
-#else
-  return lo_offset != ~0 ? (void *) lo_offset : 0;
-#endif
-}
-
-static void
-unix_physmem_free (vlib_main_t * vm, vlib_physmem_region_index_t idx, void *x)
-{
-  vlib_physmem_region_t *pr = vlib_physmem_get_region (vm, idx);
-  /* Return object to region's heap. */
-#if USE_DLMALLOC == 0
-  mheap_put (pr->heap, x - pr->heap);
-#else
-  mspace_put_no_offset (pr->heap, x);
-#endif
-}
-
-static clib_error_t *
-unix_physmem_region_alloc (vlib_main_t * vm, char *name, u32 size,
-			   u8 numa_node, u32 flags,
-			   vlib_physmem_region_index_t * idx)
-{
-  vlib_physmem_main_t *vpm = &physmem_main;
-  vlib_physmem_region_t *pr;
-  clib_error_t *error = 0;
-  clib_mem_vm_alloc_t alloc = { 0 };
-  int i;
-
-  pool_get (vpm->regions, pr);
-
-  if ((pr - vpm->regions) >= 256)
-    {
-      error = clib_error_return (0, "maximum number of regions reached");
-      goto error;
-    }
-
-  alloc.name = name;
-  alloc.size = size;
-  alloc.numa_node = numa_node;
-
-  alloc.flags = (flags & VLIB_PHYSMEM_F_SHARED) ?
-    CLIB_MEM_VM_F_SHARED : CLIB_MEM_VM_F_LOCKED;
-
-  if ((flags & VLIB_PHYSMEM_F_HUGETLB))
-    {
-      alloc.flags |= CLIB_MEM_VM_F_HUGETLB;
-      alloc.flags |= CLIB_MEM_VM_F_HUGETLB_PREALLOC;
-      alloc.flags |= CLIB_MEM_VM_F_NUMA_FORCE;
-    }
-  else
-    {
-      alloc.flags |= CLIB_MEM_VM_F_NUMA_PREFER;
-    }
-
-  error = clib_mem_vm_ext_alloc (&alloc);
-  if (error)
-    goto error;
-
-  pr->index = pr - vpm->regions;
-  pr->flags = flags;
-  pr->fd = alloc.fd;
-  pr->mem = alloc.addr;
-  pr->log2_page_size = alloc.log2_page_size;
-  pr->n_pages = alloc.n_pages;
-  pr->size = (u64) pr->n_pages << (u64) pr->log2_page_size;
-  pr->page_mask = (1ull << pr->log2_page_size) - 1;
-  pr->numa_node = numa_node;
-  pr->name = format (0, "%s%c", name, 0);
-
-  for (i = 0; i < pr->n_pages; i++)
-    {
-      void *ptr = pr->mem + ((u64) i << pr->log2_page_size);
-      int node;
-      if ((move_pages (0, 1, &ptr, 0, &node, 0) == 0) && (numa_node != node))
-	{
-	  clib_warning ("physmem page for region \'%s\' allocated on the"
-			" wrong numa node (requested %u actual %u)",
-			pr->name, pr->numa_node, node, i);
-	  break;
-	}
-    }
-
-  pr->page_table = clib_mem_vm_get_paddr (pr->mem, pr->log2_page_size,
-					  pr->n_pages);
-
-  linux_vfio_dma_map_regions (vm);
-
-  if (flags & VLIB_PHYSMEM_F_INIT_MHEAP)
-    {
-#if USE_DLMALLOC == 0
-      pr->heap = mheap_alloc_with_flags (pr->mem, pr->size,
-					 /* Don't want mheap mmap/munmap with IO memory. */
-					 MHEAP_FLAG_DISABLE_VM |
-					 MHEAP_FLAG_THREAD_SAFE);
-#else
-      pr->heap = create_mspace_with_base (pr->mem, pr->size, 1 /* locked */ );
-      mspace_disable_expand (pr->heap);
-#endif
-    }
-
-  *idx = pr->index;
-
-  goto done;
-
-error:
-  clib_memset (pr, 0, sizeof (*pr));
-  pool_put (vpm->regions, pr);
-
-done:
-  return error;
-}
-
-static void
-unix_physmem_region_free (vlib_main_t * vm, vlib_physmem_region_index_t idx)
-{
-  vlib_physmem_main_t *vpm = &physmem_main;
-  vlib_physmem_region_t *pr = vlib_physmem_get_region (vm, idx);
-
-  if (pr->fd > 0)
-    close (pr->fd);
-  munmap (pr->mem, pr->size);
-  vec_free (pr->name);
-  pool_put (vpm->regions, pr);
-}
-
-clib_error_t *
-unix_physmem_init (vlib_main_t * vm)
-{
-  vlib_physmem_main_t *vpm = &physmem_main;
-  clib_error_t *error = 0;
-  u64 *pt = 0;
-
-  /* Avoid multiple calls. */
-  if (vm->os_physmem_alloc_aligned)
-    return error;
-
-  /* check if pagemap is accessible */
-  pt = clib_mem_vm_get_paddr (&pt, min_log2 (sysconf (_SC_PAGESIZE)), 1);
-  if (pt[0])
-    vpm->flags |= VLIB_PHYSMEM_MAIN_F_HAVE_PAGEMAP;
-  vec_free (pt);
-
-  if ((error = linux_vfio_init (vm)))
-    return error;
-
-  vm->os_physmem_alloc_aligned = unix_physmem_alloc_aligned;
-  vm->os_physmem_free = unix_physmem_free;
-  vm->os_physmem_region_alloc = unix_physmem_region_alloc;
-  vm->os_physmem_region_free = unix_physmem_region_free;
-
-  return error;
-}
-
-static clib_error_t *
-show_physmem (vlib_main_t * vm,
-	      unformat_input_t * input, vlib_cli_command_t * cmd)
-{
-  vlib_physmem_main_t *vpm = &physmem_main;
-  vlib_physmem_region_t *pr;
-
-  /* *INDENT-OFF* */
-  pool_foreach (pr, vpm->regions, (
-    {
-      vlib_cli_output (vm, "index %u name '%s' page-size %uKB num-pages %d "
-		       "numa-node %u fd %d\n",
-		       pr->index, pr->name, (1 << (pr->log2_page_size -10)),
-		       pr->n_pages, pr->numa_node, pr->fd);
-      if (pr->heap)
-	vlib_cli_output (vm, "  %U", format_mheap, pr->heap, /* verbose */ 1);
-      else
-	vlib_cli_output (vm, "  no heap\n");
-    }));
-  /* *INDENT-ON* */
-  return 0;
-}
-
-/* *INDENT-OFF* */
-VLIB_CLI_COMMAND (show_physmem_command, static) = {
-  .path = "show physmem",
-  .short_help = "Show physical memory allocation",
-  .function = show_physmem,
-};
-/* *INDENT-ON* */
-
-/*
- * fd.io coding-style-patch-verification: ON
- *
- * Local Variables:
- * eval: (c-set-style "gnu")
- * End:
- */
diff --git a/src/vlib/linux/vfio.c b/src/vlib/linux/vfio.c
index e72f10388ca..d300a683dd7 100644
--- a/src/vlib/linux/vfio.c
+++ b/src/vlib/linux/vfio.c
@@ -34,52 +34,46 @@
 
 linux_vfio_main_t vfio_main;
 
-static int
-vfio_map_regions (vlib_main_t * vm, int fd)
+clib_error_t *
+vfio_map_physmem_page (vlib_main_t * vm, void *addr)
 {
-  vlib_physmem_main_t *vpm = &physmem_main;
+  vlib_physmem_main_t *vpm = &vm->physmem_main;
   linux_vfio_main_t *lvm = &vfio_main;
-  vlib_physmem_region_t *pr;
   struct vfio_iommu_type1_dma_map dm = { 0 };
-  int i;
+  uword log2_page_size = vpm->pmalloc_main->log2_page_sz;
+  uword physmem_start = pointer_to_uword (vpm->pmalloc_main->base);
+
+  if (lvm->container_fd == -1)
+    return clib_error_return (0, "No cointainer fd");
+
+  u32 page_index = vlib_physmem_get_page_index (vm, addr);
+
+  if (clib_bitmap_get (lvm->physmem_pages_mapped, page_index))
+    {
+      vlib_log_debug (lvm->log_default, "map DMA va:%p page:%u already "
+		      "mapped", addr, page_index);
+      return 0;
+    }
 
   dm.argsz = sizeof (struct vfio_iommu_type1_dma_map);
   dm.flags = VFIO_DMA_MAP_FLAG_READ | VFIO_DMA_MAP_FLAG_WRITE;
+  dm.vaddr = physmem_start + (page_index << log2_page_size);
+  dm.size = 1ULL << log2_page_size;
+  dm.iova = dm.vaddr;
+  vlib_log_debug (lvm->log_default, "map DMA page:%u va:0x%lx iova:%lx "
+		  "size:0x%lx", page_index, dm.vaddr, dm.iova, dm.size);
 
-  /* *INDENT-OFF* */
-  pool_foreach (pr, vpm->regions,
+  if (ioctl (lvm->container_fd, VFIO_IOMMU_MAP_DMA, &dm) == -1)
     {
-      vec_foreach_index (i, pr->page_table)
-        {
-	  int rv;
-	  dm.vaddr = pointer_to_uword (pr->mem) + ((u64)i << pr->log2_page_size);
-	  dm.size = 1ull << pr->log2_page_size;
-	  dm.iova = dm.vaddr;
-	  vlib_log_debug (lvm->log_default, "map DMA va:0x%lx iova:%lx "
-			  "size:0x%lx", dm.vaddr, dm.iova, dm.size);
-
-	  if ((rv = ioctl (fd, VFIO_IOMMU_MAP_DMA, &dm)) &&
-	      errno != EINVAL)
-	    {
-	      vlib_log_err (lvm->log_default, "map DMA va:0x%lx iova:%lx "
-			    "size:0x%lx failed, error %s (errno %d)",
-			    dm.vaddr, dm.iova, dm.size, strerror (errno),
-			    errno);
-	      return rv;
-	    }
-        }
-    });
-  /* *INDENT-ON* */
-  return 0;
-}
-
-void
-linux_vfio_dma_map_regions (vlib_main_t * vm)
-{
-  linux_vfio_main_t *lvm = &vfio_main;
+      vlib_log_err (lvm->log_default, "map DMA page:%u va:0x%lx iova:%lx "
+		    "size:0x%lx failed, error %s (errno %d)", page_index,
+		    dm.vaddr, dm.iova, dm.size, strerror (errno), errno);
+      return clib_error_return_unix (0, "physmem DMA map failed");
+    }
 
-  if (lvm->container_fd != -1)
-    vfio_map_regions (vm, lvm->container_fd);
+  lvm->physmem_pages_mapped = clib_bitmap_set (lvm->physmem_pages_mapped,
+					       page_index, 1);
+  return 0;
 }
 
 static linux_pci_vfio_iommu_group_t *
diff --git a/src/vlib/linux/vfio.h b/src/vlib/linux/vfio.h
index aae8e3c6ee7..c1d815664c9 100644
--- a/src/vlib/linux/vfio.h
+++ b/src/vlib/linux/vfio.h
@@ -36,6 +36,8 @@ typedef struct
   /* iommu group pool index by group id  hash */
   uword *iommu_pool_index_by_group;
 
+  clib_bitmap_t *physmem_pages_mapped;
+
   /* logging */
   vlib_log_class_t log_default;
 } linux_vfio_main_t;
@@ -43,7 +45,7 @@ typedef struct
 extern linux_vfio_main_t vfio_main;
 
 clib_error_t *linux_vfio_init (vlib_main_t * vm);
-void linux_vfio_dma_map_regions (vlib_main_t * vm);
+clib_error_t *vfio_map_physmem_page (vlib_main_t * vm, void *addr);
 clib_error_t *linux_vfio_group_get_device_fd (vlib_pci_addr_t * addr,
 					      int *fd, int *is_noiommu);