[dpdk-dev] [PATCH v2] vhost: fix mmap failure as len not aligned with hugepage size

[dpdk-dev] [PATCH v2] vhost: fix mmap failure as len not aligned with hugepage size

[Jianfeng Tan]

This patch fixes a bug under lower version linux kernel, mmap()
fails when length is not aligned with hugepage size. mmap()
without flag of MAP_ANONYMOUS, should be called with length
argument aligned with hugepagesz at older longterm version
Linux, like 2.6.32 and 3.2.72, or mmap() will fail with EINVAL.
This bug was fixed in Linux kernel by commit:
dab2d3dc45ae7343216635d981d43637e1cb7d45
To avoid failure, make sure in caller to keep length aligned.

Signed-off-by: Jianfeng Tan <jianfeng.tan@intel.com>
---
 lib/librte_vhost/vhost_user/virtio-net-user.c | 36 ++++++++++++++++-----------
 1 file changed, 21 insertions(+), 15 deletions(-)

diff --git a/lib/librte_vhost/vhost_user/virtio-net-user.c b/lib/librte_vhost/vhost_user/virtio-net-user.c
index d07452a..7ce48d0 100644
--- a/lib/librte_vhost/vhost_user/virtio-net-user.c
+++ b/lib/librte_vhost/vhost_user/virtio-net-user.c
@@ -74,7 +74,6 @@ free_mem_region(struct virtio_net *dev)
 {
 	struct orig_region_map *region;
 	unsigned int idx;
-	uint64_t alignment;
 
 	if (!dev || !dev->mem)
 		return;
@@ -82,12 +81,8 @@ free_mem_region(struct virtio_net *dev)
 	region = orig_region(dev->mem, dev->mem->nregions);
 	for (idx = 0; idx < dev->mem->nregions; idx++) {
 		if (region[idx].mapped_address) {
-			alignment = region[idx].blksz;
-			munmap((void *)(uintptr_t)
-				RTE_ALIGN_FLOOR(
-					region[idx].mapped_address, alignment),
-				RTE_ALIGN_CEIL(
-					region[idx].mapped_size, alignment));
+			munmap((void *)region[idx].mapped_address,
+					region[idx].mapped_size);
 			close(region[idx].fd);
 		}
 	}
@@ -147,6 +142,18 @@ user_set_mem_table(struct vhost_device_ctx ctx, struct VhostUserMsg *pmsg)
 		/* This is ugly */
 		mapped_size = memory.regions[idx].memory_size +
 			memory.regions[idx].mmap_offset;
+
+		/* mmap() without flag of MAP_ANONYMOUS, should be called
+		 * with length argument aligned with hugepagesz at older
+		 * longterm version Linux, like 2.6.32 and 3.2.72, or
+		 * mmap() will fail with EINVAL.
+		 *
+		 * to avoid failure, make sure in caller to keep length
+		 * aligned.
+		 */
+		alignment = get_blk_size(pmsg->fds[idx]);
+		mapped_size = RTE_ALIGN_CEIL(mapped_size, alignment);
+
 		mapped_address = (uint64_t)(uintptr_t)mmap(NULL,
 			mapped_size,
 			PROT_READ | PROT_WRITE, MAP_SHARED,
@@ -154,9 +161,11 @@ user_set_mem_table(struct vhost_device_ctx ctx, struct VhostUserMsg *pmsg)
 			0);
 
 		RTE_LOG(INFO, VHOST_CONFIG,
-			"mapped region %d fd:%d to %p sz:0x%"PRIx64" off:0x%"PRIx64"\n",
+			"mapped region %d fd:%d to:%p sz:0x%"PRIx64" "
+			"off:0x%"PRIx64" align:0x%"PRIx64"\n",
 			idx, pmsg->fds[idx], (void *)(uintptr_t)mapped_address,
-			mapped_size, memory.regions[idx].mmap_offset);
+			mapped_size, memory.regions[idx].mmap_offset,
+			alignment);
 
 		if (mapped_address == (uint64_t)(uintptr_t)MAP_FAILED) {
 			RTE_LOG(ERR, VHOST_CONFIG,
@@ -166,7 +175,7 @@ user_set_mem_table(struct vhost_device_ctx ctx, struct VhostUserMsg *pmsg)
 
 		pregion_orig[idx].mapped_address = mapped_address;
 		pregion_orig[idx].mapped_size = mapped_size;
-		pregion_orig[idx].blksz = get_blk_size(pmsg->fds[idx]);
+		pregion_orig[idx].blksz = alignment;
 		pregion_orig[idx].fd = pmsg->fds[idx];
 
 		mapped_address +=  memory.regions[idx].mmap_offset;
@@ -193,11 +202,8 @@ user_set_mem_table(struct vhost_device_ctx ctx, struct VhostUserMsg *pmsg)
 
 err_mmap:
 	while (idx--) {
-		alignment = pregion_orig[idx].blksz;
-		munmap((void *)(uintptr_t)RTE_ALIGN_FLOOR(
-			pregion_orig[idx].mapped_address, alignment),
-			RTE_ALIGN_CEIL(pregion_orig[idx].mapped_size,
-					alignment));
+		munmap((void *)pregion_orig[idx].mapped_address,
+				pregion_orig[idx].mapped_size);
 		close(pregion_orig[idx].fd);
 	}
 	free(dev->mem);
-- 
2.1.4

[dpdk-dev] [PATCH v3] vhost: fix mmap failure as len not aligned with hugepage size

[Jianfeng Tan]

This patch fixes a bug under lower version linux kernel, mmap()
fails when length is not aligned with hugepage size. mmap()
without flag of MAP_ANONYMOUS, should be called with length
argument aligned with hugepagesz at older longterm version
Linux, like 2.6.32 and 3.2.72, or mmap() will fail with EINVAL.
This bug was fixed in Linux kernel by commit:
dab2d3dc45ae7343216635d981d43637e1cb7d45
To avoid failure, make sure in caller to keep length aligned.

v3 changes:
 - fix (u64) -> (void *) convert error on 32-bit system

v2 changes:
 - add Kernel version comments and commit msg
 - remove unnecessary alignments when munmap

Signed-off-by: Jianfeng Tan <jianfeng.tan@intel.com>
---
 lib/librte_vhost/vhost_user/virtio-net-user.c | 36 ++++++++++++++++-----------
 1 file changed, 21 insertions(+), 15 deletions(-)

diff --git a/lib/librte_vhost/vhost_user/virtio-net-user.c b/lib/librte_vhost/vhost_user/virtio-net-user.c
index d07452a..99da029 100644
--- a/lib/librte_vhost/vhost_user/virtio-net-user.c
+++ b/lib/librte_vhost/vhost_user/virtio-net-user.c
@@ -74,7 +74,6 @@ free_mem_region(struct virtio_net *dev)
 {
 	struct orig_region_map *region;
 	unsigned int idx;
-	uint64_t alignment;
 
 	if (!dev || !dev->mem)
 		return;
@@ -82,12 +81,8 @@ free_mem_region(struct virtio_net *dev)
 	region = orig_region(dev->mem, dev->mem->nregions);
 	for (idx = 0; idx < dev->mem->nregions; idx++) {
 		if (region[idx].mapped_address) {
-			alignment = region[idx].blksz;
-			munmap((void *)(uintptr_t)
-				RTE_ALIGN_FLOOR(
-					region[idx].mapped_address, alignment),
-				RTE_ALIGN_CEIL(
-					region[idx].mapped_size, alignment));
+			munmap((void *)(uintptr_t)region[idx].mapped_address,
+					region[idx].mapped_size);
 			close(region[idx].fd);
 		}
 	}
@@ -147,6 +142,18 @@ user_set_mem_table(struct vhost_device_ctx ctx, struct VhostUserMsg *pmsg)
 		/* This is ugly */
 		mapped_size = memory.regions[idx].memory_size +
 			memory.regions[idx].mmap_offset;
+
+		/* mmap() without flag of MAP_ANONYMOUS, should be called
+		 * with length argument aligned with hugepagesz at older
+		 * longterm version Linux, like 2.6.32 and 3.2.72, or
+		 * mmap() will fail with EINVAL.
+		 *
+		 * to avoid failure, make sure in caller to keep length
+		 * aligned.
+		 */
+		alignment = get_blk_size(pmsg->fds[idx]);
+		mapped_size = RTE_ALIGN_CEIL(mapped_size, alignment);
+
 		mapped_address = (uint64_t)(uintptr_t)mmap(NULL,
 			mapped_size,
 			PROT_READ | PROT_WRITE, MAP_SHARED,
@@ -154,9 +161,11 @@ user_set_mem_table(struct vhost_device_ctx ctx, struct VhostUserMsg *pmsg)
 			0);
 
 		RTE_LOG(INFO, VHOST_CONFIG,
-			"mapped region %d fd:%d to %p sz:0x%"PRIx64" off:0x%"PRIx64"\n",
+			"mapped region %d fd:%d to:%p sz:0x%"PRIx64" "
+			"off:0x%"PRIx64" align:0x%"PRIx64"\n",
 			idx, pmsg->fds[idx], (void *)(uintptr_t)mapped_address,
-			mapped_size, memory.regions[idx].mmap_offset);
+			mapped_size, memory.regions[idx].mmap_offset,
+			alignment);
 
 		if (mapped_address == (uint64_t)(uintptr_t)MAP_FAILED) {
 			RTE_LOG(ERR, VHOST_CONFIG,
@@ -166,7 +175,7 @@ user_set_mem_table(struct vhost_device_ctx ctx, struct VhostUserMsg *pmsg)
 
 		pregion_orig[idx].mapped_address = mapped_address;
 		pregion_orig[idx].mapped_size = mapped_size;
-		pregion_orig[idx].blksz = get_blk_size(pmsg->fds[idx]);
+		pregion_orig[idx].blksz = alignment;
 		pregion_orig[idx].fd = pmsg->fds[idx];
 
 		mapped_address +=  memory.regions[idx].mmap_offset;
@@ -193,11 +202,8 @@ user_set_mem_table(struct vhost_device_ctx ctx, struct VhostUserMsg *pmsg)
 
 err_mmap:
 	while (idx--) {
-		alignment = pregion_orig[idx].blksz;
-		munmap((void *)(uintptr_t)RTE_ALIGN_FLOOR(
-			pregion_orig[idx].mapped_address, alignment),
-			RTE_ALIGN_CEIL(pregion_orig[idx].mapped_size,
-					alignment));
+		munmap((void *)(uintptr_t)pregion_orig[idx].mapped_address,
+				pregion_orig[idx].mapped_size);
 		close(pregion_orig[idx].fd);
 	}
 	free(dev->mem);
-- 
2.1.4

Re: [dpdk-dev] [PATCH v2] vhost: fix mmap failure as len not aligned with hugepage size

[Xie, Huawei]

On 11/12/2015 1:04 PM, Tan, Jianfeng wrote:
> This patch fixes a bug under lower version linux kernel, mmap()
> fails when length is not aligned with hugepage size. mmap()
> without flag of MAP_ANONYMOUS, should be called with length
> argument aligned with hugepagesz at older longterm version
> Linux, like 2.6.32 and 3.2.72, or mmap() will fail with EINVAL.
> This bug was fixed in Linux kernel by commit:
> dab2d3dc45ae7343216635d981d43637e1cb7d45
> To avoid failure, make sure in caller to keep length aligned.
>
> Signed-off-by: Jianfeng Tan <jianfeng.tan@intel.com>
Acked-by: Huawei Xie <huawei.xie@intel.com>

Next time please add --in-reply-to with original message id.
> ---
>  lib/librte_vhost/vhost_user/virtio-net-user.c | 36 ++++++++++++++++-----------
>  1 file changed, 21 insertions(+), 15 deletions(-)
>

Re: [dpdk-dev] [PATCH v2] vhost: fix mmap failure as len not aligned with hugepage size

[Thomas Monjalon]

2015-11-12 06:04, Jianfeng Tan:
> -			alignment = region[idx].blksz;
> -			munmap((void *)(uintptr_t)
> -				RTE_ALIGN_FLOOR(
> -					region[idx].mapped_address, alignment),
> -				RTE_ALIGN_CEIL(
> -					region[idx].mapped_size, alignment));
> +			munmap((void *)region[idx].mapped_address,
> +					region[idx].mapped_size);

Sorry, it does not compile for 32-bit:
virtio-net-user.c:84:11: error: cast to pointer from integer of different size

Re: [dpdk-dev] [PATCH v2] vhost: fix mmap failure as len not aligned with hugepage size

[Tan, Jianfeng]

> -----Original Message-----
> From: Thomas Monjalon [mailto:thomas.monjalon@6wind.com]
> Sent: Thursday, November 12, 2015 7:19 PM
> To: Tan, Jianfeng
> Cc: dev@dpdk.org
> Subject: Re: [dpdk-dev] [PATCH v2] vhost: fix mmap failure as len not aligned
> with hugepage size
> 
> 2015-11-12 06:04, Jianfeng Tan:
> > -			alignment = region[idx].blksz;
> > -			munmap((void *)(uintptr_t)
> > -				RTE_ALIGN_FLOOR(
> > -					region[idx].mapped_address,
> alignment),
> > -				RTE_ALIGN_CEIL(
> > -					region[idx].mapped_size,
> alignment));
> > +			munmap((void *)region[idx].mapped_address,
> > +					region[idx].mapped_size);
> 
> Sorry, it does not compile for 32-bit:
> virtio-net-user.c:84:11: error: cast to pointer from integer of different size

Oops, sorry, should use (void *)(uintptr_t). I'll resend this patch.

Jianfeng

Re: [dpdk-dev] [PATCH v3] vhost: fix mmap failure as len not aligned with hugepage size

[Thomas Monjalon]

2015-11-12 14:06, Jianfeng Tan:
> This patch fixes a bug under lower version linux kernel, mmap()
> fails when length is not aligned with hugepage size. mmap()
> without flag of MAP_ANONYMOUS, should be called with length
> argument aligned with hugepagesz at older longterm version
> Linux, like 2.6.32 and 3.2.72, or mmap() will fail with EINVAL.
> This bug was fixed in Linux kernel by commit:
> dab2d3dc45ae7343216635d981d43637e1cb7d45
> To avoid failure, make sure in caller to keep length aligned.
> 
> v3 changes:
>  - fix (u64) -> (void *) convert error on 32-bit system
> 
> v2 changes:
>  - add Kernel version comments and commit msg
>  - remove unnecessary alignments when munmap
> 
> Signed-off-by: Jianfeng Tan <jianfeng.tan@intel.com>

You forgot to keep Huawei's ack from v2.

Applied, thanks