[dpdk-dev] [PATCH] memalloc: check for contiguousness in external segments

[dpdk-dev] [PATCH] memalloc: check for contiguousness in external segments

[Anatoly Burakov]

For IOVA as VA mode, we assume that memory is contiguous. However,
for external segments that assumption may not necessarily hold.
Fix the code to not assume that external memory segments are
contiguous even in IOVA as VA mode.

Fixes: 5282bb1c3695 ("mem: allow memseg lists to be marked as external")

Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com>
---
 lib/librte_eal/common/eal_common_memalloc.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/librte_eal/common/eal_common_memalloc.c b/lib/librte_eal/common/eal_common_memalloc.c
index 1d41ea112..371271979 100644
--- a/lib/librte_eal/common/eal_common_memalloc.c
+++ b/lib/librte_eal/common/eal_common_memalloc.c
@@ -77,7 +77,7 @@ eal_memalloc_is_contig(const struct rte_memseg_list *msl, void *start,
 	const struct rte_memseg *ms;
 
 	/* for IOVA_VA, it's always contiguous */
-	if (rte_eal_iova_mode() == RTE_IOVA_VA)
+	if (rte_eal_iova_mode() == RTE_IOVA_VA && !msl->external)
 		return true;
 
 	/* for legacy memory, it's always contiguous */
-- 
2.17.1

[dpdk-dev] [PATCH v2] malloc: fix adjacency check to also include segment list

[Anatoly Burakov]

It may so happen that two memory locations may be adjacent in
virtual memory, but belong to different segment lists. With
current code, such segments will be concatenated. Fix the
adjacency checking code to also check if the adjacent malloc
elements belong to the same memseg list.

Fixes: 66cc45e293ed ("mem: replace memseg with memseg lists")

Cc: stable@dpdk.org

Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com>
---

Notes:
    v2: fixed typo in second comparison

 lib/librte_eal/common/malloc_elem.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/lib/librte_eal/common/malloc_elem.c b/lib/librte_eal/common/malloc_elem.c
index 1a74660de..9d3dcb6a9 100644
--- a/lib/librte_eal/common/malloc_elem.c
+++ b/lib/librte_eal/common/malloc_elem.c
@@ -316,13 +316,15 @@ remove_elem(struct malloc_elem *elem)
 static int
 next_elem_is_adjacent(struct malloc_elem *elem)
 {
-	return elem->next == RTE_PTR_ADD(elem, elem->size);
+	return elem->next == RTE_PTR_ADD(elem, elem->size) &&
+			elem->next->msl == elem->msl;
 }
 
 static int
 prev_elem_is_adjacent(struct malloc_elem *elem)
 {
-	return elem == RTE_PTR_ADD(elem->prev, elem->prev->size);
+	return elem == RTE_PTR_ADD(elem->prev, elem->prev->size) &&
+			elem->prev->msl == elem->msl;
 }
 
 /*
-- 
2.17.1

Re: [dpdk-dev] [PATCH v2] malloc: fix adjacency check to also include segment list

[Burakov, Anatoly]

On 14-Nov-18 3:00 PM, Anatoly Burakov wrote:
> It may so happen that two memory locations may be adjacent in
> virtual memory, but belong to different segment lists. With
> current code, such segments will be concatenated. Fix the
> adjacency checking code to also check if the adjacent malloc
> elements belong to the same memseg list.
> 
> Fixes: 66cc45e293ed ("mem: replace memseg with memseg lists")
> 
> Cc: stable@dpdk.org
> 
> Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com>
> ---

Oops, reply tag was set wrong.

-- 
Thanks,
Anatoly

Re: [dpdk-dev] [PATCH v2] malloc: fix adjacency check to also include segment list

[Thomas Monjalon]

14/11/2018 16:00, Anatoly Burakov:
> It may so happen that two memory locations may be adjacent in
> virtual memory, but belong to different segment lists. With
> current code, such segments will be concatenated. Fix the
> adjacency checking code to also check if the adjacent malloc
> elements belong to the same memseg list.
> 
> Fixes: 66cc45e293ed ("mem: replace memseg with memseg lists")
> 
> Cc: stable@dpdk.org
> 
> Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com>

Applied, thanks

Re: [dpdk-dev] [PATCH] memalloc: check for contiguousness in external segments

[Thomas Monjalon]

14/11/2018 15:51, Anatoly Burakov:
> For IOVA as VA mode, we assume that memory is contiguous. However,
> for external segments that assumption may not necessarily hold.
> Fix the code to not assume that external memory segments are
> contiguous even in IOVA as VA mode.
> 
> Fixes: 5282bb1c3695 ("mem: allow memseg lists to be marked as external")
> 
> Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com>

Applied, thanks

Re: [dpdk-dev] [PATCH] memalloc: check for contiguousness in external segments

[Thomas Monjalon]

14/11/2018 15:51, Anatoly Burakov:
> For IOVA as VA mode, we assume that memory is contiguous. However,
> for external segments that assumption may not necessarily hold.
> Fix the code to not assume that external memory segments are
> contiguous even in IOVA as VA mode.
> 
> Fixes: 5282bb1c3695 ("mem: allow memseg lists to be marked as external")
> 
> Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com>

Applied, thanks