* [PATCH 1/1] RDMA/cma: Fix rdma_resolve_route memory leak
@ 2021-06-24 18:55 Gerd Rausch
2021-06-25 5:49 ` Mark Zhang
2021-06-25 15:03 ` Jason Gunthorpe
0 siblings, 2 replies; 6+ messages in thread
From: Gerd Rausch @ 2021-06-24 18:55 UTC (permalink / raw)
To: linux-rdma, linux-kernel, Doug Ledford, Jason Gunthorpe
Fix a memory leak when "rmda_resolve_route" is called
more than once on the same "rdma_cm_id".
Signed-off-by: Gerd Rausch <gerd.rausch@oracle.com>
---
drivers/infiniband/core/cma.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c
index ab148a696c0c..4a76d5b4163e 100644
--- a/drivers/infiniband/core/cma.c
+++ b/drivers/infiniband/core/cma.c
@@ -2819,7 +2819,8 @@ static int cma_resolve_ib_route(struct rdma_id_private *id_priv,
cma_init_resolve_route_work(work, id_priv);
- route->path_rec = kmalloc(sizeof *route->path_rec, GFP_KERNEL);
+ if (!route->path_rec)
+ route->path_rec = kmalloc(sizeof *route->path_rec, GFP_KERNEL);
if (!route->path_rec) {
ret = -ENOMEM;
goto err1;
--
2.24.1
^ permalink raw reply related [flat|nested] 6+ messages in thread
* Re: [PATCH 1/1] RDMA/cma: Fix rdma_resolve_route memory leak
2021-06-24 18:55 [PATCH 1/1] RDMA/cma: Fix rdma_resolve_route memory leak Gerd Rausch
@ 2021-06-25 5:49 ` Mark Zhang
2021-06-25 8:15 ` Haakon Bugge
2021-06-25 16:27 ` Gerd Rausch
2021-06-25 15:03 ` Jason Gunthorpe
1 sibling, 2 replies; 6+ messages in thread
From: Mark Zhang @ 2021-06-25 5:49 UTC (permalink / raw)
To: Gerd Rausch, linux-rdma, linux-kernel, Doug Ledford, Jason Gunthorpe
On 6/25/2021 2:55 AM, Gerd Rausch wrote:
> Fix a memory leak when "rmda_resolve_route" is called
> more than once on the same "rdma_cm_id".
>
> Signed-off-by: Gerd Rausch <gerd.rausch@oracle.com>
> ---
> drivers/infiniband/core/cma.c | 3 ++-
> 1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c
> index ab148a696c0c..4a76d5b4163e 100644
> --- a/drivers/infiniband/core/cma.c
> +++ b/drivers/infiniband/core/cma.c
> @@ -2819,7 +2819,8 @@ static int cma_resolve_ib_route(struct rdma_id_private *id_priv,
>
> cma_init_resolve_route_work(work, id_priv);
>
> - route->path_rec = kmalloc(sizeof *route->path_rec, GFP_KERNEL);
> + if (!route->path_rec)
> + route->path_rec = kmalloc(sizeof *route->path_rec, GFP_KERNEL);
> if (!route->path_rec) {
> ret = -ENOMEM;
> goto err1;
If route->path_rec does exist (meaning this is not the first time
called), then it would be freed if cma_query_ib_route() below is failed,
is it good?
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH 1/1] RDMA/cma: Fix rdma_resolve_route memory leak
2021-06-25 5:49 ` Mark Zhang
@ 2021-06-25 8:15 ` Haakon Bugge
2021-06-25 16:27 ` Gerd Rausch
1 sibling, 0 replies; 6+ messages in thread
From: Haakon Bugge @ 2021-06-25 8:15 UTC (permalink / raw)
To: Mark Zhang
Cc: Gerd Rausch, OFED mailing list, linux-kernel, Doug Ledford,
Jason Gunthorpe
> On 25 Jun 2021, at 07:49, Mark Zhang <markzhang@nvidia.com> wrote:
>
> On 6/25/2021 2:55 AM, Gerd Rausch wrote:
>> Fix a memory leak when "rmda_resolve_route" is called
>> more than once on the same "rdma_cm_id".
>> Signed-off-by: Gerd Rausch <gerd.rausch@oracle.com>
>> ---
>> drivers/infiniband/core/cma.c | 3 ++-
>> 1 file changed, 2 insertions(+), 1 deletion(-)
>> diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c
>> index ab148a696c0c..4a76d5b4163e 100644
>> --- a/drivers/infiniband/core/cma.c
>> +++ b/drivers/infiniband/core/cma.c
>> @@ -2819,7 +2819,8 @@ static int cma_resolve_ib_route(struct rdma_id_private *id_priv,
>> cma_init_resolve_route_work(work, id_priv);
>> - route->path_rec = kmalloc(sizeof *route->path_rec, GFP_KERNEL);
>> + if (!route->path_rec)
>> + route->path_rec = kmalloc(sizeof *route->path_rec, GFP_KERNEL);
>> if (!route->path_rec) {
>> ret = -ENOMEM;
>> goto err1;
>
> If route->path_rec does exist (meaning this is not the first time called), then it would be freed if cma_query_ib_route() below is failed, is it good?
This may happen if rdma_resolve_route() is called after a route error event has been received. In that case, the state is set back to RDMA_CM_ADDR_RESOLVED, so it seems the API allows rdma_resolve_route() to be called again without renewing the cm_id in this case.
Another thing, if this fix is eligible, it should probably also go into cma_resolve_iboe_route() as well.
Thxs, Håkon
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH 1/1] RDMA/cma: Fix rdma_resolve_route memory leak
2021-06-24 18:55 [PATCH 1/1] RDMA/cma: Fix rdma_resolve_route memory leak Gerd Rausch
2021-06-25 5:49 ` Mark Zhang
@ 2021-06-25 15:03 ` Jason Gunthorpe
2021-06-25 16:27 ` Gerd Rausch
1 sibling, 1 reply; 6+ messages in thread
From: Jason Gunthorpe @ 2021-06-25 15:03 UTC (permalink / raw)
To: Gerd Rausch; +Cc: linux-rdma, linux-kernel, Doug Ledford
On Thu, Jun 24, 2021 at 11:55:31AM -0700, Gerd Rausch wrote:
> Fix a memory leak when "rmda_resolve_route" is called
> more than once on the same "rdma_cm_id".
>
> Signed-off-by: Gerd Rausch <gerd.rausch@oracle.com>
> ---
> drivers/infiniband/core/cma.c | 3 ++-
> 1 file changed, 2 insertions(+), 1 deletion(-)
I wondered for a while if it would be better to clear this in
cma_query_handler(), but it seems this is OK
Applied to for-next
Thanks,
Jason
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH 1/1] RDMA/cma: Fix rdma_resolve_route memory leak
2021-06-25 5:49 ` Mark Zhang
2021-06-25 8:15 ` Haakon Bugge
@ 2021-06-25 16:27 ` Gerd Rausch
1 sibling, 0 replies; 6+ messages in thread
From: Gerd Rausch @ 2021-06-25 16:27 UTC (permalink / raw)
To: Mark Zhang, linux-rdma, linux-kernel, Doug Ledford, Jason Gunthorpe
Hi Mark,
On 24/06/2021 22.49, Mark Zhang wrote:
> On 6/25/2021 2:55 AM, Gerd Rausch wrote:
>> Fix a memory leak when "rmda_resolve_route" is called
>> more than once on the same "rdma_cm_id".
>>
>> Signed-off-by: Gerd Rausch <gerd.rausch@oracle.com>
>> ---
>> drivers/infiniband/core/cma.c | 3 ++-
>> 1 file changed, 2 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c
>> index ab148a696c0c..4a76d5b4163e 100644
>> --- a/drivers/infiniband/core/cma.c
>> +++ b/drivers/infiniband/core/cma.c
>> @@ -2819,7 +2819,8 @@ static int cma_resolve_ib_route(struct rdma_id_private *id_priv,
>> cma_init_resolve_route_work(work, id_priv);
>> - route->path_rec = kmalloc(sizeof *route->path_rec, GFP_KERNEL);
>> + if (!route->path_rec)
>> + route->path_rec = kmalloc(sizeof *route->path_rec, GFP_KERNEL);
>> if (!route->path_rec) {
>> ret = -ENOMEM;
>> goto err1;
>
> If route->path_rec does exist (meaning this is not the first time called), then it would be freed if cma_query_ib_route() below is failed, is it good?
So the caller performs "rdma_resolve_route" which returns an immediate error,
but the expectation would be that the cm_id still points to
a valid (!= NULL) path record
(even though this error indicateed route lookup failed).
Which code-part and call-sequence would have such expectation?
I can't say I'm in love with this approach either, since the code
makes the assumption that "path_rec" (if != NULL) points to a space
that was allocated by "kmalloc" before and can be overwritten.
It's not permitted to point to something else.
But "__rdma_free" makes that assumption anyway, since it unconditionally
calls "kfree" on this pointer.
So there is precedence.
Just my 2ç,
Gerd
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH 1/1] RDMA/cma: Fix rdma_resolve_route memory leak
2021-06-25 15:03 ` Jason Gunthorpe
@ 2021-06-25 16:27 ` Gerd Rausch
0 siblings, 0 replies; 6+ messages in thread
From: Gerd Rausch @ 2021-06-25 16:27 UTC (permalink / raw)
To: Jason Gunthorpe; +Cc: linux-rdma, linux-kernel, Doug Ledford
Hi Jason,
On 25/06/2021 08.03, Jason Gunthorpe wrote:
> On Thu, Jun 24, 2021 at 11:55:31AM -0700, Gerd Rausch wrote:
>> Fix a memory leak when "rmda_resolve_route" is called
>> more than once on the same "rdma_cm_id".
>>
>> Signed-off-by: Gerd Rausch <gerd.rausch@oracle.com>
>> ---
>> drivers/infiniband/core/cma.c | 3 ++-
>> 1 file changed, 2 insertions(+), 1 deletion(-)
>
> I wondered for a while if it would be better to clear this in
> cma_query_handler(), but it seems this is OK
>
> Applied to for-next
>
Thank you, I appreciate it.
Gerd
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2021-06-25 16:28 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-06-24 18:55 [PATCH 1/1] RDMA/cma: Fix rdma_resolve_route memory leak Gerd Rausch
2021-06-25 5:49 ` Mark Zhang
2021-06-25 8:15 ` Haakon Bugge
2021-06-25 16:27 ` Gerd Rausch
2021-06-25 15:03 ` Jason Gunthorpe
2021-06-25 16:27 ` Gerd Rausch
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).