2022-08-01 07:58:31

by William Kucharski

[permalink] [raw]
Subject: [PATCH] RDMA/rxe: Correct error handling in routines allocating xarray entries

The current code will report an error if xa_alloc_cyclic() returns
non-zero, but it will return 1 if it wrapped indices before successfully
allocating an entry.

An error should only be reported if the call actually failed (denoted by
a return value < 0.)

Fixes: 3225717f6dfa2 ("RDMA/rxe: Replace red-black trees by xarrays")
Signed-off-by: William Kucharski <[email protected]>
---
drivers/infiniband/sw/rxe/rxe_pool.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/infiniband/sw/rxe/rxe_pool.c b/drivers/infiniband/sw/rxe/rxe_pool.c
index 19b14826385b..e9f3bbd8d605 100644
--- a/drivers/infiniband/sw/rxe/rxe_pool.c
+++ b/drivers/infiniband/sw/rxe/rxe_pool.c
@@ -139,7 +139,7 @@ void *rxe_alloc(struct rxe_pool *pool)

err = xa_alloc_cyclic(&pool->xa, &elem->index, elem, pool->limit,
&pool->next, GFP_KERNEL);
- if (err)
+ if (err < 0)
goto err_free;

return obj;
@@ -167,7 +167,7 @@ int __rxe_add_to_pool(struct rxe_pool *pool, struct rxe_pool_elem *elem)

err = xa_alloc_cyclic(&pool->xa, &elem->index, elem, pool->limit,
&pool->next, GFP_KERNEL);
- if (err)
+ if (err < 0)
goto err_cnt;

return 0;
--
2.37.1



2022-08-01 08:28:00

by Zhu Yanjun

[permalink] [raw]
Subject: Re: [PATCH] RDMA/rxe: Correct error handling in routines allocating xarray entries

On Mon, Aug 1, 2022 at 3:39 PM William Kucharski
<[email protected]> wrote:
>
> The current code will report an error if xa_alloc_cyclic() returns
> non-zero, but it will return 1 if it wrapped indices before successfully
> allocating an entry.
>
> An error should only be reported if the call actually failed (denoted by
> a return value < 0.)
>
> Fixes: 3225717f6dfa2 ("RDMA/rxe: Replace red-black trees by xarrays")
> Signed-off-by: William Kucharski <[email protected]>

Please check commit in
https://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma.git

commit 1a685940e6200e9def6e34bbaa19dd31dc5aeaf8
Author: Dongliang Mu <[email protected]>
Date: Thu Jun 9 15:06:56 2022 +0800

RDMA/rxe: fix xa_alloc_cycle() error return value check again

Currently rxe_alloc checks ret to indicate error, but 1 is also a valid
return and just indicates that the allocation succeeded with a wrap.

Fix this by modifying the check to be < 0.

Link: https://lore.kernel.org/r/[email protected]
Fixes: 3225717f6dfa ("RDMA/rxe: Replace red-black trees by xarrays")
Signed-off-by: Dongliang Mu <[email protected]>
Reviewed-by: Bob Pearson <[email protected]>
Signed-off-by: Leon Romanovsky <[email protected]>

Thanks and Regards,
Zhu Yanjun

> ---
> drivers/infiniband/sw/rxe/rxe_pool.c | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/infiniband/sw/rxe/rxe_pool.c b/drivers/infiniband/sw/rxe/rxe_pool.c
> index 19b14826385b..e9f3bbd8d605 100644
> --- a/drivers/infiniband/sw/rxe/rxe_pool.c
> +++ b/drivers/infiniband/sw/rxe/rxe_pool.c
> @@ -139,7 +139,7 @@ void *rxe_alloc(struct rxe_pool *pool)
>
> err = xa_alloc_cyclic(&pool->xa, &elem->index, elem, pool->limit,
> &pool->next, GFP_KERNEL);
> - if (err)
> + if (err < 0)
> goto err_free;
>
> return obj;
> @@ -167,7 +167,7 @@ int __rxe_add_to_pool(struct rxe_pool *pool, struct rxe_pool_elem *elem)
>
> err = xa_alloc_cyclic(&pool->xa, &elem->index, elem, pool->limit,
> &pool->next, GFP_KERNEL);
> - if (err)
> + if (err < 0)
> goto err_cnt;
>
> return 0;
> --
> 2.37.1
>

2022-08-01 20:12:31

by Bob Pearson

[permalink] [raw]
Subject: Re: [PATCH] RDMA/rxe: Correct error handling in routines allocating xarray entries

On 8/1/22 02:38, William Kucharski wrote:
> The current code will report an error if xa_alloc_cyclic() returns
> non-zero, but it will return 1 if it wrapped indices before successfully
> allocating an entry.
>
> An error should only be reported if the call actually failed (denoted by
> a return value < 0.)
>
> Fixes: 3225717f6dfa2 ("RDMA/rxe: Replace red-black trees by xarrays")
> Signed-off-by: William Kucharski <[email protected]>
> ---
> drivers/infiniband/sw/rxe/rxe_pool.c | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/infiniband/sw/rxe/rxe_pool.c b/drivers/infiniband/sw/rxe/rxe_pool.c
> index 19b14826385b..e9f3bbd8d605 100644
> --- a/drivers/infiniband/sw/rxe/rxe_pool.c
> +++ b/drivers/infiniband/sw/rxe/rxe_pool.c
> @@ -139,7 +139,7 @@ void *rxe_alloc(struct rxe_pool *pool)
>
> err = xa_alloc_cyclic(&pool->xa, &elem->index, elem, pool->limit,
> &pool->next, GFP_KERNEL);
> - if (err)
> + if (err < 0)
> goto err_free;
>
> return obj;
> @@ -167,7 +167,7 @@ int __rxe_add_to_pool(struct rxe_pool *pool, struct rxe_pool_elem *elem)
>
> err = xa_alloc_cyclic(&pool->xa, &elem->index, elem, pool->limit,
> &pool->next, GFP_KERNEL);
> - if (err)
> + if (err < 0)
> goto err_cnt;
>
> return 0;

We fixed this a while back but not sure what happened. In any case this is absolutely correct.
You can add

Reviewed-by: Bob Pearson <[email protected]>

Bob