* Re: [PATCH v2] RDMA/uverbs: Free all radix tree nodes in uverbs_destroy_api()
[not found] <20261115142356.21127-1-kartikey406@gmail.com>
@ 2026-11-15 14:24 ` syzbot
0 siblings, 0 replies; 4+ messages in thread
From: syzbot @ 2026-11-15 14:24 UTC (permalink / raw)
To: kartikey406; +Cc: kartikey406, linux-kernel, syzkaller-bugs
> #syz test: git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git master
Patch testing via #syz test is temporarily disabled.
See https://github.com/google/syzkaller/issues/8122 for details.
>
>
> syzbot reported a memory leak of radix_tree_node objects allocated
> while registering a siw device:
>
> radix_tree_node_alloc+0x81/0x1a0 lib/radix-tree.c:276
> radix_tree_insert+0x140/0x390 lib/radix-tree.c:712
> uapi_add_elm+0x61/0x140 uverbs_uapi.c:41
> uverbs_alloc_api+0x85/0xc50 uverbs_uapi.c:659
> siw_newlink+0x36a/0x3f0 siw_main.c:433
>
> radix_tree_insert() can fail with -ENOMEM after it has already linked
> intermediate nodes into the tree, leaving them with no entries.
> uverbs_destroy_api() only removes populated slots through
> uapi_remove_range(), and radix tree nodes are freed only as a side
> effect of deleting an entry, so these empty nodes are leaked when the
> uapi is destroyed.
>
> Call xa_destroy() after removing the entries so that all remaining
> nodes are freed.
>
> Reported-by: syzbot+d60be66dd00b4490fa96@syzkaller.appspotmail.com
> Closes: https://syzkaller.appspot.com/bug?extid=d60be66dd00b4490fa96
> Fixes: 9ed3e5f44772 ("IB/uverbs: Build the specs into a radix tree at runtime")
> Signed-off-by: Deepanshu Kartikey <kartikey406@gmail.com>
> ---
> Changes in v2:
> - Drop the dummy insert/delete workaround in uapi_add_elm()
> - Call xa_destroy() in uverbs_destroy_api() to free all remaining
> nodes, as suggested by Jason
> ---
> drivers/infiniband/core/uverbs_uapi.c | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/drivers/infiniband/core/uverbs_uapi.c b/drivers/infiniband/core/uverbs_uapi.c
> index b67db3790f03..77a4eb494936 100644
> --- a/drivers/infiniband/core/uverbs_uapi.c
> +++ b/drivers/infiniband/core/uverbs_uapi.c
> @@ -621,6 +621,7 @@ void uverbs_destroy_api(struct uverbs_api *uapi)
> return;
>
> uapi_remove_range(uapi, 0, U32_MAX);
> + xa_destroy(&uapi->radix);
> kfree(uapi->write_methods);
> kfree(uapi);
> }
> --
> 2.43.0
>
^ permalink raw reply [flat|nested] 4+ messages in thread
* [PATCH v2] RDMA/uverbs: Free all radix tree nodes in uverbs_destroy_api()
@ 2026-11-15 14:29 Deepanshu Kartikey
0 siblings, 0 replies; 4+ messages in thread
From: Deepanshu Kartikey @ 2026-11-15 14:29 UTC (permalink / raw)
To: jgg, leon
Cc: ynachum, yishaih, edwards, mrgolin, hemanth.selam,
michael.j.ruhl, linux-rdma, linux-kernel, Deepanshu Kartikey,
syzbot+d60be66dd00b4490fa96
syzbot reported a memory leak of radix_tree_node objects allocated
while registering a siw device:
radix_tree_node_alloc+0x81/0x1a0 lib/radix-tree.c:276
radix_tree_insert+0x140/0x390 lib/radix-tree.c:712
uapi_add_elm+0x61/0x140 uverbs_uapi.c:41
uverbs_alloc_api+0x85/0xc50 uverbs_uapi.c:659
siw_newlink+0x36a/0x3f0 siw_main.c:433
radix_tree_insert() can fail with -ENOMEM after it has already linked
intermediate nodes into the tree, leaving them with no entries.
uverbs_destroy_api() only removes populated slots through
uapi_remove_range(), and radix tree nodes are freed only as a side
effect of deleting an entry, so these empty nodes are leaked when the
uapi is destroyed.
Call xa_destroy() after removing the entries so that all remaining
nodes are freed.
Reported-by: syzbot+d60be66dd00b4490fa96@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=d60be66dd00b4490fa96
Fixes: 9ed3e5f44772 ("IB/uverbs: Build the specs into a radix tree at runtime")
Signed-off-by: Deepanshu Kartikey <kartikey406@gmail.com>
---
Changes in v2:
- Drop the dummy insert/delete workaround in uapi_add_elm()
- Call xa_destroy() in uverbs_destroy_api() to free all remaining
nodes, as suggested by Jason
Compile tested only.
---
drivers/infiniband/core/uverbs_uapi.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/drivers/infiniband/core/uverbs_uapi.c b/drivers/infiniband/core/uverbs_uapi.c
index b67db3790f03..77a4eb494936 100644
--- a/drivers/infiniband/core/uverbs_uapi.c
+++ b/drivers/infiniband/core/uverbs_uapi.c
@@ -621,6 +621,7 @@ void uverbs_destroy_api(struct uverbs_api *uapi)
return;
uapi_remove_range(uapi, 0, U32_MAX);
+ xa_destroy(&uapi->radix);
kfree(uapi->write_methods);
kfree(uapi);
}
--
2.43.0
^ permalink raw reply [flat|nested] 4+ messages in thread[parent not found: <20261115142448.21160-1-kartikey406@gmail.com>]
* Re: [PATCH v2] RDMA/uverbs: Free all radix tree nodes in uverbs_destroy_api()
[not found] <20261115142448.21160-1-kartikey406@gmail.com>
@ 2026-11-15 14:24 ` syzbot
0 siblings, 0 replies; 4+ messages in thread
From: syzbot @ 2026-11-15 14:24 UTC (permalink / raw)
To: kartikey406; +Cc: kartikey406, linux-kernel, syzkaller-bugs
> #syz test: git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git master
Patch testing via #syz test is temporarily disabled.
See https://github.com/google/syzkaller/issues/8122 for details.
>
>
> syzbot reported a memory leak of radix_tree_node objects allocated
> while registering a siw device:
>
> radix_tree_node_alloc+0x81/0x1a0 lib/radix-tree.c:276
> radix_tree_insert+0x140/0x390 lib/radix-tree.c:712
> uapi_add_elm+0x61/0x140 uverbs_uapi.c:41
> uverbs_alloc_api+0x85/0xc50 uverbs_uapi.c:659
> siw_newlink+0x36a/0x3f0 siw_main.c:433
>
> radix_tree_insert() can fail with -ENOMEM after it has already linked
> intermediate nodes into the tree, leaving them with no entries.
> uverbs_destroy_api() only removes populated slots through
> uapi_remove_range(), and radix tree nodes are freed only as a side
> effect of deleting an entry, so these empty nodes are leaked when the
> uapi is destroyed.
>
> Call xa_destroy() after removing the entries so that all remaining
> nodes are freed.
>
> Reported-by: syzbot+d60be66dd00b4490fa96@syzkaller.appspotmail.com
> Closes: https://syzkaller.appspot.com/bug?extid=d60be66dd00b4490fa96
> Fixes: 9ed3e5f44772 ("IB/uverbs: Build the specs into a radix tree at runtime")
> Signed-off-by: Deepanshu Kartikey <kartikey406@gmail.com>
> ---
> Changes in v2:
> - Drop the dummy insert/delete workaround in uapi_add_elm()
> - Call xa_destroy() in uverbs_destroy_api() to free all remaining
> nodes, as suggested by Jason
> ---
> drivers/infiniband/core/uverbs_uapi.c | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/drivers/infiniband/core/uverbs_uapi.c b/drivers/infiniband/core/uverbs_uapi.c
> index b67db3790f03..77a4eb494936 100644
> --- a/drivers/infiniband/core/uverbs_uapi.c
> +++ b/drivers/infiniband/core/uverbs_uapi.c
> @@ -621,6 +621,7 @@ void uverbs_destroy_api(struct uverbs_api *uapi)
> return;
>
> uapi_remove_range(uapi, 0, U32_MAX);
> + xa_destroy(&uapi->radix);
> kfree(uapi->write_methods);
> kfree(uapi);
> }
> --
> 2.43.0
>
^ permalink raw reply [flat|nested] 4+ messages in thread
[parent not found: <20261115141930.20935-1-kartikey406@gmail.com>]
* Re: [PATCH v2] RDMA/uverbs: Free all radix tree nodes in uverbs_destroy_api()
[not found] <20261115141930.20935-1-kartikey406@gmail.com>
@ 2026-11-15 14:19 ` syzbot
0 siblings, 0 replies; 4+ messages in thread
From: syzbot @ 2026-11-15 14:19 UTC (permalink / raw)
To: kartikey406; +Cc: kartikey406, linux-kernel, syzkaller-bugs
> #syz test: git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git master
Patch testing via #syz test is temporarily disabled.
See https://github.com/google/syzkaller/issues/8122 for details.
>
>
> syzbot reported a memory leak of radix_tree_node objects allocated
> while registering a siw device:
>
> radix_tree_node_alloc+0x81/0x1a0 lib/radix-tree.c:276
> radix_tree_insert+0x140/0x390 lib/radix-tree.c:712
> uapi_add_elm+0x61/0x140 uverbs_uapi.c:41
> uverbs_alloc_api+0x85/0xc50 uverbs_uapi.c:659
> siw_newlink+0x36a/0x3f0 siw_main.c:433
>
> radix_tree_insert() can fail with -ENOMEM after it has already linked
> intermediate nodes into the tree, leaving them with no entries.
> uverbs_destroy_api() only removes populated slots through
> uapi_remove_range(), and radix tree nodes are freed only as a side
> effect of deleting an entry, so these empty nodes are leaked when the
> uapi is destroyed.
>
> Call xa_destroy() after removing the entries so that all remaining
> nodes are freed.
>
> Reported-by: syzbot+d60be66dd00b4490fa96@syzkaller.appspotmail.com
> Closes: https://syzkaller.appspot.com/bug?extid=d60be66dd00b4490fa96
> Fixes: 9ed3e5f44772 ("IB/uverbs: Build the specs into a radix tree at runtime")
> Signed-off-by: Deepanshu Kartikey <kartikey406@gmail.com>
> ---
> Changes in v2:
> - Drop the dummy insert/delete workaround in uapi_add_elm()
> - Call xa_destroy() in uverbs_destroy_api() to free all remaining
> nodes, as suggested by Jason
> ---
> drivers/infiniband/core/uverbs_uapi.c | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/drivers/infiniband/core/uverbs_uapi.c b/drivers/infiniband/core/uverbs_uapi.c
> index b67db3790f03..77a4eb494936 100644
> --- a/drivers/infiniband/core/uverbs_uapi.c
> +++ b/drivers/infiniband/core/uverbs_uapi.c
> @@ -621,6 +621,7 @@ void uverbs_destroy_api(struct uverbs_api *uapi)
> return;
>
> uapi_remove_range(uapi, 0, U32_MAX);
> + xa_destroy(&uapi->radix);
> kfree(uapi->write_methods);
> kfree(uapi);
> }
> --
> 2.43.0
>
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2026-11-15 14:29 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
[not found] <20261115142356.21127-1-kartikey406@gmail.com>
2026-11-15 14:24 ` [PATCH v2] RDMA/uverbs: Free all radix tree nodes in uverbs_destroy_api() syzbot
2026-11-15 14:29 Deepanshu Kartikey
[not found] <20261115142448.21160-1-kartikey406@gmail.com>
2026-11-15 14:24 ` syzbot
[not found] <20261115141930.20935-1-kartikey406@gmail.com>
2026-11-15 14:19 ` syzbot
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®