| Message ID | 20230712134853.1044134-1-haowenchao2@huawei.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | NFS: Fix memleak in nfs_sysfs_move_sb_to_server | expand |
On Wed, Jul 12, 2023 at 9:48 PM Wenchao Hao <haowenchao22@gmail.com> wrote: > > call kfree() to free memory allocated in kasprintf() to fix > memleak. It's easy to recurrent this memleak, just mount and > remount nfs at NFS client. > I did not check the mail list before send this patch, Benjamin Coddingto has fixed it, so ignore this please. > The bug is detected by kmemleak with following stack: > unreferenced object 0xffffa31a91fdf2c0 (size 16): > comm "mount.nfs", pid 960, jiffies 4294724478 (age 111892.067s) > hex dump (first 16 bytes): > 73 65 72 76 65 72 2d 30 00 f2 fd 91 1a a3 ff ff server-0........ > backtrace: > [<00000000de9b21a9>] create_object+0x11/0x17 > [<00000000ed17aa96>] kmemleak_alloc+0x6c/0x7a > [<000000009f4acc69>] slab_post_alloc_hook.constprop.0+0x81/0x99 > [<00000000d2446455>] __kmem_cache_alloc_node+0x10c/0x13a > [<00000000eca4a704>] __kmalloc_node_track_caller+0x83/0xb7 > [<0000000076cdb22f>] kvasprintf+0x65/0xc3 > [<00000000f6f2e879>] kasprintf+0x4e/0x69 > [<00000000e49d03be>] nfs_sysfs_move_sb_to_server+0x24/0x5f > [<0000000039127597>] nfs_kill_super+0x25/0x49 > [<00000000e6ecf809>] deactivate_locked_super+0x35/0x74 > [<000000003b9c3350>] deactivate_super+0x3c/0x44 > [<00000000777c86be>] cleanup_mnt+0x6c/0xe7 > [<000000009f181749>] __cleanup_mnt+0x12/0x18 > [<00000000a709602e>] task_work_run+0x75/0x8c > [<0000000075da06ba>] exit_to_user_mode_prepare+0x79/0x102 > [<0000000093a53d48>] syscall_exit_to_user_mode+0x1c/0x30 > > Signed-off-by: Wenchao Hao <haowenchao2@huawei.com> > --- > fs/nfs/sysfs.c | 4 +++- > 1 file changed, 3 insertions(+), 1 deletion(-) > > diff --git a/fs/nfs/sysfs.c b/fs/nfs/sysfs.c > index acda8f033d30..bf378ecd5d9f 100644 > --- a/fs/nfs/sysfs.c > +++ b/fs/nfs/sysfs.c > @@ -345,8 +345,10 @@ void nfs_sysfs_move_sb_to_server(struct nfs_server *server) > int ret = -ENOMEM; > > s = kasprintf(GFP_KERNEL, "server-%d", server->s_sysfs_id); > - if (s) > + if (s) { > ret = kobject_rename(&server->kobj, s); > + kfree(s); > + } > if (ret < 0) > pr_warn("NFS: rename sysfs %s failed (%d)\n", > server->kobj.name, ret); > -- > 2.35.3 >
On Wed, Jul 12, 2023 at 09:48:53PM +0800, Wenchao Hao wrote: > call kfree() to free memory allocated in kasprintf() to fix > memleak. It's easy to recurrent this memleak, just mount and > remount nfs at NFS client. > > The bug is detected by kmemleak with following stack: > unreferenced object 0xffffa31a91fdf2c0 (size 16): > comm "mount.nfs", pid 960, jiffies 4294724478 (age 111892.067s) > hex dump (first 16 bytes): > 73 65 72 76 65 72 2d 30 00 f2 fd 91 1a a3 ff ff server-0........ > backtrace: > [<00000000de9b21a9>] create_object+0x11/0x17 > [<00000000ed17aa96>] kmemleak_alloc+0x6c/0x7a > [<000000009f4acc69>] slab_post_alloc_hook.constprop.0+0x81/0x99 > [<00000000d2446455>] __kmem_cache_alloc_node+0x10c/0x13a > [<00000000eca4a704>] __kmalloc_node_track_caller+0x83/0xb7 > [<0000000076cdb22f>] kvasprintf+0x65/0xc3 > [<00000000f6f2e879>] kasprintf+0x4e/0x69 > [<00000000e49d03be>] nfs_sysfs_move_sb_to_server+0x24/0x5f > [<0000000039127597>] nfs_kill_super+0x25/0x49 > [<00000000e6ecf809>] deactivate_locked_super+0x35/0x74 > [<000000003b9c3350>] deactivate_super+0x3c/0x44 > [<00000000777c86be>] cleanup_mnt+0x6c/0xe7 > [<000000009f181749>] __cleanup_mnt+0x12/0x18 > [<00000000a709602e>] task_work_run+0x75/0x8c > [<0000000075da06ba>] exit_to_user_mode_prepare+0x79/0x102 > [<0000000093a53d48>] syscall_exit_to_user_mode+0x1c/0x30 > > Signed-off-by: Wenchao Hao <haowenchao2@huawei.com> > --- > fs/nfs/sysfs.c | 4 +++- > 1 file changed, 3 insertions(+), 1 deletion(-) > > diff --git a/fs/nfs/sysfs.c b/fs/nfs/sysfs.c > index acda8f033d30..bf378ecd5d9f 100644 > --- a/fs/nfs/sysfs.c > +++ b/fs/nfs/sysfs.c > @@ -345,8 +345,10 @@ void nfs_sysfs_move_sb_to_server(struct nfs_server *server) > int ret = -ENOMEM; > > s = kasprintf(GFP_KERNEL, "server-%d", server->s_sysfs_id); > - if (s) > + if (s) { > ret = kobject_rename(&server->kobj, s); > + kfree(s); > + } > if (ret < 0) > pr_warn("NFS: rename sysfs %s failed (%d)\n", > server->kobj.name, ret); A fix was already posted: https://lore.kernel.org/linux-nfs/6702796fee0365bf399800326bbe6c88e5f73f68.1689014440.git.bcodding@redhat.com/
diff --git a/fs/nfs/sysfs.c b/fs/nfs/sysfs.c index acda8f033d30..bf378ecd5d9f 100644 --- a/fs/nfs/sysfs.c +++ b/fs/nfs/sysfs.c @@ -345,8 +345,10 @@ void nfs_sysfs_move_sb_to_server(struct nfs_server *server) int ret = -ENOMEM; s = kasprintf(GFP_KERNEL, "server-%d", server->s_sysfs_id); - if (s) + if (s) { ret = kobject_rename(&server->kobj, s); + kfree(s); + } if (ret < 0) pr_warn("NFS: rename sysfs %s failed (%d)\n", server->kobj.name, ret);
call kfree() to free memory allocated in kasprintf() to fix memleak. It's easy to recurrent this memleak, just mount and remount nfs at NFS client. The bug is detected by kmemleak with following stack: unreferenced object 0xffffa31a91fdf2c0 (size 16): comm "mount.nfs", pid 960, jiffies 4294724478 (age 111892.067s) hex dump (first 16 bytes): 73 65 72 76 65 72 2d 30 00 f2 fd 91 1a a3 ff ff server-0........ backtrace: [<00000000de9b21a9>] create_object+0x11/0x17 [<00000000ed17aa96>] kmemleak_alloc+0x6c/0x7a [<000000009f4acc69>] slab_post_alloc_hook.constprop.0+0x81/0x99 [<00000000d2446455>] __kmem_cache_alloc_node+0x10c/0x13a [<00000000eca4a704>] __kmalloc_node_track_caller+0x83/0xb7 [<0000000076cdb22f>] kvasprintf+0x65/0xc3 [<00000000f6f2e879>] kasprintf+0x4e/0x69 [<00000000e49d03be>] nfs_sysfs_move_sb_to_server+0x24/0x5f [<0000000039127597>] nfs_kill_super+0x25/0x49 [<00000000e6ecf809>] deactivate_locked_super+0x35/0x74 [<000000003b9c3350>] deactivate_super+0x3c/0x44 [<00000000777c86be>] cleanup_mnt+0x6c/0xe7 [<000000009f181749>] __cleanup_mnt+0x12/0x18 [<00000000a709602e>] task_work_run+0x75/0x8c [<0000000075da06ba>] exit_to_user_mode_prepare+0x79/0x102 [<0000000093a53d48>] syscall_exit_to_user_mode+0x1c/0x30 Signed-off-by: Wenchao Hao <haowenchao2@huawei.com> --- fs/nfs/sysfs.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)