On 4/5/2019 6:01 PM, Mukesh Ojha wrote:
On 4/5/2019 5:40 PM, Greg KH wrote:
On Fri, Apr 05, 2019 at 05:13:00PM +0530, Gaurav Kohli wrote:
On 4/5/2019 5:03 PM, Greg KH wrote:
On Fri, Apr 05, 2019 at 04:51:07PM +0530, Gaurav Kohli wrote:
While adding kernfs node for child to the parent kernfs
node and when child node founds that parent kn count is
zero, then below comes like:
WARNING: fs/kernfs/dir.c:494 kernfs_get+0x64/0x88
This indicates that parent is in kernfs_put path/ or already
freed, and if the child node keeps continue to
make new kernfs node, then there is chance of
below race for parent node:
CPU0 CPU1
//Parent node //child node
kernfs_put
atomic_dec_and_test(&kn->count)
//count is 0, so continue
kernfs_new_node(child)
kernfs_get(parent);
//increment parent count to 1
//warning come as parent count is 0
/* link in */
kernfs_add_one(kn);
// this should fail as parent is
//in free path.
kernfs_put(child)
kmem_cache_free(parent)
kmem_cache_free(child)
kn = parent
atomic_dec_and_test(&kn->count))
//this is 0 now, so release will
continue for parent.
kmem_cache_free(parent)
To prevent this race, child simply has to decrement count of parent
kernfs node and keep continue the free path for itself.
Signed-off-by: Gaurav Kohli <gkohli@xxxxxxxxxxxxxx>
Signed-off-by: Mukesh Ojha <mojha@xxxxxxxxxxxxxx>
diff --git a/fs/kernfs/dir.c b/fs/kernfs/dir.c
index b84d635..d5a36e8 100644
--- a/fs/kernfs/dir.c
+++ b/fs/kernfs/dir.c
@@ -515,7 +515,6 @@ void kernfs_put(struct kernfs_node *kn)
if (!kn || !atomic_dec_and_test(&kn->count))
return;
root = kernfs_root(kn);
- repeat:
/*
* Moving/renaming is always done while holding reference.
* kn->parent won't change beneath us.
@@ -545,8 +544,8 @@ void kernfs_put(struct kernfs_node *kn)
kn = parent;
if (kn) {
- if (atomic_dec_and_test(&kn->count))
- goto repeat;
+ /* Parent may be on free path, so simply decrement the count */
That's the wrong indentation :(
And how are you hitting this issue? What user of kernfs is causing
this?
Hi Greg,
Thanks, will fix comment indentation, seen during sys-executor
running:
We have only one instance , In logs below warning also came:
WARNING: CPU: 4 kernel/msm-4.14/fs/kernfs/dir.c:494
kernfs_get+0x64/0x88
which indicated parent is in put path.
[ 160.125151] Disabling lock debugging due to kernel taint
[ 160.130626] INFO: Allocated in __kernfs_new_node+0x8c/0x3c0
age=11 cpu=2
pid=7098
[ 160.138314] kmem_cache_alloc+0x358/0x388
[ 160.142445] __kernfs_new_node+0x8c/0x3c0
[ 160.146590] kernfs_new_node+0x80/0xc8
[ 160.150462] kernfs_create_dir_ns+0x44/0xfc
[ 160.154777] sysfs_create_dir_ns+0xa8/0x130
[ 160.158416] CPU5: update max cpu_capacity 1024
[ 160.159085] kobject_add_internal+0x278/0x650
[ 160.163567] kobject_add_varg+0xe0/0x130
[ 160.167606] kobject_add+0x15c/0x1d0
[ 160.168452] CPU5: update max cpu_capacity 780
[ 160.171287] get_device_parent+0x2d0/0x34c
[ 160.175510] device_add+0x240/0xde0
[ 160.178371] CPU6: update max cpu_capacity 916
[ 160.179108] input_register_device+0x5f4/0xa0c
[ 160.183686] uinput_ioctl_handler+0x1184/0x2198
[ 160.202436] INFO: Freed in kernfs_put+0x2c8/0x434 age=14 cpu=0
pid=7096
[ 160.209230] kernfs_put+0x2c8/0x434
[ 160.212825] kobject_del+0x50/0xcc
[ 160.216332] cleanup_glue_dir+0x124/0x16c
[ 160.220456] device_del+0x55c/0x5c8
[ 160.224047] __input_unregister_device+0x274/0x2a8
[ 160.228974] input_unregister_device+0x90/0xd0
[ 160.233553] uinput_destroy_device+0x15c/0x1dc
[ 160.238131] uinput_release+0x44/0x5c
[ 160.241898] __fput+0x1f4/0x4e4
[ 160.245127] ____fput+0x20/0x2c
during code review, I have found race between kernfs parent put call
and
child get call.
So this is a sysfs usage of this?
yes
Using input devices or cpu devices
for the stress test?
input devices ..
[ 1714.090310] input: syz1 as /devices/virtual/input/input191
[ 1714.223037] input: syz1 as /devices/virtual/input/input192
..
[ 1714.428228] input: syz1 as /devices/virtual/input/input193
..
[ 1714.528256] input: syz1 as /devices/virtual/input/input194
..
[ 1714.756481] input: syz1 as /devices/virtual/input/input195
..
[ 1714.831920] input: syz1 as /devices/virtual/input/input196
..
Cheers,
Mukesh
Hi Greg, Tejun,
With earlier patch that i have posted, there is chance of memory leak
for parent, if below case occurs:
Add parent
Add child
put parent
put child -> this will skip the freeing of parent node with above patch.
So to avoid race mentioned in earlier mail, creation of kernfs should
not proceed, if count of parent is zero, Instead
of warning, we should return from that place.
Can you please check below patch for the same, or please let us know
some other way to fix the race.
diff --git a/fs/kernfs/dir.c b/fs/kernfs/dir.c
index 89d1dc1..c41085a 100644
--- a/fs/kernfs/dir.c
+++ b/fs/kernfs/dir.c
@@ -676,6 +676,11 @@ struct kernfs_node *kernfs_new_node(struct
kernfs_node *parent,
{
struct kernfs_node *kn;
+ if (!atomic_read(&parent->count)) {
+ WARN_ON(1);
+ return NULL;
+ }
+
kn = __kernfs_new_node(kernfs_root(parent), name, mode, flags);
if (kn) {
kernfs_get(parent);
Regards
Gaurav
thanks,
greg k-h
--
Qualcomm India Private Limited, on behalf of Qualcomm Innovation Center,
Inc. is a member of the Code Aurora Forum,
a Linux Foundation Collaborative Project.
