possible deadlock in display_open

[syzbot]

Hello,

syzbot found the following crash on:

HEAD commit: e96407b4 usb-fuzzer: main usb gadget fuzzer driver
git tree: https://github.com/google/kasan.git usb-fuzzer
console output: https://syzkaller.appspot.com/x/log.txt?x=13b29b26600000
kernel config: https://syzkaller.appspot.com/x/.config?x=cfa2c18fb6a8068e
dashboard link: https://syzkaller.appspot.com/bug?extid=c558267ad910fc494497
compiler: gcc (GCC) 9.0.0 20181231 (experimental)
syz repro: https://syzkaller.appspot.com/x/repro.syz?x=15427002600000
C reproducer: https://syzkaller.appspot.com/x/repro.c?x=111cb61c600000

IMPORTANT: if you fix the bug, please add the following tag to the commit:
Reported-by: syzbot+c55826...@syzkaller.appspotmail.com

======================================================
WARNING: possible circular locking dependency detected
5.3.0-rc2+ #25 Not tainted
------------------------------------------------------
syz-executor626/1723 is trying to acquire lock:
000000001a0d74d7 (driver_lock#2){+.+.}, at: display_open+0x1f/0x1d0
drivers/media/rc/imon.c:475

but task is already holding lock:
00000000076a0058 (minor_rwsem){++++}, at: usb_open+0x23/0x270
drivers/usb/core/file.c:39

which lock already depends on the new lock.


the existing dependency chain (in reverse order) is:

-> #2 (minor_rwsem){++++}:
down_write+0x92/0x150 kernel/locking/rwsem.c:1500
usb_register_dev drivers/usb/core/file.c:187 [inline]
usb_register_dev+0x131/0x6a0 drivers/usb/core/file.c:156
imon_init_display drivers/media/rc/imon.c:2343 [inline]
imon_probe+0x244d/0x2af0 drivers/media/rc/imon.c:2426
usb_probe_interface+0x305/0x7a0 drivers/usb/core/driver.c:361
really_probe+0x281/0x650 drivers/base/dd.c:548
driver_probe_device+0x101/0x1b0 drivers/base/dd.c:709
__device_attach_driver+0x1c2/0x220 drivers/base/dd.c:816
bus_for_each_drv+0x15c/0x1e0 drivers/base/bus.c:454
__device_attach+0x217/0x360 drivers/base/dd.c:882
bus_probe_device+0x1e4/0x290 drivers/base/bus.c:514
device_add+0xae6/0x16f0 drivers/base/core.c:2114
usb_set_configuration+0xdf6/0x1670 drivers/usb/core/message.c:2023
generic_probe+0x9d/0xd5 drivers/usb/core/generic.c:210
usb_probe_device+0x99/0x100 drivers/usb/core/driver.c:266
really_probe+0x281/0x650 drivers/base/dd.c:548
driver_probe_device+0x101/0x1b0 drivers/base/dd.c:709
__device_attach_driver+0x1c2/0x220 drivers/base/dd.c:816
bus_for_each_drv+0x15c/0x1e0 drivers/base/bus.c:454
__device_attach+0x217/0x360 drivers/base/dd.c:882
bus_probe_device+0x1e4/0x290 drivers/base/bus.c:514
device_add+0xae6/0x16f0 drivers/base/core.c:2114
usb_new_device.cold+0x6a4/0xe79 drivers/usb/core/hub.c:2536
hub_port_connect drivers/usb/core/hub.c:5098 [inline]
hub_port_connect_change drivers/usb/core/hub.c:5213 [inline]
port_event drivers/usb/core/hub.c:5359 [inline]
hub_event+0x1b5c/0x3640 drivers/usb/core/hub.c:5441
process_one_work+0x92b/0x1530 kernel/workqueue.c:2269
worker_thread+0x96/0xe20 kernel/workqueue.c:2415
kthread+0x318/0x420 kernel/kthread.c:255
ret_from_fork+0x24/0x30 arch/x86/entry/entry_64.S:352

-> #1 (&ictx->lock){+.+.}:
__mutex_lock_common kernel/locking/mutex.c:930 [inline]
__mutex_lock+0x158/0x1360 kernel/locking/mutex.c:1077
imon_init_intf0 drivers/media/rc/imon.c:2188 [inline]
imon_probe+0xf0c/0x2af0 drivers/media/rc/imon.c:2387
usb_probe_interface+0x305/0x7a0 drivers/usb/core/driver.c:361
really_probe+0x281/0x650 drivers/base/dd.c:548
driver_probe_device+0x101/0x1b0 drivers/base/dd.c:709
__device_attach_driver+0x1c2/0x220 drivers/base/dd.c:816
bus_for_each_drv+0x15c/0x1e0 drivers/base/bus.c:454
__device_attach+0x217/0x360 drivers/base/dd.c:882
bus_probe_device+0x1e4/0x290 drivers/base/bus.c:514
device_add+0xae6/0x16f0 drivers/base/core.c:2114
usb_set_configuration+0xdf6/0x1670 drivers/usb/core/message.c:2023
generic_probe+0x9d/0xd5 drivers/usb/core/generic.c:210
usb_probe_device+0x99/0x100 drivers/usb/core/driver.c:266
really_probe+0x281/0x650 drivers/base/dd.c:548
driver_probe_device+0x101/0x1b0 drivers/base/dd.c:709
__device_attach_driver+0x1c2/0x220 drivers/base/dd.c:816
bus_for_each_drv+0x15c/0x1e0 drivers/base/bus.c:454
__device_attach+0x217/0x360 drivers/base/dd.c:882
bus_probe_device+0x1e4/0x290 drivers/base/bus.c:514
device_add+0xae6/0x16f0 drivers/base/core.c:2114
usb_new_device.cold+0x6a4/0xe79 drivers/usb/core/hub.c:2536
hub_port_connect drivers/usb/core/hub.c:5098 [inline]
hub_port_connect_change drivers/usb/core/hub.c:5213 [inline]
port_event drivers/usb/core/hub.c:5359 [inline]
hub_event+0x1b5c/0x3640 drivers/usb/core/hub.c:5441
process_one_work+0x92b/0x1530 kernel/workqueue.c:2269
worker_thread+0x96/0xe20 kernel/workqueue.c:2415
kthread+0x318/0x420 kernel/kthread.c:255
ret_from_fork+0x24/0x30 arch/x86/entry/entry_64.S:352

-> #0 (driver_lock#2){+.+.}:
check_prev_add kernel/locking/lockdep.c:2405 [inline]
check_prevs_add kernel/locking/lockdep.c:2507 [inline]
validate_chain kernel/locking/lockdep.c:2897 [inline]
__lock_acquire+0x1f7c/0x3b50 kernel/locking/lockdep.c:3880
lock_acquire+0x127/0x320 kernel/locking/lockdep.c:4412
__mutex_lock_common kernel/locking/mutex.c:930 [inline]
__mutex_lock+0x158/0x1360 kernel/locking/mutex.c:1077
display_open+0x1f/0x1d0 drivers/media/rc/imon.c:475
usb_open+0x1df/0x270 drivers/usb/core/file.c:48
chrdev_open+0x219/0x5c0 fs/char_dev.c:414
do_dentry_open+0x494/0x1120 fs/open.c:797
do_last fs/namei.c:3416 [inline]
path_openat+0x1430/0x3f50 fs/namei.c:3533
do_filp_open+0x1a1/0x280 fs/namei.c:3563
do_sys_open+0x3c0/0x580 fs/open.c:1089
do_syscall_64+0xb7/0x580 arch/x86/entry/common.c:296
entry_SYSCALL_64_after_hwframe+0x49/0xbe

other info that might help us debug this:

Chain exists of:
driver_lock#2 --> &ictx->lock --> minor_rwsem

Possible unsafe locking scenario:

CPU0 CPU1
---- ----
lock(minor_rwsem);
lock(&ictx->lock);
lock(minor_rwsem);
lock(driver_lock#2);

*** DEADLOCK ***

1 lock held by syz-executor626/1723:
#0: 00000000076a0058 (minor_rwsem){++++}, at: usb_open+0x23/0x270
drivers/usb/core/file.c:39

stack backtrace:
CPU: 1 PID: 1723 Comm: syz-executor626 Not tainted 5.3.0-rc2+ #25
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS
Google 01/01/2011
Call Trace:
__dump_stack lib/dump_stack.c:77 [inline]
dump_stack+0xca/0x13e lib/dump_stack.c:113
check_noncircular+0x345/0x3e0 kernel/locking/lockdep.c:1741
check_prev_add kernel/locking/lockdep.c:2405 [inline]
check_prevs_add kernel/locking/lockdep.c:2507 [inline]
validate_chain kernel/locking/lockdep.c:2897 [inline]
__lock_acquire+0x1f7c/0x3b50 kernel/locking/lockdep.c:3880
lock_acquire+0x127/0x320 kernel/locking/lockdep.c:4412
__mutex_lock_common kernel/locking/mutex.c:930 [inline]
__mutex_lock+0x158/0x1360 kernel/locking/mutex.c:1077
display_open+0x1f/0x1d0 drivers/media/rc/imon.c:475
usb_open+0x1df/0x270 drivers/usb/core/file.c:48
chrdev_open+0x219/0x5c0 fs/char_dev.c:414
do_dentry_open+0x494/0x1120 fs/open.c:797
do_last fs/namei.c:3416 [inline]
path_openat+0x1430/0x3f50 fs/namei.c:3533
do_filp_open+0x1a1/0x280 fs/namei.c:3563
do_sys_open+0x3c0/0x580 fs/open.c:1089
do_syscall_64+0xb7/0x580 arch/x86/entry/common.c:296
entry_SYSCALL_64_after_hwframe+0x49/0xbe
RIP: 0033:0x401300
Code: 01 f0 ff ff 0f 83 00 0b 00 00 c3 66 2e 0f 1f 84 00 00 00 00 00 0f 1f
44 00 00 83 3d 8d 0a 2d 00 00 75 14 b8 02 00 00 00 0f 05 <48> 3d 01 f0 ff
ff 0f 83 d4 0a 00 00 c3 48 83 ec 08 e8 3a 00 00 00
RSP: 002b:00007ffc59dcdec8 EFLAGS: 00000246 ORIG_RAX: 0000000000000002
RAX: ffffffffffffffda RBX: 00000000004002c8 RCX: 0000000000401300
RDX: 0000000000000000 RSI: 0000000000000002 RDI: 00007ffc59dcdee0
RBP: 00000


---
This bug is generated by a bot. It may contain errors.
See https://goo.gl/tpsmEJ for more information about syzbot.
syzbot engineers can be reached at syzk...@googlegroups.com.

syzbot will keep track of this bug report. See:
https://goo.gl/tpsmEJ#status for how to communicate with syzbot.
syzbot can test patches for this bug, for details see:
https://goo.gl/tpsmEJ#testing-patches

[Tetsuo Handa]

Since usb_register_dev() from imon_init_display() from imon_probe() holds
minor_rwsem while display_open() which holds driver_lock and ictx->lock is
called with minor_rwsem held from usb_open(), holding driver_lock or
ictx->lock when calling usb_register_dev() causes circular locking
dependency problem.

Since usb_deregister_dev() from imon_disconnect() holds minor_rwsem while
display_open() which holds driver_lock is called with minor_rwsem held,
holding driver_lock when calling usb_deregister_dev() also causes circular
locking dependency problem.

But actually do we need to hold these locks?
Could you explain possible race scenario if we do below?

drivers/media/rc/imon.c | 21 ---------------------
1 file changed, 21 deletions(-)

diff --git a/drivers/media/rc/imon.c b/drivers/media/rc/imon.c
index 54da6f60079b..e0e893d96cf3 100644
--- a/drivers/media/rc/imon.c
+++ b/drivers/media/rc/imon.c
@@ -439,9 +439,6 @@ static struct usb_driver imon_driver = {
.id_table = imon_usb_id_table,
};

-/* to prevent races between open() and disconnect(), probing, etc */
-static DEFINE_MUTEX(driver_lock);
-
/* Module bookkeeping bits */
MODULE_AUTHOR(MOD_AUTHOR);
MODULE_DESCRIPTION(MOD_DESC);
@@ -499,9 +496,6 @@ static int display_open(struct inode *inode, struct file *file)
int subminor;
int retval = 0;

- /* prevent races with disconnect */
- mutex_lock(&driver_lock);
-
subminor = iminor(inode);
interface = usb_find_interface(&imon_driver, subminor);
if (!interface) {
@@ -534,7 +528,6 @@ static int display_open(struct inode *inode, struct file *file)
mutex_unlock(&ictx->lock);

exit:
- mutex_unlock(&driver_lock);
return retval;
}

@@ -2416,9 +2409,6 @@ static int imon_probe(struct usb_interface *interface,
dev_dbg(dev, "%s: found iMON device (%04x:%04x, intf%d)\n",
__func__, vendor, product, ifnum);

- /* prevent races probing devices w/multiple interfaces */
- mutex_lock(&driver_lock);
-
first_if = usb_ifnum_to_if(usbdev, 0);
if (!first_if) {
ret = -ENODEV;
@@ -2456,8 +2446,6 @@ static int imon_probe(struct usb_interface *interface,
usb_set_intfdata(interface, ictx);

if (ifnum == 0) {
- mutex_lock(&ictx->lock);
-
if (product == 0xffdc && ictx->rf_device) {
sysfs_err = sysfs_create_group(&interface->dev.kobj,
&imon_rf_attr_group);
@@ -2468,21 +2456,17 @@ static int imon_probe(struct usb_interface *interface,

if (ictx->display_supported)
imon_init_display(ictx, interface);
-
- mutex_unlock(&ictx->lock);
}

dev_info(dev, "iMON device (%04x:%04x, intf%d) on usb<%d:%d> initialized\n",
vendor, product, ifnum,
usbdev->bus->busnum, usbdev->devnum);

- mutex_unlock(&driver_lock);
usb_put_dev(usbdev);

return 0;

fail:
- mutex_unlock(&driver_lock);
usb_put_dev(usbdev);
dev_err(dev, "unable to register, err %d\n", ret);

@@ -2498,9 +2482,6 @@ static void imon_disconnect(struct usb_interface *interface)
struct device *dev;
int ifnum;

- /* prevent races with multi-interface device probing and display_open */
- mutex_lock(&driver_lock);
-
ictx = usb_get_intfdata(interface);
dev = ictx->dev;
ifnum = interface->cur_altsetting->desc.bInterfaceNumber;
@@ -2545,8 +2526,6 @@ static void imon_disconnect(struct usb_interface *interface)
if (!ictx->dev_present_intf0 && !ictx->dev_present_intf1)
free_imon_context(ictx);

- mutex_unlock(&driver_lock);
-
dev_dbg(dev, "%s: iMON device (intf%d) disconnected\n",
__func__, ifnum);
}
--
2.34.1

[Sean Young]

Hello,

On Mon, Apr 25, 2022 at 02:29:26PM +0900, Tetsuo Handa wrote:
> Since usb_register_dev() from imon_init_display() from imon_probe() holds
> minor_rwsem while display_open() which holds driver_lock and ictx->lock is
> called with minor_rwsem held from usb_open(), holding driver_lock or
> ictx->lock when calling usb_register_dev() causes circular locking
> dependency problem.
>
> Since usb_deregister_dev() from imon_disconnect() holds minor_rwsem while
> display_open() which holds driver_lock is called with minor_rwsem held,
> holding driver_lock when calling usb_deregister_dev() also causes circular
> locking dependency problem.
>
> But actually do we need to hold these locks?
> Could you explain possible race scenario if we do below?

The problem is there are imon devices which have two usb interfaces, even
though it is one device. The probe and disconnect function of both usb
interfaces can run concurrently.

If the imon_probe is running for interface 1, and the probe for interface 0
has not completed yet, then the driver may erronously think the probe for
interface 0 failed at `if (!first_if_ctx) {` on line 2442.

Of course, this depends on probe/disconnect functions being allowed to run
concurrently on different interfaces of the same usb device.

This code is rather tricky, and I'm sure there must be a better way of
dealing with multiple interfaces for a usb driver than what imon.c does.

Thanks
Sean

[Tetsuo Handa]

On 2022/04/25 18:20, Sean Young wrote:
> The problem is there are imon devices which have two usb interfaces, even
> though it is one device. The probe and disconnect function of both usb
> interfaces can run concurrently.
>

> Of course, this depends on probe/disconnect functions being allowed to run
> concurrently on different interfaces of the same usb device.

I don't have real hardware to confirm. If you have an imon device which has
two usb interfaces, please try below debug printk() patch in order to see
whether the caller is already holding locks for serialization.

------------------------------------------------------------
drivers/media/rc/imon.c | 25 +++++--------------------
1 file changed, 5 insertions(+), 20 deletions(-)

diff --git a/drivers/media/rc/imon.c b/drivers/media/rc/imon.c
index 54da6f60079b..a309c28e63d8 100644
--- a/drivers/media/rc/imon.c
+++ b/drivers/media/rc/imon.c
@@ -30,6 +30,7 @@
#include <media/rc-core.h>

#include <linux/timer.h>
+#include <linux/sched/debug.h>

#define MOD_AUTHOR "Jarod Wilson <ja...@wilsonet.com>"
#define MOD_DESC "Driver for SoundGraph iMON MultiMedia IR/Display"
@@ -439,9 +440,6 @@ static struct usb_driver imon_driver = {

.id_table = imon_usb_id_table,
};

-/* to prevent races between open() and disconnect(), probing, etc */
-static DEFINE_MUTEX(driver_lock);
-
/* Module bookkeeping bits */
MODULE_AUTHOR(MOD_AUTHOR);
MODULE_DESCRIPTION(MOD_DESC);

@@ -499,9 +497,6 @@ static int display_open(struct inode *inode, struct file *file)

int subminor;
int retval = 0;

- /* prevent races with disconnect */
- mutex_lock(&driver_lock);
-
subminor = iminor(inode);
interface = usb_find_interface(&imon_driver, subminor);
if (!interface) {

@@ -534,7 +529,6 @@ static int display_open(struct inode *inode, struct file *file)

mutex_unlock(&ictx->lock);

exit:
- mutex_unlock(&driver_lock);
return retval;
}

@@ -2407,6 +2401,8 @@ static int imon_probe(struct usb_interface *interface,
struct imon_context *first_if_ctx = NULL;
u16 vendor, product;

+ dump_stack();
+ debug_show_held_locks(current);
usbdev = usb_get_dev(interface_to_usbdev(interface));
iface_desc = interface->cur_altsetting;
ifnum = iface_desc->desc.bInterfaceNumber;
@@ -2416,9 +2412,6 @@ static int imon_probe(struct usb_interface *interface,

dev_dbg(dev, "%s: found iMON device (%04x:%04x, intf%d)\n",
__func__, vendor, product, ifnum);

- /* prevent races probing devices w/multiple interfaces */
- mutex_lock(&driver_lock);
-
first_if = usb_ifnum_to_if(usbdev, 0);
if (!first_if) {
ret = -ENODEV;

@@ -2456,8 +2449,6 @@ static int imon_probe(struct usb_interface *interface,

usb_set_intfdata(interface, ictx);

if (ifnum == 0) {
- mutex_lock(&ictx->lock);
-
if (product == 0xffdc && ictx->rf_device) {
sysfs_err = sysfs_create_group(&interface->dev.kobj,
&imon_rf_attr_group);

@@ -2468,21 +2459,17 @@ static int imon_probe(struct usb_interface *interface,

if (ictx->display_supported)
imon_init_display(ictx, interface);
-
- mutex_unlock(&ictx->lock);
}

dev_info(dev, "iMON device (%04x:%04x, intf%d) on usb<%d:%d> initialized\n",
vendor, product, ifnum,
usbdev->bus->busnum, usbdev->devnum);

- mutex_unlock(&driver_lock);
usb_put_dev(usbdev);

return 0;

fail:
- mutex_unlock(&driver_lock);
usb_put_dev(usbdev);
dev_err(dev, "unable to register, err %d\n", ret);

@@ -2498,8 +2485,8 @@ static void imon_disconnect(struct usb_interface *interface)

struct device *dev;
int ifnum;

- /* prevent races with multi-interface device probing and display_open */
- mutex_lock(&driver_lock);

+ dump_stack();
+ debug_show_held_locks(current);

ictx = usb_get_intfdata(interface);
dev = ictx->dev;

@@ -2545,8 +2532,6 @@ static void imon_disconnect(struct usb_interface *interface)

if (!ictx->dev_present_intf0 && !ictx->dev_present_intf1)
free_imon_context(ictx);

- mutex_unlock(&driver_lock);
-
dev_dbg(dev, "%s: iMON device (intf%d) disconnected\n",
__func__, ifnum);
}

------------------------------------------------------------

Below is a result with syzkaller-generated reproducer. If an imon device
which has two usb interfaces shares locks held, "prevent races with
multi-interface device probing" part can be removed.

------------------------------------------------------------
[ 193.854446][ T4177] usb 2-1: new high-speed USB device number 2 using dummy_hcd
[ 194.381949][ T4177] usb 2-1: config 0 interface 0 altsetting 0 endpoint 0x84 has an invalid bInterval 0, changing to 7
[ 194.389192][ T4177] usb 2-1: New USB device found, idVendor=15c2, idProduct=0039, bcdDevice=d2.65
[ 194.394772][ T4177] usb 2-1: New USB device strings: Mfr=0, Product=0, SerialNumber=0
[ 194.415288][ T4177] usb 2-1: config 0 descriptor??
[ 194.458149][ T4177] CPU: 0 PID: 4177 Comm: kworker/0:3 Not tainted 5.18.0-rc4-dirty #772
[ 194.460825][ T4177] Hardware name: innotek GmbH VirtualBox/VirtualBox, BIOS VirtualBox 12/01/2006
[ 194.463534][ T4177] Workqueue: usb_hub_wq hub_event
[ 194.465133][ T4177] Call Trace:
[ 194.466158][ T4177] <TASK>
[ 194.467087][ T4177] dump_stack_lvl+0xcd/0x134
[ 194.468655][ T4177] imon_probe+0x26/0x10da
[ 194.470109][ T4177] ? _raw_spin_unlock_irqrestore+0x3d/0x70
[ 194.471973][ T4177] ? __pm_runtime_set_status+0x29c/0x5d0
[ 194.473543][ T4177] ? _raw_spin_unlock_irqrestore+0x3d/0x70
[ 194.475219][ T4177] usb_probe_interface+0x19b/0x3e0
[ 194.476806][ T4177] ? usb_match_dynamic_id+0xe0/0xe0
[ 194.478282][ T4177] really_probe+0x138/0x4c0
[ 194.479511][ T4177] __driver_probe_device+0x191/0x220
[ 194.481069][ T4177] driver_probe_device+0x2a/0x120
[ 194.482622][ T4177] __device_attach_driver+0x105/0x1a0
[ 194.484609][ T4177] ? driver_allows_async_probing+0x90/0x90
[ 194.486344][ T4177] bus_for_each_drv+0xba/0x100
[ 194.487985][ T4177] __device_attach+0x130/0x290
[ 194.489305][ T4177] bus_probe_device+0xdb/0xf0
[ 194.490604][ T4177] device_add+0x635/0xdf0
[ 194.492076][ T4177] ? __mutex_unlock_slowpath+0x37/0x280
[ 194.493649][ T4177] usb_set_configuration+0x936/0xc20
[ 194.495389][ T4177] usb_generic_driver_probe+0x8c/0xc0
[ 194.496883][ T4177] usb_probe_device+0x6c/0x180
[ 194.498207][ T4177] ? usb_driver_release_interface+0xc0/0xc0
[ 194.500193][ T4177] really_probe+0x138/0x4c0
[ 194.501561][ T4177] __driver_probe_device+0x191/0x220
[ 194.503154][ T4177] driver_probe_device+0x2a/0x120
[ 194.504596][ T4177] __device_attach_driver+0x105/0x1a0
[ 194.506200][ T4177] ? driver_allows_async_probing+0x90/0x90
[ 194.507925][ T4177] bus_for_each_drv+0xba/0x100
[ 194.509331][ T4177] __device_attach+0x130/0x290
[ 194.510671][ T4177] bus_probe_device+0xdb/0xf0
[ 194.512115][ T4177] device_add+0x635/0xdf0
[ 194.513392][ T4177] usb_new_device.cold+0x10f/0x58e
[ 194.514905][ T4177] hub_event+0x15f9/0x2650
[ 194.516333][ T4177] process_one_work+0x381/0x940
[ 194.517826][ T4177] worker_thread+0x5b/0x5d0
[ 194.519247][ T4177] ? process_one_work+0x940/0x940
[ 194.527532][ T4177] kthread+0x135/0x170
[ 194.535432][ T4177] ? kthread_complete_and_exit+0x30/0x30
[ 194.543576][ T4177] ret_from_fork+0x1f/0x30
[ 194.551127][ T4177] </TASK>
[ 194.558820][ T4177] 5 locks held by kworker/0:3/4177:
[ 194.566387][ T4177] #0: ffff88800a0fed38 ((wq_completion)usb_hub_wq){+.+.}-{0:0}, at: process_one_work+0x2b5/0x940
[ 194.575897][ T4177] #1: ffffc90002fe3e70 ((work_completion)(&hub->events)){+.+.}-{0:0}, at: process_one_work+0x2b5/0x940
[ 194.585166][ T4177] #2: ffff888105129220 (&dev->mutex){....}-{3:3}, at: hub_event+0xc4/0x2650
[ 194.593581][ T4177] #3: ffff88810cc40220 (&dev->mutex){....}-{3:3}, at: __device_attach+0x41/0x290
[ 194.603139][ T4177] #4: ffff88810cc409a8 (&dev->mutex){....}-{3:3}, at: __device_attach+0x41/0x290
[ 194.618658][ T4177] input: iMON Panel, Knob and Mouse(15c2:0039) as /devices/platform/dummy_hcd.0/usb2/2-1/2-1:0.0/input/input6
[ 194.781071][ T4177] rc_core: IR keymap rc-imon-pad not found
[ 194.789123][ T4177] Registered IR keymap rc-empty
[ 194.796624][ T4177] imon 2-1:0.0: Looks like you're trying to use an IR protocol this device does not support
[ 194.806014][ T4177] imon 2-1:0.0: Unsupported IR protocol specified, overriding to iMON IR protocol
[ 194.965830][ T4177] rc rc0: iMON Remote (15c2:0039) as /devices/platform/dummy_hcd.0/usb2/2-1/2-1:0.0/rc/rc0
[ 194.998797][ T4177] input: iMON Remote (15c2:0039) as /devices/platform/dummy_hcd.0/usb2/2-1/2-1:0.0/rc/rc0/input7
[ 195.033215][ T4177] imon 2-1:0.0: iMON device (15c2:0039, intf0) on usb<2:2> initialized
[ 195.119182][ T3553] usb 2-1: USB disconnect, device number 2
[ 195.131624][ C0] imon 2-1:0.0: imon usb_rx_callback_intf0: status(-71): ignored
[ 195.157933][ T3553] CPU: 2 PID: 3553 Comm: kworker/2:3 Not tainted 5.18.0-rc4-dirty #772
[ 195.171455][ T3553] Hardware name: innotek GmbH VirtualBox/VirtualBox, BIOS VirtualBox 12/01/2006
[ 195.184454][ T3553] Workqueue: usb_hub_wq hub_event
[ 195.198577][ T3553] Call Trace:
[ 195.208174][ T3553] <TASK>
[ 195.216716][ T3553] dump_stack_lvl+0xcd/0x134
[ 195.225712][ T3553] imon_disconnect+0x1c/0x271
[ 195.234653][ T3553] usb_unbind_interface+0xe2/0x3e0
[ 195.243813][ T3553] ? usb_unbind_device+0xc0/0xc0
[ 195.253236][ T3553] device_remove+0x6d/0x80
[ 195.262130][ T3553] device_release_driver_internal+0x275/0x300
[ 195.271507][ T3553] bus_remove_device+0x12f/0x1d0
[ 195.281785][ T3553] device_del+0x1e5/0x4e0
[ 195.290748][ T3553] usb_disable_device+0x15f/0x310
[ 195.300276][ T3553] usb_disconnect.cold+0xd8/0x2fd
[ 195.309417][ T3553] ? __mutex_unlock_slowpath+0x37/0x280
[ 195.318605][ T3553] hub_event+0x190b/0x2650
[ 195.327623][ T3553] process_one_work+0x381/0x940
[ 195.336256][ T3553] worker_thread+0x5b/0x5d0
[ 195.345289][ T3553] ? process_one_work+0x940/0x940
[ 195.354733][ T3553] kthread+0x135/0x170
[ 195.363529][ T3553] ? kthread_complete_and_exit+0x30/0x30
[ 195.373531][ T3553] ret_from_fork+0x1f/0x30
[ 195.382434][ T3553] </TASK>
[ 195.397291][ T3553] 5 locks held by kworker/2:3/3553:
[ 195.408408][ T3553] #0: ffff88800a0fed38 ((wq_completion)usb_hub_wq){+.+.}-{0:0}, at: process_one_work+0x2b5/0x940
[ 195.417251][ T3553] #1: ffffc90002963e70 ((work_completion)(&hub->events)){+.+.}-{0:0}, at: process_one_work+0x2b5/0x940
[ 195.429307][ T3553] #2: ffff888105129220 (&dev->mutex){....}-{3:3}, at: hub_event+0xc4/0x2650
[ 195.439282][ T3553] #3: ffff88810cc40220 (&dev->mutex){....}-{3:3}, at: usb_disconnect+0x8e/0xa0
[ 195.450701][ T3553] #4: ffff88810cc409a8 (&dev->mutex){....}-{3:3}, at: device_release_driver_internal+0x55/0x300
------------------------------------------------------------

[Sean Young]

On Mon, Apr 25, 2022 at 08:14:12PM +0900, Tetsuo Handa wrote:
> On 2022/04/25 18:20, Sean Young wrote:
> > The problem is there are imon devices which have two usb interfaces, even
> > though it is one device. The probe and disconnect function of both usb
> > interfaces can run concurrently.
> >
> > Of course, this depends on probe/disconnect functions being allowed to run
> > concurrently on different interfaces of the same usb device.
>
> I don't have real hardware to confirm. If you have an imon device which has
> two usb interfaces, please try below debug printk() patch in order to see
> whether the caller is already holding locks for serialization.

I am afraid calling debug_show_held_locks() is not really going to tell us
this information. This should be figured out from understanding the usb
stack, not from seeing if the probe happens to be concurrent.

Just because the locks were not held, does not mean that the usb interface
initialization is not concurrent.

Thanks
Sean

[Alan Stern]

On Mon, Apr 25, 2022 at 12:56:19PM +0100, Sean Young wrote:
> On Mon, Apr 25, 2022 at 08:14:12PM +0900, Tetsuo Handa wrote:
> > On 2022/04/25 18:20, Sean Young wrote:
> > > The problem is there are imon devices which have two usb interfaces, even
> > > though it is one device. The probe and disconnect function of both usb
> > > interfaces can run concurrently.
> > >
> > > Of course, this depends on probe/disconnect functions being allowed to run
> > > concurrently on different interfaces of the same usb device.
> >
> > I don't have real hardware to confirm. If you have an imon device which has
> > two usb interfaces, please try below debug printk() patch in order to see
> > whether the caller is already holding locks for serialization.
>
> I am afraid calling debug_show_held_locks() is not really going to tell us
> this information. This should be figured out from understanding the usb
> stack, not from seeing if the probe happens to be concurrent.
>
> Just because the locks were not held, does not mean that the usb interface
> initialization is not concurrent.

The driver and USB cores guarantee that when an interface is probed,
both the interface and its USB device are locked. Ditto for when the
disconnect callback gets run. So concurrent probing/disconnection of
multiple interfaces on the same device is not possible.

Alan Stern

[Sean Young]

Hi Alan, Tetsuo,

That is very helpful, thank you Alan. I think the original patch as proposed
by Tetsuo Handa is correct.

However, the commit message does need to say that the lock is not needed
because usb core already guarantees this. Please submit a v2 for this.

Thank you both,

Sean

[syzbot]

Hello,

syzbot has tested the proposed patch but the reproducer is still triggering an issue:
KASAN: use-after-free Read in display_close

==================================================================
BUG: KASAN: use-after-free in __mutex_lock_common+0x11b/0x26e0 kernel/locking/mutex.c:579
Read of size 8 at addr ffff88806ec3c080 by task syz-executor.4/5279

CPU: 1 PID: 5279 Comm: syz-executor.4 Not tainted 5.18.0-rc4-syzkaller-00006-gd615b5416f8a-dirty #0

Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
Call Trace:

<TASK>
__dump_stack lib/dump_stack.c:88 [inline]
dump_stack_lvl+0x1e3/0x2cb lib/dump_stack.c:106
print_address_description+0x65/0x4b0 mm/kasan/report.c:313
print_report+0xf4/0x210 mm/kasan/report.c:429
kasan_report+0xfb/0x130 mm/kasan/report.c:491
__mutex_lock_common+0x11b/0x26e0 kernel/locking/mutex.c:579
__mutex_lock kernel/locking/mutex.c:733 [inline]
mutex_lock_nested+0x17/0x20 kernel/locking/mutex.c:785
display_close+0x53/0x1d0 drivers/media/rc/imon.c:550
__fput+0x3b9/0x820 fs/file_table.c:317
task_work_run+0x146/0x1c0 kernel/task_work.c:164
resume_user_mode_work include/linux/resume_user_mode.h:49 [inline]
exit_to_user_mode_loop+0x134/0x160 kernel/entry/common.c:169
exit_to_user_mode_prepare+0xad/0x110 kernel/entry/common.c:201
__syscall_exit_to_user_mode_work kernel/entry/common.c:283 [inline]
syscall_exit_to_user_mode+0x2e/0x70 kernel/entry/common.c:294
entry_SYSCALL_64_after_hwframe+0x44/0xae
RIP: 0033:0x7eff0683bd2b
Code: 0f 05 48 3d 00 f0 ff ff 77 45 c3 0f 1f 40 00 48 83 ec 18 89 7c 24 0c e8 63 fc ff ff 8b 7c 24 0c 41 89 c0 b8 03 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 35 44 89 c7 89 44 24 0c e8 a1 fc ff ff 8b 44
RSP: 002b:00007ffc4436ecd0 EFLAGS: 00000293 ORIG_RAX: 0000000000000003
RAX: 0000000000000000 RBX: 0000000000000005 RCX: 00007eff0683bd2b
RDX: 0000000000000000 RSI: 0000000000000001 RDI: 0000000000000004
RBP: 00007eff0699d960 R08: 0000000000000000 R09: 00007eff069a01c0
R10: 0000000000000000 R11: 0000000000000293 R12: 0000000000031ea9
R13: 00007ffc4436edd0 R14: 00007eff0699c030 R15: 0000000000000032
</TASK>

Allocated by task 14:
kasan_save_stack mm/kasan/common.c:38 [inline]
kasan_set_track mm/kasan/common.c:45 [inline]
set_alloc_info mm/kasan/common.c:436 [inline]
____kasan_kmalloc+0xdc/0x110 mm/kasan/common.c:515
kasan_kmalloc include/linux/kasan.h:234 [inline]
kmem_cache_alloc_trace+0x94/0x310 mm/slub.c:3258
kmalloc include/linux/slab.h:581 [inline]
kzalloc include/linux/slab.h:714 [inline]
imon_init_intf0 drivers/media/rc/imon.c:2209 [inline]
imon_probe+0x1af/0x2ef0 drivers/media/rc/imon.c:2421
usb_probe_interface+0x66e/0xb60 drivers/usb/core/driver.c:396
call_driver_probe+0x96/0x250
really_probe+0x222/0x8e0 drivers/base/dd.c:621
__driver_probe_device+0x1f4/0x3f0 drivers/base/dd.c:752
driver_probe_device+0x50/0x240 drivers/base/dd.c:782
__device_attach_driver+0x254/0x3a0 drivers/base/dd.c:899
bus_for_each_drv+0x18a/0x210 drivers/base/bus.c:427
__device_attach+0x32e/0x570 drivers/base/dd.c:970
bus_probe_device+0xb8/0x1f0 drivers/base/bus.c:487
device_add+0xb20/0xf90 drivers/base/core.c:3405
usb_set_configuration+0x1a5f/0x20e0 drivers/usb/core/message.c:2170
usb_generic_driver_probe+0x83/0x140 drivers/usb/core/generic.c:238
usb_probe_device+0x131/0x260 drivers/usb/core/driver.c:293
call_driver_probe+0x96/0x250
really_probe+0x222/0x8e0 drivers/base/dd.c:621
__driver_probe_device+0x1f4/0x3f0 drivers/base/dd.c:752
driver_probe_device+0x50/0x240 drivers/base/dd.c:782
__device_attach_driver+0x254/0x3a0 drivers/base/dd.c:899
bus_for_each_drv+0x18a/0x210 drivers/base/bus.c:427
__device_attach+0x32e/0x570 drivers/base/dd.c:970
bus_probe_device+0xb8/0x1f0 drivers/base/bus.c:487
device_add+0xb20/0xf90 drivers/base/core.c:3405
usb_new_device+0xbfc/0x18a0 drivers/usb/core/hub.c:2566
hub_port_connect+0x106b/0x2930 drivers/usb/core/hub.c:5363
hub_port_connect_change+0x619/0xbe0 drivers/usb/core/hub.c:5507
port_event+0xe72/0x13a0 drivers/usb/core/hub.c:5665
hub_event+0x5be/0xd70 drivers/usb/core/hub.c:5747
process_one_work+0x81c/0xd10 kernel/workqueue.c:2289
worker_thread+0xb14/0x1330 kernel/workqueue.c:2436
kthread+0x266/0x300 kernel/kthread.c:376
ret_from_fork+0x1f/0x30

Freed by task 14:
kasan_save_stack mm/kasan/common.c:38 [inline]
kasan_set_track+0x4c/0x70 mm/kasan/common.c:45
kasan_set_free_info+0x1f/0x40 mm/kasan/generic.c:370
____kasan_slab_free+0xd8/0x110 mm/kasan/common.c:366
kasan_slab_free include/linux/kasan.h:200 [inline]
slab_free_hook mm/slub.c:1728 [inline]
slab_free_freelist_hook+0x12e/0x1a0 mm/slub.c:1754
slab_free mm/slub.c:3510 [inline]
kfree+0xc6/0x210 mm/slub.c:4552
free_imon_context drivers/media/rc/imon.c:483 [inline]
imon_disconnect+0x473/0x5d0 drivers/media/rc/imon.c:2527
usb_unbind_interface+0x1f2/0x860 drivers/usb/core/driver.c:458
device_remove drivers/base/dd.c:532 [inline]
__device_release_driver drivers/base/dd.c:1200 [inline]
device_release_driver_internal+0x56b/0x820 drivers/base/dd.c:1223
bus_remove_device+0x2fd/0x410 drivers/base/bus.c:529
device_del+0x6ec/0xbe0 drivers/base/core.c:3592
usb_disable_device+0x3dd/0x820 drivers/usb/core/message.c:1419
usb_disconnect+0x346/0x890 drivers/usb/core/hub.c:2228
hub_port_connect+0x2a2/0x2930 drivers/usb/core/hub.c:5207
hub_port_connect_change+0x619/0xbe0 drivers/usb/core/hub.c:5507
port_event+0xe72/0x13a0 drivers/usb/core/hub.c:5665
hub_event+0x5be/0xd70 drivers/usb/core/hub.c:5747
process_one_work+0x81c/0xd10 kernel/workqueue.c:2289
worker_thread+0xb14/0x1330 kernel/workqueue.c:2436
kthread+0x266/0x300 kernel/kthread.c:376
ret_from_fork+0x1f/0x30

Last potentially related work creation:
kasan_save_stack+0x3b/0x60 mm/kasan/common.c:38
__kasan_record_aux_stack+0xaf/0xc0 mm/kasan/generic.c:348
call_rcu+0x163/0x9c0 kernel/rcu/tree.c:3074
in6_dev_put include/net/addrconf.h:380 [inline]
addrconf_ifdown+0x1799/0x1a00 net/ipv6/addrconf.c:3888
addrconf_notify+0x403/0xf80
notifier_call_chain kernel/notifier.c:84 [inline]
raw_notifier_call_chain+0xe7/0x170 kernel/notifier.c:392
call_netdevice_notifiers_info net/core/dev.c:1938 [inline]
call_netdevice_notifiers_extack net/core/dev.c:1976 [inline]
call_netdevice_notifiers net/core/dev.c:1990 [inline]
unregister_netdevice_many+0xeee/0x1950 net/core/dev.c:10751
unregister_netdevice_queue+0x2e6/0x350 net/core/dev.c:10683
unregister_netdevice include/linux/netdevice.h:2954 [inline]
nsim_destroy+0x44/0x140 drivers/net/netdevsim/netdev.c:382
__nsim_dev_port_del+0x152/0x1b0 drivers/net/netdevsim/dev.c:1427
nsim_dev_port_del_all drivers/net/netdevsim/dev.c:1440 [inline]
nsim_dev_reload_destroy+0x2c2/0x4d0 drivers/net/netdevsim/dev.c:1659
nsim_dev_reload_down+0xfb/0x150 drivers/net/netdevsim/dev.c:962
devlink_reload+0x200/0x6c0 net/core/devlink.c:4070
devlink_pernet_pre_exit+0x1ab/0x350 net/core/devlink.c:11476
ops_pre_exit_list net/core/net_namespace.c:152 [inline]
cleanup_net+0x608/0xc50 net/core/net_namespace.c:582
process_one_work+0x81c/0xd10 kernel/workqueue.c:2289
worker_thread+0xb14/0x1330 kernel/workqueue.c:2436
kthread+0x266/0x300 kernel/kthread.c:376
ret_from_fork+0x1f/0x30

Second to last potentially related work creation:
kasan_save_stack+0x3b/0x60 mm/kasan/common.c:38
__kasan_record_aux_stack+0xaf/0xc0 mm/kasan/generic.c:348
insert_work+0x54/0x3e0 kernel/workqueue.c:1358
__queue_work+0x969/0xc20 kernel/workqueue.c:1517
call_timer_fn+0xf5/0x210 kernel/time/timer.c:1421
expire_timers kernel/time/timer.c:1461 [inline]
__run_timers+0x7a2/0x980 kernel/time/timer.c:1737
run_timer_softirq+0x63/0xf0 kernel/time/timer.c:1750
__do_softirq+0x382/0x793 kernel/softirq.c:558

The buggy address belongs to the object at ffff88806ec3c000
which belongs to the cache kmalloc-2k of size 2048
The buggy address is located 128 bytes inside of
2048-byte region [ffff88806ec3c000, ffff88806ec3c800)

The buggy address belongs to the physical page:
page:ffffea0001bb0e00 refcount:1 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x6ec38
head:ffffea0001bb0e00 order:3 compound_mapcount:0 compound_pincount:0
flags: 0xfff00000010200(slab|head|node=0|zone=1|lastcpupid=0x7ff)
raw: 00fff00000010200 dead000000000100 dead000000000122 ffff888011042000
raw: 0000000000000000 0000000000080008 00000001ffffffff 0000000000000000
page dumped because: kasan: bad access detected
page_owner tracks the page as allocated
page last allocated via order 3, migratetype Unmovable, gfp_mask 0xd20c0(__GFP_IO|__GFP_FS|__GFP_NOWARN|__GFP_NORETRY|__GFP_COMP|__GFP_NOMEMALLOC), pid 3597, tgid 3597 (syz-executor.0), ts 50108796928, free_ts 11730170768
prep_new_page mm/page_alloc.c:2441 [inline]
get_page_from_freelist+0x72e/0x7a0 mm/page_alloc.c:4182
__alloc_pages+0x26c/0x5f0 mm/page_alloc.c:5408
alloc_slab_page+0x70/0xf0 mm/slub.c:1799
allocate_slab+0x5e/0x560 mm/slub.c:1944
new_slab mm/slub.c:2004 [inline]
___slab_alloc+0x41e/0xcd0 mm/slub.c:3005
__slab_alloc mm/slub.c:3092 [inline]
slab_alloc_node mm/slub.c:3183 [inline]
slab_alloc mm/slub.c:3225 [inline]
__kmalloc+0x2ba/0x370 mm/slub.c:4410
kmalloc include/linux/slab.h:586 [inline]
kzalloc include/linux/slab.h:714 [inline]
__register_sysctl_table+0xdd/0x1380 fs/proc/proc_sysctl.c:1335
__addrconf_sysctl_register+0x26e/0x3c0 net/ipv6/addrconf.c:7052
addrconf_sysctl_register+0x128/0x180 net/ipv6/addrconf.c:7099
ipv6_add_dev+0xda5/0x12c0 net/ipv6/addrconf.c:450
addrconf_notify+0x629/0xf80 net/ipv6/addrconf.c:3521
notifier_call_chain kernel/notifier.c:84 [inline]
raw_notifier_call_chain+0xe7/0x170 kernel/notifier.c:392
call_netdevice_notifiers_info net/core/dev.c:1938 [inline]
call_netdevice_notifiers_extack net/core/dev.c:1976 [inline]
call_netdevice_notifiers+0x14e/0x1d0 net/core/dev.c:1990
register_netdevice+0x158f/0x19a0 net/core/dev.c:9994
nsim_init_netdevsim drivers/net/netdevsim/netdev.c:317 [inline]
nsim_create+0x349/0x3f0 drivers/net/netdevsim/netdev.c:365
__nsim_dev_port_add+0x6c2/0xb30 drivers/net/netdevsim/dev.c:1390
page last free stack trace:
reset_page_owner include/linux/page_owner.h:24 [inline]
free_pages_prepare mm/page_alloc.c:1356 [inline]
free_pcp_prepare+0x812/0x900 mm/page_alloc.c:1406
free_unref_page_prepare mm/page_alloc.c:3328 [inline]
free_unref_page+0x7d/0x390 mm/page_alloc.c:3423
free_contig_range+0xa3/0x160 mm/page_alloc.c:9418
destroy_args+0xfe/0x971 mm/debug_vm_pgtable.c:1018
debug_vm_pgtable+0x44c/0x4a5 mm/debug_vm_pgtable.c:1332
do_one_initcall+0xbd/0x2b0 init/main.c:1298
do_initcall_level+0x168/0x218 init/main.c:1371
do_initcalls+0x4b/0x8c init/main.c:1387
kernel_init_freeable+0x43a/0x5c3 init/main.c:1613
kernel_init+0x19/0x2b0 init/main.c:1502
ret_from_fork+0x1f/0x30

Memory state around the buggy address:
ffff88806ec3bf80: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
ffff88806ec3c000: fa fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
>ffff88806ec3c080: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
^
ffff88806ec3c100: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
ffff88806ec3c180: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
==================================================================


Tested on:

commit: d615b541 Merge tag 'f2fs-fix-5.18' of git://git.kernel..
git tree: upstream
console output: https://syzkaller.appspot.com/x/log.txt?x=1232f89f700000
kernel config: https://syzkaller.appspot.com/x/.config?x=a935ec57fc0fd0ed
dashboard link: https://syzkaller.appspot.com/bug?extid=c558267ad910fc494497
compiler: Debian clang version 13.0.1-++20220126092033+75e33f71c2da-1~exp1~20220126212112.63, GNU ld (GNU Binutils for Debian) 2.35.2
patch: https://syzkaller.appspot.com/x/patch.diff?x=13f74658f00000

[Tetsuo Handa]

Since usb_register_dev() from imon_init_display() from imon_probe() holds
minor_rwsem while display_open() which holds driver_lock and ictx->lock is
called with minor_rwsem held from usb_open(), holding driver_lock or
ictx->lock when calling usb_register_dev() causes circular locking
dependency problem.

Since usb_deregister_dev() from imon_disconnect() holds minor_rwsem while
display_open() which holds driver_lock is called with minor_rwsem held,
holding driver_lock when calling usb_deregister_dev() also causes circular
locking dependency problem.

Sean Young explained that the problem is there are imon devices which have

two usb interfaces, even though it is one device. The probe and disconnect
function of both usb interfaces can run concurrently.

Alan Stern responded that the driver and USB cores guarantee that when an

interface is probed, both the interface and its USB device are locked.
Ditto for when the disconnect callback gets run. So concurrent probing/
disconnection of multiple interfaces on the same device is not possible.

Therefore, simply remove redundant serialization.

Link: https://syzkaller.appspot.com/bug?extid=c558267ad910fc494497
Reported-by: syzbot <syzbot+c55826...@syzkaller.appspotmail.com>
Signed-off-by: Tetsuo Handa <penguin...@I-love.SAKURA.ne.jp>
Cc: Sean Young <se...@mess.org>
Cc: Alan Stern <st...@rowland.harvard.edu>
---
This patch might be OK for solving circular locking dependency problem.
But https://syzkaller.appspot.com/text?tag=CrashReport&x=1632f89f700000 says
free_imon_context() from imon_disconnect() can call kfree(ictx) before
display_close() is called, resulting in use-after-free problem. I guess that
we need to defer free_imon_context() using refcount till display_close() is
called. Fix as a separate patch or include into this patch?

drivers/media/rc/imon.c | 21 ---------------------
1 file changed, 21 deletions(-)

diff --git a/drivers/media/rc/imon.c b/drivers/media/rc/imon.c
index 54da6f60079b..e0e893d96cf3 100644
--- a/drivers/media/rc/imon.c
+++ b/drivers/media/rc/imon.c
@@ -439,9 +439,6 @@ static struct usb_driver imon_driver = {

.id_table = imon_usb_id_table,
};

-/* to prevent races between open() and disconnect(), probing, etc */
-static DEFINE_MUTEX(driver_lock);
-
/* Module bookkeeping bits */
MODULE_AUTHOR(MOD_AUTHOR);
MODULE_DESCRIPTION(MOD_DESC);

@@ -499,9 +496,6 @@ static int display_open(struct inode *inode, struct file *file)

int subminor;
int retval = 0;

- /* prevent races with disconnect */
- mutex_lock(&driver_lock);
-
subminor = iminor(inode);
interface = usb_find_interface(&imon_driver, subminor);
if (!interface) {

@@ -534,7 +528,6 @@ static int display_open(struct inode *inode, struct file *file)

mutex_unlock(&ictx->lock);

exit:
- mutex_unlock(&driver_lock);
return retval;
}

@@ -2416,9 +2409,6 @@ static int imon_probe(struct usb_interface *interface,

dev_dbg(dev, "%s: found iMON device (%04x:%04x, intf%d)\n",
__func__, vendor, product, ifnum);

- /* prevent races probing devices w/multiple interfaces */
- mutex_lock(&driver_lock);
-
first_if = usb_ifnum_to_if(usbdev, 0);
if (!first_if) {
ret = -ENODEV;

@@ -2456,8 +2446,6 @@ static int imon_probe(struct usb_interface *interface,

usb_set_intfdata(interface, ictx);

if (ifnum == 0) {
- mutex_lock(&ictx->lock);
-
if (product == 0xffdc && ictx->rf_device) {
sysfs_err = sysfs_create_group(&interface->dev.kobj,
&imon_rf_attr_group);

@@ -2468,21 +2456,17 @@ static int imon_probe(struct usb_interface *interface,

if (ictx->display_supported)
imon_init_display(ictx, interface);
-
- mutex_unlock(&ictx->lock);
}

dev_info(dev, "iMON device (%04x:%04x, intf%d) on usb<%d:%d> initialized\n",
vendor, product, ifnum,
usbdev->bus->busnum, usbdev->devnum);

- mutex_unlock(&driver_lock);
usb_put_dev(usbdev);

return 0;

fail:
- mutex_unlock(&driver_lock);
usb_put_dev(usbdev);
dev_err(dev, "unable to register, err %d\n", ret);

@@ -2498,9 +2482,6 @@ static void imon_disconnect(struct usb_interface *interface)

struct device *dev;
int ifnum;

- /* prevent races with multi-interface device probing and display_open */
- mutex_lock(&driver_lock);

-

ictx = usb_get_intfdata(interface);
dev = ictx->dev;

ifnum = interface->cur_altsetting->desc.bInterfaceNumber;

@@ -2545,8 +2526,6 @@ static void imon_disconnect(struct usb_interface *interface)

if (!ictx->dev_present_intf0 && !ictx->dev_present_intf1)
free_imon_context(ictx);

- mutex_unlock(&driver_lock);
-
dev_dbg(dev, "%s: iMON device (intf%d) disconnected\n",
__func__, ifnum);
}
--

2.34.1

[syzbot]

Hello,

syzbot has tested the proposed patch and the reproducer did not trigger any issue:

Reported-and-tested-by: syzbot+c55826...@syzkaller.appspotmail.com

Tested on:

commit: d615b541 Merge tag 'f2fs-fix-5.18' of git://git.kernel..
git tree: upstream

kernel config: https://syzkaller.appspot.com/x/.config?x=a935ec57fc0fd0ed
dashboard link: https://syzkaller.appspot.com/bug?extid=c558267ad910fc494497
compiler: Debian clang version 13.0.1-++20220126092033+75e33f71c2da-1~exp1~20220126212112.63, GNU ld (GNU Binutils for Debian) 2.35.2

patch: https://syzkaller.appspot.com/x/patch.diff?x=1340ac1f700000

Note: testing is done by a robot and is best-effort only.

[Sean Young]

On Tue, Apr 26, 2022 at 09:54:18AM +0900, Tetsuo Handa wrote:
> Since usb_register_dev() from imon_init_display() from imon_probe() holds
> minor_rwsem while display_open() which holds driver_lock and ictx->lock is
> called with minor_rwsem held from usb_open(), holding driver_lock or
> ictx->lock when calling usb_register_dev() causes circular locking
> dependency problem.
>
> Since usb_deregister_dev() from imon_disconnect() holds minor_rwsem while
> display_open() which holds driver_lock is called with minor_rwsem held,
> holding driver_lock when calling usb_deregister_dev() also causes circular
> locking dependency problem.
>
> Sean Young explained that the problem is there are imon devices which have
> two usb interfaces, even though it is one device. The probe and disconnect
> function of both usb interfaces can run concurrently.
>
> Alan Stern responded that the driver and USB cores guarantee that when an
> interface is probed, both the interface and its USB device are locked.
> Ditto for when the disconnect callback gets run. So concurrent probing/
> disconnection of multiple interfaces on the same device is not possible.
>
> Therefore, simply remove redundant serialization.
>
> Link: https://syzkaller.appspot.com/bug?extid=c558267ad910fc494497
> Reported-by: syzbot <syzbot+c55826...@syzkaller.appspotmail.com>
> Signed-off-by: Tetsuo Handa <penguin...@I-love.SAKURA.ne.jp>
> Cc: Sean Young <se...@mess.org>
> Cc: Alan Stern <st...@rowland.harvard.edu>

Thanks, looks good!

> ---
> This patch might be OK for solving circular locking dependency problem.
> But https://syzkaller.appspot.com/text?tag=CrashReport&x=1632f89f700000 says
> free_imon_context() from imon_disconnect() can call kfree(ictx) before
> display_close() is called, resulting in use-after-free problem. I guess that
> we need to defer free_imon_context() using refcount till display_close() is
> called. Fix as a separate patch or include into this patch?

That is a separate issue, so that needs a separate patch.

Sean

[Tetsuo Handa]

On 2022/04/26 16:57, Sean Young wrote:
>> This patch might be OK for solving circular locking dependency problem.
>> But https://syzkaller.appspot.com/text?tag=CrashReport&x=1632f89f700000 says
>> free_imon_context() from imon_disconnect() can call kfree(ictx) before
>> display_close() is called, resulting in use-after-free problem. I guess that
>> we need to defer free_imon_context() using refcount till display_close() is
>> called. Fix as a separate patch or include into this patch?
>
> That is a separate issue, so that needs a separate patch.
>

I think this should be addressed in this fix, for this patch needs to keep
"prevent races between open() and disconnect()" role of driver_lock.

It seems to me that "struct imon_context" is shared between intf0 and intf1.
Is imon_disconnect() called on the same "struct imon_context" for two times
if ictx->dev_present_intf0 = true and ictx->dev_present_intf1 = true ? Below
patch assumes that imon_disconnect() is called on the same "struct imon_context"
for two times if ictx->dev_present_intf0 = true and ictx->dev_present_intf1 = true.
Is this assumption correct?

From 73df086af9b4786cfcee81a7e82ee8e71be82290 Mon Sep 17 00:00:00 2001
From: Tetsuo Handa <penguin...@I-love.SAKURA.ne.jp>
Date: Tue, 26 Apr 2022 10:15:24 +0000
Subject: [PATCH v2] media: imon: reorganize serialization

Since usb_register_dev() from imon_init_display() from imon_probe() holds
minor_rwsem while display_open() which holds driver_lock and ictx->lock is
called with minor_rwsem held from usb_open(), holding driver_lock or
ictx->lock when calling usb_register_dev() causes circular locking
dependency problem.

Since usb_deregister_dev() from imon_disconnect() holds minor_rwsem while
display_open() which holds driver_lock is called with minor_rwsem held,
holding driver_lock when calling usb_deregister_dev() also causes circular
locking dependency problem.

Sean Young explained that the problem is there are imon devices which have
two usb interfaces, even though it is one device. The probe and disconnect
function of both usb interfaces can run concurrently.

Alan Stern responded that the driver and USB cores guarantee that when an
interface is probed, both the interface and its USB device are locked.
Ditto for when the disconnect callback gets run. So concurrent probing/
disconnection of multiple interfaces on the same device is not possible.

Therefore, we don't need locks for handling race between imon_probe() and
imon_disconnect(). But we still need to handle race between display_open()
/vfd_write()/lcd_write()/display_close() and imon_disconnect(), for
disconnect event can happen while file descriptors are in use.

Since "struct file"->private_data is set by display_open(), vfd_write()/
lcd_write()/display_close() can assume that "struct file"->private_data
is not NULL even after usb_set_intfdata(interface, NULL) was called.

Replace insufficiently held driver_lock with refcount_t based management.
Add a boolean flag for recording whether imon_disconnect() was already
called. Use RCU for accessing this boolean flag and refcount_t.

Link: https://syzkaller.appspot.com/bug?extid=c558267ad910fc494497
Reported-by: syzbot <syzbot+c55826...@syzkaller.appspotmail.com>
Signed-off-by: Tetsuo Handa <penguin...@I-love.SAKURA.ne.jp>
Cc: Sean Young <se...@mess.org>
Cc: Alan Stern <st...@rowland.harvard.edu>

---
drivers/media/rc/imon.c | 99 +++++++++++++++++++----------------------
1 file changed, 47 insertions(+), 52 deletions(-)

diff --git a/drivers/media/rc/imon.c b/drivers/media/rc/imon.c
index 54da6f60079b..9a4f24e294bc 100644
--- a/drivers/media/rc/imon.c
+++ b/drivers/media/rc/imon.c
@@ -153,6 +153,24 @@ struct imon_context {
const struct imon_usb_dev_descr *dev_descr;
/* device description with key */
/* table for front panels */
+ /*
+ * Fields for deferring free_imon_context().
+ *
+ * Since reference to "struct imon_context" is stored into
+ * "struct file_operations"->private_data, we need to remember
+ * how many file descriptors might access this "struct imon_context".
+ */
+ refcount_t users;
+ /*
+ * Use a flag for telling display_open()/vfd_write()/lcd_write() that
+ * imon_disconnect() was already called.
+ */
+ bool disconnected;
+ /*
+ * We need to wait for RCU grace period in order to allow
+ * display_open() to safely check ->disconnected and increment ->users.
+ */
+ struct rcu_head rcu;
};

#define TOUCH_TIMEOUT (HZ/30)
@@ -160,18 +178,18 @@ struct imon_context {
/* vfd character device file operations */
static const struct file_operations vfd_fops = {
.owner = THIS_MODULE,
- .open = &display_open,
- .write = &vfd_write,
- .release = &display_close,
+ .open = display_open,
+ .write = vfd_write,
+ .release = display_close,
.llseek = noop_llseek,
};

/* lcd character device file operations */
static const struct file_operations lcd_fops = {
.owner = THIS_MODULE,
- .open = &display_open,
- .write = &lcd_write,
- .release = &display_close,
+ .open = display_open,
+ .write = lcd_write,
+ .release = display_close,
.llseek = noop_llseek,
};

@@ -439,9 +457,6 @@ static struct usb_driver imon_driver = {

.id_table = imon_usb_id_table,
};

-/* to prevent races between open() and disconnect(), probing, etc */
-static DEFINE_MUTEX(driver_lock);
-
/* Module bookkeeping bits */
MODULE_AUTHOR(MOD_AUTHOR);
MODULE_DESCRIPTION(MOD_DESC);

@@ -481,9 +496,11 @@ static void free_imon_context(struct imon_context *ictx)
struct device *dev = ictx->dev;

usb_free_urb(ictx->tx_urb);
+ WARN_ON(ictx->dev_present_intf0);
usb_free_urb(ictx->rx_urb_intf0);
+ WARN_ON(ictx->dev_present_intf1);
usb_free_urb(ictx->rx_urb_intf1);
- kfree(ictx);
+ kfree_rcu(ictx, rcu);

dev_dbg(dev, "%s: iMON context freed\n", __func__);
}
@@ -499,9 +516,6 @@ static int display_open(struct inode *inode, struct file *file)

int subminor;
int retval = 0;

- /* prevent races with disconnect */
- mutex_lock(&driver_lock);
-
subminor = iminor(inode);
interface = usb_find_interface(&imon_driver, subminor);
if (!interface) {

@@ -509,13 +523,16 @@ static int display_open(struct inode *inode, struct file *file)
retval = -ENODEV;
goto exit;
}
- ictx = usb_get_intfdata(interface);

- if (!ictx) {
+ rcu_read_lock();
+ ictx = usb_get_intfdata(interface);
+ if (!ictx || ictx->disconnected || !refcount_inc_not_zero(&ictx->users)) {
+ rcu_read_unlock();
pr_err("no context found for minor %d\n", subminor);
retval = -ENODEV;
goto exit;
}
+ rcu_read_unlock();

mutex_lock(&ictx->lock);

@@ -533,8 +550,10 @@ static int display_open(struct inode *inode, struct file *file)

mutex_unlock(&ictx->lock);

+ if (retval && refcount_dec_and_test(&ictx->users))
+ free_imon_context(ictx);
+

exit:
- mutex_unlock(&driver_lock);
return retval;
}

@@ -544,16 +563,9 @@ static int display_open(struct inode *inode, struct file *file)
*/
static int display_close(struct inode *inode, struct file *file)
{
- struct imon_context *ictx = NULL;
+ struct imon_context *ictx = file->private_data;
int retval = 0;

- ictx = file->private_data;
-
- if (!ictx) {
- pr_err("no context for device\n");
- return -ENODEV;
- }
-
mutex_lock(&ictx->lock);

if (!ictx->display_supported) {
@@ -568,6 +580,8 @@ static int display_close(struct inode *inode, struct file *file)
}

mutex_unlock(&ictx->lock);
+ if (refcount_dec_and_test(&ictx->users))
+ free_imon_context(ictx);
return retval;
}

@@ -934,15 +948,12 @@ static ssize_t vfd_write(struct file *file, const char __user *buf,
int offset;
int seq;
int retval = 0;
- struct imon_context *ictx;
+ struct imon_context *ictx = file->private_data;
static const unsigned char vfd_packet6[] = {
0x01, 0x00, 0x00, 0x00, 0x00, 0xFF, 0xFF };

- ictx = file->private_data;
- if (!ictx) {
- pr_err_ratelimited("no context for device\n");
+ if (ictx->disconnected)
return -ENODEV;
- }

mutex_lock(&ictx->lock);

@@ -1018,13 +1029,10 @@ static ssize_t lcd_write(struct file *file, const char __user *buf,
size_t n_bytes, loff_t *pos)
{
int retval = 0;
- struct imon_context *ictx;
+ struct imon_context *ictx = file->private_data;

- ictx = file->private_data;
- if (!ictx) {
- pr_err_ratelimited("no context for device\n");
+ if (ictx->disconnected)
return -ENODEV;
- }

mutex_lock(&ictx->lock);

@@ -2404,7 +2412,6 @@ static int imon_probe(struct usb_interface *interface,
int ifnum, sysfs_err;
int ret = 0;
struct imon_context *ictx = NULL;
- struct imon_context *first_if_ctx = NULL;
u16 vendor, product;

usbdev = usb_get_dev(interface_to_usbdev(interface));
@@ -2416,17 +2423,12 @@ static int imon_probe(struct usb_interface *interface,

dev_dbg(dev, "%s: found iMON device (%04x:%04x, intf%d)\n",
__func__, vendor, product, ifnum);

- /* prevent races probing devices w/multiple interfaces */
- mutex_lock(&driver_lock);
-
first_if = usb_ifnum_to_if(usbdev, 0);
if (!first_if) {
ret = -ENODEV;

goto fail;
}

- first_if_ctx = usb_get_intfdata(first_if);
-
if (ifnum == 0) {
ictx = imon_init_intf0(interface, id);
if (!ictx) {
@@ -2434,9 +2436,11 @@ static int imon_probe(struct usb_interface *interface,
ret = -ENODEV;
goto fail;
}
+ refcount_set(&ictx->users, 1);

} else {
/* this is the secondary interface on the device */
+ struct imon_context *first_if_ctx = usb_get_intfdata(first_if);

/* fail early if first intf failed to register */
if (!first_if_ctx) {
@@ -2450,14 +2454,13 @@ static int imon_probe(struct usb_interface *interface,
ret = -ENODEV;
goto fail;
}
+ refcount_inc(&ictx->users);

}

usb_set_intfdata(interface, ictx);

if (ifnum == 0) {
- mutex_lock(&ictx->lock);
-
if (product == 0xffdc && ictx->rf_device) {
sysfs_err = sysfs_create_group(&interface->dev.kobj,
&imon_rf_attr_group);

@@ -2468,21 +2471,17 @@ static int imon_probe(struct usb_interface *interface,

if (ictx->display_supported)
imon_init_display(ictx, interface);
-
- mutex_unlock(&ictx->lock);
}

dev_info(dev, "iMON device (%04x:%04x, intf%d) on usb<%d:%d> initialized\n",
vendor, product, ifnum,
usbdev->bus->busnum, usbdev->devnum);

- mutex_unlock(&driver_lock);
usb_put_dev(usbdev);

return 0;

fail:
- mutex_unlock(&driver_lock);
usb_put_dev(usbdev);
dev_err(dev, "unable to register, err %d\n", ret);

@@ -2498,10 +2497,8 @@ static void imon_disconnect(struct usb_interface *interface)

struct device *dev;
int ifnum;

- /* prevent races with multi-interface device probing and display_open */
- mutex_lock(&driver_lock);
-
ictx = usb_get_intfdata(interface);

+ ictx->disconnected = true;

dev = ictx->dev;
ifnum = interface->cur_altsetting->desc.bInterfaceNumber;

@@ -2542,11 +2539,9 @@ static void imon_disconnect(struct usb_interface *interface)
}
}

- if (!ictx->dev_present_intf0 && !ictx->dev_present_intf1)
+ if (refcount_dec_and_test(&ictx->users))

[syzbot]

Hello,

syzbot has tested the proposed patch and the reproducer did not trigger any issue:

Reported-and-tested-by: syzbot+c55826...@syzkaller.appspotmail.com

Tested on:

commit: d4323059 usb: rework usb_maxpacket() using usb_pipe_en..
git tree: git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb.git
kernel config: https://syzkaller.appspot.com/x/.config?x=c36430635c209df2

dashboard link: https://syzkaller.appspot.com/bug?extid=c558267ad910fc494497
compiler: Debian clang version 13.0.1-++20220126092033+75e33f71c2da-1~exp1~20220126212112.63, GNU ld (GNU Binutils for Debian) 2.35.2

patch: https://syzkaller.appspot.com/x/patch.diff?x=1203600ef00000

[Tetsuo Handa]

Since the boolean flag for imon_disconnect() is shared, disconnect event
on either intf0 or intf1 affects both interfaces. But I assume that this
change does not matter, for usually disconnect event would not happen
while interfaces are in use.

Link: https://syzkaller.appspot.com/bug?extid=c558267ad910fc494497
Reported-by: syzbot <syzbot+c55826...@syzkaller.appspotmail.com>
Signed-off-by: Tetsuo Handa <penguin...@I-love.SAKURA.ne.jp>

Tested-by: syzbot <syzbot+c55826...@syzkaller.appspotmail.com>

Cc: Sean Young <se...@mess.org>
Cc: Alan Stern <st...@rowland.harvard.edu>
---

Changes in v2:
Defer free_imon_context() using refcount till display_close() is called.

[Sean Young]

Hello,

Sorry for not responding earlier.

On Mon, May 02, 2022 at 12:49:04PM +0900, Tetsuo Handa wrote:
> Since usb_register_dev() from imon_init_display() from imon_probe() holds
> minor_rwsem while display_open() which holds driver_lock and ictx->lock is
> called with minor_rwsem held from usb_open(), holding driver_lock or
> ictx->lock when calling usb_register_dev() causes circular locking
> dependency problem.
>
> Since usb_deregister_dev() from imon_disconnect() holds minor_rwsem while
> display_open() which holds driver_lock is called with minor_rwsem held,
> holding driver_lock when calling usb_deregister_dev() also causes circular
> locking dependency problem.
>
> Sean Young explained that the problem is there are imon devices which have
> two usb interfaces, even though it is one device. The probe and disconnect
> function of both usb interfaces can run concurrently.
>
> Alan Stern responded that the driver and USB cores guarantee that when an
> interface is probed, both the interface and its USB device are locked.
> Ditto for when the disconnect callback gets run. So concurrent probing/
> disconnection of multiple interfaces on the same device is not possible.

So this part of the patch address the issue of driver_lock being
unnecessary. This should be in its own patch, so I am going to merge:

https://patchwork.linuxtv.org/project/linux-media/patch/349f3e34-41ed-f832...@I-love.SAKURA.ne.jp/

Is it possible to modify the users/disconnected fields while holding the
lock mutex in imon_context? This would make it unnecessary to use rcu and
simplify the code.

Thanks

Sean

[Tetsuo Handa]

On 2022/05/02 18:39, Sean Young wrote:
> So this part of the patch address the issue of driver_lock being
> unnecessary. This should be in its own patch, so I am going to merge:
>
> https://patchwork.linuxtv.org/project/linux-media/patch/349f3e34-41ed-f832...@I-love.SAKURA.ne.jp/

driver_lock is not unnecessary, unless combined with kfree_rcu() approach.

>> + /*
>> + * We need to wait for RCU grace period in order to allow
>> + * display_open() to safely check ->disconnected and increment ->users.
>> + */
>> + struct rcu_head rcu;
>
> Is it possible to modify the users/disconnected fields while holding the
> lock mutex in imon_context? This would make it unnecessary to use rcu and
> simplify the code.

I don't think it is possible, and I don't think it simplifies the code.

Unless we revive global driver_lock lock (untested delta diff is shown below),
nothing prevents from calling kfree(ictx) before holding ictx->lock or
checking ->disconnected or incrementing ->users.

As a whole, I think kfree_rcu() is simpler while removing global lock.

diff --git a/drivers/media/rc/imon.c b/drivers/media/rc/imon.c
index 9a4f24e294bc..469a2f869572 100644
--- a/drivers/media/rc/imon.c
+++ b/drivers/media/rc/imon.c
@@ -166,11 +166,6 @@ struct imon_context {

* imon_disconnect() was already called.

*/
bool disconnected;
- /*
- * We need to wait for RCU grace period in order to allow
- * display_open() to safely check ->disconnected and increment ->users.
- */
- struct rcu_head rcu;
};

#define TOUCH_TIMEOUT (HZ/30)
@@ -457,6 +452,9 @@ static struct usb_driver imon_driver = {
.id_table = imon_usb_id_table,
};

+/* to prevent races between open() and kfree() */
+static DEFINE_MUTEX(driver_lock);
+

/* Module bookkeeping bits */
MODULE_AUTHOR(MOD_AUTHOR);
MODULE_DESCRIPTION(MOD_DESC);

@@ -516,6 +514,8 @@ static int display_open(struct inode *inode, struct file *file)

int subminor;
int retval = 0;

+ mutex_lock(&driver_lock);
+

subminor = iminor(inode);
interface = usb_find_interface(&imon_driver, subminor);
if (!interface) {

@@ -524,15 +524,13 @@ static int display_open(struct inode *inode, struct file *file)
goto exit;
}

- rcu_read_lock();
ictx = usb_get_intfdata(interface);

if (!ictx || ictx->disconnected || !refcount_inc_not_zero(&ictx->users)) {

- rcu_read_unlock();

pr_err("no context found for minor %d\n", subminor);
retval = -ENODEV;
goto exit;
}

- rcu_read_unlock();
+ mutex_unlock(&driver_lock);

mutex_lock(&ictx->lock);

@@ -550,10 +548,15 @@ static int display_open(struct inode *inode, struct file *file)

mutex_unlock(&ictx->lock);

- if (retval && refcount_dec_and_test(&ictx->users))
- free_imon_context(ictx);
-
+ if (retval) {
+ mutex_unlock(&driver_lock);

+ if (refcount_dec_and_test(&ictx->users))
+ free_imon_context(ictx);

+ mutex_unlock(&driver_lock);
+ }
+ return retval;
exit:
+ mutex_unlock(&driver_lock);
return retval;
}

@@ -2497,6 +2500,8 @@ static void imon_disconnect(struct usb_interface *interface)

struct device *dev;
int ifnum;

+ mutex_lock(&driver_lock);
+
ictx = usb_get_intfdata(interface);

ictx->disconnected = true;
dev = ictx->dev;

@@ -2542,6 +2547,8 @@ static void imon_disconnect(struct usb_interface *interface)
if (refcount_dec_and_test(&ictx->users))
free_imon_context(ictx);

+ mutex_unlock(&driver_lock);
+

[Tetsuo Handa]

On 2022/05/02 19:26, Tetsuo Handa wrote:
> @@ -550,10 +548,15 @@ static int display_open(struct inode *inode, struct file *file)
>
> mutex_unlock(&ictx->lock);
>
> - if (retval && refcount_dec_and_test(&ictx->users))
> - free_imon_context(ictx);
> -
> + if (retval) {
> + mutex_unlock(&driver_lock);

Oops. This is mutex_lock(&driver_lock);.

> + if (refcount_dec_and_test(&ictx->users))
> + free_imon_context(ictx);
> + mutex_unlock(&driver_lock);
> + }
> + return retval;
> exit:
> + mutex_unlock(&driver_lock);
> return retval;
> }
>

But if you merely want to avoid adding "struct rcu_head rcu;" to "struct imon_context",
we can use synchronize_rcu(); kfree(ictx); sequence (at the cost of waiting for RCU
grace period inside free_imon_context()).

[Oliver Neukum]

On 02.05.22 05:49, Tetsuo Handa wrote:
Hi,

there is one open question with this patch I am afraid.

>
> @@ -533,8 +550,10 @@ static int display_open(struct inode *inode, struct file *file)
>
> mutex_unlock(&ictx->lock);
>
> + if (retval && refcount_dec_and_test(&ictx->users))
> + free_imon_context(ictx);
> +
>

When could this ever happen? Either the device is disconnected, then
you'll go to 'exit' or the refcount will go back to something >0, won't it?

    Regards
        Oliver

[Tetsuo Handa]

(Step 0) Say, ictx->users is initially 1.

(Step 1) display_open() increments via refcount_inc_not_zero(), now is 2.

(Step 2) imon_disconnect() decrements via refcount_dec_and_test(), now is 1.

(Step 3) if retval != 0, display_open() needs to undo (Step 1) via
refcount_dec_and_test(), now is 0.

because imon_disconnect() can be called while display_open() is in progress...

[Greg KH]

On Mon, May 02, 2022 at 12:49:04PM +0900, Tetsuo Handa wrote:

Are you sure this is going to work properly?

How do you handle userspace passing around file descriptors to other
processes?

You really should not ever have to count this.

Why not just put a kref into your larger structure?

I think trying to count users of open/close is never going to work, just
allow the normal open/close logic to work properly and track your data
structure based on reference counts like it should be doing already.

thanks,

greg k-h

[Tetsuo Handa]

On 2022/05/02 21:53, Greg KH wrote:
>> @@ -153,6 +153,24 @@ struct imon_context {
>> const struct imon_usb_dev_descr *dev_descr;
>> /* device description with key */
>> /* table for front panels */
>> + /*
>> + * Fields for deferring free_imon_context().
>> + *
>> + * Since reference to "struct imon_context" is stored into
>> + * "struct file_operations"->private_data, we need to remember
>> + * how many file descriptors might access this "struct imon_context".
>> + */
>> + refcount_t users;
>
> Are you sure this is going to work properly?
>
> How do you handle userspace passing around file descriptors to other
> processes?
>
> You really should not ever have to count this.

Passing around file descriptors results in nothing but delay of freeing memory.
Is this so fatal problem?

static void free_imon_context(struct imon_context *ictx)
{
struct device *dev = ictx->dev;

usb_free_urb(ictx->tx_urb);

usb_free_urb(ictx->rx_urb_intf0);
usb_free_urb(ictx->rx_urb_intf1);
kfree(ictx);

dev_dbg(dev, "%s: iMON context freed\n", __func__);
}

If this is so fatal, can we call usb_free_urb() upon imon_disconnect() ?

[Greg KH]

On Mon, May 02, 2022 at 10:04:22PM +0900, Tetsuo Handa wrote:
> On 2022/05/02 21:53, Greg KH wrote:
> >> @@ -153,6 +153,24 @@ struct imon_context {
> >> const struct imon_usb_dev_descr *dev_descr;
> >> /* device description with key */
> >> /* table for front panels */
> >> + /*
> >> + * Fields for deferring free_imon_context().
> >> + *
> >> + * Since reference to "struct imon_context" is stored into
> >> + * "struct file_operations"->private_data, we need to remember
> >> + * how many file descriptors might access this "struct imon_context".
> >> + */
> >> + refcount_t users;
> >
> > Are you sure this is going to work properly?
> >
> > How do you handle userspace passing around file descriptors to other
> > processes?
> >
> > You really should not ever have to count this.
>
> Passing around file descriptors results in nothing but delay of freeing memory.
> Is this so fatal problem?

Not at all, it's just that any driver that tries to count open calls
usually is wrong as they forget about this type of thing, which is why I
asked.

thanks,

greg k-h

[Tetsuo Handa]

On 2022/05/02 12:49, Tetsuo Handa wrote:
> @@ -153,6 +153,24 @@ struct imon_context {
> const struct imon_usb_dev_descr *dev_descr;
> /* device description with key */
> /* table for front panels */
> + /*
> + * Fields for deferring free_imon_context().
> + *
> + * Since reference to "struct imon_context" is stored into
> + * "struct file_operations"->private_data, we need to remember

Oops. This is "struct file"->private_data. Please correct when applying. ;-)

[Oliver Neukum]

Hi,

thank you. I did not think of error handling. You are correct.

    Regards
        Oliver