Skip to contentSkip to conversation

tty: tty_open can return to userspace holding tty_mutex

PATCH Eric Paris 4 participants 4 messages Jan 26, 2009 Char and misc drivers

lore
Share & tools
lore

Eric Paris

author
#0 lore raw

__tty_open could return (to userspace) holding the tty_mutex thanks to a regression introduced by 4a2b5fddd53b80efcb3266ee36e23b8de28e761a. This was found by bisecting an fsfuzzer problem. Admittedly I have no idea how it managed to tickle this 100% reliably, but it is clearly a regression and when hit leaves the box in a completely unusable state. This patch lets the fsfuzzer test complete every time.

---
 drivers/char/tty_io.c |    4 +++-
 1 files changed, 3 insertions(+), 1 deletions(-)
drivers/char/tty_io.c +3 -1
index d33e5ab..bc84e12 100644
@@ -1817,8 +1817,10 @@ got_driver:
 		/* check whether we're reopening an existing tty */
 		tty = tty_driver_lookup_tty(driver, inode, index);
 
-		if (IS_ERR(tty))
+		if (IS_ERR(tty)) {
+			mutex_unlock(&tty_mutex);
 			return PTR_ERR(tty);
+		}
 	}
 
 	if (tty) {
#1 lore raw
Quoted 6 lines · On Mon, 26 Jan 2009 16:02:40 -0500 Eric Paris <[email protected]> wrote:

__tty_open could return (to userspace) holding the tty_mutex thanks to a regression introduced by 4a2b5fddd53b80efcb3266ee36e23b8de28e761a. This was found by bisecting an fsfuzzer problem. Admittedly I have no idea how it managed to tickle this 100% reliably, but it is clearly a regression and when hit leaves the box in a completely unusable state. This patch lets the fsfuzzer test complete every time.

Well spotted, and will go fix now. I think your fsfuzzer created a device node for an out of range tty subdevice

#2 lore raw
In reply to Eric Paris
Quoted 6 lines · Eric Paris

__tty_open could return (to userspace) holding the tty_mutex thanks to a regression introduced by 4a2b5fddd53b80efcb3266ee36e23b8de28e761a. This was found by bisecting an fsfuzzer problem. Admittedly I have no idea how it managed to tickle this 100% reliably, but it is clearly a regression and when hit leaves the box in a completely unusable state. This patch lets the fsfuzzer test complete every time.

This should go to -stable too, right?
It's present in 2.6.28 (but not 2.6.27).

/mjt

Quoted 22 lines
---
 drivers/char/tty_io.c |    4 +++-
 1 files changed, 3 insertions(+), 1 deletions(-)
drivers/char/tty_io.c +3 -1
index d33e5ab..bc84e12 100644
@@ -1817,8 +1817,10 @@ got_driver:
 		/* check whether we're reopening an existing tty */
 		tty = tty_driver_lookup_tty(driver, inode, index);
 
-		if (IS_ERR(tty))
+		if (IS_ERR(tty)) {
+			mutex_unlock(&tty_mutex);
 			return PTR_ERR(tty);
+		}
 	}
 
 	if (tty) {
#3 lore raw
In reply to Eric Paris

Eric Paris [[email protected]] wrote:

| __tty_open could return (to userspace) holding the tty_mutex thanks to a
| regression introduced by 4a2b5fddd53b80efcb3266ee36e23b8de28e761a.  This was
| found by bisecting an fsfuzzer problem.  Admittedly I have no idea how it
| managed to tickle this 100% reliably, but it is clearly a regression and
| when hit leaves the box in a completely unusable state.  This patch lets
| the fsfuzzer test complete every time.

Good catch.

|
| Signed-off-by: Eric Paris <[email protected]>

| 
| ---
|  drivers/char/tty_io.c |    4 +++-
|  1 files changed, 3 insertions(+), 1 deletions(-)
| 
| diff --git a/drivers/char/tty_io.c b/drivers/char/tty_io.c
| index d33e5ab..bc84e12 100644
| --- a/drivers/char/tty_io.c
| +++ b/drivers/char/tty_io.c
| @@ -1817,8 +1817,10 @@ got_driver:
|  		/* check whether we're reopening an existing tty */
|  		tty = tty_driver_lookup_tty(driver, inode, index);
| 
| -		if (IS_ERR(tty))
| +		if (IS_ERR(tty)) {
| +			mutex_unlock(&tty_mutex);
|  			return PTR_ERR(tty);
| +		}
|  	}
| 
|  	if (tty) {
|