Re: [PATCH] tty: fix data race in n_tty_receive_buf_common

From: Kohli, Gaurav
Date: Fri Jan 05 2018 - 15:28:24 EST


HI Alan,

Sorry correcting the typo here:


On 1/6/2018 1:44 AM, Kohli, Gaurav wrote:
Hi Alan,


On 1/5/2018 7:45 PM, Alan Cox wrote:
But in above case , there we can hit another race, if we have a sequence
like this
tty_init_dev->alloc_tty_struct -> tty_ldisc_init -> this will initialize
ldisc ,
but at this moment disc_data is still NULL

And if flush_to_ldisc comes in between, it will take ldisc reference and
proceeds receive buffer.
So you need to move the lock up one line to protect the assignment to
tty->port->itty. We can do that.

At that point your flush_to_ldisc should see either port->itty = NULL or a
valid initialized ldisc.



Yes , with little modification this should work.

+retval =Â tty_ldisc_lock(tty, 5 * HZ);
+if (retval)
+ÂÂÂÂ ÂÂÂ goto err_release_lock;
tty->port->itty = tty;
/*
* Structures all installed ... call the ldisc open routines.
* If we fail here just call release_tty to clean up. No need
* to decrement the use counts, as release_tty doesn't care.
*/
retval = tty_ldisc_setup(tty, tty->link);
ÂÂÂ ÂÂÂ if (retval)
ÂÂÂ ÂÂÂ ÂÂÂ goto err_release_tty;
tty_ldisc_unlock(tty);
err_release_tty:
tty_info_ratelimited(tty, "ldisc open failed (%d), clearing slot %d\n",
ÂÂÂ retval, idx);
+err_release_lock;
ÂÂÂ +tty_unlock(tty);
ÂÂÂ +release_tty(tty, idx);
+tty_ldisc_unlock(tty);
+return ERR_PTR(retval);

Please let me know if above modification seems fine , then we can upload this and try reproduction of Bug.

Regards
Gaurav


--
Qualcomm India Private Limited, on behalf of Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum,
a Linux Foundation Collaborative Project.