linux-kernel - Re: [PATCH 16/20] futex: Implement sys_futex

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [day] [month] [year] [list]

Date:   Wed, 15 Sep 2021 13:29:30 -0300
From:   André Almeida <andrealmeid@...labora.com>
To:     Peter Zijlstra <peterz@...radead.org>
Cc:     linux-kernel@...r.kernel.org, kernel@...labora.com,
        bigeasy@...utronix.de, krisman@...labora.com,
        linux-api@...r.kernel.org, libc-alpha@...rceware.org,
        mingo@...hat.com, rostedt@...dmis.org, dvhart@...radead.org,
        mtk.manpages@...il.com, dave@...olabs.net, arnd@...db.de,
        tglx@...utronix.de
Subject: Re: [PATCH 16/20] futex: Implement sys_futex_waitv()

Às 11:07 de 15/09/21, Peter Zijlstra escreveu:
> From: André Almeida <andrealmeid@...labora.com>
> 
> Add support to wait on multiple futexes. This is the interface
> implemented by this syscall:
> 
> futex_waitv(struct futex_waitv *waiters, unsigned int nr_futexes,
> 	    unsigned int flags, struct timespec *timo)
> 
> +/**
> + * futex_wait_multiple_setup - Prepare to wait and enqueue multiple futexes
> + * @vs:		The futex list to wait on
> + * @count:	The size of the list
> + * @awaken:	Index of the last awoken futex, if any. Used to notify the
> + *		caller that it can return this index to userspace (return parameter)
> + *
> + * Prepare multiple futexes in a single step and enqueue them. This may fail if
> + * the futex list is invalid or if any futex was already awoken. On success the
> + * task is ready to interruptible sleep.
> + *
> + * Return:
> + *  -  1 - One of the futexes was awaken by another thread
> + *  -  0 - Success
> + *  - <0 - -EFAULT, -EWOULDBLOCK or -EINVAL
> + */
> +static int futex_wait_multiple_setup(struct futex_vector *vs, int count, int *awaken)
> +{
> +	struct futex_hash_bucket *hb;
> +	bool retry = false;
> +	int ret, i;
> +	u32 uval;
> +
> +	/*
> +	 * Enqueuing multiple futexes is tricky, because we need to enqueue
> +	 * each futex in the list before dealing with the next one to avoid
> +	 * deadlocking on the hash bucket. But, before enqueuing, we need to
> +	 * make sure that current->state is TASK_INTERRUPTIBLE, so we don't
> +	 * absorb any awake events, which cannot be done before the
> +	 * get_futex_key of the next key, because it calls get_user_pages,
> +	 * which can sleep. Thus, we fetch the list of futexes keys in two
> +	 * steps, by first pinning all the memory keys in the futex key, and
> +	 * only then we read each key and queue the corresponding futex.
> +	 *
> +	 * Private futexes doesn't need to recalculate hash in retry, so skip
> +	 * get_futex_key() when retrying.
> +	 */
> +retry:
> +	for (i = 0; i < count; i++) {
> +		if ((vs[i].w.flags & FUTEX_PRIVATE_FLAG) && retry)
> +			continue;
> +
> +		ret = get_futex_key(u64_to_user_ptr(vs[i].w.uaddr),
> +				    !(vs[i].w.flags & FUTEX_PRIVATE_FLAG),
> +				    &vs[i].q.key, FUTEX_READ);
> +
> +		if (unlikely(ret))
> +			return ret;
> +	}
> +
> +	set_current_state(TASK_INTERRUPTIBLE);
> +
> +	for (i = 0; i < count; i++) {
> +		u32 __user *uaddr = (u32 __user *)(unsigned long)vs[i].w.uaddr;
> +		struct futex_q *q = &vs[i].q;
> +		u32 val = (u32)vs[i].w.val;
> +
> +		hb = futex_q_lock(q);
> +		ret = futex_get_value_locked(&uval, uaddr);
> +
> +		if (!ret && uval == val) {
> +			/*
> +			 * The bucket lock can't be held while dealing with the
> +			 * next futex. Queue each futex at this moment so hb can
> +			 * be unlocked.
> +			 */
> +			futex_queue(q, hb);
> +			continue;
> +		}
> +
> +		futex_q_unlock(hb);
> +		__set_current_state(TASK_RUNNING);
> +
> +		/*
> +		 * Even if something went wrong, if we find out that a futex
> +		 * was awaken, we don't return error and return this index to
> +		 * userspace
> +		 */
> +		*awaken = unqueue_multiple(vs, i);
> +		if (*awaken >= 0)
> +			return 1;
> +
> +		if (uval != val)
> +			return -EWOULDBLOCK;
> +
> +		if (ret) {
> +			/*
> +			 * If we need to handle a page fault, we need to do so
> +			 * without any lock and any enqueued futex (otherwise
> +			 * we could lose some wakeup). So we do it here, after
> +			 * undoing all the work done so far. In success, we
> +			 * retry all the work.
> +			 */
> +			if (get_user(uval, uaddr))
> +				return -EFAULT;
> +
> +			retry = true;
> +			goto retry;
> +		}

My bad again: the last two if's should be in the reserve order. If ret
!= 0, the user copy didn't succeed and the value wasn't copied to uval,
thus the comparison (uval != val) should happen only if ret == 0.


> +	}
> +
> +	return 0;
> +}