From patchwork Thu Mar 28 11:59:40 2013 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: David Herrmann X-Patchwork-Id: 2356181 Return-Path: X-Original-To: patchwork-linux-input@patchwork.kernel.org Delivered-To: patchwork-process-083081@patchwork2.kernel.org Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by patchwork2.kernel.org (Postfix) with ESMTP id A6E39DF2A1 for ; Thu, 28 Mar 2013 11:56:46 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1754513Ab3C1L4o (ORCPT ); Thu, 28 Mar 2013 07:56:44 -0400 Received: from mail-wi0-f175.google.com ([209.85.212.175]:59491 "EHLO mail-wi0-f175.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752588Ab3C1L4m (ORCPT ); Thu, 28 Mar 2013 07:56:42 -0400 Received: by mail-wi0-f175.google.com with SMTP id c10so3213343wiw.8 for ; Thu, 28 Mar 2013 04:56:41 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113; h=x-received:from:to:cc:subject:date:message-id:x-mailer; bh=7lXR0p3Z8ipcP6PvwaqeUy6sBYNjg2k4WTesP8cW+SE=; b=hMUs6UJIVfQj7uDK2HGcONScvHSMhK/iBSb6h3iM6fX5c+QB5SXS2YRkE6pzuebyKL GOE6Jl89rpZAByKKi+sCTQZTeCP2wVQ8f+QU6IOV27mtgXjGLyQUwTmF54JMT4IXjClY oDh+k7iIrtjEtWoz87I91GctjPX0pMB3h2sveAgWWPjUXTfLz/3YnJa/hNh2JqNE+z+2 gulTfx/2EThpQEyDUmbhL8drvuWNqXN6nZE7EGvVA4T6+uqK17bAXBobEMAhfsjWlC1n sZxFttS6ywg7LowOLpMIWiqgwhxuZdcDQVFGdgg6dhVkJcPKbsbLMZNv9PCNriZFCsTE 9JFg== X-Received: by 10.180.72.165 with SMTP id e5mr15669108wiv.7.1364471800920; Thu, 28 Mar 2013 04:56:40 -0700 (PDT) Received: from localhost.localdomain (stgt-5f71a373.pool.mediaWays.net. [95.113.163.115]) by mx.google.com with ESMTPS id fp2sm13817257wib.7.2013.03.28.04.56.38 (version=TLSv1.2 cipher=ECDHE-RSA-RC4-SHA bits=128/128); Thu, 28 Mar 2013 04:56:39 -0700 (PDT) From: David Herrmann To: linux-input@vger.kernel.org Cc: Dmitry Torokhov , David Herrmann Subject: [PATCH v3] Input: evdev - Flush queues during EVIOCGKEY-like ioctls Date: Thu, 28 Mar 2013 12:59:40 +0100 Message-Id: <1364471980-2426-1-git-send-email-dh.herrmann@gmail.com> X-Mailer: git-send-email 1.8.2 Sender: linux-input-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-input@vger.kernel.org If userspace requests current KEY-state, they very likely assume that no such events are pending in the output queue of the evdev device. Otherwise, they will parse events which they already handled via EVIOCGKEY(). For XKB applications this can cause irreversible keyboard states if a modifier is locked multiple times because a CTRL-DOWN event is handled once via EVIOCGKEY() and once from the queue via read(), even though it should handle it only once. Therefore, lets do the only logical thing and flush the evdev queue atomically during this ioctl. We only flush events that are affected by the given ioctl. This only affects boolean events like KEY, SND, SW and LED. ABS, REL and others are not affected as duplicate events can be handled gracefully by user-space. Note: This actually breaks semantics of the evdev ABI. However, investigations showed that userspace already expects the new semantics and we end up fixing at least all XKB applications. All applications that are aware of this race-condition mirror the KEY state for each open-file and detect/drop duplicate events. Hence, they do not care whether duplicates are posted or not and work fine with this fix. Also note that we need proper locking to guarantee atomicity and avoid dead-locks. event_lock must be locked before queue_lock (see input-core). However, we can safely release event_lock while flushing the queue. This allows the input-core to proceed with pending events and only stop if it needs our queue_lock to post new events. This should guarantee that we don't block event-dispatching for too long while flushing a single event queue. Signed-off-by: David Herrmann Acked-by: Peter Hutterer --- Hi Dmitry I hope this time everything is right. I tested this on my local machine with uinput. You can find the selftests here: https://gist.github.com/dvdhrm/5262589/raw/a3802ebfea0c319e42842664b0d7d1e7580197f1/inputtest.c With syntax-highlighting: https://gist.github.com/dvdhrm/5262589 I'm currently not entirely sure how the kernel selftests in tools/testing/selftests/ work so I haven't added it. However, feel free to add this. It's public domain. The script tests normal input-queue operation, SYN_DROPPED, SYN_REPORT, EVIOCGKEY+FLUSH, EVIOCKGEY+FLUSH+SYN_DROPPED. Everything worked as expected and the EVIOCGKEY tests fail without this patch applied (as expected). Thanks David drivers/input/evdev.c | 133 ++++++++++++++++++++++++++++++++++++++++++++++++-- 1 file changed, 129 insertions(+), 4 deletions(-) diff --git a/drivers/input/evdev.c b/drivers/input/evdev.c index f0f8928..e5ceebc 100644 --- a/drivers/input/evdev.c +++ b/drivers/input/evdev.c @@ -52,6 +52,82 @@ struct evdev_client { struct input_event buffer[]; }; +/* flush queued events of type @type, caller must hold client->buffer_lock */ +static void __flush_queue(struct evdev_client *client, unsigned int type) +{ + unsigned int i, head, num; + unsigned int mask = client->bufsize - 1; + bool is_report; + struct input_event *ev; + + BUG_ON(type == EV_SYN); + + head = client->tail; + client->packet_head = client->tail; + + /* init to 1 so a leading SYN_REPORT will not be dropped */ + num = 1; + + for (i = client->tail; i != client->head; i = (i + 1) & mask) { + ev = &client->buffer[i]; + is_report = ev->type == EV_SYN && ev->code == SYN_REPORT; + + if (ev->type == type) { + /* drop matched entry */ + continue; + } else if (is_report && !num) { + /* drop empty SYN_REPORT groups */ + continue; + } else if (head != i) { + /* move entry to fill the gap */ + client->buffer[head].time = ev->time; + client->buffer[head].type = ev->type; + client->buffer[head].code = ev->code; + client->buffer[head].value = ev->value; + } + + num++; + head = (head + 1) & mask; + + if (is_report) { + num = 0; + client->packet_head = head; + } + } + + client->head = head; +} + +/* queue SYN_DROPPED event */ +static void queue_syn_dropped(struct evdev_client *client) +{ + unsigned long flags; + struct input_event ev; + ktime_t time; + + time = ktime_get(); + if (client->clkid != CLOCK_MONOTONIC) + time = ktime_sub(time, ktime_get_monotonic_offset()); + + ev.time = ktime_to_timeval(time); + ev.type = EV_SYN; + ev.code = SYN_DROPPED; + ev.value = 0; + + spin_lock_irqsave(&client->buffer_lock, flags); + + client->buffer[client->head++] = ev; + client->head &= client->bufsize - 1; + + if (unlikely(client->head == client->tail)) { + /* drop queue but keep our SYN_DROPPED event */ + client->tail = (client->head - 1) & (client->bufsize - 1); + client->packet_head = client->tail; + } + + spin_unlock_irqrestore(&client->buffer_lock, flags); +} + static void __pass_event(struct evdev_client *client, const struct input_event *event) { @@ -650,6 +726,51 @@ static int evdev_handle_set_keycode_v2(struct input_dev *dev, void __user *p) return input_set_keycode(dev, &ke); } +/* + * If we transfer state to the user, we should flush all pending events + * from the client's queue. Otherwise, they might end up with duplicate + * events, which can screw up client's state tracking. + * If bits_to_user fails after flushing the queue, we queue a SYN_DROPPED event + * so user-space will notice missing events. + * + * LOCKING: + * We need to take event_lock before buffer_lock to avoid dead-locks. But we + * need the even_lock only to guarantee consistent state. We can safely release + * it while flushing the queue. This allows input-core to handle filters while + * we flush the queue. + */ +static int evdev_handle_get_val(struct evdev_client *client, + struct input_dev *dev, unsigned int type, + unsigned long *bits, unsigned int max, + unsigned int size, void __user *p, int compat) +{ + int ret; + unsigned long *mem; + + mem = kmalloc(sizeof(unsigned long) * max, GFP_KERNEL); + if (!mem) + return -ENOMEM; + + spin_lock_irq(&dev->event_lock); + spin_lock(&client->buffer_lock); + + memcpy(mem, bits, sizeof(unsigned long) * max); + + spin_unlock(&dev->event_lock); + + __flush_queue(client, type); + + spin_unlock_irq(&client->buffer_lock); + + ret = bits_to_user(mem, max, size, p, compat); + if (ret < 0) + queue_syn_dropped(client); + + kfree(mem); + + return ret; +} + static int evdev_handle_mt_request(struct input_dev *dev, unsigned int size, int __user *ip) @@ -771,16 +892,20 @@ static long evdev_do_ioctl(struct file *file, unsigned int cmd, return evdev_handle_mt_request(dev, size, ip); case EVIOCGKEY(0): - return bits_to_user(dev->key, KEY_MAX, size, p, compat_mode); + return evdev_handle_get_val(client, dev, EV_KEY, dev->key, + KEY_MAX, size, p, compat_mode); case EVIOCGLED(0): - return bits_to_user(dev->led, LED_MAX, size, p, compat_mode); + return evdev_handle_get_val(client, dev, EV_LED, dev->led, + LED_MAX, size, p, compat_mode); case EVIOCGSND(0): - return bits_to_user(dev->snd, SND_MAX, size, p, compat_mode); + return evdev_handle_get_val(client, dev, EV_SND, dev->snd, + SND_MAX, size, p, compat_mode); case EVIOCGSW(0): - return bits_to_user(dev->sw, SW_MAX, size, p, compat_mode); + return evdev_handle_get_val(client, dev, EV_SW, dev->sw, + SW_MAX, size, p, compat_mode); case EVIOCGNAME(0): return str_to_user(dev->name, size, p);