From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-10.0 required=3.0 tests=BAYES_00,DKIM_SIGNED, DKIM_VALID,DKIM_VALID_AU,FREEMAIL_FORGED_FROMDOMAIN,FREEMAIL_FROM, HEADER_FROM_DIFFERENT_DOMAINS,INCLUDES_PATCH,MAILING_LIST_MULTI,SIGNED_OFF_BY, SPF_HELO_NONE,SPF_PASS,URIBL_BLOCKED autolearn=unavailable autolearn_force=no version=3.4.0 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id E9762C433E3 for ; Sat, 25 Jul 2020 17:45:27 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by mail.kernel.org (Postfix) with ESMTP id BCDD7206D7 for ; Sat, 25 Jul 2020 17:45:27 +0000 (UTC) Authentication-Results: mail.kernel.org; dkim=pass (2048-bit key) header.d=gmail.com header.i=@gmail.com header.b="AkfC4AmE" Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1727910AbgGYRp0 (ORCPT ); Sat, 25 Jul 2020 13:45:26 -0400 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:46480 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1726995AbgGYRpZ (ORCPT ); Sat, 25 Jul 2020 13:45:25 -0400 Received: from mail-wr1-x444.google.com (mail-wr1-x444.google.com [IPv6:2a00:1450:4864:20::444]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 64CB7C08C5C0 for ; Sat, 25 Jul 2020 10:45:24 -0700 (PDT) Received: by mail-wr1-x444.google.com with SMTP id y3so11042791wrl.4 for ; Sat, 25 Jul 2020 10:45:24 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20161025; h=date:from:to:cc:subject:message-id:references:mime-version :content-disposition:in-reply-to; bh=IbabLhexFI8pOt8xU63JQH30ITPxZNOZxsnTB/GJm18=; b=AkfC4AmE2FbyLy5BfpzsvgLRd0zuSKadnjM88dORGkeDJDmxsKMBzrzq+dMKCikUZ0 itvsTnF2RH4+6HmbZbAb4ZAB14SdyF0ifnQlXSw/DprDnBshRBQ2AC62zpupmE55GESK HolAYBW3oaa4FJUahR0RpnOLEZ6uK/EtbRsi14sOwJrc7zoKPQlftgheZXsJ50xDSHK1 +G1IWYKbosurdWfT7eWZ4qrhpPGWpyMwYJSNGU/tlRgJghMnbzvg0oXV2IPq6Ebczcq6 8+EZOLqI8kEccWPO0cky8TDVuMi/vWwV1LOdLPqu20WkCSf9Z/HV/mLWYKc9H+z/BdPY 5b+Q== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:date:from:to:cc:subject:message-id:references :mime-version:content-disposition:in-reply-to; bh=IbabLhexFI8pOt8xU63JQH30ITPxZNOZxsnTB/GJm18=; b=GQAARhycOkQtDrKvO4zVbCYYPSJJPCrpJGoA1YXuiOy00MfRv8UfyVM+curKSqs4Qc uGjHRPE/SKrOtqsHiAzAzngXpu107hidkSQRFCAFKTukEjxdkDrhitlE2qFhX6LIxESF 77zysPr8n3hqlUJjVsMpO0Sgn4H/rIEI0VOoM35Bza7MB1ShvCBU5n3HIQ1P4nIu2aqN 5jVaD9lUX10t2N6OOBgfLbzc8lF95oz0ifCiFDKQ0DeiMN3gorfRJZ9tmsHmpUMQaULz 8Yba+lmzi5tHajsYNuZTgcnmbElfP8lPqhLFeb9Fgg8G5MsNE3KFbMe8cX7PNn/Ua/xX j4ug== X-Gm-Message-State: AOAM531KfOE5PJpxnYdyU+J3Uw9oaMLTi0tyL5sHNwhSw9WLDOTZLLrz 0MpPiKsSyYEesaEBP5oD6X1OsCYJ X-Google-Smtp-Source: ABdhPJwBWNZQW0ZxAaF3mpFlVk3Hk6HJNDMQf3XpxGFT860FKVZe+RqqSPfGSETZx6oiyIzCR2Tk/g== X-Received: by 2002:adf:bb07:: with SMTP id r7mr7760304wrg.102.1595699123058; Sat, 25 Jul 2020 10:45:23 -0700 (PDT) Received: from smtp.gmail.com (a95-92-181-29.cpe.netcabo.pt. [95.92.181.29]) by smtp.gmail.com with ESMTPSA id z15sm4966128wrn.89.2020.07.25.10.45.21 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Sat, 25 Jul 2020 10:45:22 -0700 (PDT) Date: Sat, 25 Jul 2020 14:45:14 -0300 From: Melissa Wen To: Daniel Vetter Cc: Sidong Yang , Haneen Mohammed , Rodrigo Siqueira , David Airlie , Rodrigo Siqueira , Linux Kernel Mailing List , dri-devel , kernel-usp@googlegroups.com Subject: Re: [PATCH] drm/vkms: add missing drm_crtc_vblank_put to the get/put pair on flush Message-ID: <20200725174514.esh4gqek6oddizvh@smtp.gmail.com> References: <20200722110411.ebkv6knvc6mzw5uf@smtp.gmail.com> <20200722120502.GK6419@phenom.ffwll.local> <20200722140604.27dfzfnzug5vb75r@smtp.gmail.com> <20200725011737.GA2851@realwakka> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: Sender: linux-kernel-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 07/25, Daniel Vetter wrote: > On Sat, Jul 25, 2020 at 5:12 AM Sidong Yang wrote: > > > > On Wed, Jul 22, 2020 at 05:17:05PM +0200, Daniel Vetter wrote: > > > On Wed, Jul 22, 2020 at 4:06 PM Melissa Wen wrote: > > > > > > > > On 07/22, daniel@ffwll.ch wrote: > > > > > On Wed, Jul 22, 2020 at 08:04:11AM -0300, Melissa Wen wrote: > > > > > > This patch adds a missing drm_crtc_vblank_put op to the pair > > > > > > drm_crtc_vblank_get/put (inc/decrement counter to guarantee vblanks). > > > > > > > > > > > > It clears the execution of the following kms_cursor_crc subtests: > > > > > > 1. pipe-A-cursor-[size,alpha-opaque, NxN-(on-screen, off-screen, sliding, > > > > > > random, fast-moving])] - successful when running individually. > > > > > > 2. pipe-A-cursor-dpms passes again > > > > > > 3. pipe-A-cursor-suspend also passes > > > > > > > > > > > > The issue was initially tracked in the sequential execution of IGT > > > > > > kms_cursor_crc subtest: when running the test sequence or one of its > > > > > > subtests twice, the odd execs complete and the pairs get stuck in an > > > > > > endless wait. In the IGT code, calling a wait_for_vblank before the start > > > > > > of CRC capture prevented the busy-wait. But the problem persisted in the > > > > > > pipe-A-cursor-dpms and -suspend subtests. > > > > > > > > > > > > Checking the history, the pipe-A-cursor-dpms subtest was successful when, > > > > > > in vkms_atomic_commit_tail, instead of using the flip_done op, it used > > > > > > wait_for_vblanks. Another way to prevent blocking was wait_one_vblank when > > > > > > enabling crtc. However, in both cases, pipe-A-cursor-suspend persisted > > > > > > blocking in the 2nd start of CRC capture, which may indicate that > > > > > > something got stuck in the step of CRC setup. Indeed, wait_one_vblank in > > > > > > the crc setup was able to sync things and free all kms_cursor_crc > > > > > > subtests. > > > > > > > > > > > > Tracing and comparing a clean run with a blocked one: > > > > > > - in a clean one, vkms_crtc_atomic_flush enables vblanks; > > > > > > - when blocked, only in next op, vkms_crtc_atomic_enable, the vblanks > > > > > > started. Moreover, a series of vkms_vblank_simulate flow out until > > > > > > disabling vblanks. > > > > > > Also watching the steps of vkms_crtc_atomic_flush, when the very first > > > > > > drm_crtc_vblank_get returned an error, the subtest crashed. On the other > > > > > > hand, when vblank_get succeeded, the subtest completed. Finally, checking > > > > > > the flush steps: it increases counter to hold a vblank reference (get), > > > > > > but there isn't a op to decreased it and release vblanks (put). > > > > > > > > > > > > Cc: Daniel Vetter > > > > > > Cc: Rodrigo Siqueira > > > > > > Cc: Haneen Mohammed > > > > > > Signed-off-by: Melissa Wen > > > > > > --- > > > > > > drivers/gpu/drm/vkms/vkms_crtc.c | 1 + > > > > > > 1 file changed, 1 insertion(+) > > > > > > > > > > > > diff --git a/drivers/gpu/drm/vkms/vkms_crtc.c b/drivers/gpu/drm/vkms/vkms_crtc.c > > > > > > index ac85e17428f8..a99d6b4a92dd 100644 > > > > > > --- a/drivers/gpu/drm/vkms/vkms_crtc.c > > > > > > +++ b/drivers/gpu/drm/vkms/vkms_crtc.c > > > > > > @@ -246,6 +246,7 @@ static void vkms_crtc_atomic_flush(struct drm_crtc *crtc, > > > > > > > > > > > > spin_unlock(&crtc->dev->event_lock); > > > > > > > > > > > > + drm_crtc_vblank_put(crtc); > > > > > > > > > > Uh so I reviewed this a bit more carefully now, and I dont think this is > > > > > the correct bugfix. From the kerneldoc of drm_crtc_arm_vblank_event(): > > > > > > > > > > * Caller must hold a vblank reference for the event @e acquired by a > > > > > * drm_crtc_vblank_get(), which will be dropped when the next vblank arrives. > > > > > > > > > > So when we call drm_crtc_arm_vblank_event then the vblank_put gets called > > > > > for us. And that's the only case where we successfully acquired a vblank > > > > > interrupt reference since on failure of drm_crtc_vblank_get (0 indicates > > > > > success for that function, failure negative error number) we directly send > > > > > out the event. > > > > > > > > > > So something else fishy is going on, and now I'm totally confused why this > > > > > even happens. > > > > > > > > > > We also have a pile of WARN_ON checks in drm_crtc_vblank_put to make sure > > > > > we don't underflow the refcount, so it's also not that I think (except if > > > > > this patch creates more WARNING backtraces). > > > > > > > > > > But clearly it changes behaviour somehow ... can you try to figure out > > > > > what changes? Maybe print out the vblank->refcount at various points in > > > > > the driver, and maybe also trace when exactly the fake vkms vblank hrtimer > > > > > is enabled/disabled ... > > > > > > > > :( > > > > > > > > I can check these, but I also have other suspicions. When I place the > > > > drm_crct_vblank_put out of the if (at the end of flush), it not only solve > > > > the issue of blocking on kms_cursor_crc, but also the WARN_ON on kms_flip > > > > doesn't appear anymore (a total cleanup). Just after: > > > > > > > > vkms_output->composer_state = to_vkms_crtc_state(crtc->state); > > > > > > > > looks like there is something stuck around here. > > > > > > Hm do you have the full WARNING for this? Maybe this gives me an idea > > > what's going wrong. > > > > > > > Besides, there is a lock at atomic_begin: > > > > > > > > /* This lock is held across the atomic commit to block vblank timer > > > > * from scheduling vkms_composer_worker until the composer is updated > > > > */ > > > > spin_lock_irq(&vkms_output->lock); > > > > > > > > that seems to be released on atomic_flush and make me suspect something > > > > missing on the composer update. > > > > > > atomic_begin/atomic_flush are symmetric functions an always called > > > around all the plane updates. So having the spin_lock in _begin and > > > the spin_unlock in _flush should be symmetric and correct. > > > > > > If you want to make sure, recompile with CONFIG_PROVE_LOCKING, which > > > should immmediately give you a huge splat in dmesg if there's anything > > > unbalanced with locking. > > > > > > > I'll check all these things and come back with news (hope) :) > > > > > > Have fun chasing stuff :-) > > > > > > Cheers, Daniel > > > > > > > > > > > > > > Thanks, > > > > > > > > Melissa > > > > > > > > > > I'm totally confused about what's going on here now. > > > > > -Daniel > > > > Hi Daniel, Melissa. > > I found something about this problem. > > I traced vblank->refcount that it's important in the problem. > > In normal case, first test run calls commit_tail() and enable vblank in > > atomic_flush(). in drm_vblank_get(), it enable vblank when refcount was zero. > > > > in first test run, it disable crtc for cleanup test. drm_crtc_vblank_off() was > > called by atomic_disable. in this function vblank's refcount was increased for > > prevent subsequent drm_vblank_get() from re-enabling the vblank interrupt. > > and refcount goes one not zero for next test run. > > > > and next test run, drm_vblank_get() was called but it didn't enable vblank > > because refcount was already one. drm_crtc_vblank_on() was called in next. but > > it didn't enable vblank but just increase refcount only. > > > > I think this is why this problem happen. don't know how to fix this correctly. > > should we force to enable vblank after enabling crtc? > > Hm, between drm_crtc_vblank_off and drm_crtc_vblank_on > drm_crtc_vblank_get should fail (and leave the refcount unchanged). > It's convoluted logic, but the check for vblank->enabled should catch > that and return -EINVAL for this case. Does that not happen? > > It would indeed explain the bug (I think, I've been wrong way too many > times with this). > -Daniel > Hi Daniel and Sidong, I don't know if it will be confusing, but I will try to explain in a little more detail (and newbie way) what I saw in this behavior of the refcount (similar to what Sidong evaluated). 1. Starting with the loading of vkms is: In vkms_init: After drm_vblank_init (refcount=0), it calls: vkms_modeset_init --> vkms_output_init ----> drm_mode_config_reset -------> vkms_atomic_crtc_reset (even more inside)--> drm_crtc_vblank_reset that bumps the refcount to prevent vblank_get to enable vblank (refcount=1) 2. So, when we start a subtest, vblank is still disabled and in commit_tail, commit_planes triggers a atomic_begin/flush->vblank_get that return -EINVAL because !vblank->enabled (refcount ends 1) and send_vblank; however the test fails before atomic_enable decrements refcount to 0 and reset timestamp. ** This warning also appears in this very first running: WARNING: CPU: 0 PID: 708 at drivers/gpu/drm/vkms/vkms_crtc.c:91 vkms_get_vblank_timestamp+0x41/0x50 [vkms] In the end, this sequence modeset_disable -> atomic_begin -> atomic_flush: refcount going from 0 to 1 and than drm_vblank_enable prepares to everything going well in the next subtest (because atomic_disable is not called). 3. It could be nice, but in the next subtest (with success), as refcount + vblank_enabled ok, after doind its job, it calls atomic_disable->vblank_off and here refcount ends 1 and vblank disabled (the problem returns). So, we have a kind of good turn and bad turn. I tried different things, but the only relatively stable result was putting the sequence modeset_disable + modeset_enables + commit_planes in the commit_tail. That didn't convince me and then I keep trying things. > > > > Thanks > > -Sidong > > > > > > > > > > > > > crtc->state->event = NULL; > > > > > > } > > > > > > > > > > > > -- > > > > > > 2.27.0 > > > > > > > > > > > > > > > > -- > > > > > Daniel Vetter > > > > > Software Engineer, Intel Corporation > > > > > http://blog.ffwll.ch > > > > > > > > > > > > -- > > > Daniel Vetter > > > Software Engineer, Intel Corporation > > > http://blog.ffwll.ch > > > _______________________________________________ > > > dri-devel mailing list > > > dri-devel@lists.freedesktop.org > > > https://lists.freedesktop.org/mailman/listinfo/dri-devel > > > > -- > Daniel Vetter > Software Engineer, Intel Corporation > http://blog.ffwll.ch