Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id ACBD7C433FE for ; Mon, 29 Nov 2021 21:24:32 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S231331AbhK2V1t (ORCPT ); Mon, 29 Nov 2021 16:27:49 -0500 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:44186 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S230440AbhK2VZs (ORCPT ); Mon, 29 Nov 2021 16:25:48 -0500 Received: from mail-pj1-x1036.google.com (mail-pj1-x1036.google.com [IPv6:2607:f8b0:4864:20::1036]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id A0F1AC214FCD for ; Mon, 29 Nov 2021 11:36:04 -0800 (PST) Received: by mail-pj1-x1036.google.com with SMTP id iq11so13495747pjb.3 for ; Mon, 29 Nov 2021 11:36:04 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gateworks-com.20210112.gappssmtp.com; s=20210112; h=mime-version:references:in-reply-to:from:date:message-id:subject:to :cc:content-transfer-encoding; bh=xohg7JdVvkvO2SUDiXwKowHodphNxP500HwNzMb+bfU=; b=3mJajd4V1DdvWtEF+wYEzCoqrnMMQRznFjD4nvIf9I7+pySjX6zaxFf/1dX6yvFAVo Ud79y+0FkUQvZzWVEB4fcm6/U8zeGg/8/7JodJ4XQYY/HzLufwHcVWafaORCNlApyIsg jmsV9GSaaqh5LWKI/Zr9+xCY4BkfnQzJ2kxrV54UnaNnob6DE9lZ0N7bJI3SeLWeGQL3 n8zH1ni2jqgGratb3Ipr0t3zw+MCdIgUFiEsuDtJ5gjs5e3aSnvyWBCXApLwNVAXQc7Z XVSWfMm6umx0mJmJPaTWV9WDLvRagwtJ8Z++9FtebkJ+kLJGr9DohAH5v1jNl39MKijx imTA== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20210112; h=x-gm-message-state:mime-version:references:in-reply-to:from:date :message-id:subject:to:cc:content-transfer-encoding; bh=xohg7JdVvkvO2SUDiXwKowHodphNxP500HwNzMb+bfU=; b=vJEmKs4B9RV0WH9EQn7agmgx7OX3PxeMrRj3JvSct5wI/Rq2h+vjuulL9WXc1yDIbM 2t+FJHC5ri/OrPkFp4qFWkJGxAhH7XNZ9i1Fx0THdtaNNsZcs8O1c6gdnIS+Q6fi9Lmn 5xh0xducJspU6TFc/xNB/SbeeNqB6slkklH5oMsnYQh5BojHDD5msTrbCWloN2Y8S3/C +r3t6HG7uQ/hGEiQ6R7xHby/tIvbPWpwN2tMzqME3c8gW11on20vRuv+i6abDUobIHBz lSiTQun5zNHFbHWeCqc8PCuvWvgG73mWV2QQh/P1A5kKWelNDwVnJ7jnUcEckaqDowY4 qVew== X-Gm-Message-State: AOAM532M0dXN4EZk/WEmIhZaE2Gc6HTr6kx03fT1469bkWJZV/dDRuYY jFpuk+VShr5lxELNlz92TcjCCJC/ypzQ3+CeW3G4xw== X-Google-Smtp-Source: ABdhPJxuNaUbjV4bLmcxiXLKB+HIvSn9asc0AxMeUnpXUU1L5OoHO+ZOt1JvoDAaN4JiDPsEAouzXlKexW49JuVqeFU= X-Received: by 2002:a17:902:7c8a:b0:143:bb4a:7bb3 with SMTP id y10-20020a1709027c8a00b00143bb4a7bb3mr63117830pll.46.1638214563723; Mon, 29 Nov 2021 11:36:03 -0800 (PST) MIME-Version: 1.0 References: <20211106183802.893285-1-aford173@gmail.com> <718f7f6d6cd564d031c1963f1590c62d549ae725.camel@ndufresne.ca> <8db00a4b6faa99c940d9bc86e17161eb0db5efe3.camel@ndufresne.ca> <7f94eaacfddb8c5434c17f1e069ea87a17657ce9.camel@ndufresne.ca> In-Reply-To: From: Tim Harvey Date: Mon, 29 Nov 2021 11:35:52 -0800 Message-ID: Subject: Re: [RFC 0/5] arm64: imx8mm: Enable Hantro VPUs To: Adam Ford Cc: Ezequiel Garcia , Nicolas Dufresne , linux-media , Schrempf Frieder , Marek Vasut , Jagan Teki , Adam Ford-BE , cstevens@beaconembedded.com, Philipp Zabel , Mauro Carvalho Chehab , Rob Herring , Shawn Guo , Sascha Hauer , Pengutronix Kernel Team , Fabio Estevam , NXP Linux Team , Greg Kroah-Hartman , Heiko Stuebner , Lucas Stach , Joakim Zhang , Alice Guo , Peng Fan , "open list:HANTRO VPU CODEC DRIVER" , "open list:OPEN FIRMWARE AND FLATTENED DEVICE TREE BINDINGS" , "moderated list:ARM/FREESCALE IMX / MXC ARM ARCHITECTURE" , open list , "open list:STAGING SUBSYSTEM" Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Mon, Nov 29, 2021 at 10:59 AM Adam Ford wrote: > > On Mon, Nov 29, 2021 at 10:54 AM Ezequiel Garcia > wrote: > > > > On Mon, 29 Nov 2021 at 13:48, Adam Ford wrote: > > > > > > On Tue, Nov 23, 2021 at 2:07 PM Nicolas Dufresne wrote: > > > > > > > > Le lundi 22 novembre 2021 =C3=A0 09:25 -0800, Tim Harvey a =C3=A9cr= it : > > > > > On Sat, Nov 20, 2021 at 7:36 AM Adam Ford wr= ote: > > > > > > > > > > > > On Fri, Nov 19, 2021 at 5:37 PM Adam Ford = wrote: > > > > > > > > > > > > > > On Fri, Nov 19, 2021 at 10:29 AM Nicolas Dufresne wrote: > > > > > > > > > > > > > > > > Hi Adam, Tim, > > > > > > > > > > > > > > > > [...] > > > > > > > > > > > > Nicolas and Adam, > > > > > > > > > > > > > > > > > > > > > > > > For the H1 patches in this series: I've been able t= o test the IMX8MM > > > > > > > > > > > > H1 JPEG encode using GStreamer 1.18.5: > > > > > > > > > > > > $ gst-inspect-1.0 | grep -e "v4l2.*enc" > > > > > > > > > > > > video4linux2: v4l2jpegenc: V4L2 JPEG Encoder > > > > > > > > > > > > $ gst-launch-1.0 videotestsrc ! jpegenc ! rtpjpegpa= y ! udpsink > > > > > > > > > > > ^ v4l2jpegenc > > > > > > > > > > > > > > > > > > > > > > This is just a transcript error ? > > > > > > > > > > > > > > > > > > > > Nicolas, > > > > > > > > > > > > > > > > > > > > No! Thanks for catching my mistake. I was testing with = software encode... ooops! > > > > > > > > > > > > > > > > > > > > 'gst-launch-1.0 videotestsrc ! v4l2jpegenc ! fakesink' = actually hangs > > > > > > > > > > the board so likely a power-domain issue there? > > > > > > > > > > > > > > > > > > The v4l2-compliance tests fail on the h1 decoder with a h= ang, but I > > > > > > > > > think we're writing to registers which are not documented= in the Mini > > > > > > > > > TRM. The Mini TRM doesn't explicitly show the JPEG encod= ing as a > > > > > > > > > feature, but some of the registers state JPEG, but becaus= e some of the > > > > > > > > > registers written for the H1 are not documented in the TR= M. If those > > > > > > > > > registers are restricted or not in this SoC, I am concern= ed that it > > > > > > > > > might be related. I'll try to run some more tests this w= eekend to > > > > > > > > > check on the status of the power-domain stuff. > > > > > > > > > > > > > > > > To verify if the HW support JPEG encoding you can read SWRE= G63 bit 25. This is > > > > > > > > in the TRM, just not labelled properly. To mimic the decodi= ng side, would be "HW > > > > > > > > synthesis config register X" with the bit labelled SW_ENC_J= PEG_PROF (but > > > > > > > > PROF/profile is on or off). If your board hang while readin= g this, you likely > > > > > > > > didn't get the power bit right. > > > > > > > > > > > > > > > > IMX8 has an undocumented control block thing that we have b= een fighting with in > > > > > > > > imx8q, perhaps that's your issue. Few driver was proposed,= we are still pending > > > > > > > > on NXP solution to be submitted (they asked us to wait, sti= ll waiting =3D)). > > > > > > > > > > > > > > Nicolas, > > > > > > > > > > > > > > Thanks for the suggestion to read offset FC. There was an at= tempt > > > > > > > made by Lucas Stach to develop a VPU blk-ctrl driver to coord= inate the > > > > > > > power-domains with the GPC driver. Unfortunately, it does app= ear to > > > > > > > hang, so it might not be operating correctly. > > > > > > > > > > > > > > Lucas, > > > > > > > > > > > > > > Do you have any idea of stuff I can try to see if the power d= omain is > > > > > > > coming online correctly? > > > > > > > > > > > > > > [ 10.434727] imx-pgc imx-pgc-domain.6: request the vpumix d= omain to power up > > > > > > > [ 10.463647] imx-pgc imx-pgc-domain.6: request the vpumix A= DB400 to power up > > > > > > > [ 10.517155] imx-pgc imx-pgc-domain.6: genpd vpumix success > > > > > > > [ 10.728927] vpu: set fuse bits to enable > > > > > > > [ 10.825500] imx8m-blk-ctrl 38330000.blk-ctrl: power vpublk= -g1 GPC domain > > > > > > > [ 10.878986] imx-pgc imx-pgc-domain.7: request the vpu-g1 d= omain to power up > > > > > > > [ 10.932429] imx-pgc imx-pgc-domain.7: genpd vpu-g1 success > > > > > > > [ 10.971988] imx8m-blk-ctrl 38330000.blk-ctrl: genpd vpublk= -g1 success > > > > > > > [ 11.004726] hantro-vpu 38300000.video-codec: registered > > > > > > > nxp,imx8mm-vpu-dec as /dev/video0 > > > > > > > [ 11.040760] imx8m-blk-ctrl 38330000.blk-ctrl: power vpublk= -g2 GPC domain > > > > > > > [ 11.066181] imx-pgc imx-pgc-domain.8: request the vpu-g2 d= omain to power up > > > > > > > [ 11.087887] imx-pgc imx-pgc-domain.8: genpd vpu-g2 success > > > > > > > [ 11.113808] imx8m-blk-ctrl 38330000.blk-ctrl: genpd vpublk= -g2 success > > > > > > > [ 11.139634] hantro-vpu 38310000.video-codec: registered > > > > > > > nxp,imx8mm-vpu-g2-dec as /dev/video1 > > > > > > > [ 11.156463] imx8m-blk-ctrl 38330000.blk-ctrl: power vpublk= -h1 GPC domain > > > > > > > [ 11.170817] imx-pgc imx-pgc-domain.9: request the vpu-h1 d= omain to power up > > > > > > > [ 11.232990] imx-pgc imx-pgc-domain.9: genpd vpu-h1 success > > > > > > > [ 11.252546] imx8m-blk-ctrl 38330000.blk-ctrl: genpd vpublk= -h1 success > > > > > > > [ 11.266152] hantro-vpu 38320000.video-codec: Checking vpu-= >enc_base + 0xfc > > > > > > > > > > > > > > > > > > > > > > > > > > > > adam > > > > > > > > > > > > > > > > > > > Nicolas, Tim, and Lucas, > > > > > > > > > > > > I think I have the hanging resolved in the power domains, and I= 'll be > > > > > > pushing the fix to the GPCv2. > > > > > > > > > > > > For the H1 Encoder, I added some debugging code to read the off= set > > > > > > 0xfc and print some data based on the findings of that VPU-h1 o= ffset. > > > > > > I basically check the various bits per the TRM to see if they a= re set > > > > > > and print some splat to indicate whether or not the function is > > > > > > supported. > > > > > > > > > > > > [ 8.861865] hantro-vpu 38320000.video-codec: Checking vpu->e= nc_base + 0xfc > > > > > > [ 8.870594] hantro-vpu 38320000.video-codec: Stabilization s= upported by HW > > > > > > [ 8.889341] hantro-vpu 38320000.video-codec: VP8 encoding su= pported by HW > > > > > > [ 8.899386] hantro-vpu 38320000.video-codec: H.264 encoding = supported by HW > > > > > > [ 8.918171] hantro-vpu 38320000.video-codec: RGB to YUV conv= ersion > > > > > > supported by HW > > > > > > [ 8.934067] hantro-vpu 38320000.video-codec: registered > > > > > > nxp,imx8mm-vpu-h1-enc as /dev/video2 > > > > > > > > > > > > Unfortunately, JPEG is not listed as supported. :-( > > > > > > > > > > Adam, > > > > > > > > > > Well not having JPEG encode support is unfortunate, and unexpecte= d. Do > > > > > we not have hantro support yet for VP8/H264 encode? > > > > > > > > There is no mainline support yet. You can derive from RK3288 suppor= t using Google ChromeOS method (a v4l2 plugins that simulate in userspace a= stateful encoder): > > > > > > > > - libv4l plugins / https://chromium.googlesource.com/chromiumos/thi= rd_party/libv4lplugins/+/refs/heads/master > > > > - Kernel Driver / https://chromium.googlesource.com/chromiumos/thir= d_party/kernel/+/chromeos-4.4/drivers/media/platform/rockchip-vpu/ > > > > > > > > > > > > > > I haven't quite figured out how to build a modern mono-repo gstre= amer > > > > > on the ubuntu 20.04 rootfs I'm using so I haven't been able to te= st > > > > > VPU encode/decode properly. I'll keep working on it when I'm back= in > > > > > the office the following week. > > > > > > > > Did a quick test to make sure there isn't any ubuntu specific block= ers, here's a > > > > dirty script that produce a minimal GStreamer, there was really not= hing special > > > > compare to other meson projects. Note that I use --wrap-mode=3Dnofa= llback to avoid > > > > letting GStreamer complete it's feature-set by downloading the plan= et. This > > > > already build quite a lot and could likely be made smaller by avoid= plugins-good > > > > build-dep call, but then you need to check for v4l2odecs and video4= linux devs > > > > (mostly gudev a glib udev binding). > > > > > > > > # Install ubuntu > > > > podman run -it --rm ubuntu:20.04 > > > > sed -i "s/# deb-src/deb-src/" /etc/apt/sources.list > > > > apt update > > > > apt build-dep gstreamer1.0-plugins-good > > > > apt install git python3-pip flex bison > > > > > > > > # Need a newer meson > > > > pip3 install --user meson > > > > export PATH=3D$PATH:~/.local/bin > > > > > > > > # Build GStreamer > > > > git clone https://gitlab.freedesktop.org/gstreamer/gstreamer.git > > > > cd gstreamer > > > > meson setup build --wrap-mode=3Dnofallback > > > > ninja -C build > > > > > > > > # Run in-place > > > > ./gst-env.py > > > > gst-inspect-1.0 v4l2codecs > > > > gst-inspect 1.0 video4linux2 > > > > > > > Thanks for the suggestions. > > > > > > I downloaded what's in the master repo: > > > > > > [gst-main] root@localhost:~/gstreamer# gst-inspect-1.0 v4l2codecs > > > > > > ** (gst-plugin-scanner:7317): CRITICAL **: 10:29:51.847: can't find > > > gi.repository.Gst > > > Plugin Details: > > > Name v4l2codecs > > > Description V4L2 CODEC Accelerators plugin > > > Filename > > > /root/gstreamer/builddir/subprojects/gst-plugins-bad/sys/v4l2codecs/l= ibgstv4l2codecs.so > > > Version 1.19.3.1 > > > License LGPL > > > Source module gst-plugins-bad > > > Binary package GStreamer Bad Plug-ins git > > > Origin URL Unknown package origin > > > > > > v4l2slh264dec: V4L2 Stateless H.264 Video Decoder > > > v4l2slmpeg2dec: V4L2 Stateless Mpeg2 Video Decoder > > > v4l2slvp8alphadecodebin: VP8 Alpha Decoder > > > v4l2slvp8dec: V4L2 Stateless VP8 Video Decoder > > > > > > 4 features: > > > +-- 4 elements > > > > > > [gst-main] root@localhost:~/gstreamer# gst-inspect-1.0 video4linux2 > > > Plugin Details: > > > Name video4linux2 > > > Description elements for Video 4 Linux > > > Filename > > > /root/gstreamer/builddir/subprojects/gst-plugins-good/sys/v4l2/libgst= video4linux2.so > > > Version 1.19.3.1 > > > License LGPL > > > Source module gst-plugins-good > > > Binary package GStreamer Good Plug-ins git > > > Origin URL Unknown package origin > > > > > > v4l2deviceprovider: Video (video4linux2) Device Provider > > > v4l2jpegenc: V4L2 JPEG Encoder > > > v4l2radio: Radio (video4linux2) Tuner > > > v4l2sink: Video (video4linux2) Sink > > > v4l2src: Video (video4linux2) Source > > > > > > 5 features: > > > +-- 4 elements > > > +-- 1 device providers > > > > > > I still have the H1 encoder enabled, but I know JPEG isn't supported, > > > so I'm going to attempt to do some decoding and pipe to fakesink sinc= e > > > I don't have a functional display yet. > > > > > > gst-launch-1.0 -ev filesrc location=3Dtrailer_1080p_h264_mp3.avi ! > > > decodebin3 ! fakesink > > > > > > Redistribute latency... > > > /GstPipeline:pipeline0/GstDecodebin3:decodebin3-0/v4l2slh264dec:v4l2s= lh264dec0.GstPad:src: > > > caps =3D video/x-raw, format=3D(string)NV12, width=3D(int)1920, > > > height=3D(int)1080, interlace-mode=3D(string)progressive, > > > multiview-mode=3D(string)mono, > > > multiview-flags=3D(GstVideoMultiviewFlagsSet)0:ffffffff:/right-view-f= irst/left-flipped/left-flopped/right-flipped/right-flopped/half-aspect/mixe= d-mono, > > > pixel-aspect-ratio=3D(fraction)1/1, framerate=3D(fraction)25/1 > > > /GstPipeline:pipeline0/GstDecodebin3:decodebin3-0.GstGhostPad:video_0= : > > > caps =3D video/x-raw, format=3D(string)NV12, width=3D(int)1920, > > > height=3D(int)1080, interlace-mode=3D(string)progressive, > > > multiview-mode=3D(string)mono, > > > multiview-flags=3D(GstVideoMultiviewFlagsSet)0:ffffffff:/right-view-f= irst/left-flipped/left-flopped/right-flipped/right-flopped/half-aspect/mixe= d-mono, > > > pixel-aspect-ratio=3D(fraction)1/1, framerate=3D(fraction)25/1 > > > /GstPipeline:pipeline0/GstDecodebin3:decodebin3-0.GstGhostPad:video_0= .GstProxyPad:proxypad6: > > > caps =3D video/x-raw, format=3D(string)NV12, width=3D(int)1920, > > > height=3D(int)1080, interlace-mode=3D(string)progressive, > > > multiview-mode=3D(string)mono, > > > multiview-flags=3D(GstVideoMultiviewFlagsSet)0:ffffffff:/right-view-f= irst/left-flipped/left-flopped/right-flipped/right-flopped/half-aspect/mixe= d-mono, > > > pixel-aspect-ratio=3D(fraction)1/1, framerate=3D(fraction)25/1 > > > /GstPipeline:pipeline0/GstDecodebin3:decodebin3-0/GstMultiQueue:multi= queue0: > > > min-interleave-time =3D 300000000 > > > Redistribute latency... > > > /GstPipeline:pipeline0/GstDecodebin3:decodebin3-0/v4l2slh264dec:v4l2s= lh264dec0.GstPad:sink: > > > caps =3D video/x-h264, variant=3D(string)itu, framerate=3D(fraction)2= 5/1, > > > width=3D(int)1920, height=3D(int)1080, chroma-format=3D(string)4:2:0, > > > bit-depth-luma=3D(uint)8, bit-depth-chroma=3D(uint)8, > > > parsed=3D(boolean)true, stream-format=3D(string)avc, alignment=3D(str= ing)au, > > > profile=3D(string)high, level=3D(string)4, > > > codec_data=3D(buffer)01640028ffe1001a67640028acd940780227e58400000300= 0400000300c83c60c65801000668ebe3cb22c0 > > > New clock: GstSystemClock > > > > > > And it appears to stream, because the counter increases. I haven't > > > checked the CPU utilization, but the fact that it shows v4l2slh264dec > > > is good. > > > > > > Is there a way to know if/how the decoder is using the proper VPU? I > > > assume if it wasn't using the proper one, it would fail, but was just > > > curious. > > > > > > > A few ways. You can check /proc/interrupts, which should have > > VPU activity. > > > > Or enable debug messages for the module, > > using the debug hantro parameter. V4L2 has debug messages > > that you can enable, see /sys/class/video4linux/video0/dev_debug. > > > > Instead of fakesink you can output to pngenc/jpegenc and check the outp= ut > > is visually correct. If at all possible, the proper way is to use Flust= er, > > and report the score you get: > > > > https://github.com/fluendo/fluster > > > > I ran fluster on the VP8 decoder, but only 55/61 passed. > > *************************************************************************= *************************** > Running test suite VP8-TEST-VECTORS with decoder GStreamer-VP8-V4L2SL-Gst= 1.0 > Using 4 parallel job(s) > *************************************************************************= *************************** > > [TEST SUITE ] (DECODER ) TEST VECTOR > ... RESULT > ---------------------------------------------------------------------- > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-00-comprehensive-004 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-00-comprehensive-001 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-00-comprehensive-002 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-00-comprehensive-003 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-00-comprehensive-005 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-00-comprehensive-006 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-00-comprehensive-007 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-00-comprehensive-008 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-00-comprehensive-011 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-00-comprehensive-009 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-00-comprehensive-012 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-00-comprehensive-013 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-00-comprehensive-014 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-00-comprehensive-010 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-00-comprehensive-016 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-00-comprehensive-017 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-00-comprehensive-018 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) vp80-01-intra-1400 > ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) vp80-01-intra-1416 > ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) vp80-01-intra-1417 > ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) vp80-01-intra-1411 > ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) vp80-02-inter-1402 > ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) vp80-02-inter-1412 > ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) vp80-02-inter-1424 > ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-03-segmentation-01 ... Fail > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-03-segmentation-02 ... Fail > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-03-segmentation-03 ... Fail > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-03-segmentation-04 ... Fail > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-03-segmentation-1401 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) vp80-02-inter-1418 > ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-03-segmentation-1403 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-03-segmentation-1407 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-03-segmentation-1408 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-03-segmentation-1409 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-03-segmentation-1413 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-03-segmentation-1415 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-03-segmentation-1425 ... Fail > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-03-segmentation-1426 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-03-segmentation-1427 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-03-segmentation-1432 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-03-segmentation-1435 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-03-segmentation-1436 ... Fail > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-00-comprehensive-015 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-03-segmentation-1441 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-03-segmentation-1437 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-04-partitions-1404 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-03-segmentation-1442 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-04-partitions-1405 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-04-partitions-1406 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-05-sharpness-1428 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-05-sharpness-1429 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-05-sharpness-1431 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-03-segmentation-1410 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-03-segmentation-1414 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-05-sharpness-1430 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-05-sharpness-1433 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-05-sharpness-1438 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-05-sharpness-1434 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-05-sharpness-1439 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-05-sharpness-1440 ... Success > [VP8-TEST-VECTORS] (GStreamer-VP8-V4L2SL-Gst1.0) > vp80-05-sharpness-1443 ... Success > > > =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D > FAIL: vp80-03-segmentation-01 (GStreamer-VP8-V4L2SL-Gst1.0.VP8-TEST-VECTO= RS) > ---------------------------------------------------------------------- > Traceback (most recent call last): > File "/root/gstreamer/fluster/fluster/test.py", line 104, in _test > self.assertEqual( > AssertionError: 'db954c077b7a3f34a448ceaacf8f525c' !=3D > '8bbb396a9bdf8afa250d3b2e45e6b367' > - db954c077b7a3f34a448ceaacf8f525c > + 8bbb396a9bdf8afa250d3b2e45e6b367 > : vp80-03-segmentation-01 > > =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D > FAIL: vp80-03-segmentation-02 (GStreamer-VP8-V4L2SL-Gst1.0.VP8-TEST-VECTO= RS) > ---------------------------------------------------------------------- > Traceback (most recent call last): > File "/root/gstreamer/fluster/fluster/test.py", line 104, in _test > self.assertEqual( > AssertionError: '4d2d65efeee1c83772c33a13446bd1a4' !=3D > '1b2061d4a74549228769f8e292bcb15f' > - 4d2d65efeee1c83772c33a13446bd1a4 > + 1b2061d4a74549228769f8e292bcb15f > : vp80-03-segmentation-02 > > =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D > FAIL: vp80-03-segmentation-03 (GStreamer-VP8-V4L2SL-Gst1.0.VP8-TEST-VECTO= RS) > ---------------------------------------------------------------------- > Traceback (most recent call last): > File "/root/gstreamer/fluster/fluster/test.py", line 104, in _test > self.assertEqual( > AssertionError: '73d864433691f8db43257b88495ac8c3' !=3D > 'fd1eb6ebd7100995bad11042a9bea048' > - 73d864433691f8db43257b88495ac8c3 > + fd1eb6ebd7100995bad11042a9bea048 > : vp80-03-segmentation-03 > > =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D > FAIL: vp80-03-segmentation-04 (GStreamer-VP8-V4L2SL-Gst1.0.VP8-TEST-VECTO= RS) > ---------------------------------------------------------------------- > Traceback (most recent call last): > File "/root/gstreamer/fluster/fluster/test.py", line 104, in _test > self.assertEqual( > AssertionError: '7f846c8bd7cdfe61f8542f382f9d8eeb' !=3D > '0c27a47c4fd8bbfce173d005bef8be6a' > - 7f846c8bd7cdfe61f8542f382f9d8eeb > + 0c27a47c4fd8bbfce173d005bef8be6a > : vp80-03-segmentation-04 > > =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D > FAIL: vp80-03-segmentation-1425 (GStreamer-VP8-V4L2SL-Gst1.0.VP8-TEST-VEC= TORS) > ---------------------------------------------------------------------- > Traceback (most recent call last): > File "/root/gstreamer/fluster/fluster/test.py", line 104, in _test > self.assertEqual( > AssertionError: '96ffacf0c3eae59b58252be24a60e9b2' !=3D > '83e8a322e8ab23e60ba16430aacad827' > - 96ffacf0c3eae59b58252be24a60e9b2 > + 83e8a322e8ab23e60ba16430aacad827 > : vp80-03-segmentation-1425 > > =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D > FAIL: vp80-03-segmentation-1436 (GStreamer-VP8-V4L2SL-Gst1.0.VP8-TEST-VEC= TORS) > ---------------------------------------------------------------------- > Traceback (most recent call last): > File "/root/gstreamer/fluster/fluster/test.py", line 104, in _test > self.assertEqual( > AssertionError: 'bfd17a557ee1ba347c755a18ce5a64a6' !=3D > '5bca61a733c1936205f82de1492a1b2b' > - bfd17a557ee1ba347c755a18ce5a64a6 > + 5bca61a733c1936205f82de1492a1b2b > : vp80-03-segmentation-1436 > > Ran 55/61 tests successfully in 12.104 secs > > I am not that familiar with this tool, but I assume failures are bad. > However these look like Python errors and not gst errors. > > The H264 decoder resulted in: > > Ran 85/135 tests successfully in 57.821 secs > > I can provide the splat if you want. Those looked like gst errors, > because most of the error messages state the gst-launch-1.0 returned > non-zero exit status 1. > > > > It should be easy to use. > > It was. > > > > > I think I'll redo the patch without the RFC and without the H1 encode= r > > > unless anyone has any objections. I know I need to rebase on > > > linux-next anyway because the patches don't apply cleanly. Is there = a > > > specific branch I should use? I don't know if this goes through > > > Shawn's IMX tree or the media tree (or a combination) > > > > > > > You should rebase on media's master branch: > > > > https://git.linuxtv.org/media_tree.git/log/ > > I'll submit the patch with a cover letter with the results of the VP8 > and H264 fluster test in the cover letter. Is there a stateless > decoder for the VP9 decoder? gst-inspect only shows the following > v4l2codecs. > > v4l2slh264dec: V4L2 Stateless H.264 Video Decoder > v4l2slmpeg2dec: V4L2 Stateless Mpeg2 Video Decoder > v4l2slvp8alphadecodebin: VP8 Alpha Decoder > v4l2slvp8dec: V4L2 Stateless VP8 Video Decoder > > thanks for all your help. Hopefully we can get this integrated soon. > Adam, What deps did you install in order to get v4l2codecs building? I installed libgudev-1.0-dev based on Nicolas' suggestion and rebuilt (not sure if I needed to re-configure somehow) but there is still nothing in build/subprojects/gst-plugins-bad/sys/v4l2codecs/. A 'meson configure' tells me that v4l2codecs is set to 'auto' but I'm not sure how to find out what dependencies are needed or what may be missing. Best regards, Tim