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=-1.0 required=3.0 tests=HEADER_FROM_DIFFERENT_DOMAINS, MAILING_LIST_MULTI,SPF_PASS,UNPARSEABLE_RELAY autolearn=ham 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 E0747C43219 for ; Fri, 26 Apr 2019 18:19:53 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id BF355208CA for ; Fri, 26 Apr 2019 18:19:53 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1726116AbfDZSTx (ORCPT ); Fri, 26 Apr 2019 14:19:53 -0400 Received: from bhuna.collabora.co.uk ([46.235.227.227]:40338 "EHLO bhuna.collabora.co.uk" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1726083AbfDZSTx (ORCPT ); Fri, 26 Apr 2019 14:19:53 -0400 Received: from [127.0.0.1] (localhost [127.0.0.1]) (Authenticated sender: ezequiel) with ESMTPSA id B4D49273331 Message-ID: <52effdcfcbacac8a6df56d0d43591326d314a894.camel@collabora.com> Subject: Re: [PATCH 4/5] media: coda: Remove pic_run_work worker From: Ezequiel Garcia To: Philipp Zabel , linux-media@vger.kernel.org, Hans Verkuil Cc: kernel@collabora.com Date: Fri, 26 Apr 2019 15:19:46 -0300 In-Reply-To: <1556266306.3664.5.camel@pengutronix.de> References: <20190425183546.16244-1-ezequiel@collabora.com> <20190425183546.16244-5-ezequiel@collabora.com> <1556266306.3664.5.camel@pengutronix.de> Organization: Collabora Content-Type: text/plain; charset="UTF-8" User-Agent: Evolution 3.30.5-1 MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Sender: linux-media-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-media@vger.kernel.org On Fri, 2019-04-26 at 10:11 +0200, Philipp Zabel wrote: > On Thu, 2019-04-25 at 15:35 -0300, Ezequiel Garcia wrote: > > There isn't any reason to run the mem2mem job on a separate worker, > > because the mem2mem framework guarantees that device_run will never > > run in interrupt context. > > The purpose of the workqueue is to serialize BIT processor commands, > currently the PIC_RUN commands issued by the mem2mem framework (as well > as SEQ_INIT, SET_FRAME_BUF, and ENCODE_HEADER) against the SEQ_END > command issued directly from the STREAMOFF ioctl. Right, but that's serialized by the coda_mutex, not by the worker, right? > Further, to fully support the stateful decoder API we'll have to move > SEQ_INIT out of the mem2mem device_run as well, since that should be > called on queued OUTPUT buffers before the CAPTURE side is even > streaming. > Isn't that already done? I see SEQ_INIT being issued in start_streaming. In what sense is this driver currently violating the decoder spec? So, returning to the serialization. I believe commands are still serialized after this change. The pic_run_work worker is only queued from .device_run. Only one job can run on a context at any given time, but multiple contexts can run in parallel. Inside the worker, the coda_mutex serializes the commands. The worker waits until the command has finished execution. So, with this commit, there is no longer a worker, but commands are still serialized by the mutex and the fact that the command is completed in .device_run. That being said, the worker does makes the serialization more clear, so I think dropping this patch is better. Perhaps adding a small comment so the purpose of pic_run_work is clear. Thanks, Eze