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 Received: from gabe.freedesktop.org (gabe.freedesktop.org [131.252.210.177]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.lore.kernel.org (Postfix) with ESMTPS id C89D0C433EF for ; Mon, 22 Nov 2021 15:16:19 +0000 (UTC) Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 106A489B99; Mon, 22 Nov 2021 15:16:19 +0000 (UTC) Received: from aposti.net (aposti.net [89.234.176.197]) by gabe.freedesktop.org (Postfix) with ESMTPS id 3DD8E89B99 for ; Mon, 22 Nov 2021 15:16:18 +0000 (UTC) Date: Mon, 22 Nov 2021 15:16:06 +0000 From: Paul Cercueil Subject: Re: [PATCH 01/15] iio: buffer-dma: Get rid of incoming/outgoing queues To: Lars-Peter Clausen Message-Id: In-Reply-To: References: <20211115141925.60164-1-paul@crapouillou.net> <20211115141925.60164-2-paul@crapouillou.net> <0COX2R.BSNX3NW8N48T@crapouillou.net> <332d001d-8b5a-bba2-c490-ed2e5efd0b1d@metafoo.de> MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1; format=flowed Content-Transfer-Encoding: quoted-printable X-BeenThere: dri-devel@lists.freedesktop.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: Direct Rendering Infrastructure - Development List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Cc: Michael Hennerich , linux-iio@vger.kernel.org, linux-kernel@vger.kernel.org, dri-devel@lists.freedesktop.org, linaro-mm-sig@lists.linaro.org, Alexandru Ardelean , Christian =?iso-8859-1?b?S/ZuaWc=?= , Jonathan Cameron , linux-media@vger.kernel.org Errors-To: dri-devel-bounces@lists.freedesktop.org Sender: "dri-devel" Hi Lars, Le lun., nov. 22 2021 at 16:08:51 +0100, Lars-Peter Clausen=20 a =E9crit : > On 11/21/21 9:08 PM, Paul Cercueil wrote: >>=20 >>=20 >> Le dim., nov. 21 2021 at 19:49:03 +0100, Lars-Peter Clausen=20 >> =7F a =E9crit : >>> On 11/21/21 6:52 PM, Paul Cercueil wrote: >>>> Hi Lars, >>>>=20 >>>> Le dim., nov. 21 2021 at 17:23:35 +0100, Lars-Peter Clausen=20 >>>> =7F=7F=7F=7F a =E9crit : >>>>> On 11/15/21 3:19 PM, Paul Cercueil wrote: >>>>>> The buffer-dma code was using two queues, incoming and outgoing,=20 >>>>>> to >>>>>> manage the state of the blocks in use. >>>>>>=20 >>>>>> While this totally works, it adds some complexity to the code, >>>>>> especially since the code only manages 2 blocks. It is much=20 >>>>>> easier to >>>>>> just check each block's state manually, and keep a counter for=20 >>>>>> the =7F=7F=7F=7F=7Fnext >>>>>> block to dequeue. >>>>>>=20 >>>>>> Since the new DMABUF based API wouldn't use these incoming and=20 >>>>>> =7F=7F=7F=7F=7Foutgoing >>>>>> queues anyway, getting rid of them now makes the upcoming changes >>>>>> simpler. >>>>>>=20 >>>>>> Signed-off-by: Paul Cercueil >>>>> The outgoing queue is going to be replaced by fences, but I think=20 >>>>> =7F=7F=7F=7Fwe =7F=7Fneed to keep the incoming queue. >>>>=20 >>>> Blocks are always accessed in sequential order, so we now have a=20 >>>> =7F=7F=7F=7F"queue->next_dequeue" that cycles between the buffers=20 >>>> allocated for =7F=7F=7F=7Ffileio. >>>>=20 >>>>>> [...] >>>>>> @@ -442,28 +435,33 @@ EXPORT_SYMBOL_GPL(iio_dma_buffer_disable); >>>>>> static void iio_dma_buffer_enqueue(struct iio_dma_buffer_queue=20 >>>>>> =7F=7F=7F=7F=7F=7F=7F=7F*queue, >>>>>> struct iio_dma_buffer_block *block) >>>>>> { >>>>>> - if (block->state =3D=3D IIO_BLOCK_STATE_DEAD) { >>>>>> + if (block->state =3D=3D IIO_BLOCK_STATE_DEAD) >>>>>> iio_buffer_block_put(block); >>>>>> - } else if (queue->active) { >>>>>> + else if (queue->active) >>>>>> iio_dma_buffer_submit_block(queue, block); >>>>>> - } else { >>>>>> + else >>>>>> block->state =3D IIO_BLOCK_STATE_QUEUED; >>>>>> - list_add_tail(&block->head, &queue->incoming); >>>>> If iio_dma_buffer_enqueue() is called with a dmabuf and the=20 >>>>> buffer =7F=7F=7F=7Fis =7F=7Fnot active, it will be marked as queued, = but we=20 >>>>> don't actually =7F=7F=7F=7Fkeep a =7F=7Freference to it anywhere. It = will=20 >>>>> never be submitted to =7F=7F=7F=7Fthe DMA, and =7F=7Fit will never be= =20 >>>>> signaled as completed. >>>>=20 >>>> We do keep a reference to the buffers, in the queue->fileio.blocks=20 >>>> =7F=7F=7F=7Farray. When the buffer is enabled, all the blocks in that=20 >>>> array =7F=7F=7Fthat =7Fare in the "queued" state will be submitted to = the=20 >>>> DMA. >>>>=20 >>> But not when used in combination with the DMA buf changes later in=20 >>> =7F=7Fthis series. >>>=20 >>=20 >> That's still the case after the DMABUF changes of the series. Or can=20 >> =7Fyou point me exactly what you think is broken? >>=20 > When you allocate a DMABUF with the allocate IOCTL and then submit it=20 > with the enqueue IOCTL before the buffer is enabled it will end up=20 > marked as queued, but not actually be queued anywhere. >=20 Ok, it works for me because I never enqueue blocks before enabling the=20 buffer. I can add a requirement that blocks must be enqueued only after=20 the buffer is enabled. Cheers, -Paul