From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1754320AbcEWN6w (ORCPT ); Mon, 23 May 2016 09:58:52 -0400 Received: from pegasos-out.vodafone.de ([80.84.1.38]:56931 "EHLO pegasos-out.vodafone.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752335AbcEWN6v (ORCPT ); Mon, 23 May 2016 09:58:51 -0400 X-Spam-Flag: NO X-Spam-Score: -0.045 Authentication-Results: rohrpostix2.prod.vfnet.de (amavisd-new); dkim=pass header.i=@vodafone.de X-DKIM: OpenDKIM Filter v2.6.8 pegasos-out.vodafone.de 2DF285C7B87 Subject: Re: [PATCH 3/3] dma-buf/fence: add signal_on_any to the fence array To: dri-devel@lists.freedesktop.org, linux-kernel@vger.kernel.org, gustavo@padovan.org, chris@chris-wilson.co.uk References: <1464002722-4989-1-git-send-email-deathsimple@vodafone.de> <1464002722-4989-3-git-send-email-deathsimple@vodafone.de> <20160523135049.GA27098@phenom.ffwll.local> From: =?UTF-8?Q?Christian_K=c3=b6nig?= Message-ID: <57430C91.9060106@vodafone.de> Date: Mon, 23 May 2016 15:58:41 +0200 User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:38.0) Gecko/20100101 Thunderbird/38.7.2 MIME-Version: 1.0 In-Reply-To: <20160523135049.GA27098@phenom.ffwll.local> Content-Type: text/plain; charset=windows-1252; format=flowed Content-Transfer-Encoding: 8bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Am 23.05.2016 um 15:50 schrieb Daniel Vetter: > On Mon, May 23, 2016 at 01:25:22PM +0200, Christian König wrote: >> From: Christian König >> >> If @signal_on_any is true the fence array signals if any fence in the array >> signals, otherwise it signals when all fences in the array signal. > What do you need this for? Sounds neat, but I can't come up with a use > case. And that use case imo should be both in the commit message and > kerneldoc. Good point, going to fix this when if we really use the patch. Just wanted to get opinions on this. I'm currently working on making some of the resource management in amdgpu more friendly to our GPU scheduler. Waiting for resources in the GPU scheduler is implemented as waiting for a fence to fire. So when you want to wait for any resource to become available (like in our case the VMIDs) you could make good use of something like this. Christian. > -Daniel > >> Signed-off-by: Christian König >> --- >> drivers/dma-buf/fence-array.c | 23 ++++++++++++++++------- >> include/linux/fence-array.h | 3 ++- >> 2 files changed, 18 insertions(+), 8 deletions(-) >> >> diff --git a/drivers/dma-buf/fence-array.c b/drivers/dma-buf/fence-array.c >> index bbcf79f..726e840 100644 >> --- a/drivers/dma-buf/fence-array.c >> +++ b/drivers/dma-buf/fence-array.c >> @@ -41,6 +41,7 @@ static void fence_array_cb_func(struct fence *f, struct fence_cb *cb) >> >> if (atomic_dec_and_test(&array->num_pending)) >> fence_signal(&array->base); >> + fence_put(&array->base); >> } >> >> static bool fence_array_enable_signaling(struct fence *fence) >> @@ -51,10 +52,13 @@ static bool fence_array_enable_signaling(struct fence *fence) >> >> for (i = 0; i < array->num_fences; ++i) { >> cb[i].array = array; >> + fence_get(&array->base); >> if (!fence_add_callback(array->fences[i], &cb[i].cb, >> - fence_array_cb_func)) >> + fence_array_cb_func)) { >> + fence_put(&array->base); >> if (atomic_dec_and_test(&array->num_pending)) >> return false; >> + } >> } >> >> return true; >> @@ -90,10 +94,11 @@ const struct fence_ops fence_array_ops = { >> >> /** >> * fence_array_create - Create a custom fence array >> - * @num_fences: [in] number of fences to add in the array >> - * @fences: [in] array containing the fences >> - * @context: [in] fence context to use >> - * @seqno: [in] sequence number to use >> + * @num_fences: [in] number of fences to add in the array >> + * @fences: [in] array containing the fences >> + * @context: [in] fence context to use >> + * @seqno: [in] sequence number to use >> + * @signal_on_any [in] signal on any fence in the array >> * >> * Allocate a fence_array object and initialize the base fence with fence_init(). >> * In case of error it returns NULL. >> @@ -101,9 +106,13 @@ const struct fence_ops fence_array_ops = { >> * The caller should allocte the fences array with num_fences size >> * and fill it with the fences it wants to add to the object. Ownership of this >> * array is take and fence_put() is used on each fence on release. >> + * >> + * If @signal_on_any is true the fence array signals if any fence in the array >> + * signals, otherwise it signals when all fences in the array signal. >> */ >> struct fence_array *fence_array_create(int num_fences, struct fence **fences, >> - u64 context, unsigned seqno) >> + u64 context, unsigned seqno, >> + bool signal_on_any) >> { >> struct fence_array *array; >> size_t size = sizeof(*array); >> @@ -119,7 +128,7 @@ struct fence_array *fence_array_create(int num_fences, struct fence **fences, >> context, seqno); >> >> array->num_fences = num_fences; >> - atomic_set(&array->num_pending, num_fences); >> + atomic_set(&array->num_pending, signal_on_any ? 1 : num_fences); >> array->fences = fences; >> >> return array; >> diff --git a/include/linux/fence-array.h b/include/linux/fence-array.h >> index 593ab98..86baaa4 100644 >> --- a/include/linux/fence-array.h >> +++ b/include/linux/fence-array.h >> @@ -67,6 +67,7 @@ static inline struct fence_array *to_fence_array(struct fence *fence) >> } >> >> struct fence_array *fence_array_create(int num_fences, struct fence **fences, >> - u64 context, unsigned seqno); >> + u64 context, unsigned seqno, >> + bool signal_on_any); >> >> #endif /* __LINUX_FENCE_ARRAY_H */ >> -- >> 2.5.0 >>