Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1754346AbaGJV5e (ORCPT ); Thu, 10 Jul 2014 17:57:34 -0400 Received: from mail-wi0-f173.google.com ([209.85.212.173]:55784 "EHLO mail-wi0-f173.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753432AbaGJV5b (ORCPT ); Thu, 10 Jul 2014 17:57:31 -0400 From: Oded Gabbay X-Google-Original-From: Oded Gabbay To: David Airlie , Alex Deucher , Jerome Glisse Cc: linux-kernel@vger.kernel.org, dri-devel@lists.freedesktop.org, John Bridgman , Andrew Lewycky , Joerg Roedel , Oded Gabbay Subject: [PATCH 76/83] hsa/radeon: Check oversubscription before destroying runlist Date: Fri, 11 Jul 2014 00:54:32 +0300 Message-Id: <1405029279-6894-48-git-send-email-oded.gabbay@amd.com> X-Mailer: git-send-email 1.9.1 In-Reply-To: <1405029279-6894-1-git-send-email-oded.gabbay@amd.com> References: <1405029279-6894-1-git-send-email-oded.gabbay@amd.com> Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org This patch fixes a bug when using the mode of CP hardware scheduling without oversubscription. The bug was that the oversubscription check was performed _after_ the current runlist was destroyed, which caused the current HSA application to stop working. This patch moves the oversubscription check before the call to destroy the current runlist. If there is oversubscription, the function prints an error to dmesg and simply exits. Signed-off-by: Oded Gabbay --- drivers/gpu/hsa/radeon/kfd_packet_manager.c | 3 --- drivers/gpu/hsa/radeon/kfd_process_queue_manager.c | 9 +++++++++ 2 files changed, 9 insertions(+), 3 deletions(-) diff --git a/drivers/gpu/hsa/radeon/kfd_packet_manager.c b/drivers/gpu/hsa/radeon/kfd_packet_manager.c index 5cd23b0..0aef907 100644 --- a/drivers/gpu/hsa/radeon/kfd_packet_manager.c +++ b/drivers/gpu/hsa/radeon/kfd_packet_manager.c @@ -88,9 +88,6 @@ static int pm_allocate_runlist_ib(struct packet_manager *pm, unsigned int **rl_b BUG_ON(is_over_subscription == NULL); pm_calc_rlib_size(pm, rl_buffer_size, is_over_subscription); - if (*is_over_subscription && - sched_policy == KFD_SCHED_POLICY_HWS_NO_OVERSUBSCRIPTION) - return -EFAULT; retval = radeon_kfd_vidmem_alloc_map(pm->dqm->dev, &pm->ib_buffer_obj, (void **)rl_buffer, rl_gpu_buffer, ALIGN(*rl_buffer_size, PAGE_SIZE)); diff --git a/drivers/gpu/hsa/radeon/kfd_process_queue_manager.c b/drivers/gpu/hsa/radeon/kfd_process_queue_manager.c index 5d7c46d..97b3cc6 100644 --- a/drivers/gpu/hsa/radeon/kfd_process_queue_manager.c +++ b/drivers/gpu/hsa/radeon/kfd_process_queue_manager.c @@ -174,6 +174,15 @@ int pqm_create_queue(struct process_queue_manager *pqm, switch (type) { case KFD_QUEUE_TYPE_COMPUTE: + /* check if there is over subscription */ + if ((sched_policy == KFD_SCHED_POLICY_HWS_NO_OVERSUBSCRIPTION) && + ((dev->dqm->processes_count >= VMID_PER_DEVICE) || + (dev->dqm->queue_count >= PIPE_PER_ME_CP_SCHEDULING * QUEUES_PER_PIPE))) { + pr_err("kfd: over-subscription is not allowed in radeon_kfd.sched_policy == 1\n"); + retval = -EPERM; + goto err_create_queue; + } + retval = create_cp_queue(pqm, dev, &q, &q_properties, f, *qid); if (retval != 0) goto err_create_queue; -- 1.9.1 -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/