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 85296C77B7A for ; Wed, 17 May 2023 15:50:16 +0000 (UTC) Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 21DDB10E44B; Wed, 17 May 2023 15:50:16 +0000 (UTC) Received: from dfw.source.kernel.org (dfw.source.kernel.org [139.178.84.217]) by gabe.freedesktop.org (Postfix) with ESMTPS id 8BEB210E44B for ; Wed, 17 May 2023 15:50:13 +0000 (UTC) Received: from smtp.kernel.org (relay.kernel.org [52.25.139.140]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by dfw.source.kernel.org (Postfix) with ESMTPS id 0E16E614E9; Wed, 17 May 2023 15:50:13 +0000 (UTC) Received: from rdvivi-mobl4 (fmdmzpr02-ext.fm.intel.com [192.55.54.37]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.kernel.org (Postfix) with ESMTPSA id E9708C433EF; Wed, 17 May 2023 15:50:10 +0000 (UTC) Date: Wed, 17 May 2023 11:50:08 -0400 From: Rodrigo Vivi To: Matthew Auld Message-ID: References: <20230517152244.348171-1-matthew.auld@intel.com> <20230517152244.348171-3-matthew.auld@intel.com> MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Disposition: inline Content-Transfer-Encoding: 8bit In-Reply-To: <20230517152244.348171-3-matthew.auld@intel.com> Subject: Re: [Intel-xe] [PATCH v5 3/7] drm/xe: don't allocate under ct->lock X-BeenThere: intel-xe@lists.freedesktop.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: Intel Xe graphics driver List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Cc: intel-xe@lists.freedesktop.org, Rodrigo Vivi Errors-To: intel-xe-bounces@lists.freedesktop.org Sender: "Intel-xe" On Wed, May 17, 2023 at 04:22:40PM +0100, Matthew Auld wrote: > Seems to be a sensitive lock, where ct->lock looks to be primed with > fs_reclaim, so holding that and then allocating memory will cause > lockdep to complain. We need to change the ordering wrt to grabbing the > ct->lock and potentially grabbing the runtime_pm, since some of the > runtime_pm routines can allocate memory (or at least that's what lockdep > seems to suggest). > > Signed-off-by: Matthew Auld > Cc: Thomas Hellström > Cc: Rodrigo Vivi It is a good idea anyway to get the mem_access as early as possible. Different from the locks that we want to limit the time of the critical area. Reviewed-by: Rodrigo Vivi > --- > drivers/gpu/drm/xe/xe_gt_tlb_invalidation.c | 4 ++++ > drivers/gpu/drm/xe/xe_guc_ct.c | 13 +++++++------ > 2 files changed, 11 insertions(+), 6 deletions(-) > > diff --git a/drivers/gpu/drm/xe/xe_gt_tlb_invalidation.c b/drivers/gpu/drm/xe/xe_gt_tlb_invalidation.c > index c815a42e2cdb..20f8f0aae6b4 100644 > --- a/drivers/gpu/drm/xe/xe_gt_tlb_invalidation.c > +++ b/drivers/gpu/drm/xe/xe_gt_tlb_invalidation.c > @@ -5,6 +5,7 @@ > > #include "xe_gt_tlb_invalidation.h" > > +#include "xe_device.h" > #include "xe_gt.h" > #include "xe_guc.h" > #include "xe_guc_ct.h" > @@ -112,6 +113,8 @@ static int send_tlb_invalidation(struct xe_guc *guc, > * in order which they currently are, if that changes the algorithm will > * need to be updated. > */ > + > + xe_device_mem_access_get(gt->xe); > mutex_lock(&guc->ct.lock); > seqno = gt->tlb_invalidation.seqno; > if (fence) { > @@ -140,6 +143,7 @@ static int send_tlb_invalidation(struct xe_guc *guc, > if (ret < 0 && fence) > invalidation_fence_signal(fence); > mutex_unlock(&guc->ct.lock); > + xe_device_mem_access_put(gt->xe); > > return ret; > } > diff --git a/drivers/gpu/drm/xe/xe_guc_ct.c b/drivers/gpu/drm/xe/xe_guc_ct.c > index 9055ff133a7c..579d7f341f13 100644 > --- a/drivers/gpu/drm/xe/xe_guc_ct.c > +++ b/drivers/gpu/drm/xe/xe_guc_ct.c > @@ -494,26 +494,22 @@ static int __guc_ct_send_locked(struct xe_guc_ct *ct, const u32 *action, > } > } > > - xe_device_mem_access_get(ct_to_xe(ct)); > retry: > ret = has_room(ct, len + GUC_CTB_HDR_LEN, g2h_len); > if (unlikely(ret)) > - goto put_wa; > + goto out; > > ret = h2g_write(ct, action, len, g2h_fence ? g2h_fence->seqno : 0, > !!g2h_fence); > if (unlikely(ret)) { > if (ret == -EAGAIN) > goto retry; > - goto put_wa; > + goto out; > } > > g2h_reserve_space(ct, g2h_len, num_g2h); > xe_guc_notify(ct_to_guc(ct)); > -put_wa: > - xe_device_mem_access_put(ct_to_xe(ct)); > out: > - > return ret; > } > > @@ -535,6 +531,7 @@ static int guc_ct_send_locked(struct xe_guc_ct *ct, const u32 *action, u32 len, > > XE_BUG_ON(g2h_len && g2h_fence); > lockdep_assert_held(&ct->lock); > + xe_device_assert_mem_access(ct_to_xe(ct)); > > try_again: > ret = __guc_ct_send_locked(ct, action, len, g2h_len, num_g2h, > @@ -602,10 +599,14 @@ static int guc_ct_send(struct xe_guc_ct *ct, const u32 *action, u32 len, > > XE_BUG_ON(g2h_len && g2h_fence); > > + xe_device_mem_access_get(ct_to_xe(ct)); > + > mutex_lock(&ct->lock); > ret = guc_ct_send_locked(ct, action, len, g2h_len, num_g2h, g2h_fence); > mutex_unlock(&ct->lock); > > + xe_device_mem_access_put(ct_to_xe(ct)); > + > return ret; > } > > -- > 2.40.1 >