Received-SPF: pass (google.com: domain of linux-kernel+bounces-156443-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:4601:e00::3 as permitted sender) client-ip=2604:1380:4601:e00::3;
Date: Wed, 24 Apr 2024 09:04:29 +0200
From: Boris Brezillon <boris.brezillon@collabora.com>
To: =?UTF-8?B?QWRyacOhbg==?= Larumbe <adrian.larumbe@collabora.com>
Cc: Qiang Yu <yuq825@gmail.com>, Maarten Lankhorst
 <maarten.lankhorst@linux.intel.com>, Maxime Ripard <mripard@kernel.org>,
 Thomas Zimmermann <tzimmermann@suse.de>, David Airlie <airlied@gmail.com>,
 Daniel Vetter <daniel@ffwll.ch>, Rob Herring <robh@kernel.org>, Steven
 Price <steven.price@arm.com>, Sumit Semwal <sumit.semwal@linaro.org>,
 Christian =?UTF-8?B?S8O2bmln?= <christian.koenig@amd.com>, Dmitry Osipenko
 <dmitry.osipenko@collabora.com>, Zack Rusin <zack.rusin@broadcom.com>,
 kernel@collabora.com, dri-devel@lists.freedesktop.org,
 lima@lists.freedesktop.org, linux-kernel@vger.kernel.org,
 linux-media@vger.kernel.org, linaro-mm-sig@lists.linaro.org
Subject: Re: [PATCH v2] drm/panfrost: Fix dma_resv deadlock at drm object
 pin time
Message-ID: <20240424090429.57de7d1c@collabora.com>
In-Reply-To: <20240423224651.138163-1-adrian.larumbe@collabora.com>
References: <20240423224651.138163-1-adrian.larumbe@collabora.com>
Organization: Collabora
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

On Tue, 23 Apr 2024 23:46:23 +0100
Adri=C3=A1n Larumbe <adrian.larumbe@collabora.com> wrote:

> When Panfrost must pin an object that is being prepared a dma-buf
> attachment for on behalf of another driver, the core drm gem object pinni=
ng
> code already takes a lock on the object's dma reservation.
>=20
> However, Panfrost GEM object's pinning callback would eventually try taki=
ng
> the lock on the same dma reservation when delegating pinning of the object
> onto the shmem subsystem, which led to a deadlock.
>=20
> This can be shown by enabling CONFIG_DEBUG_WW_MUTEX_SLOWPATH, which throws
> the following recursive locking situation:
>=20
> weston/3440 is trying to acquire lock:
> ffff000000e235a0 (reservation_ww_class_mutex){+.+.}-{3:3}, at: drm_gem_sh=
mem_pin+0x34/0xb8 [drm_shmem_helper]
> but task is already holding lock:
> ffff000000e235a0 (reservation_ww_class_mutex){+.+.}-{3:3}, at: drm_gem_pi=
n+0x2c/0x80 [drm]
>=20
> Fix it by assuming the object's reservation had already been locked by the
> time we reach panfrost_gem_pin.
>=20
> Do the same thing for the Lima driver, as it most likely suffers from the
> same issue.
>=20
> Cc: Thomas Zimmermann <tzimmermann@suse.de>
> Cc: Dmitry Osipenko <dmitry.osipenko@collabora.com>
> Cc: Boris Brezillon <boris.brezillon@collabora.com>
> Cc: Steven Price <steven.price@arm.com>
> Fixes: a78027847226 ("drm/gem: Acquire reservation lock in drm_gem_{pin/u=
npin}()")
> Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
> Signed-off-by: Adri=C3=A1n Larumbe <adrian.larumbe@collabora.com>
> ---
>  drivers/gpu/drm/lima/lima_gem.c         | 9 +++++++--
>  drivers/gpu/drm/panfrost/panfrost_gem.c | 8 +++++++-
>  2 files changed, 14 insertions(+), 3 deletions(-)
>=20
> diff --git a/drivers/gpu/drm/lima/lima_gem.c b/drivers/gpu/drm/lima/lima_=
gem.c
> index 7ea244d876ca..8a5bcf498ef6 100644
> --- a/drivers/gpu/drm/lima/lima_gem.c
> +++ b/drivers/gpu/drm/lima/lima_gem.c
> @@ -184,8 +184,13 @@ static int lima_gem_pin(struct drm_gem_object *obj)
> =20
>  	if (bo->heap_size)
>  		return -EINVAL;
> -
> -	return drm_gem_shmem_pin(&bo->base);
> +	/*
> +	 * Pinning can only happen in response to a prime attachment request
> +	 * from another driver, but dma reservation locking is already being
> +	 * handled by drm_gem_pin

This comment looks a bit weird now that you call a function that
doesn't have the _locked suffix. I'd be tempted to drop it, or clarify
the fact drm_gem_shmem_object_pin() expects the resv lock to be held.

> +	 */
> +	drm_WARN_ON(obj->dev, obj->import_attach);

Should this be moved to drm_gem_shmem_[un]pin_locked() instead of being
duplicated in all overloads of ->[un]pin()?

> +	return drm_gem_shmem_object_pin(obj);
>  }
> =20
>  static int lima_gem_vmap(struct drm_gem_object *obj, struct iosys_map *m=
ap)
> diff --git a/drivers/gpu/drm/panfrost/panfrost_gem.c b/drivers/gpu/drm/pa=
nfrost/panfrost_gem.c
> index d47b40b82b0b..e3fbcb020617 100644
> --- a/drivers/gpu/drm/panfrost/panfrost_gem.c
> +++ b/drivers/gpu/drm/panfrost/panfrost_gem.c
> @@ -192,7 +192,13 @@ static int panfrost_gem_pin(struct drm_gem_object *o=
bj)
>  	if (bo->is_heap)
>  		return -EINVAL;
> =20
> -	return drm_gem_shmem_pin(&bo->base);
> +	/*
> +	 * Pinning can only happen in response to a prime attachment request
> +	 * from another driver, but dma reservation locking is already being
> +	 * handled by drm_gem_pin
> +	 */
> +	drm_WARN_ON(obj->dev, obj->import_attach);
> +	return drm_gem_shmem_object_pin(obj);
>  }
> =20
>  static enum drm_gem_object_status panfrost_gem_status(struct drm_gem_obj=
ect *obj)