[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <7c41b175-6da2-6801-43c2-c4038c08b228@amd.com>
Date: Thu, 25 Feb 2021 13:59:11 +0100
From: Christian König <christian.koenig@....com>
To: Arnd Bergmann <arnd@...nel.org>, Ben Skeggs <bskeggs@...hat.com>,
David Airlie <airlied@...ux.ie>,
Daniel Vetter <daniel@...ll.ch>
Cc: Arnd Bergmann <arnd@...db.de>, Lyude Paul <lyude@...hat.com>,
Thomas Zimmermann <tzimmermann@...e.de>,
dri-devel@...ts.freedesktop.org, nouveau@...ts.freedesktop.org,
linux-kernel@...r.kernel.org
Subject: Re: [PATCH] drm/nouveau/pci: rework AGP dependency
Am 25.02.21 um 13:52 schrieb Arnd Bergmann:
> From: Arnd Bergmann <arnd@...db.de>
>
> I noticed a warning from 'nm' when CONFIG_TRIM_UNUSED_KSYMS is set
> and IS_REACHABLE(CONFIG_AGP) is false:
>
> drivers/gpu/drm/nouveau/nvkm/subdev/pci/agp.o: no symbols
>
> I later found this is completely harmless and we should find a way
> to suppress the warning, but at that point I had already done a
> cleanup patch to address this instance.
>
> It turns out this code could be improved anyway, as the current version
> behaves unexpectedly when AGP is a loadable module but nouveau is built-in
> itself, in which case it silently omits agp support.
>
> A better way to handle this is with a Kconfig dependency that requires
> AGP either to be disabled, or forces nouveau to be a module for AGP=m.
> With this change, the compile-time hack can be removed and lld no
> longer warns.
>
> Fixes: 340b0e7c500a ("drm/nouveau/pci: merge agp handling from nouveau drm")
> Signed-off-by: Arnd Bergmann <arnd@...db.de>
> ---
> drivers/gpu/drm/nouveau/Kbuild | 1 +
> drivers/gpu/drm/nouveau/Kconfig | 1 +
> drivers/gpu/drm/nouveau/nvkm/subdev/pci/Kbuild | 2 +-
> drivers/gpu/drm/nouveau/nvkm/subdev/pci/agp.c | 2 --
> drivers/gpu/drm/nouveau/nvkm/subdev/pci/agp.h | 9 +++++----
> 5 files changed, 8 insertions(+), 7 deletions(-)
>
> diff --git a/drivers/gpu/drm/nouveau/Kbuild b/drivers/gpu/drm/nouveau/Kbuild
> index 60586fb8275e..173b8d9d85e3 100644
> --- a/drivers/gpu/drm/nouveau/Kbuild
> +++ b/drivers/gpu/drm/nouveau/Kbuild
> @@ -15,6 +15,7 @@ nouveau-y := $(nvif-y)
> #- code also used by various userspace tools/tests
> include $(src)/nvkm/Kbuild
> nouveau-y += $(nvkm-y)
> +nouveau-m += $(nvkm-m)
>
> # DRM - general
> ifdef CONFIG_X86
> diff --git a/drivers/gpu/drm/nouveau/Kconfig b/drivers/gpu/drm/nouveau/Kconfig
> index 278e048235a9..90276a557a70 100644
> --- a/drivers/gpu/drm/nouveau/Kconfig
> +++ b/drivers/gpu/drm/nouveau/Kconfig
> @@ -2,6 +2,7 @@
> config DRM_NOUVEAU
> tristate "Nouveau (NVIDIA) cards"
> depends on DRM && PCI && MMU
> + depends on AGP || !AGP
My first thought was WTF? But then I realized that this totally makes sense.
We should probably have the same for radeon as well.
Apart from that the patch is Acked-by: Christian König
<christian.koenig@....com>
> select IOMMU_API
> select FW_LOADER
> select DRM_KMS_HELPER
> diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/pci/Kbuild b/drivers/gpu/drm/nouveau/nvkm/subdev/pci/Kbuild
> index 174bdf995271..a400c680cf65 100644
> --- a/drivers/gpu/drm/nouveau/nvkm/subdev/pci/Kbuild
> +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/pci/Kbuild
> @@ -1,5 +1,5 @@
> # SPDX-License-Identifier: MIT
> -nvkm-y += nvkm/subdev/pci/agp.o
> +nvkm-$(CONFIG_AGP) += nvkm/subdev/pci/agp.o
> nvkm-y += nvkm/subdev/pci/base.o
> nvkm-y += nvkm/subdev/pci/pcie.o
> nvkm-y += nvkm/subdev/pci/nv04.o
> diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/pci/agp.c b/drivers/gpu/drm/nouveau/nvkm/subdev/pci/agp.c
> index 385a90f91ed6..86c9e1d658af 100644
> --- a/drivers/gpu/drm/nouveau/nvkm/subdev/pci/agp.c
> +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/pci/agp.c
> @@ -20,7 +20,6 @@
> * OTHER DEALINGS IN THE SOFTWARE.
> */
> #include "agp.h"
> -#ifdef __NVKM_PCI_AGP_H__
> #include <core/option.h>
>
> struct nvkm_device_agp_quirk {
> @@ -172,4 +171,3 @@ nvkm_agp_ctor(struct nvkm_pci *pci)
>
> pci->agp.mtrr = arch_phys_wc_add(pci->agp.base, pci->agp.size);
> }
> -#endif
> diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/pci/agp.h b/drivers/gpu/drm/nouveau/nvkm/subdev/pci/agp.h
> index ad4d3621d02b..041fe1fbf093 100644
> --- a/drivers/gpu/drm/nouveau/nvkm/subdev/pci/agp.h
> +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/pci/agp.h
> @@ -1,15 +1,14 @@
> -/* SPDX-License-Identifier: MIT */
> -#include "priv.h"
> -#if defined(CONFIG_AGP) || (defined(CONFIG_AGP_MODULE) && defined(MODULE))
> #ifndef __NVKM_PCI_AGP_H__
> #define __NVKM_PCI_AGP_H__
>
> +/* SPDX-License-Identifier: MIT */
> +#include "priv.h"
> +#if IS_ENABLED(CONFIG_AGP)
> void nvkm_agp_ctor(struct nvkm_pci *);
> void nvkm_agp_dtor(struct nvkm_pci *);
> void nvkm_agp_preinit(struct nvkm_pci *);
> int nvkm_agp_init(struct nvkm_pci *);
> void nvkm_agp_fini(struct nvkm_pci *);
> -#endif
> #else
> static inline void nvkm_agp_ctor(struct nvkm_pci *pci) {}
> static inline void nvkm_agp_dtor(struct nvkm_pci *pci) {}
> @@ -17,3 +16,5 @@ static inline void nvkm_agp_preinit(struct nvkm_pci *pci) {}
> static inline int nvkm_agp_init(struct nvkm_pci *pci) { return -ENOSYS; }
> static inline void nvkm_agp_fini(struct nvkm_pci *pci) {}
> #endif
> +
> +#endif
Powered by blists - more mailing lists