Am 14.11.2016 um 00:03 schrieb Stefan Christ:
Hi, this is the second version of the refactoring work suggested by Daniel Vetter in the email: https://lists.freedesktop.org/archives/dri-devel/2016-July/113237.html The define DRM_FB_HELPER_DEFAULT_OPS provides the drm_fb_helper default implementations for functions in struct fb_ops. A drm driver can use it like: static struct fb_ops drm_fbdev_cma_ops = { .owner = THIS_MODULE, DRM_FB_HELPER_DEFAULT_OPS, /* driver specific implementations */ };
Looks good to me in general, but I've got one question. Why didn't you put the owner field into the macro as well?
I have strong doubts that anybody would set anything else than THIS_MODULE here.
Regards, Christian.
The first patch that adds the helper define was already accepted into the kernel release candidate v4.9-rc1. For the new stuff: Implementing fb_debug_enter and fb_debug_leave callbacks is not without pitfalls. It was necessary that driver also implements the mode_set_base_atomic callback in struct drm_crtc_helper_funcs. Otherwise it may segfaults. The first patch in this series changes that and introduces a check for it. Now functions drm_fb_helper_debug_(enter|leave) can be used without implementing callback mode_set_base_atomic. At least the drm drivers armada, ast, qxl, udl and virtio are affected by this issue. It seems that the code path was never executed for these drivers and nobody noticed. Refactoring patches for ast, cirrus and mgag200 are dropped. I leave it to Daniel if additional drivers should not be converted. This series is based on tag v4.9-rc1 and applies cleanly onto drm-next, too. Kind regards, Stefan Christ Changes in v2: - add patch 'fix segfaults in drm_fb_helper_debug_*'. So functions drm_fb_helper_debug_(enter|leave) can be used unconditionally. - add patch 'add fb_debug_* to DRM_FB_HELPER_DEFAULT_OPS' that uses drm_fb_helper_debug_(enter|leave) to implement fb_debug_(enter|leave). It depends on the previous patch to avoid regressions. - Drop patches for drm drivers 'ast', 'cirrus' and 'mgag200' since they reimplemented most of the fb_ops callbacks anyway. - Also remove fb_debug_(enter|leave) callbacks in fb_ops all refactoring patches. Stefan Christ (18): drm/fb-helper: fix segfaults in drm_fb_helper_debug_* drm/fb-helper: add fb_debug_* to DRM_FB_HELPER_DEFAULT_OPS drm/amdgpu: use DRM_FB_HELPER_DEFAULT_OPS for fb_ops drm/armada: use DRM_FB_HELPER_DEFAULT_OPS for fb_ops drm/bochs: use DRM_FB_HELPER_DEFAULT_OPS for fb_ops drm/fb_cma_helper: use DRM_FB_HELPER_DEFAULT_OPS for fb_ops drm/exynos: use DRM_FB_HELPER_DEFAULT_OPS for fb_ops drm/nouveau: use DRM_FB_HELPER_DEFAULT_OPS for fb_ops drm/qxl: use DRM_FB_HELPER_DEFAULT_OPS for fb_ops drm/rockchip: use DRM_FB_HELPER_DEFAULT_OPS for fb_ops drm/radeon: use DRM_FB_HELPER_DEFAULT_OPS for fb_ops drm/tegra: use DRM_FB_HELPER_DEFAULT_OPS for fb_ops drm/udl: use DRM_FB_HELPER_DEFAULT_OPS for fb_ops drm/msm: use DRM_FB_HELPER_DEFAULT_OPS for fb_ops drm/virtio: use DRM_FB_HELPER_DEFAULT_OPS for fb_ops drm/omapdrm: use DRM_FB_HELPER_DEFAULT_OPS for fb_ops drm/i915: use DRM_FB_HELPER_DEFAULT_OPS for fb_ops drm/gma500: use DRM_FB_HELPER_DEFAULT_OPS for fb_ops drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c | 8 +------- drivers/gpu/drm/armada/armada_fbdev.c | 8 +------- drivers/gpu/drm/bochs/bochs_fbdev.c | 6 +----- drivers/gpu/drm/drm_fb_cma_helper.c | 6 +----- drivers/gpu/drm/drm_fb_helper.c | 6 ++++++ drivers/gpu/drm/exynos/exynos_drm_fbdev.c | 6 +----- drivers/gpu/drm/gma500/framebuffer.c | 12 +++--------- drivers/gpu/drm/i915/intel_fbdev.c | 5 +---- drivers/gpu/drm/msm/msm_fbdev.c | 7 +------ drivers/gpu/drm/nouveau/nouveau_fbcon.c | 16 ++-------------- drivers/gpu/drm/omapdrm/omap_fbdev.c | 5 +---- drivers/gpu/drm/qxl/qxl_fb.c | 8 +------- drivers/gpu/drm/radeon/radeon_fb.c | 8 +------- drivers/gpu/drm/rockchip/rockchip_drm_fbdev.c | 6 +----- drivers/gpu/drm/tegra/fb.c | 6 +----- drivers/gpu/drm/udl/udl_fb.c | 8 +------- drivers/gpu/drm/virtio/virtgpu_fb.c | 8 +------- include/drm/drm_fb_helper.h | 4 +++- 18 files changed, 28 insertions(+), 105 deletions(-)
_______________________________________________ dri-devel mailing list dri-devel@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/dri-devel