<div dir="ltr">Hi Daniel, sorry for being late. <div><div class="gmail_extra"> <div class="gmail_quote">2013/7/16 Daniel Vetter <<a href="mailto:daniel.vetter@ffwll.ch" target="_blank">daniel.vetter@ffwll.ch</a>> <blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Hi all, This patch series is my 2nd real stab at fixing up the locking issues around our two buffer sharing mechanisms in gem: flink and prime. I think the approach taken here is much better than my first stab, and it also seems to no longer leak buffers ;-) There some assorted cleanup and prep work (and one i915 fix) thrown into the mix, it's all stuff I've stumbled over while digging through the code. Open issues left in prime-land after these patches: - exynos probably wants a similar patch to "drm/i915: explicit store base gem object in dma_buf->priv". The current code should be correct, but it's a bit </blockquote><div> </div><div>How about using stuffs of drm_prime instead of specific ones? Seem like that we could replace specific dmabuf stuffs with common ones of drm_prime, at least in case of Exynos: i.e. each driver can export a gem to a dmabuf through drm_gem_prime_export function of drm_prime instead of specific one. By doing so, I think we could remove duplicated codes of drivers, specific dmabuf stuffs. I'm not sure but it seems like that there is any reason you try to use existing stuffs with a little change: maybe the stuffs of drm_prime couldn't be used for all drm drivers commonly. </div><div>Thanks, Inki Dae </div><div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"> tricky. I've opted not to do that since last time around I've touched exynos a </blockquote><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"> bit it broke horribly ;-) </blockquote><div></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"> - The prime core should now no longer depend upon obj->import_attach being set by drivers in their prime_import callback. This should allos us to fix udl which really doesn't need (nor want, it confuses swiotlb among other things) a device attachment. Didn't write that patch since my displaylink seems to have died. - There's still the issue Inki's team pointed out where if you import a foreign object on different fds you'll get different gem objects. So we need some form of a per-device import cache (on top of the per-file-priv dma-buf cache we already have). Didn't do this yet since I want to have good test coverage (already started a bit), it looks like a bit more work and I'm not sure about the exact design of the code yet. Review and testing highly welcome. Cheers, Daniel Daniel Vetter (20): drm: use common drm_gem_dmabuf_release in i915/exynos drivers drm/i915: unpin backing storage in dmabuf_unmap drm/i915: explicit store base gem object in dma_buf->priv drm/prime: add a bit of documentation about gem_obj->import_attach drm/gem: remove drm_gem_object_handle_unreference drm/gem: inline drm_gem_object_handle_reference drm/gem: move drm_gem_object_handle_unreference_unlocked into drm_gem.c drm/gem: remove bogus NULL check from drm_gem_object_handle_unreference_unlocked drm/gem: WARN about unbalanced handle refcounts drm/gem: fix up flink name create race drm/prime: fix error path in drm_gem_prime_fd_to_handle drm/gem: make drm_gem_object_handle_unreference_unlocked static drm/gem: create drm_gem_dumb_destroy drm/prime: use proper pointer in drm_gem_prime_handle_to_fd drm/prime: shrink critical section protected by prime lock drm/prime: clarify logic a bit in drm_gem_prime_fd_to_handle drm/gem: switch dev->object_name_lock to a mutex drm/gem: completely close gem_open vs. gem_close races drm/prime: proper locking+refcounting for obj->dma_buf link drm/prime: Simplify drm_gem_remove_prime_handles drivers/gpu/drm/ast/ast_drv.c | 2 +- drivers/gpu/drm/ast/ast_drv.h | 3 - drivers/gpu/drm/ast/ast_main.c | 7 -- drivers/gpu/drm/cirrus/cirrus_drv.c | 2 +- drivers/gpu/drm/cirrus/cirrus_drv.h | 3 - drivers/gpu/drm/cirrus/cirrus_main.c | 7 -- drivers/gpu/drm/drm_fops.c | 1 + drivers/gpu/drm/drm_gem.c | 192 ++++++++++++++++++++--------- drivers/gpu/drm/drm_gem_cma_helper.c | 10 -- drivers/gpu/drm/drm_info.c | 2 +- drivers/gpu/drm/drm_prime.c | 96 ++++++++++----- drivers/gpu/drm/exynos/exynos_drm_dmabuf.c | 23 +--- drivers/gpu/drm/exynos/exynos_drm_drv.c | 2 +- drivers/gpu/drm/exynos/exynos_drm_gem.c | 22 +--- drivers/gpu/drm/exynos/exynos_drm_gem.h | 9 -- drivers/gpu/drm/gma500/gem.c | 17 --- drivers/gpu/drm/gma500/psb_drv.c | 2 +- drivers/gpu/drm/gma500/psb_drv.h | 2 - drivers/gpu/drm/i915/i915_drv.c | 2 +- drivers/gpu/drm/i915/i915_drv.h | 2 - drivers/gpu/drm/i915/i915_gem.c | 7 -- drivers/gpu/drm/i915/i915_gem_dmabuf.c | 34 +++-- drivers/gpu/drm/mgag200/mgag200_drv.c | 2 +- drivers/gpu/drm/mgag200/mgag200_drv.h | 3 - drivers/gpu/drm/mgag200/mgag200_main.c | 7 -- drivers/gpu/drm/nouveau/nouveau_display.c | 7 -- drivers/gpu/drm/nouveau/nouveau_display.h | 2 - drivers/gpu/drm/nouveau/nouveau_drm.c | 2 +- drivers/gpu/drm/omapdrm/omap_drv.c | 2 +- drivers/gpu/drm/omapdrm/omap_drv.h | 2 - drivers/gpu/drm/omapdrm/omap_gem.c | 15 --- drivers/gpu/drm/qxl/qxl_drv.c | 2 +- drivers/gpu/drm/qxl/qxl_drv.h | 3 - drivers/gpu/drm/qxl/qxl_dumb.c | 7 -- drivers/gpu/drm/radeon/radeon.h | 3 - drivers/gpu/drm/radeon/radeon_drv.c | 5 +- drivers/gpu/drm/radeon/radeon_gem.c | 7 -- drivers/gpu/drm/rcar-du/rcar_du_drv.c | 2 +- drivers/gpu/drm/shmobile/shmob_drm_drv.c | 2 +- drivers/gpu/drm/tilcdc/tilcdc_drv.c | 2 +- drivers/gpu/drm/udl/udl_drv.c | 2 +- drivers/gpu/drm/udl/udl_drv.h | 2 - drivers/gpu/drm/udl/udl_gem.c | 6 - drivers/gpu/host1x/drm/drm.c | 2 +- drivers/gpu/host1x/drm/gem.c | 6 - drivers/gpu/host1x/drm/gem.h | 2 - drivers/staging/imx-drm/imx-drm-core.c | 2 +- include/drm/drmP.h | 94 +++++++------- include/drm/drm_gem_cma_helper.h | 8 -- 49 files changed, 279 insertions(+), 367 deletions(-) -- 1.8.3.2 _______________________________________________ dri-devel mailing list <a href="mailto:dri-devel@lists.freedesktop.org">dri-devel@lists.freedesktop.org</a> <a href="http://lists.freedesktop.org/mailman/listinfo/dri-devel" target="_blank">http://lists.freedesktop.org/mailman/listinfo/dri-devel</a> </blockquote></div> </div></div></div>