Import drm from Linux v5.6-rc3 (commit f8788d86ab28f61f7b46eb6be375f8a726783636)

drivers/gpu/drm -> sys/external/bsd/drm2/dist/drm
include/drm -> sys/external/bsd/drm2/dist/include/drm
include/uapi/drm -> sys/external/bsd/drm2/dist/include/uapi/drm

GPL exclusions in dist/drm:

- amd/amdgpu/amdgpu_atpx_handler.c
- arc/
- arm/
- armada/
- aspeed/
- atmel-hlcdc/
- bochs/
- bridge/
- cirrus/
- drm_dp_cec.c
- drm_dp_mst_topology_internal.h
- drm_edid_load.c
- drm_format_helper.c
- drm_gem_cma_helper.c
- drm_gem_framebuffer_helper.c
- drm_gem_shmem_helper.c
- drm_gem_ttm_helper.c
- drm_gem_vram_helper.c
- drm_hdcp.c
- drm_lease.c
- drm_mipi_dbi.c
- drm_simple_kms_helper.c
- drm_sysfs.c
- drm_trace.h
- drm_vram_helper_common.c
- drm_writeback.c
- etnaviv/
- exynos/
- fsl-dcu/
- gma500/
- hisilicon/
- i2c/tda9950.c
- i2c/tda998x_drv.c
- i915/gt/selftest_context.c
- i915/gt/selftest_engine.c
- i915/gt/selftest_engine.h
- i915/gt/selftest_engine_cs.c
- i915/gt/selftest_engine_pm.c
- i915/i915_trace.h
- i915/selftests/i915_live_selftests.h
- i915/selftests/i915_mock_selftests.h
- i915/selftests/i915_perf_selftests.h
- i915/selftests/lib_sw_fence.h
- imx/
- ingenic/
- lima/
- mcde/
- mediatek/
- meson/
- mgag200/
- msm/
- mxsfb/
- omapdrm/
- panel/
- panfrost/
- pl111/
- radeon/radeon_atpx_handler.c
- rcar-du/
- rockchip/
- selftests/drm_cmdline_selftests.h
- selftests/drm_modeset_selftests.h
- selftests/test-drm_cmdline_parser.c
- selftests/test-drm_damage_helper.c
- selftests/test-drm_dp_mst_helper.c
- selftests/test-drm_format.c
- selftests/test-drm_framebuffer.c
- selftests/test-drm_modeset_common.c
- selftests/test-drm_modeset_common.h
- selftests/test-drm_plane_helper.c
- selftests/test-drm_rect.c
- shmobile/
- sti/
- stm/
- sun4i/
- tegra/
- tilcdc/
- tiny/
- tve200/
- udl/
- v3d/
- vc4/
- virtio/virtgpu_trace.h
- virtio/virtgpu_trace_points.c
- vkms/
- vmwgfx/device_include/vmware_pack_begin.h
- vmwgfx/device_include/vmware_pack_end.h
- zte/

Exceptions -- these all appear to be files to which someone added
an SPDX license header automatically by a script that is not aware
of the default MIT licensing under drivers/gpu/drm:

- ast/ast_dp501.c
- ast/ast_dram_tables.h
- lib/drm_random.c
- lib/drm_random.h
- i915/display/intel_acpi.c
- i915/selftests/mock_gem_device.h
- i915/selftests/i915_mock_selftests.h
- i915/selftests/i915_live_selftests.h
- r128/ati_pcigart.h
- selftests/drm_mm_selftests.h
- selftests/test-drm_mm.c
- vmwgfx/device_include/vm_basic_types.h

GPL exclusions in dist/include/drm:

- bridge
- drm_client.h
- drm_fb_cma_helper.h
- drm_format_helper.h
- drm_gem_cma_helper.h
- drm_gem_shmem_helper.h
- drm_gem_ttm_helper.h
- drm_gem_vram_helper.h
- drm_lease.h
- drm_mipi_dbi.h
- drm_mipi_dsi.h
- drm_of.h
- drm_simple_kms_helper.h
- drm_sysfs.h
- drm_writeback.h
- gma_drm.h
- i2c/tda998x.h
- i915_mei_hdcp_interface.h
- intel-gtt.h

Exceptions:
- drm_agpsupport.h - was in original drm; wrong spdx header auto-added

GPL exclusions in dist/include/uapi/drm:

- armada_drm.h
- etnaviv_drm.h
- exynos_drm.h
- lima_drm.h
- omap_drm.h

Exceptions:
- i810_drm.h - was in original drm; spdx header is wrong

Merge linux-drm-v5-6-rc3

merge linux-drm-v4-4-143

branches: 1.1.1;
Initial revision

Import drm from Linux v5.6-rc3 (commit f8788d86ab28f61f7b46eb6be375f8a726783636)

drivers/gpu/drm -> sys/external/bsd/drm2/dist/drm
include/drm -> sys/external/bsd/drm2/dist/include/drm
include/uapi/drm -> sys/external/bsd/drm2/dist/include/uapi/drm

GPL exclusions in dist/drm:

- amd/amdgpu/amdgpu_atpx_handler.c
- arc/
- arm/
- armada/
- aspeed/
- atmel-hlcdc/
- bochs/
- bridge/
- cirrus/
- drm_dp_cec.c
- drm_dp_mst_topology_internal.h
- drm_edid_load.c
- drm_format_helper.c
- drm_gem_cma_helper.c
- drm_gem_framebuffer_helper.c
- drm_gem_shmem_helper.c
- drm_gem_ttm_helper.c
- drm_gem_vram_helper.c
- drm_hdcp.c
- drm_lease.c
- drm_mipi_dbi.c
- drm_simple_kms_helper.c
- drm_sysfs.c
- drm_trace.h
- drm_vram_helper_common.c
- drm_writeback.c
- etnaviv/
- exynos/
- fsl-dcu/
- gma500/
- hisilicon/
- i2c/tda9950.c
- i2c/tda998x_drv.c
- i915/gt/selftest_context.c
- i915/gt/selftest_engine.c
- i915/gt/selftest_engine.h
- i915/gt/selftest_engine_cs.c
- i915/gt/selftest_engine_pm.c
- i915/i915_trace.h
- i915/selftests/i915_live_selftests.h
- i915/selftests/i915_mock_selftests.h
- i915/selftests/i915_perf_selftests.h
- i915/selftests/lib_sw_fence.h
- imx/
- ingenic/
- lima/
- mcde/
- mediatek/
- meson/
- mgag200/
- msm/
- mxsfb/
- omapdrm/
- panel/
- panfrost/
- pl111/
- radeon/radeon_atpx_handler.c
- rcar-du/
- rockchip/
- selftests/drm_cmdline_selftests.h
- selftests/drm_modeset_selftests.h
- selftests/test-drm_cmdline_parser.c
- selftests/test-drm_damage_helper.c
- selftests/test-drm_dp_mst_helper.c
- selftests/test-drm_format.c
- selftests/test-drm_framebuffer.c
- selftests/test-drm_modeset_common.c
- selftests/test-drm_modeset_common.h
- selftests/test-drm_plane_helper.c
- selftests/test-drm_rect.c
- shmobile/
- sti/
- stm/
- sun4i/
- tegra/
- tilcdc/
- tiny/
- tve200/
- udl/
- v3d/
- vc4/
- virtio/virtgpu_trace.h
- virtio/virtgpu_trace_points.c
- vkms/
- vmwgfx/device_include/vmware_pack_begin.h
- vmwgfx/device_include/vmware_pack_end.h
- zte/

Exceptions -- these all appear to be files to which someone added
an SPDX license header automatically by a script that is not aware
of the default MIT licensing under drivers/gpu/drm:

- ast/ast_dp501.c
- ast/ast_dram_tables.h
- lib/drm_random.c
- lib/drm_random.h
- i915/display/intel_acpi.c
- i915/selftests/mock_gem_device.h
- i915/selftests/i915_mock_selftests.h
- i915/selftests/i915_live_selftests.h
- r128/ati_pcigart.h
- selftests/drm_mm_selftests.h
- selftests/test-drm_mm.c
- vmwgfx/device_include/vm_basic_types.h

GPL exclusions in dist/include/drm:

- bridge
- drm_client.h
- drm_fb_cma_helper.h
- drm_format_helper.h
- drm_gem_cma_helper.h
- drm_gem_shmem_helper.h
- drm_gem_ttm_helper.h
- drm_gem_vram_helper.h
- drm_lease.h
- drm_mipi_dbi.h
- drm_mipi_dsi.h
- drm_of.h
- drm_simple_kms_helper.h
- drm_sysfs.h
- drm_writeback.h
- gma_drm.h
- i2c/tda998x.h
- i915_mei_hdcp_interface.h
- intel-gtt.h

Exceptions:
- drm_agpsupport.h - was in original drm; wrong spdx header auto-added

GPL exclusions in dist/include/uapi/drm:

- armada_drm.h
- etnaviv_drm.h
- exynos_drm.h
- lima_drm.h
- omap_drm.h

Exceptions:
- i810_drm.h - was in original drm; spdx header is wrong

Import drm from Linux v5.6-rc3 (commit f8788d86ab28f61f7b46eb6be375f8a726783636)

drivers/gpu/drm -> sys/external/bsd/drm2/dist/drm
include/drm -> sys/external/bsd/drm2/dist/include/drm
include/uapi/drm -> sys/external/bsd/drm2/dist/include/uapi/drm

GPL exclusions in dist/drm:

- amd/amdgpu/amdgpu_atpx_handler.c
- arc/
- arm/
- armada/
- aspeed/
- atmel-hlcdc/
- bochs/
- bridge/
- cirrus/
- drm_dp_cec.c
- drm_dp_mst_topology_internal.h
- drm_edid_load.c
- drm_format_helper.c
- drm_gem_cma_helper.c
- drm_gem_framebuffer_helper.c
- drm_gem_shmem_helper.c
- drm_gem_ttm_helper.c
- drm_gem_vram_helper.c
- drm_hdcp.c
- drm_lease.c
- drm_mipi_dbi.c
- drm_simple_kms_helper.c
- drm_sysfs.c
- drm_trace.h
- drm_vram_helper_common.c
- drm_writeback.c
- etnaviv/
- exynos/
- fsl-dcu/
- gma500/
- hisilicon/
- i2c/tda9950.c
- i2c/tda998x_drv.c
- i915/gt/selftest_context.c
- i915/gt/selftest_engine.c
- i915/gt/selftest_engine.h
- i915/gt/selftest_engine_cs.c
- i915/gt/selftest_engine_pm.c
- i915/i915_trace.h
- i915/selftests/i915_live_selftests.h
- i915/selftests/i915_mock_selftests.h
- i915/selftests/i915_perf_selftests.h
- i915/selftests/lib_sw_fence.h
- imx/
- ingenic/
- lima/
- mcde/
- mediatek/
- meson/
- mgag200/
- msm/
- mxsfb/
- omapdrm/
- panel/
- panfrost/
- pl111/
- radeon/radeon_atpx_handler.c
- rcar-du/
- rockchip/
- selftests/drm_cmdline_selftests.h
- selftests/drm_modeset_selftests.h
- selftests/test-drm_cmdline_parser.c
- selftests/test-drm_damage_helper.c
- selftests/test-drm_dp_mst_helper.c
- selftests/test-drm_format.c
- selftests/test-drm_framebuffer.c
- selftests/test-drm_modeset_common.c
- selftests/test-drm_modeset_common.h
- selftests/test-drm_plane_helper.c
- selftests/test-drm_rect.c
- shmobile/
- sti/
- stm/
- sun4i/
- tegra/
- tilcdc/
- tiny/
- tve200/
- udl/
- v3d/
- vc4/
- virtio/virtgpu_trace.h
- virtio/virtgpu_trace_points.c
- vkms/
- vmwgfx/device_include/vmware_pack_begin.h
- vmwgfx/device_include/vmware_pack_end.h
- zte/

Exceptions -- these all appear to be files to which someone added
an SPDX license header automatically by a script that is not aware
of the default MIT licensing under drivers/gpu/drm:

- ast/ast_dp501.c
- ast/ast_dram_tables.h
- lib/drm_random.c
- lib/drm_random.h
- i915/display/intel_acpi.c
- i915/selftests/mock_gem_device.h
- i915/selftests/i915_mock_selftests.h
- i915/selftests/i915_live_selftests.h
- r128/ati_pcigart.h
- selftests/drm_mm_selftests.h
- selftests/test-drm_mm.c
- vmwgfx/device_include/vm_basic_types.h

GPL exclusions in dist/include/drm:

- bridge
- drm_client.h
- drm_fb_cma_helper.h
- drm_format_helper.h
- drm_gem_cma_helper.h
- drm_gem_shmem_helper.h
- drm_gem_ttm_helper.h
- drm_gem_vram_helper.h
- drm_lease.h
- drm_mipi_dbi.h
- drm_mipi_dsi.h
- drm_of.h
- drm_simple_kms_helper.h
- drm_sysfs.h
- drm_writeback.h
- gma_drm.h
- i2c/tda998x.h
- i915_mei_hdcp_interface.h
- intel-gtt.h

Exceptions:
- drm_agpsupport.h - was in original drm; wrong spdx header auto-added

GPL exclusions in dist/include/uapi/drm:

- armada_drm.h
- etnaviv_drm.h
- exynos_drm.h
- lima_drm.h
- omap_drm.h

Exceptions:
- i810_drm.h - was in original drm; spdx header is wrong

Import drm from Linux v5.6-rc3 (commit f8788d86ab28f61f7b46eb6be375f8a726783636)

drivers/gpu/drm -> sys/external/bsd/drm2/dist/drm
include/drm -> sys/external/bsd/drm2/dist/include/drm
include/uapi/drm -> sys/external/bsd/drm2/dist/include/uapi/drm

GPL exclusions in dist/drm:

- amd/amdgpu/amdgpu_atpx_handler.c
- arc/
- arm/
- armada/
- aspeed/
- atmel-hlcdc/
- bochs/
- bridge/
- cirrus/
- drm_dp_cec.c
- drm_dp_mst_topology_internal.h
- drm_edid_load.c
- drm_format_helper.c
- drm_gem_cma_helper.c
- drm_gem_framebuffer_helper.c
- drm_gem_shmem_helper.c
- drm_gem_ttm_helper.c
- drm_gem_vram_helper.c
- drm_hdcp.c
- drm_lease.c
- drm_mipi_dbi.c
- drm_simple_kms_helper.c
- drm_sysfs.c
- drm_trace.h
- drm_vram_helper_common.c
- drm_writeback.c
- etnaviv/
- exynos/
- fsl-dcu/
- gma500/
- hisilicon/
- i2c/tda9950.c
- i2c/tda998x_drv.c
- i915/gt/selftest_context.c
- i915/gt/selftest_engine.c
- i915/gt/selftest_engine.h
- i915/gt/selftest_engine_cs.c
- i915/gt/selftest_engine_pm.c
- i915/i915_trace.h
- i915/selftests/i915_live_selftests.h
- i915/selftests/i915_mock_selftests.h
- i915/selftests/i915_perf_selftests.h
- i915/selftests/lib_sw_fence.h
- imx/
- ingenic/
- lima/
- mcde/
- mediatek/
- meson/
- mgag200/
- msm/
- mxsfb/
- omapdrm/
- panel/
- panfrost/
- pl111/
- radeon/radeon_atpx_handler.c
- rcar-du/
- rockchip/
- selftests/drm_cmdline_selftests.h
- selftests/drm_modeset_selftests.h
- selftests/test-drm_cmdline_parser.c
- selftests/test-drm_damage_helper.c
- selftests/test-drm_dp_mst_helper.c
- selftests/test-drm_format.c
- selftests/test-drm_framebuffer.c
- selftests/test-drm_modeset_common.c
- selftests/test-drm_modeset_common.h
- selftests/test-drm_plane_helper.c
- selftests/test-drm_rect.c
- shmobile/
- sti/
- stm/
- sun4i/
- tegra/
- tilcdc/
- tiny/
- tve200/
- udl/
- v3d/
- vc4/
- virtio/virtgpu_trace.h
- virtio/virtgpu_trace_points.c
- vkms/
- vmwgfx/device_include/vmware_pack_begin.h
- vmwgfx/device_include/vmware_pack_end.h
- zte/

Exceptions -- these all appear to be files to which someone added
an SPDX license header automatically by a script that is not aware
of the default MIT licensing under drivers/gpu/drm:

- ast/ast_dp501.c
- ast/ast_dram_tables.h
- lib/drm_random.c
- lib/drm_random.h
- i915/display/intel_acpi.c
- i915/selftests/mock_gem_device.h
- i915/selftests/i915_mock_selftests.h
- i915/selftests/i915_live_selftests.h
- r128/ati_pcigart.h
- selftests/drm_mm_selftests.h
- selftests/test-drm_mm.c
- vmwgfx/device_include/vm_basic_types.h

GPL exclusions in dist/include/drm:

- bridge
- drm_client.h
- drm_fb_cma_helper.h
- drm_format_helper.h
- drm_gem_cma_helper.h
- drm_gem_shmem_helper.h
- drm_gem_ttm_helper.h
- drm_gem_vram_helper.h
- drm_lease.h
- drm_mipi_dbi.h
- drm_mipi_dsi.h
- drm_of.h
- drm_simple_kms_helper.h
- drm_sysfs.h
- drm_writeback.h
- gma_drm.h
- i2c/tda998x.h
- i915_mei_hdcp_interface.h
- intel-gtt.h

Exceptions:
- drm_agpsupport.h - was in original drm; wrong spdx header auto-added

GPL exclusions in dist/include/uapi/drm:

- armada_drm.h
- etnaviv_drm.h
- exynos_drm.h
- lima_drm.h
- omap_drm.h

Exceptions:
- i810_drm.h - was in original drm; spdx header is wrong

Merge linux-drm-v5-6-rc3

merge linux-drm-v4-4-143

branches: 1.1.1;
Initial revision

Merge linux-drm-v5-6-rc3

merge linux-drm-v4-4-143

branches: 1.1.1;
Initial revision

Merge linux-drm-v5-6-rc3

merge linux-drm-v4-4-143

branches: 1.1.1;
Initial revision

Merge linux-drm-v5-6-rc3

merge linux-drm-v4-4-143

branches: 1.4.4; 1.4.28; 1.4.30;
Make it build and boot on my test machines.

Screen blanks on boot on the Ivy Bridge system with

DRM error in cpt_serr_int_handler: PCH transcoder A FIFO underrun

But after that everything is OK. Appears to be an upstream problem.
To investigate...

I think there's a cache flushing issue somewhere -- there are little
display artefacts on my T60.

fix merge conflicts

branches: 1.2.2; 1.2.4;
Merge riastradh-drm2 to HEAD.

branches: 1.1.1;
Initial revision

Merge linux-drm-v5-6-rc3

merge linux-drm-v4-4-143

branches: 1.1.1;
Initial revision

Merge linux-drm-v5-6-rc3

merge linux-drm-v4-4-143

branches: 1.1.1;
Initial revision

Merge linux-drm-v5-6-rc3

merge linux-drm-v4-4-143

branches: 1.1.1;
Initial revision

Merge linux-drm-v5-6-rc3

merge linux-drm-v4-4-143

branches: 1.1.1;
Initial revision

i915: Avoid freeing anything under a spin lock.

i915: Use new rb_move to relocate the active tree.

Revert "i915: Defer final wakeup on active until after retirement."

This reverts i915_active.c 1.11. ref->retire might free the object,
so touching it is not allowed -- that would use-after-free. Linux
uses the object only for its address with wake_up_var.

The reason I made this change was that I guessed i915_active_wait has
to wait until after ref->retire finishes -- after all, Linux seems to
defer the wakeup until then. However, even the Linux code doesn't
guarantee this, because i915_active_wait could be called _during_
ref->retire, and would witness ref->count == 0, and would not wait
until it has completed in that case. So maybe my guess was wrong,
and it is OK for i915_active_wait to return while ref->retire is
still in flight -- I don't see any logic that obviously requires it
to wait for ref->retire, in any case.

Or maybe something does rely on i915_active_wait to wait for
ref->retire to finish, in which case we need a different mechanism
for i915_active_release itself to wait until i915_active_retire has
woken up, without dereferencing ref since it might be dead after
ref->retire.

i915: Defer final wakeup on active until after retirement.

Not sure what I was thinking when I moved this earlier!

i915: Fix memory leak in active node allocation under #ifdef NetBSD.

Potential fix for PR kern/56573.

i915: Allow duplicates in active tree, sorted by kva.

Upstream looks like it uses insertion order, but that's not really
convenient.

i915: Abuse abstractions gruesomely but with less lock screwitude.

i915: Destroy locks in i915_active objects on free.

i915: Missed a spot in flailing the bad thing going wrong.

i915: Flail harder to work around the bad thing that went wrong.

i915: intel_engine_pm.c and i915_active.c

...with a little less heinous abuse of C, and an attempt to
disentangle the tentacular abstraction violations rampant in these
components

Hack i915_vma.c: rbtree, unmap by pa rather than vm object, PRIx64.

Merge linux-drm-v5-6-rc3

branches: 1.1.1;
Initial revision

i915: Destroy locks in i915_active objects on free.

Merge linux-drm-v5-6-rc3

branches: 1.1.1;
Initial revision

i915: intel_engine_pm.c and i915_active.c

...with a little less heinous abuse of C, and an attempt to
disentangle the tentacular abstraction violations rampant in these
components

Merge linux-drm-v5-6-rc3

branches: 1.1.1;
Initial revision

i915: LIST_HEAD -> struct list_head = LIST_HEAD_INIT

i915: Call i915_global_buddy_init from i915 module init.

i915: More includes with nonobvious side-loads.

Merge linux-drm-v5-6-rc3

branches: 1.1.1;
Initial revision

Merge linux-drm-v5-6-rc3

branches: 1.1.1;
Initial revision

i915: Omit more unnecessary local diffs.

Give up on figuring side-loading, we need bitmap_* functions.


Author: Maya Rashish <maya@NetBSD.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

PRIx64 for u64


Author: Maya Rashish <maya@NetBSD.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

NetBSD-ify clflush size


Author: Maya Rashish <maya@NetBSD.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

First pass at i915, far from complete.

HEAVILY MODIFIED IN MERGE - maya

Merge linux-drm-v5-6-rc3

Reduce diffs by side-loading some header files like Linux.

From riastradh

Eliminate -Wpointer-arith patches; use -Wno-pointer-arith instead.

branches: 1.20.2;
Use the original linux function rather than my wrong translation.

...Include the header to have it.

Thanks Riastradh!

Add what appears to be the fixes to CVE-2019-0154, CVE-2019-0155.
This commit requires review, but I'd also like it to be tested by others
while it is being reviewed.

CVE-2019-0155:
It was discovered that the Intel i915 graphics chipsets allowed userspace
to modify page table entries via writes to MMIO from the Blitter Command
Streamer and expose kernel memory information. A local attacker could use
this to expose sensitive information or possibly elevate privileges.

CVE-2019-0154:
It was discovered that the Intel i915 graphics chipsets could cause
a system hang when userspace performed a read from GT memory mapped
input output (MMIO) when the product is in certain low power states.
A local attacker could use this to cause a denial of service.

From upstream commits to linux-4.4.y:

branches: 1.18.4;
Add __diagused attribute to some diagnostic routines.

OK riastradh@

Handle uvm object reference counts in uvm_map more carefully.

Acquire a reference unconditionally first; then let uvm_map consume
it on success, and release it ourselves on failure.

As we did before -- acquiring a fresh reference on success -- another
thread might release the reference with a concurrent uvm_unmap before
we could acquire it back, thereby destroying the object.

XXX pullup-7 in part (i915_gem_fault)
XXX pullup-8 in part (i915_gem_fault)

Fix order of arguments to kcopy. @#&!$!*#

Add const qualifiers to src to compiler will detect this.

Use kcopy, not memcpy, in case of failure to page in dst. From chs@.

Forgot again to restore reference consumed by uvm_map.

Add missing errno comment while here.

For reasons beyond me, map does size/offset but unmap does start/end.

Use uvm_map directly and make sure we unmap correct length.

Fix mapping of non-page-aligned stuff in vmap_batch.

uvm_map consumes reference on success, so be sure to acquire another.

Remove wrong comment.

Just use uvm_map on the uvm object instead of futzing with pages.

Rework how we map i915 drm gem objects.

The old way made no sense at all. This way might make sense.

No arithmetic on void *.

merge linux-drm-v4-4-143

branches: 1.4.18; 1.4.20;
Avoid missing field initializers.

branches: 1.3.2; 1.3.4; 1.3.6;
Don't cast ptrdiff_t to unsigned long before formatting it with %td.

Noted by Robert Swindells and Patrick Welche.

Make it build and boot on my test machines.

Screen blanks on boot on the Ivy Bridge system with

DRM error in cpt_serr_int_handler: PCH transcoder A FIFO underrun

But after that everything is OK. Appears to be an upstream problem.
To investigate...

I think there's a cache flushing issue somewhere -- there are little
display artefacts on my T60.

branches: 1.1.1;
Initial revision

Merge linux-drm-v5-6-rc3

Undo the `enum pipe' -> `enum i915_pipe' rename.

branches: 1.3.6;
merge linux-drm-v4-4-143

branches: 1.2.18; 1.2.20;
Change enum pipe to enum i915_pipe
(enum pipe uses too global of a tag and conflicts with struct pipe).

branches: 1.1.1;
Initial revision

Merge linux-drm-v5-6-rc3

branches: 1.1.1;
Initial revision

Distribute local changes from i915_dma.c

First pass at i915, far from complete.

HEAVILY MODIFIED IN MERGE - maya

drm: Temporarily restore some local files deleted upstream.

These had local changes which needed to be distributed into other
files upstream, but that didn't happen in the merge commit. It will
happen, and the files will be deleted, in a later commit.

Merge linux-drm-v5-6-rc3

Reduce diffs by nixing DRM_SUSER macro, deleted upstream long ago.

Instead stub Linux capable(CAP_SYS_ADMIN).

Undo the `enum pipe' -> `enum i915_pipe' rename.

Use a header file hack to cope with Linux/NetBSD namespace clashes.

Currently serves for:

ALIGN
mutex_init
mutex_destroy

branches: 1.28.2;
Add what appears to be the fixes to CVE-2019-0154, CVE-2019-0155.
This commit requires review, but I'd also like it to be tested by others
while it is being reviewed.

CVE-2019-0155:
It was discovered that the Intel i915 graphics chipsets allowed userspace
to modify page table entries via writes to MMIO from the Blitter Command
Streamer and expose kernel memory information. A local attacker could use
this to expose sensitive information or possibly elevate privileges.

CVE-2019-0154:
It was discovered that the Intel i915 graphics chipsets could cause
a system hang when userspace performed a read from GT memory mapped
input output (MMIO) when the product is in certain low power states.
A local attacker could use this to cause a denial of service.

From upstream commits to linux-4.4.y:

branches: 1.27.4;
Avoid stack protector violation

'const int' is not a real constant in C and it cannot be used to define an
array with a number of elements, as it's turned into a variable length
array.

Triggered by GCC when building with more debug options.

add support for kabylake and skylake GT4 (untested) GPUs.
largely taken from openbsd and linux 4.13 trees (which
have this code identical), with mimimal porting to netbsd.

i have not installed (and thus tested) the newer referenced
firmware files.

only real local change is to fix IS_BROXTON() macro to check
the things valid in this era of drm. previous match would
attach on KBL, and then a loop would never exit.

tested on kabylake P630. needs mesa 11.x or newer for GL
to work.

ok @riastradh.

Rework PCI and AGP conditionalization.

- Push drm_pci_set_unique into driver callback.
- Eliminate drm_pci_set_unique_hook.
- Gather all drm_agp_* functions into struct drm_agp_hooks.
- Replace the nonsensical old atomic garbage by serious locking.
- Make drm_agpsupport.c its own module.
- Eliminate NDRMKMS_PCI.
- Use NAGP from "agp.h" only in drm_module.c for horrible hack.
=> See comment in file for rationale.
- Always define CONFIG_PCI=1 and CONFIG_AGP=1.
- Always go through the drm_agp_* function hooks.
- Ifdef out nouveau agp stuff that doesn't go through drm_agp_*
for reasons that I'm too frustrated to figure out tonight.
- pci_iomap no longer automagically does agp_i810_borrow.
=> Use drm_agp_borrow instead.

Work around ioctl ABI mistake in i915 drm.

A flags argument was added to the I915_GEM_MMAP ioctl. The version
of xf86-video-intel we imported was compiled to use the new ioctl
structure including the flags argument, but without initializing the
flags argument, so it would be userland stack garbage causing
i915_gem_mmap_ioctl to fail.

Newer xf86-video-intel will query the mmap version first before
trying to use either the older ioctl structure, with the original
shorter argument and thus no stack garbage (which will be padded with
to the full length in the kernel by drm_ioctl), or the newer ioctl
structure, with the argument initialized.

libdrm does not appear to be affected by this -- it zeroes the longer
structure before passing it in.

Patch from mrg@, diagnosis and explanation by me.

Fold local intel_uncore_destroy into upstream intel_uncore_fini.

Add some shrinker and oom notifier stubs.

tyop

provide pr_notice
avoid needing unregister_oom_notifier
use intel_register_dsm_handler(dev), as older drm does
provide dummy async_schedule
provide atomic_or
match side-loading


Author: coypu <coypu@sdf.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

merge linux-drm-v4-4-143

branches: 1.18.16; 1.18.18;
fix the same bug on the linux side, print the error, and return the -tive
error to mimick linux.

Zero out the guard for bus_space_unmap before calling i915_dma_cleanup() which
calls i915_free_hws(), which then tries to unmap. Perhaps this fixes PR/50060.

Use pmap_pv(9) to remove mappings of Intel graphics aperture pages.

Proposed on tech-kern with no objections:

https://mail-index.netbsd.org/tech-kern/2015/03/26/msg018561.html

Further background at:

https://mail-index.netbsd.org/tech-kern/2014/07/23/msg017392.html

New macro DRM_SPIN_WAIT_ON better reflects DRM_WAIT_ON.

We still need to adapt all waits from upstream to use an interlock,
so we can't implement DRM_WAIT_ON verbatim, but this more closely
reflects the API of DRM_WAIT_ON than DRM_*WAIT*_UNTIL do.

Major difference is that this polls every tick, like DRM_WAIT_ON,
unlike DRM_*WAIT*_UNTIL. So it will mask missing wakeups, but it
wouldn't surprise me if there were such things upstream.

And another timeout return value mistake.

Initialize mchdev_lock earlier. Fixes i915drmkms boot on Ironlake.

Derived from a patch sent to me by degroote@ longer ago than I care
to admit.

branches: 1.12.2;
i915drmkms(4): Enable CONFIG_ACPI if NACPICA > 0.
Now brightness can be adjusted via hotkey on Mouse Computer LB-J300X (Clevo W330SU2).

Change enum pipe to enum i915_pipe
(enum pipe uses too global of a tag and conflicts with struct pipe).

branches: 1.10.2; 1.10.4;
Get suspend/resume sorta working for i915drm.

Make it build and boot on my test machines.

Screen blanks on boot on the Ivy Bridge system with

DRM error in cpt_serr_int_handler: PCH transcoder A FIFO underrun

But after that everything is OK. Appears to be an upstream problem.
To investigate...

I think there's a cache flushing issue somewhere -- there are little
display artefacts on my T60.

fix merge conflicts

Call teardown_timer in i915_driver_unload.

Mark another place in i915drmkms where paddr constraints are set.

branches: 1.5.2;
Convert pending_flip_lock to spin lock -- interrupt handlers take it.

branches: 1.4.2;
Don't ifdef out unmapping of the GTT on NetBSD.

Fix unused variable warnings in drm2.

Merge riastradh-drm2 to HEAD.

branches: 1.1.1;
Initial revision

i915: Make sure to call i915_drm_prepare too on suspend.

i915_drm_resume_early: initialize ret

clang -Wsometimes-uninitialized says uninitialized use
on line 2043 when if statement on line 2037 is false

drm: Switch sense of #if __linux__ conditionals to use __NetBSD__.

Keep it consistent with the rest of drm.

i915: mutex_init/destroy and spin_lock_init/destroy audit

i915: Omit needless ifdef and do call intel_bios_init.

i915: Define i915_gem_uvm_ops.

i915: Omit needless reachover of i915_drm_driver.

i915: Nix drm_dev_alloc, and do drm_pci_attach in i915_driver_probe.

i915 uses a struct drm_device embedded in struct drm_i915_private
now, so we can't allocate it separately.

i915: machete

i915: Fix drm_dev_init argument.

Confusingly the parameter is named `parent' but means the struct
device that corresponds to the struct pci_dev. Use pci_dev_dev here
since it works in both APIs.

pdev is ifdef'd out, so ifdef out the declaration too


Author: Maya Rashish <maya@NetBSD.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

Move local diff's location to above where it's used


Author: Maya Rashish <maya@NetBSD.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

Emulate pdev->dev for NetBSD.


Author: Maya Rashish <maya@NetBSD.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

Exclude call to dma_set_max_seg_size

We probably don't need to replicate this.


Author: Maya Rashish <maya@NetBSD.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

Provide declarations, we use these from different files.


Author: Maya Rashish <maya@NetBSD.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

Shuffle ifdefs for build. Remove local diff that doesn't build.


Author: Maya Rashish <maya@NetBSD.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

Remove local diff


Author: Maya Rashish <maya@NetBSD.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

restore lost ifdef of vga_switcheroo code


Author: Maya Rashish <maya@NetBSD.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

Remove duplicate declaration of i915_device_ids

The one that compiles is in i915_pci.c now. No `pciidlist` in this
file!


Author: Maya Rashish <maya@NetBSD.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

nbsd-namespace.h for the right mutex_init


Author: Maya Rashish <maya@NetBSD.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

Remove local diff, side effect of backporting kabylake support.


Author: Maya Rashish <maya@NetBSD.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

drm: Remove local diffs for drm_pci_set_busid.

Instead, move it to its own file that can be safely used without
bringing in a link/run-time dependency on pci.

Empty out drm_os_netbsd.h.

Distribute local changes from i915_dma.c

Restore local intel_acpi changes.

Restore #ifdef CONFIG_PNP tweak lost in merge.

Cleanup ifdef goo that got mangled.

Endifs without matching ifdefs and so on.


Author: Maya Rashish <maya@NetBSD.org>

First pass at i915, far from complete.

HEAVILY MODIFIED IN MERGE - maya

Merge linux-drm-v5-6-rc3

Reduce diffs by side-loading some header files like Linux.

From riastradh

Stub out console_lock/unlock and drm_fb_helper_set_suspend.

Reduce some more diffs.

branches: 1.17.2;
Add what appears to be the fixes to CVE-2019-0154, CVE-2019-0155.
This commit requires review, but I'd also like it to be tested by others
while it is being reviewed.

CVE-2019-0155:
It was discovered that the Intel i915 graphics chipsets allowed userspace
to modify page table entries via writes to MMIO from the Blitter Command
Streamer and expose kernel memory information. A local attacker could use
this to expose sensitive information or possibly elevate privileges.

CVE-2019-0154:
It was discovered that the Intel i915 graphics chipsets could cause
a system hang when userspace performed a read from GT memory mapped
input output (MMIO) when the product is in certain low power states.
A local attacker could use this to cause a denial of service.

From upstream commits to linux-4.4.y:

branches: 1.16.4;
add support for kabylake and skylake GT4 (untested) GPUs.
largely taken from openbsd and linux 4.13 trees (which
have this code identical), with mimimal porting to netbsd.

i have not installed (and thus tested) the newer referenced
firmware files.

only real local change is to fix IS_BROXTON() macro to check
the things valid in this era of drm. previous match would
attach on KBL, and then a loop would never exit.

tested on kabylake P630. needs mesa 11.x or newer for GL
to work.

ok @riastradh.

Draft support for drm prime.

Call i915_drm_suspend_late too.

Attempt to hook up i915 suspend/resume callbacks.

Disable legacy mmap in i915.

Tweak ifdefs in i915_drv.c.

Disable and enable some i915 pm code.

struct drm_bus abstraction is no more. Cope.

struct drm_driver now has the bus-specific intr establish routine
(called request_irq/free_irq to match Linux's style).

don't provide dev_to_i915 and ifdef out all users (unused)
provide nsecs_to_jiffies64, hopefully msec_to_jiffies * 10^6
provide BUILD_BUG as a no-op
include linux/err.h instead of matching side-loading, too hard to
figure out what gets it.


Author: coypu <coypu@sdf.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

merge linux-drm-v4-4-143

branches: 1.6.4; 1.6.28; 1.6.30;
Let the drm driver decide what part of what object gets mmapped.

Pass the file around too so radeon/ttm can get at it to verify access.

Add drm_gem_or_legacy_mmap_object for drivers to choose the previous
behaviour, like i915.

Get suspend/resume sorta working for i915drm.

Make it build and boot on my test machines.

Screen blanks on boot on the Ivy Bridge system with

DRM error in cpt_serr_int_handler: PCH transcoder A FIFO underrun

But after that everything is OK. Appears to be an upstream problem.
To investigate...

I think there's a cache flushing issue somewhere -- there are little
display artefacts on my T60.

fix merge conflicts

branches: 1.2.2; 1.2.4;
Merge riastradh-drm2 to HEAD.

branches: 1.1.1;
Initial revision

drm: Set CONFIG_ACPI in linux/acpi.h and make it build.

Leave a little ACPI-related functionality disabled for now, like
getting EDID out of ACPI -- needs a bit more work to make this work,
and I don't have hardware to work on that.

Should help with failures of the forms:

- unable to locate a BIOS ROM
- bios: unable to locate usable image

on various machines.

branches: 1.48.4;
i915: Make sure to call i915_drm_prepare too on suspend.

drm: Omit needless conditionals around #include "acpica.h".

i915: Rework atomic commit wakeups to work with condvars.

Lock order is intel_state->commit_ready.wait.lock then
dev_priv->atomic_commit_lock.

i915: Disentangle uabi engines rb/llist/list spaghetti.

drm: Another pass over i915. Most of i915 gem builds now.

i915: bus_space_read/write moved to intel_uncore.h; nix diff.


Author: Maya Rashish <maya@NetBSD.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

Provide declarations, we use these from different files.


Author: Maya Rashish <maya@NetBSD.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

Remove local change to constify, and remove const qualifier

We certainly write to this pointer in i915_driver_create with
memcpy.


Author: Maya Rashish <maya@NetBSD.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

Provide and use an rb_next2 which takes a pointer to the tree as well

Easier for us to implement.


Author: Maya Rashish <maya@NetBSD.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

const const const


Author: Maya Rashish <maya@NetBSD.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

Our __always_inline doesn't make the function inline.

Avoid compiler warnings about the function being defined but not used


Author: Maya Rashish <maya@NetBSD.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

Revert vestigial local patch.

Disable some unimportant broken BUILD_BUG_ONs.

First pass at i915, far from complete.

HEAVILY MODIFIED IN MERGE - maya

Merge linux-drm-v5-6-rc3

i915drmkms: Fix LOCKDEBUG panic and potential deadlock.

This path is taken with a spin lock held, and possibly even in
interrupt context, where taking vmobjlock is not kosher, but we are
guaranteed to have the queue populated and unchanging.

XXX pullup-9

branches: 1.32.10; 1.32.12;
UVM locking changes, proposed on tech-kern:

- Change the lock on uvm_object, vm_amap and vm_anon to be a RW lock.
- Break v_interlock and vmobjlock apart. v_interlock remains a mutex.
- Do partial PV list locking in the x86 pmap. Others to follow later.

Undo the `enum pipe' -> `enum i915_pipe' rename.

branches: 1.30.2;
Add what appears to be the fixes to CVE-2019-0154, CVE-2019-0155.
This commit requires review, but I'd also like it to be tested by others
while it is being reviewed.

CVE-2019-0155:
It was discovered that the Intel i915 graphics chipsets allowed userspace
to modify page table entries via writes to MMIO from the Blitter Command
Streamer and expose kernel memory information. A local attacker could use
this to expose sensitive information or possibly elevate privileges.

CVE-2019-0154:
It was discovered that the Intel i915 graphics chipsets could cause
a system hang when userspace performed a read from GT memory mapped
input output (MMIO) when the product is in certain low power states.
A local attacker could use this to cause a denial of service.

From upstream commits to linux-4.4.y:

branches: 1.29.2;
Use unsigned. Found by KUBSan.

add support for kabylake and skylake GT4 (untested) GPUs.
largely taken from openbsd and linux 4.13 trees (which
have this code identical), with mimimal porting to netbsd.

i have not installed (and thus tested) the newer referenced
firmware files.

only real local change is to fix IS_BROXTON() macro to check
the things valid in this era of drm. previous match would
attach on KBL, and then a loop would never exit.

tested on kabylake P630. needs mesa 11.x or newer for GL
to work.

ok @riastradh.

Draft support for drm prime.

Use dtrace probes to implement i915_trace.h.

Call i915_drm_suspend_late too.

Work around ioctl ABI mistake in i915 drm.

A flags argument was added to the I915_GEM_MMAP ioctl. The version
of xf86-video-intel we imported was compiled to use the new ioctl
structure including the flags argument, but without initializing the
flags argument, so it would be userland stack garbage causing
i915_gem_mmap_ioctl to fail.

Newer xf86-video-intel will query the mmap version first before
trying to use either the older ioctl structure, with the original
shorter argument and thus no stack garbage (which will be padded with
to the full length in the kernel by drm_ioctl), or the newer ioctl
structure, with the argument initialized.

libdrm does not appear to be affected by this -- it zeroes the longer
structure before passing it in.

Patch from mrg@, diagnosis and explanation by me.

Fold local intel_uncore_destroy into upstream intel_uncore_fini.

Attempt to hook up i915 suspend/resume callbacks.

Add __UNCONST. Used only for fetches, sometimes const.

Use a prefetchable linear bus space mapping. No __iomem.

Rename drm_gem_object::gemo_shm_uao -> drm_gem_object::filp.

The Linux member that is basically analogous to this is named so,
which means we can eliminate a few diffs this way.

Rework how we map i915 drm gem objects.

The old way made no sense at all. This way might make sense.

Use bus_space_read/write_4 instead of readl/writel.

Need vaddr_t, not pointer, for pmap_extract.

Use the variable, don't shadow it.

don't provide dev_to_i915 and ifdef out all users (unused)
provide nsecs_to_jiffies64, hopefully msec_to_jiffies * 10^6
provide BUILD_BUG as a no-op
include linux/err.h instead of matching side-loading, too hard to
figure out what gets it.


Author: coypu <coypu@sdf.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

un-static i915_disable_vga, used in intelfb.c.

Now I reach linking, and have a lot of undefined functions.
-Werror build still not handled.


Author: coypu <coypu@sdf.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

merge linux-drm-v4-4-143

branches: 1.11.18; 1.11.20;
fix build failure.

> /tmp/bracket/build/2014.11.10.22.43.46-i386/src/sys/external/bsd/drm2/include/linux/pci.h:36:20: fatal error: acpica.h: No such file or directory

i915drmkms(4): Enable CONFIG_ACPI if NACPICA > 0.
Now brightness can be adjusted via hotkey on Mouse Computer LB-J300X (Clevo W330SU2).

Change enum pipe to enum i915_pipe
(enum pipe uses too global of a tag and conflicts with struct pipe).

branches: 1.8.2;
Restore Intel opregion stuff.

branches: 1.7.2;
Rework Intel framebuffer attachment.

Turns out genfb_softc must be at the start of its own device_private,
so we can't have it in the middle of a structure, and certainly can't
have potentially multiple genfbs per graphics device. Failing to do
this is why entering ddb didn't work with an i915drmkms console.

Instead of putting a genfb_softc into each drm_fb_helper, put a
device_t into each drm_fb_helper and create a new intelfb(4) device
for each Intel framebuffer. This will be more flexible, too, in case
we want accelerated framebuffers later on. (XXX Need to adapt the
radeon code, which I'm disabling until that happens to avoid breaking
the build.)

While here, defer disabling VGA, vga_cndetach, and genfb_attach until
we're actually ready to do them all together along with the mode
switch. This should reduce the amount of time during which the
screen is blanked, in case it hangs.

Get suspend/resume sorta working for i915drm.

Make it build and boot on my test machines.

Screen blanks on boot on the Ivy Bridge system with

DRM error in cpt_serr_int_handler: PCH transcoder A FIFO underrun

But after that everything is OK. Appears to be an upstream problem.
To investigate...

I think there's a cache flushing issue somewhere -- there are little
display artefacts on my T60.

fix merge conflicts

branches: 1.3.2;
Convert pending_flip_lock to spin lock -- interrupt handlers take it.

branches: 1.2.2;
Merge riastradh-drm2 to HEAD.

branches: 1.1.1;
Initial revision

Merge linux-drm-v5-6-rc3

branches: 1.1.1;
Initial revision

i915: mutex_init/destroy and spin_lock_init/destroy audit

drm: Allow multiple concurrent io_mapping_map_wc on the same space.

(This should maybe check that there aren't overlapping maps of the
same parts of the space.)

i915: Fill out i915_gem_phys_pwrite.

i915: Omit more unnecessary local diffs.

i915: Restore i915_gem.c shmem_pwrite.

This was ifdef'd out back when it was a fast path shmem_pwrite_fast
using __copy_in_user_inatomic, but the ifdef got merged anyway into a
different function that is not a fast path.

i915: machete

Add missing header, NetBSD-ify


Author: Maya Rashish <maya@NetBSD.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

Stub out some functions.

We can probably do without the fast path, and let's see whether
we get to i915_gem_phys_pwrite at all.


Author: Maya Rashish <maya@NetBSD.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

i915: Move linux/nbsd-namespace.h to end of includes.

Distribute local changes from i915_dma.c

Don't ifdef __NetBSD__ around ALIGN; just patch to round_up.


Author: Maya Rashish <maya@NetBSD.org>

Sync i915_gem_fault. Hack up the rest of i915_gem.c enough to build.

First pass at i915, far from complete.

HEAVILY MODIFIED IN MERGE - maya

Merge linux-drm-v5-6-rc3

UVM locking changes, proposed on tech-kern:

- Change the lock on uvm_object, vm_amap and vm_anon to be a RW lock.
- Break v_interlock and vmobjlock apart. v_interlock remains a mutex.
- Do partial PV list locking in the x86 pmap. Others to follow later.

do not wait for memory in pgo_fault methods, just return ENOMEM
and let the uvm_fault code wait if it is appropriate.

Reduce diffs by side-loading some header files like Linux.

From riastradh

Use a header file hack to cope with Linux/NetBSD namespace clashes.

Currently serves for:

ALIGN
mutex_init
mutex_destroy

Eliminate -Wpointer-arith patches; use -Wno-pointer-arith instead.

Apply patch from kamil@: vmobjlock needs to be held for uvm_pagemarkdirty().

Merge from yamt-pagecache (after much testing):

- Reduce unnecessary page scan in putpages esp. when an object has a ton of
pages cached but only a few of them are dirty.

- Reduce the number of pmap operations by tracking page dirtiness more
precisely in uvm layer.

branches: 1.54.6;
Draft support for drm prime.

Handle uvm object reference counts in uvm_map more carefully.

Acquire a reference unconditionally first; then let uvm_map consume
it on success, and release it ourselves on failure.

As we did before -- acquiring a fresh reference on success -- another
thread might release the reference with a concurrent uvm_unmap before
we could acquire it back, thereby destroying the object.

XXX pullup-7 in part (i915_gem_fault)
XXX pullup-8 in part (i915_gem_fault)

Work around ioctl ABI mistake in i915 drm.

A flags argument was added to the I915_GEM_MMAP ioctl. The version
of xf86-video-intel we imported was compiled to use the new ioctl
structure including the flags argument, but without initializing the
flags argument, so it would be userland stack garbage causing
i915_gem_mmap_ioctl to fail.

Newer xf86-video-intel will query the mmap version first before
trying to use either the older ioctl structure, with the original
shorter argument and thus no stack garbage (which will be padded with
to the full length in the kernel by drm_ioctl), or the newer ioctl
structure, with the argument initialized.

libdrm does not appear to be affected by this -- it zeroes the longer
structure before passing it in.

Patch from mrg@, diagnosis and explanation by me.

Need irq_lock to wait on irq_queue. From mrg@.

Sprinkle idr_preload/idr_preload_end.

Try to do these outside all mutex locks.

Tricky case is i915_gem_context, where the struct mutex is held for a
long time.

With <asm/cpufeature.h>, we can now use cpu_has_pat.

Guess what uvm criterion means we should discard backing pages.

Don't forget to set .uio_offset and to call UIO_SETUP_SYSSPACE.

Reduce diff by going back to just using ACCESS_ONCE.

Implement i915_gem_object_invalidate with PGO_DEACTIVATE.

Implement set_page_dirty. Reduce diff a little.

Rename drm_gem_object::gemo_shm_uao -> drm_gem_object::filp.

The Linux member that is basically analogous to this is named so,
which means we can eliminate a few diffs this way.

Compute the number of bytes written as expected.

Rework how we map i915 drm gem objects.

The old way made no sense at all. This way might make sense.

Missed a missed_irq.

Fix some merge mistakes, 64-bit printfs, &c.

#ifdef custodial service

bring in old drm changes
panic(XXX) out parts that need attention


Author: coypu <coypu@sdf.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

merge linux-drm-v4-4-143

Handle uvm object reference counts in uvm_map more carefully.

Acquire a reference unconditionally first; then let uvm_map consume
it on success, and release it ourselves on failure.

As we did before -- acquiring a fresh reference on success -- another
thread might release the reference with a concurrent uvm_unmap before
we could acquire it back, thereby destroying the object.

XXX pullup-7
XXX pullup-8

branches: 1.34.2; 1.34.4;
Update the kernhist(9) kernel history code to address issues identified
in PR kern/52639, as well as some general cleaning-up...

(As proposed on tech-kern@ with additional changes and enhancements.)

Details of changes:

* All history arguments are now stored as uintmax_t values[1], both in
the kernel and in the structures used for exporting the history data
to userland via sysctl(9). This avoids problems on some architectures
where passing a 64-bit (or larger) value to printf(3) can cause it to
process the value as multiple arguments. (This can be particularly
problematic when printf()'s format string is not a literal, since in
that case the compiler cannot know how large each argument should be.)

* Update the data structures used for exporting kernel history data to
include a version number as well as the length of history arguments.

* All [2] existing users of kernhist(9) have had their format strings
updated. Each format specifier now includes an explicit length
modifier 'j' to refer to numeric values of the size of uintmax_t.

* All [2] existing users of kernhist(9) have had their format strings
updated to replace uses of "%p" with "%#jx", and the pointer
arguments are now cast to (uintptr_t) before being subsequently cast
to (uintmax_t). This is needed to avoid compiler warnings about
casting "pointer to integer of a different size."

* All [2] existing users of kernhist(9) have had instances of "%s" or
"%c" format strings replaced with numeric formats; several instances
of mis-match between format string and argument list have been fixed.

* vmstat(1) has been modified to handle the new size of arguments in the
history data as exported by sysctl(9).

* vmstat(1) now provides a warning message if the history requested with
the -u option does not exist (previously, this condition was silently
ignored, with only a single blank line being printed).

* vmstat(1) now checks the version and argument length included in the
data exported via sysctl(9) and exits if they do not match the values
with which vmstat was built.

* The kernhist(9) man-page has been updated to note the additional
requirements imposed on the format strings, along with several other
minor changes and enhancements.

[1] It would have been possible to use an explicit length (for example,
uint64_t) for the history arguments. But that would require another
"rototill" of all the users in the future when we add support for an
architecture that supports a larger size. Also, the printf(3) format
specifiers for explicitly-sized values, such as "%"PRIu64, are much
more verbose (and less aesthetically appealing, IMHO) than simply
using "%ju".

[2] I've tried very hard to find "all [the] existing users of kernhist(9)"
but it is possible that I've missed some of them. I would be glad to
update any stragglers that anyone identifies.

branches: 1.33.10;
We never exec(2) with a kernel vmspace, so do not test for that, but instead
KASSERT() that we don't.
When calculating the load address for the interpreter (e.g. ld.elf_so),
we need to take into account wether the exec'd process will run with
topdown memory or bottom up. We can not use the current vmspace's flags
to test for that, as this happens too early. Luckily the execpack already
knows what the new state will be later, so instead of testing the current
vmspace, pass the info as additional argument to struct emul
e_vm_default_addr.
Fix all such functions and adopt all callers.

Pull upstream fix to avoid kernel panic on starting X on Intel 855GM machines.

Reported in PR kern/49875, and ok'ed to commit by riastradh@.

Should be pulled up to netbsd-7.

fix Xorg coredumps that have started happening recently.
the problem is that we get a SIGALRM while we're sleeping during a page fault
on a mapping of a GEM object, and since we're sleeping interruptibly,
the GEM operation fails with EINTR. this error is returned all the way back
through uvm_fault() to the trap handler, which responds to that error
by delivering a SIGSEGV.

fix this by doing like the linux version of the GEM fault handler and converting
EINTR into success, which results in delivering the original signal and
retrying the fault.

Remove workaround for PR kern/49195.

Problem seems to have been properly fixed in rev. 1.25.

Use pmap_pv(9) to remove mappings of Intel graphics aperture pages.

Proposed on tech-kern with no objections:

https://mail-index.netbsd.org/tech-kern/2015/03/26/msg018561.html

Further background at:

https://mail-index.netbsd.org/tech-kern/2014/07/23/msg017392.html

Dedup the NetBSD portion of the code (ok Riastradh), no functional change.

Mark pages dirty after use for write in the aperture (`GTT').

Unwire pages when we're done in i915_gem_object_attach_phys.

Re-enable prints to diagnose PR 49195.

I believe I fixed this problem by fixing __wait_seqno, but before I
remove the workaround I'll try to make sure it's not getting hit any
more.

Make sure wedged actually gets set on all code paths.

Otherwise GCC helpfully realizes we're invoking undefined behaviour
and optimizes away the possibility that this routine will return
success, without saying a word about it.

Make timed and untimed cases of __wait_seqno agree on return value.

...aaaaand one more fix for __wait_seqno return value.

Fix return code of __wait_seqno.

MAX(ret, 0) is 0 if ret is negative, but if ret is negative we want
to return that negative value, meaning error. Should've been
MIN(ret, 0), but I'll just rewrite it to clarify a wee bit.

If the GPU reset, call i915_gem_check_wedge and always return failure
like Linux does. Caller must retry in that case.

Fix returned timeout in wait_seqno: remaining time, not time slept.

Also need to flush an object from CPU domain.

branches: 1.18.2;
When moving an object to inactive, make sure that we first flush it from
the GTT domain in case it belonged there. Also, fix some compilation issues
when turning on WATCH_GTT and WATCH_LISTS, although this is a lost cause;
most of the code has rotted beyond repair.

fix uninitialized

Avoid container_of-inspired &obj->base == NULL nonsense.

Drop take the {ttm,gem} vmobjlock in the fault handler.

- We don't need this lock.
- uvm does nothing between taking it and calling the fault handler.
- Now that the uvm_aobj shares vmobjlock with the {ttm,gem} uvm
object, we must not hold the lock when we call uvm_obj_wirepages on
the uvm_aobj.

XXX pullup to netbsd-7

branches: 1.14.2; 1.14.4;
Explain why i915_gem_release_mmap is broken.

Make it build and boot on my test machines.

Screen blanks on boot on the Ivy Bridge system with

DRM error in cpt_serr_int_handler: PCH transcoder A FIFO underrun

But after that everything is OK. Appears to be an upstream problem.
To investigate...

I think there's a cache flushing issue somewhere -- there are little
display artefacts on my T60.

fix merge conflicts

Constrain addresses of pages backing i915 GEM objects.

Use the new uao_set_pgfl and x86_select_freelist for the purpose.

Ensure we call uvmfault_unlockall on every exit from i915_gem_fault.

branches: 1.9.2;
Oops -- drop the GEM object reference on error too.

Don't map the GEM uvm_aobj copy-on-write -- what was I thinking?

Do transfer the GEM object reference to the uvm_aobj reference --
these are not the same thing. (There's another uvm object whose
references are the same thing as the GEM object references, but
that's not the uao.)

With these changes, it looks like the GPU is no longer trying to draw
graphics all over kernel data structures. Wish I had that month of
debugging back!

Fix sense of test in last commit, noted by Robert Swindells.

Reject 32-bit paddrs on 965.

XXX Doing the check here is wrong; it serves only to report an
earlier problem, which is that there's on way to express constraints
on paddrs to uvm_obj_wirepages. bus_dmamem_alloc can do this, but it
gives us pages out of thin air, not pages backing a uvm object. I
was hoping this wouldn't manifest as a real problem, but evidently it
does.

Fix >40-bit paddr error branch in i915_gem_object_get_pages_gtt.

Tweak some DRM GEM page indexing crap.

- Fix order of subtraction in drm_mmap_paddr_locked.
- Address GEM objects' pages from 0, not from the mmap cookie.
- Check page alignment earlier in mmap code paths.
- Sprinkle kasserts throughout.

Still doesn't fix the garbage that is sometimes being scribbled all
over kernel memory!

Convert pending_flip_lock to spin lock -- interrupt handlers take it.

branches: 1.2.2;
Merge riastradh-drm2 to HEAD.

branches: 1.1.1;
Initial revision

nix the NetBSD specific GEM_BUG_ON().

avoids GCC 12 warnings, and matches upstream closer.

branches: 1.7.4;
i915: Obviate need for __diagused on variables in GEM_BUG_ON.

i915: Make GEM_BUG_ON a little more useful here.

Ifdef out i915's local hacks reaching into tasklet's private parts.

Moved i915 tasklet hacks into tasklet proper.

<linux/stringify.h> for __stringify; gave up on guessing side loads.

Use the slightly less grody tasklet API violation.

Merge linux-drm-v5-6-rc3

branches: 1.1.1;
Initial revision

Merge linux-drm-v5-6-rc3

branches: 1.2.2; 1.2.6;
merge linux-drm-v4-4-143

branches: 1.1.1;
Initial revision

Merge linux-drm-v5-6-rc3

branches: 1.2.2; 1.2.6;
merge linux-drm-v4-4-143

branches: 1.1.1;
Initial revision

Merge linux-drm-v5-6-rc3

Reduce diffs by side-loading some header files like Linux.

From riastradh

Reduce diffs by nixing DRM_SUSER macro, deleted upstream long ago.

Instead stub Linux capable(CAP_SYS_ADMIN).

branches: 1.10.2;
Add what appears to be the fixes to CVE-2019-0154, CVE-2019-0155.
This commit requires review, but I'd also like it to be tested by others
while it is being reviewed.

CVE-2019-0155:
It was discovered that the Intel i915 graphics chipsets allowed userspace
to modify page table entries via writes to MMIO from the Blitter Command
Streamer and expose kernel memory information. A local attacker could use
this to expose sensitive information or possibly elevate privileges.

CVE-2019-0154:
It was discovered that the Intel i915 graphics chipsets could cause
a system hang when userspace performed a read from GT memory mapped
input output (MMIO) when the product is in certain low power states.
A local attacker could use this to cause a denial of service.

From upstream commits to linux-4.4.y:

branches: 1.9.4;
Sprinkle idr_preload/idr_preload_end.

Try to do these outside all mutex locks.

Tricky case is i915_gem_context, where the struct mutex is held for a
long time.

bring in old drm changes
panic(XXX) out parts that need attention


Author: coypu <coypu@sdf.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

merge linux-drm-v4-4-143

branches: 1.6.4; 1.6.28; 1.6.30;
dctx->obj may still be null here!

Make it build and boot on my test machines.

Screen blanks on boot on the Ivy Bridge system with

DRM error in cpt_serr_int_handler: PCH transcoder A FIFO underrun

But after that everything is OK. Appears to be an upstream problem.
To investigate...

I think there's a cache flushing issue somewhere -- there are little
display artefacts on my T60.

fix merge conflicts

Mark variables __diagused.

branches: 1.2.2; 1.2.4;
Merge riastradh-drm2 to HEAD.

branches: 1.1.1;
Initial revision

Merge linux-drm-v5-6-rc3

merge linux-drm-v4-4-143

branches: 1.1.1;
Initial revision

Merge linux-drm-v5-6-rc3

Draft support for drm prime.

merge linux-drm-v4-4-143

branches: 1.1.1;
Initial revision

i915: Omit more unnecessary local diffs.

drm/i915: Misc build fixes.

First pass at i915, far from complete.

HEAVILY MODIFIED IN MERGE - maya

Merge linux-drm-v5-6-rc3

merge linux-drm-v4-4-143

branches: 1.1.1;
Initial revision

Merge linux-drm-v5-6-rc3

Reduce diffs by side-loading some header files like Linux.

From riastradh

Reduce diffs by nixing DRM_SUSER macro, deleted upstream long ago.

Instead stub Linux capable(CAP_SYS_ADMIN).

branches: 1.9.2;
Add what appears to be the fixes to CVE-2019-0154, CVE-2019-0155.
This commit requires review, but I'd also like it to be tested by others
while it is being reviewed.

CVE-2019-0155:
It was discovered that the Intel i915 graphics chipsets allowed userspace
to modify page table entries via writes to MMIO from the Blitter Command
Streamer and expose kernel memory information. A local attacker could use
this to expose sensitive information or possibly elevate privileges.

CVE-2019-0154:
It was discovered that the Intel i915 graphics chipsets could cause
a system hang when userspace performed a read from GT memory mapped
input output (MMIO) when the product is in certain low power states.
A local attacker could use this to cause a denial of service.

From upstream commits to linux-4.4.y:

branches: 1.8.4;
No arithmetic on void *.

Fix pointer arithmetic and include asm/cpufeature.h for cpu_has_clflush.

merge linux-drm-v4-4-143

branches: 1.5.16; 1.5.18;
Fix typo: __gtt_iomem, not __acpi_iomem. No functional change.

branches: 1.4.2; 1.4.4; 1.4.6;
Make it build and boot on my test machines.

Screen blanks on boot on the Ivy Bridge system with

DRM error in cpt_serr_int_handler: PCH transcoder A FIFO underrun

But after that everything is OK. Appears to be an upstream problem.
To investigate...

I think there's a cache flushing issue somewhere -- there are little
display artefacts on my T60.

fix merge conflicts

branches: 1.2.2; 1.2.4;
Merge riastradh-drm2 to HEAD.

branches: 1.1.1;
Initial revision

Merge linux-drm-v5-6-rc3

UVM locking changes, proposed on tech-kern:

- Change the lock on uvm_object, vm_amap and vm_anon to be a RW lock.
- Break v_interlock and vmobjlock apart. v_interlock remains a mutex.
- Do partial PV list locking in the x86 pmap. Others to follow later.

Apply patch from kamil@: vmobjlock needs to be held for uvm_pagemarkdirty().

Merge from yamt-pagecache (after much testing):

- Reduce unnecessary page scan in putpages esp. when an object has a ton of
pages cached but only a few of them are dirty.

- Reduce the number of pmap operations by tracking page dirtiness more
precisely in uvm layer.

branches: 1.5.2; 1.5.6; 1.5.10;
Use dtrace probes to implement i915_trace.h.

Rework how we map i915 drm gem objects.

The old way made no sense at all. This way might make sense.

Fix 64-bit printfs.

merge linux-drm-v4-4-143

branches: 1.1.1;
Initial revision

drm: Another pass over i915. Most of i915 gem builds now.

drm: Nix use of uvm pglist. Just use arrays of page pointers.

linux/bitmap.h for bitmap_zalloc


Author: Maya Rashish <maya@NetBSD.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

NetBSD-ify sglist, update netbsd goo to newer struct


Author: Maya Rashish <maya@NetBSD.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

Merge linux-drm-v5-6-rc3

branches: 1.1.1;
Initial revision

drm: Another pass over i915. Most of i915 gem builds now.

drm: Nix use of uvm pglist. Just use arrays of page pointers.

Sync i915_gem_fault. Hack up the rest of i915_gem.c enough to build.

Merge linux-drm-v5-6-rc3

branches: 1.1.1;
Initial revision

x86: Forbid using x86/pte.h directly; use machine/pte.h.

machine/pte.h already used outside sys/arch, so let's make it the
primary thing and make sure to use x86/pte.h only as a subroutine.

drm: Another pass over i915 and some supporting logic.

This makes a shim around sg_table, which essentially represents two
things:

1. an array of pages (roughly corresponding to bus_dma_segment_t[])
2. an array of DMA addresses stored in a bus_dmamap_t

Both parts are optional; different parts of i915 use sg_tables to
pass around one or both of the two parts. This helps to reduce the
ifdefs by quite a bit, although it's not always clear which part of
an sg_table any particular interface is actually using which is why I
was reluctant to do this before.

drm: Nix use of uvm pglist. Just use arrays of page pointers.

Work over i915_gem_gtt.c enough to make it build.

First pass at i915, far from complete.

HEAVILY MODIFIED IN MERGE - maya

Merge linux-drm-v5-6-rc3

Switch to the new PTE naming. The old naming is now unused, remove it.

branches: 1.17.4;
Reduce diffs by side-loading some header files like Linux.

From riastradh

branches: 1.16.2;
Add what appears to be the fixes to CVE-2019-0154, CVE-2019-0155.
This commit requires review, but I'd also like it to be tested by others
while it is being reviewed.

CVE-2019-0155:
It was discovered that the Intel i915 graphics chipsets allowed userspace
to modify page table entries via writes to MMIO from the Blitter Command
Streamer and expose kernel memory information. A local attacker could use
this to expose sensitive information or possibly elevate privileges.

CVE-2019-0154:
It was discovered that the Intel i915 graphics chipsets could cause
a system hang when userspace performed a read from GT memory mapped
input output (MMIO) when the product is in certain low power states.
A local attacker could use this to cause a denial of service.

From upstream commits to linux-4.4.y:

branches: 1.15.4;
Scale offset by size of unit, gen6_pte_t.

Pass the vaddr we mapped to kunmap_px. Thanks, void *.

Zero the scratch page.

Pass waitok/nowait flags to bus_dma. Use boundary=0.

Rework how we map i915 drm gem objects.

The old way made no sense at all. This way might make sense.

Some tweaks en route to i915_gem_gtt.

Disable rotated and partial object views for now.

#ifdef custodial service

So that's where the missing ifdef wound up!

some more definitions necessary

ifdef out all the i915 gtt virtual memory stuff that errors,
unless it looks exactly like the old code and then merge the
netbsd ifdefs for it.

we don't want to use their ALIGN(, which has more arguments,
use the old alt function.
merge in the old intel_acpi.c code.
don't duplicate DECLARE_BITMAP


Author: coypu <coypu@sdf.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

merge linux-drm-v4-4-143

branches: 1.4.16; 1.4.18;
Fix kassert: npdp may be GEN8_LEGACY_PDPS exactly.

Tripped over by msaitoh, our first Broadwell user, apparently.

branches: 1.3.2; 1.3.4; 1.3.6;
Avoid bus_space_read/write_8 on i386.

Would be nice if there were a clearer way to detect it than _LP64,
and if its absence caused a compile error instead of a link error.

Make it build and boot on my test machines.

Screen blanks on boot on the Ivy Bridge system with

DRM error in cpt_serr_int_handler: PCH transcoder A FIFO underrun

But after that everything is OK. Appears to be an upstream problem.
To investigate...

I think there's a cache flushing issue somewhere -- there are little
display artefacts on my T60.

branches: 1.1.1;
Initial revision

drm: Another pass over i915 and some supporting logic.

This makes a shim around sg_table, which essentially represents two
things:

1. an array of pages (roughly corresponding to bus_dma_segment_t[])
2. an array of DMA addresses stored in a bus_dmamap_t

Both parts are optional; different parts of i915 use sg_tables to
pass around one or both of the two parts. This helps to reduce the
ifdefs by quite a bit, although it's not always clear which part of
an sg_table any particular interface is actually using which is why I
was reluctant to do this before.

Add netbsd-ified i915_gem_gtt_{prepare,finish}_pages prototypes


Author: Maya Rashish <maya@NetBSD.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

Merge linux-drm-v5-6-rc3

Add what appears to be the fixes to CVE-2019-0154, CVE-2019-0155.
This commit requires review, but I'd also like it to be tested by others
while it is being reviewed.

CVE-2019-0155:
It was discovered that the Intel i915 graphics chipsets allowed userspace
to modify page table entries via writes to MMIO from the Blitter Command
Streamer and expose kernel memory information. A local attacker could use
this to expose sensitive information or possibly elevate privileges.

CVE-2019-0154:
It was discovered that the Intel i915 graphics chipsets could cause
a system hang when userspace performed a read from GT memory mapped
input output (MMIO) when the product is in certain low power states.
A local attacker could use this to cause a denial of service.

From upstream commits to linux-4.4.y:

branches: 1.7.2; 1.7.6; 1.7.8;
Some tweaks en route to i915_gem_gtt.

Typo: dm_segs, not dm_seg.

Fix misnamed parameter.

some more definitions necessary

ifdef out all the i915 gtt virtual memory stuff that errors,
unless it looks exactly like the old code and then merge the
netbsd ifdefs for it.

we don't want to use their ALIGN(, which has more arguments,
use the old alt function.
merge in the old intel_acpi.c code.
don't duplicate DECLARE_BITMAP


Author: coypu <coypu@sdf.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

Define px_dma for our bus_dmaish i915_dma_page.

merge linux-drm-v4-4-143

branches: 1.1.1;
Initial revision

Merge linux-drm-v5-6-rc3

Use a header file hack to cope with Linux/NetBSD namespace clashes.

Currently serves for:

ALIGN
mutex_init
mutex_destroy

branches: 1.9.2; 1.9.6; 1.9.10;
Handle uvm object reference counts in uvm_map more carefully.

Acquire a reference unconditionally first; then let uvm_map consume
it on success, and release it ourselves on failure.

As we did before -- acquiring a fresh reference on success -- another
thread might release the reference with a concurrent uvm_unmap before
we could acquire it back, thereby destroying the object.

XXX pullup-7 in part (i915_gem_fault)
XXX pullup-8 in part (i915_gem_fault)

uvm_map consumes reference on success, so be sure to acquire another.

Move hint setting nearer to uvm_map for clarity. CTASSERT page size.

Elsewhere this code says 4096, rather than PAGE_SIZE. PAGE_SIZE is
the smallest unit that uvm_map accepts, so it had better match.

Initialize kva hint for uvm_map to avoid using uninitialized memory.

Rename drm_gem_object::gemo_shm_uao -> drm_gem_object::filp.

The Linux member that is basically analogous to this is named so,
which means we can eliminate a few diffs this way.

Rework how we map i915 drm gem objects.

The old way made no sense at all. This way might make sense.

Use bus_dmamem_map instead of kmap(sg_page(...)).

merge linux-drm-v4-4-143

branches: 1.1.1;
Initial revision

Merge linux-drm-v5-6-rc3

branches: 1.2.2; 1.2.6;
merge linux-drm-v4-4-143

branches: 1.1.1;
Initial revision

Merge linux-drm-v5-6-rc3

branches: 1.6.2; 1.6.6;
Rename drm_gem_object::gemo_shm_uao -> drm_gem_object::filp.

The Linux member that is basically analogous to this is named so,
which means we can eliminate a few diffs this way.

Use gemo_shm_uao rather than filp.

Use kpause for schedule_timeout_killable/fatal_signal_pending.

Horrible kludgey mutex_is_locked_by that should work here.

merge linux-drm-v4-4-143

branches: 1.1.1;
Initial revision

Merge linux-drm-v5-6-rc3

Reduce diffs by side-loading some header files like Linux.

From riastradh

Use a header file hack to cope with Linux/NetBSD namespace clashes.

Currently serves for:

ALIGN
mutex_init
mutex_destroy

branches: 1.12.2;
Avoid undefined behavior in g4x_get_stolen_reserved().
The change is the same as newer i915_gem_stolen.c.

branches: 1.11.4;
add support for kabylake and skylake GT4 (untested) GPUs.
largely taken from openbsd and linux 4.13 trees (which
have this code identical), with mimimal porting to netbsd.

i have not installed (and thus tested) the newer referenced
firmware files.

only real local change is to fix IS_BROXTON() macro to check
the things valid in this era of drm. previous match would
attach on KBL, and then a loop would never exit.

tested on kabylake P630. needs mesa 11.x or newer for GL
to work.

ok @riastradh.

Re-enable i915 GSM now that pci_bus_config_read/write_* is here.

Rework how we map i915 drm gem objects.

The old way made no sense at all. This way might make sense.

Ifdef out more graphics stolen memory stuff for now.

Destroy mutex if we init it.

some more definitions necessary

ifdef out all the i915 gtt virtual memory stuff that errors,
unless it looks exactly like the old code and then merge the
netbsd ifdefs for it.

we don't want to use their ALIGN(, which has more arguments,
use the old alt function.
merge in the old intel_acpi.c code.
don't duplicate DECLARE_BITMAP


Author: coypu <coypu@sdf.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

merge linux-drm-v4-4-143

branches: 1.4.4; 1.4.28; 1.4.30;
Disable reusing graphics stolen memory in i915drmkms for now.

Sometimes the GTT is located in graphics stolen memory, and who knows
what else the BIOS might have done with it. Upstream has a patch to
sneak around the GTT, but disabling it altogether is simpler and
probably safer.

Don't double-free obj->pages, with two different allocators to boot!

Make it build and boot on my test machines.

Screen blanks on boot on the Ivy Bridge system with

DRM error in cpt_serr_int_handler: PCH transcoder A FIFO underrun

But after that everything is OK. Appears to be an upstream problem.
To investigate...

I think there's a cache flushing issue somewhere -- there are little
display artefacts on my T60.

branches: 1.1.1;
Initial revision

Merge linux-drm-v5-6-rc3

Forgot to nix conflicts after moving changes to i915_gem_fence.c.

merge linux-drm-v4-4-143

branches: 1.4.18; 1.4.20;
Fix little C issues in i915drmkms hindering the Clang build.

- Test `x == NULL', not `&container_of(x, t, base)->base == NULL'.
- Fix format strings in intel_dp_probe_oui to use %hhx for uint8_t.
- Omit needless ioread32 in intel_overlay.c.
- Ifdef out unused definitions in intel_tv.c.

branches: 1.3.2; 1.3.4;
fix merge conflicts

branches: 1.2.2; 1.2.4;
Merge riastradh-drm2 to HEAD.

branches: 1.1.1;
Initial revision

Merge linux-drm-v5-6-rc3

branches: 1.3.2; 1.3.6;
Don't care about obj->get_page.

merge linux-drm-v4-4-143

branches: 1.1.1;
Initial revision

Update for new name, fallthrough, reorder to make GCC happy about fallthrough comment


Author: Maya Rashish <maya@NetBSD.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

Restore mmap ioctl version ABI hack that was lost in merge.

Merge linux-drm-v5-6-rc3

branches: 1.1.1;
Initial revision

Make intel_psr.c build
netbsd-ify
Stub out intel_lpe_audio.c (for now?).
Hammer intel_opregion.c into shape.
Hack up intel_lrc.c
Adapt intel_reset.c
Stub out intel_region_lmem.c
Adapt intel_ring_submission.c
Adapt intel_rps.c
Adapt intel_sdvo.c
Adapt intel_runtime_pm.c
Adapt intel_sideband.c
Adapt intel_sprite.c
Adapt intel_tv.c
Adapt intel_tc.c
Adapt intel_sseu.c
Adapt intel_timeline.c
Adapt intel_uc_fw.c and fix typo
Fix up intel_uncore.c
intel_vga.c
intel_wakeref.c
intel_vdsc.c
intel_wopcm.c
intel_workarounds.c
i915_globals.c
Fix up intelfb.c
Stub out intel_guc_log.c
Stub out intel_dsi_dcs_backlight.c
i915_perf.c
intel_hdmi.c
intel_hdcp.c


Author: Maya Rashish <maya@NetBSD.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

Merge linux-drm-v5-6-rc3

branches: 1.1.1;
Initial revision

Merge linux-drm-v5-6-rc3

branches: 1.1.1;
Initial revision

i915: Suggest gnats.n.o, not fdo gitlab, for bugs.

Merge linux-drm-v5-6-rc3

Reduce diffs by side-loading some header files like Linux.

From riastradh

branches: 1.10.6;
Omit print with wrong format -- vestige of merge.

Lock around use of irq_queue.

Cast suseconds_t to long for %ld. Use PRIx64, not %llx, for uint64_t.

some more definitions necessary

ifdef out all the i915 gtt virtual memory stuff that errors,
unless it looks exactly like the old code and then merge the
netbsd ifdefs for it.

we don't want to use their ALIGN(, which has more arguments,
use the old alt function.
merge in the old intel_acpi.c code.
don't duplicate DECLARE_BITMAP


Author: coypu <coypu@sdf.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

mild modifications to avoid needing more headers


Author: coypu <coypu@sdf.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

merge linux-drm-v4-4-143

branches: 1.4.10; 1.4.16; 1.4.18;
Avoid taking locks during interrupts and explain why we are doing it this way.

branches: 1.3.2; 1.3.4; 1.3.6; 1.3.8; 1.3.12;
Fix formatting and locking errors.

Format time_t by casting to intmax_t and using PRIdMAX, since we have
no PRIxyz for time_t.

We already hold the irq lock on entry to i915_record_ring_state, so
don't take it again.

Make it build and boot on my test machines.

Screen blanks on boot on the Ivy Bridge system with

DRM error in cpt_serr_int_handler: PCH transcoder A FIFO underrun

But after that everything is OK. Appears to be an upstream problem.
To investigate...

I think there's a cache flushing issue somewhere -- there are little
display artefacts on my T60.

branches: 1.1.1;
Initial revision

Merge linux-drm-v5-6-rc3

branches: 1.1.1;
Initial revision

Merge linux-drm-v5-6-rc3

branches: 1.2.2; 1.2.6;
merge linux-drm-v4-4-143

branches: 1.1.1;
Initial revision

Merge linux-drm-v5-6-rc3

Eliminate -Wpointer-arith patches; use -Wno-pointer-arith instead.

branches: 1.8.2; 1.8.6; 1.8.10;
Use dtrace probes to implement i915_trace.h.

Rework how we map i915 drm gem objects.

The old way made no sense at all. This way might make sense.

Implement a bus_dma version of Linux sg_pcopy_from_buffer.

Include <linux/bitmap.h> for bitmap operations.

No arithmetic on void *.

<asm/cpufeature.h> for cache_line_size.

merge linux-drm-v4-4-143

branches: 1.1.1;
Initial revision

Merge linux-drm-v5-6-rc3

merge linux-drm-v4-4-143

branches: 1.1.1;
Initial revision

i915: more work

drm: Another pass over i915. Most of i915 gem builds now.

Undo harmful local diffs


Author: Maya Rashish <maya@NetBSD.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

More crap to make this thing get closer to building.

bus-spaceify raw_reg_read/write.

XXX heavily modified - maya

Merge linux-drm-v5-6-rc3

Reduce diffs by side-loading some header files like Linux.

From riastradh

Undo the `enum pipe' -> `enum i915_pipe' rename.

branches: 1.17.6;
add support for kabylake and skylake GT4 (untested) GPUs.
largely taken from openbsd and linux 4.13 trees (which
have this code identical), with mimimal porting to netbsd.

i have not installed (and thus tested) the newer referenced
firmware files.

only real local change is to fix IS_BROXTON() macro to check
the things valid in this era of drm. previous match would
attach on KBL, and then a loop would never exit.

tested on kabylake P630. needs mesa 11.x or newer for GL
to work.

ok @riastradh.

Advance by 4-byte quantities, not by 1-byte quantities.

Narrow scope of locks in i915 error checks to reduce deadlocks.

Not really sure why I had put some of these locks here...

Fix up i915_irq.c.

- <linux/hardirq.h> for synchronize_irq
- DRM_IRQ_ARGS
- PRIx64, not %llx, for 64-bit printf

enum pipe -> enum i915_pipe (why did we inflict this on ourselves, again?)

struct drm_bus abstraction is no more. Cope.

struct drm_driver now has the bus-specific intr establish routine
(called request_irq/free_irq to match Linux's style).

merge linux-drm-v4-4-143

branches: 1.10.18; 1.10.20;
Change enum pipe to enum i915_pipe
(enum pipe uses too global of a tag and conflicts with struct pipe).

Lock irq_lock around i915_error_wake_up in i915_error_work_func.

Fix two mistakes in previous.

- i915_error_wake_up is given irq_lock, so it need not take that.
- Unlock irq_lock on exit from i915_hangcheck_elapsed if ring_hung.

XXX This introduces the lock order irq_lock -> pending_flip_lock.
How about just using irq_lock for pending flips?

Fix i915 locking around error handling.

branches: 1.6.2; 1.6.4;
Make it build and boot on my test machines.

Screen blanks on boot on the Ivy Bridge system with

DRM error in cpt_serr_int_handler: PCH transcoder A FIFO underrun

But after that everything is OK. Appears to be an upstream problem.
To investigate...

I think there's a cache flushing issue somewhere -- there are little
display artefacts on my T60.

fix merge conflicts

Work around bug on one-pipe 845 devices until we update drm again.

branches: 1.3.2; 1.3.4;
Fix unused variable warnings in drm2.

Merge riastradh-drm2 to HEAD.

branches: 1.1.1;
Initial revision

Merge linux-drm-v5-6-rc3

branches: 1.1.1;
Initial revision

drm: Another pass over i915. Most of i915 gem builds now.

No need to port this until we deal with movntdqa (and kernel FPU use).

Merge linux-drm-v5-6-rc3

branches: 1.1.1;
Initial revision

Merge linux-drm-v5-6-rc3

branches: 1.1.1;
Initial revision

Merge linux-drm-v5-6-rc3

branches: 1.1.1;
Initial revision

More crap to make this thing get closer to building.