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: 1.10.4;
Let radeon DRM driver build on ports that lack ACPI

linux: Define acpi_size as alias for ACPI_SIZE.

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

drm: Move acpi_check_dsm &c. from intel_acpi.c to new linux_acpi.c.

i915: another whack at it

Pull in acpica.h on aarch64

branches: 1.4.18; 1.4.24;
only include acpivar.h when NACPICA > 0

branches: 1.3.2;
i915drmkms(4): Enable CONFIG_ACPI if NACPICA > 0.
Now brightness can be adjusted via hotkey on Mouse Computer LB-J300X (Clevo W330SU2).

branches: 1.2.4; 1.2.6; 1.2.8; 1.2.10;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file acpi.h was initially added on branch riastradh-drm2.

branches: 1.2.4; 1.2.6; 1.2.10;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file acpi_io.h was initially added on branch riastradh-drm2.

Make <linux/agp_backend.h> yield the AGP APIish.

branches: 1.2.4; 1.2.6; 1.2.10;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file agp_backend.h was initially added on branch riastradh-drm2.

More Linux header stubs.

Let's just use empty header files until we actually populate these.
Save a bit of disk space...

linux: stub async_synchronize_cookie

branches: 1.4.2; 1.4.6;
Implement <linux/async.h> synchronously.

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>

include guard for async.h


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

local versions of headers required


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

amdgpu: Another pass through.

amdgpu_object.c
amdgpu_pci.c
amdgpu_pm.c
amdgpu_polaris10_smumgr.c
amdgpu_pp_psm.c
amdgpu_ppatomctrl.c
amdgpu_ppatomfwctrl.c
amdgpu_process_pptables_v1_0.o
amdgpu_psp.o
amdgpu_psp_v10_0.c
amdgpu_psp_v11_0.c
amdgpu_psp_v12_0.c
amdgpu_psp_v3_1.c
amdgpu_ras.c
amdgpu_ras_eeprom.c
amdgpu_rc_calc.c
amdgpu_rc_calc_dpi.c
amdgpu_ring.c
amdgpu_rlc.c
amdgpu_rn_clk_mgr.c
amdgpu_sdma_v3_0.c
amdgpu_sdma_v4_0.c
amdgpu_sdma_v5_0.c
amdgpu_si.c
amdgpu_si_dma.c
amdgpu_si_dpm.c
amdgpu_smu.c
amdgpu_smu10_hwmgr.c
amdgpu_smu7_hwmgr.c
amdgpu_smu8_hwmgr.c
amdgpu_smu8_smumgr.c
amdgpu_smu_v11_0.c (and a sprinkling of consts)
amdgpu_smu_v11_0_i2c.c
amdgpu_soc15.c
amdgpu_tonga_smumgr.c
amdgpu_ttm.c
amdgpu_ucode.c
amdgpu_uvd.c
amdgpu_uvd_v5_0.c
amdgpu_uvd_v6_0.c
amdgpu_uvd_v7_0.c
amdgpu_vce.c
amdgpu_vce_v2_0.c
amdgpu_vce_v4_0.c
amdgpu_vcn_v1_0.c
amdgpu_vcn_v2_0.c
amdgpu_vcn_v2_5.c
amdgpu_vega10_hwmgr.c
amdgpu_vega10_powertune.c
amdgpu_vega10_processpptables.c
amdgpu_vega10_reg_init.c
amdgpu_vega12_hwmgr.c
amdgpu_vega20_hwmgr.c
amdgpu_vega20_ppt.c
amdgpu_vega20_reg_init.c
amdgpu_vi.c
amdgpu_virt.c
amdgpu_vram_mgr.c
amdgpu_xgmi.c -- disable (for now?)
amdgpufb.c

i915: more work

i915: More progress.

linux: Just use __sync_* for xchg and cmpxchg.

Easier than figuring out the dance with __builtin_choose_expr to make
these work with all types they need to work with.

linux: Make xchg and cmpxchg work on pointers too.

drm/linux: atomic_fetch_add/inc

drm/linux: Add memset32, SZ_64K, atomic_sub_return.

cmpxchg

Move Linux atomic bitops from linux/atomic.h to linux/bitops.h.

Move smp_mb_before/after_atomic from linux/atomic.h to asm/barrier.h.

Linux clear_bit_unlock

atomic_set_release

atomic_long_xchg

Omit __insn_barrier where implied or subsumed by volatile.

atomic_dec_if_positive

Allow atomic_read of const.

Guarantee no repeated loads.

atomic_fetch_xor

atomic_andnot

Kludgey xchg.

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

From riastradh

branches: 1.21.2;
Fix signed/unsigned comparison

Merge the [pgoyette-compat] branch

Implement atomic64_add_return and atomic64_inc_return.

Sort.

Fix argument types of atomic64_add/sub to be sensible.

Technically Linux does use long long rather than int64_t, but that's
silly.

Implement ATOMIC64_INIT.

Hash-locked atomic64 to give ports like powerpc32 a chance at radeon.

Add atomic_long.

Attempt to match Linux semantics for membars implied by atomics.

This is kind of moot at the moment because we're mostly x86-only for
drmkms, but this might help in the future if we ever went beyond x86.

Nix memory barriers that were nixed upstream. Add smp_mb__after_atomic.

Fix return types of test_and_{set,clear,change}_bit.

Add Linux smp_mb__before_atomic.

Implement atomic64_cmpxchg.

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>

branches: 1.7.4; 1.7.28; 1.7.30;
Correct return values of Linux test-and-set operations (PR 48999).

Linux's Documentation/atomic_ops.txt says in no uncertain terms these
must return 0 or 1, not zero or nonzero. I don't think this has
caused an issue for drm (yet), but better to have it right.

First whack at radeon driver.

No hardware to test yet, but it builds.

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 atomic_dec_and_test in <linux/atomic.h>: test 0, not -1.

branches: 1.3.2; 1.3.4; 1.3.6;
Fix volatile qualifiers in drm2 <linux/atomic.h>.

Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file atomic.h was initially added on branch riastradh-drm2.

Implement ewma API, exponentially weighted moving average.

Create stub headers for files included by drm code.


Author: Maya Rashish <maya@NetBSD.org>

Backlight enable/disable stubs.

branches: 1.2.4; 1.2.6; 1.2.10;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file backlight.h was initially added on branch riastradh-drm2.

i915: another whack at it

Create stub headers for files included by drm code.


Author: Maya Rashish <maya@NetBSD.org>

amdgpu: Another pass through.

amdgpu_object.c
amdgpu_pci.c
amdgpu_pm.c
amdgpu_polaris10_smumgr.c
amdgpu_pp_psm.c
amdgpu_ppatomctrl.c
amdgpu_ppatomfwctrl.c
amdgpu_process_pptables_v1_0.o
amdgpu_psp.o
amdgpu_psp_v10_0.c
amdgpu_psp_v11_0.c
amdgpu_psp_v12_0.c
amdgpu_psp_v3_1.c
amdgpu_ras.c
amdgpu_ras_eeprom.c
amdgpu_rc_calc.c
amdgpu_rc_calc_dpi.c
amdgpu_ring.c
amdgpu_rlc.c
amdgpu_rn_clk_mgr.c
amdgpu_sdma_v3_0.c
amdgpu_sdma_v4_0.c
amdgpu_sdma_v5_0.c
amdgpu_si.c
amdgpu_si_dma.c
amdgpu_si_dpm.c
amdgpu_smu.c
amdgpu_smu10_hwmgr.c
amdgpu_smu7_hwmgr.c
amdgpu_smu8_hwmgr.c
amdgpu_smu8_smumgr.c
amdgpu_smu_v11_0.c (and a sprinkling of consts)
amdgpu_smu_v11_0_i2c.c
amdgpu_soc15.c
amdgpu_tonga_smumgr.c
amdgpu_ttm.c
amdgpu_ucode.c
amdgpu_uvd.c
amdgpu_uvd_v5_0.c
amdgpu_uvd_v6_0.c
amdgpu_uvd_v7_0.c
amdgpu_vce.c
amdgpu_vce_v2_0.c
amdgpu_vce_v4_0.c
amdgpu_vcn_v1_0.c
amdgpu_vcn_v2_0.c
amdgpu_vcn_v2_5.c
amdgpu_vega10_hwmgr.c
amdgpu_vega10_powertune.c
amdgpu_vega10_processpptables.c
amdgpu_vega10_reg_init.c
amdgpu_vega12_hwmgr.c
amdgpu_vega20_hwmgr.c
amdgpu_vega20_ppt.c
amdgpu_vega20_reg_init.c
amdgpu_vi.c
amdgpu_virt.c
amdgpu_vram_mgr.c
amdgpu_xgmi.c -- disable (for now?)
amdgpufb.c

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/linux: Fix bitmap_zalloc size.

Provide bitmap_alloc


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

bitmap_complement

Split bitmap_set/clear into begin/middle/end sections.

For clarity and speed. Largely from uwe@ with tweaks from me.

Name the number of words in bitmap_zero too.

Be consistent about nbits >= bpl.

Name bpl in bitmap_and/or. Fix and clarify comments.

Try setting and clearing the bits we meant, not other bits.

Fix comments to match intent while we're fixing reality too.

Implement Linux bitmap utilities.

branches: 1.2.4; 1.2.6; 1.2.10; 1.2.34; 1.2.36;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file bitmap.h was initially added on branch riastradh-drm2.

move bitops.h so we can include it from kernel.h

match linux side-loading of this header.


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

Fix find_first_zero_bit to find the high bits of 64-bit words...oops.

XXX pullup

hweight32 should take uint32_t, not uint16_t. OOPS.

XXX pullup

branches: 1.11.10; 1.11.16; 1.11.18;
Fix comment in previous.

Fix indexing convention of Linux __ffs/__ffs64 shims.

They reject zero input, and yield zero-based indexing, unlike our
ffs/ffs64, which return zero for zero and yield one-based indexing.

May let nouveau make a little more progress toward booting!

Found by mrg@.

Add hweight8 (hamming weight of 8-bit quantity).

branches: 1.8.2;
Another round of nouveau whack-a-mole.

branches: 1.7.2; 1.7.4;
Add Linux __ffs64.

Correct return values of Linux test-and-set operations (PR 48999).

Linux's Documentation/atomic_ops.txt says in no uncertain terms these
must return 0 or 1, not zero or nonzero. I don't think this has
caused an issue for drm (yet), but better to have it right.

Add non-atomic Linux test-and-set operations.

First whack at radeon driver.

No hardware to test yet, but it builds.

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: 1.2.2; 1.2.4; 1.2.6;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file bitops.h was initially added on branch riastradh-drm2.

Add a BIT_MASK in bits.h, move it to common so bitops.h can sideload

match linux.


Author: Maya Rashish <maya@NetBSD.org>

Create stub headers for files included by drm code.


Author: Maya Rashish <maya@NetBSD.org>

Move linux/bug.h and asm/bug.h to common.

branches: 1.3.4;
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.

branches: 1.2.2; 1.2.4; 1.2.6;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file bug.h was initially added on branch riastradh-drm2.

linux: Move BUILD_BUG_* to common linux/build_bug.h.

Create stub headers for files included by drm code.


Author: Maya Rashish <maya@NetBSD.org>

More Linux header stubs.

Let's just use empty header files until we actually populate these.
Save a bit of disk space...

CAP_SYS_NICE

branches: 1.2.2; 1.2.6;
fix a typo, caught by -Werror,-Wheader-guard

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

Instead stub Linux capable(CAP_SYS_ADMIN).

branches: 1.2.4; 1.2.6; 1.2.10;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file cdev.h was initially added on branch riastradh-drm2.

branches: 1.2.2; 1.2.6;
Populate circ_buf.h with CIRC_SPACE.

local versions of headers required


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

branches: 1.3.2; 1.3.6;
Stub out clk API if no FBT.

Our clk API already echoes Linux's, it seems.

Stub <linux/clk.h>.

branches: 1.2.4; 1.2.6; 1.2.10;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file compat.h was initially added on branch riastradh-drm2.

Move <linux/compiler.h> to common.

Stub linux compiler.h, scatterlist.h, seqlock.h, sort.h.

Move <linux/completion.h> from drm2 to external/bsd/common.

Add some more Linux completion operations.

Adapted from a patch by nick@.

Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file completion.h was initially added on branch riastradh-drm2.

branches: 1.1.2; 1.1.6;
Add stub headers to be filled in as needed.

Stub out console_lock/unlock and drm_fb_helper_set_suspend.

Reduce some more diffs.

branches: 1.2.4; 1.2.6; 1.2.10; 1.2.36; 1.2.42;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file console.h was initially added on branch riastradh-drm2.

linux: Use sys/cpu.h, not machine/cpu.h.

Provide cpufreq.h


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

branches: 1.2.4; 1.2.6; 1.2.10;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file cpufreq.h was initially added on branch riastradh-drm2.

branches: 1.2.2; 1.2.6;
The lua 5.1 to 5.3 update added isprint to libkern.h

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: 1.2.4; 1.2.6; 1.2.10;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file debugfs.h was initially added on branch riastradh-drm2.

Create stub headers for files included by drm code.


Author: Maya Rashish <maya@NetBSD.org>

match linux side-loading


Author: Maya Rashish <maya@NetBSD.org>

drm: sys/param.h, not machine/param.h, for DELAY.

branches: 1.6.10; 1.6.12;
Placate -Wsign-compare.

branches: 1.5.6;
Round nsec up for usec delay.

XXX pullup

branches: 1.4.10; 1.4.16; 1.4.18;
In msleep use mdelay if cold

branches: 1.3.2; 1.3.4; 1.3.6;
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.

branches: 1.2.2; 1.2.4; 1.2.6;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file delay.h was initially added on branch riastradh-drm2.

drm: Reduce verbosity of drm kernel messages.

drm: Ifdef out devm_* stuff rather than stub it out.

Stub is not helpful, and this pollutes the symbol namespace.

drm: Downgrade warnings from aprint_error to aprint_normal.

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>

Alias dev_WARN=dev_warn.

Revert "Give up on finding how linux side-loads these"

Found it.

Add some sideloads.

Provide dev_warn_ratelimited


Author: Maya Rashish <maya@NetBSD.org>

stub devm_add_action


Author: Maya Rashish <maya@NetBSD.org>

provide get_device, XXX might want to revisit this later


Author: Maya Rashish <maya@NetBSD.org>

Empty dev_pm_set_driver_flags stub.

Paranoia: let dev be null, in match routines.

Add some more device print routines.

branches: 1.4.4; 1.4.28; 1.4.30;
First whack at radeon driver.

No hardware to test yet, but it builds.

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: 1.2.2; 1.2.4; 1.2.6;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file device.h was initially added on branch riastradh-drm2.

amdgpu: amdgpu_dma_buf.c

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.

Fix up drm_prime.c.

dma_buf_ops::map and unmap are no more.

Rename reservation_object -> dma_resv.

match linux side-loading


Author: Maya Rashish <maya@NetBSD.org>

Update type signatures for begin/end_cpu_access.

Update dma-buf API enough to compile drm_prime.c.

Implement poll for reservations and dma bufs.

Draft support for drm prime.

branches: 1.2.4; 1.2.6; 1.2.10; 1.2.34; 1.2.36;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file dma-buf.h was initially added on branch riastradh-drm2.

drm: Draft dma-fence-array.

amdgpu: amdgpu_dma_buf.c

<sys/stdbool.h> for bool

dma-fence-array.h stub declarations.

Start dma_fence_array.

Bunch more Linux header file stubs.

drm: Implement dma fence chains.

dma fence chain stubs

dma fence chain stubs

Create stub headers for files included by drm code.


Author: Maya Rashish <maya@NetBSD.org>

drm: Fix dma fence stub so it works with locking operations.

drm: Support 64-bit fence context and sequence numbers.

drm: Sprinkle assertions about fence lifetime.

amdgpu_fb.c
amdgpu_fence.c
amdgpu_gart.c
amdgpu_fixpt31_32.c
amdgpu_fiji_smumgr.c
amdgpu_gem.c
amdgpu_gfx_v10_0.c
amdgpu_gfx_v6_0.c through amdgpu_gfx_v9_0.c
amdgpu_gmc.c
amdgpu_gmc_v6_0.c through amdgpu_gmc_v10_0.c
amdgpu_gtt_mgr.c
some amdgpu_h* files
some amdgpu_i* files
some amdgpu_j* files
amdgpu_kms.c
some amdgpu_m* and amdgpu_n* files

i915: Defer destroying request lock and fence until page destruction.

Move grody i915 dma-fence hacks into dma-fence proper.

Teach dma_fence_any_wait_timeout to return index of signalled fence.

dma_fence_get_status

More side loads.

First approximation of dma_fence_get_stub.

Allow const volatile fence pointer pointers in dma_fence_get_rcu_safe.

Namespace for dma_fence_get_rcu_safe.

dma_fence_set_error

Linux wants to violate abstractions, so fine, let them.

Add dma_fence_get_rcu_safe.

Rename fence -> dma_fence, step 3 of 3: code.

Rename fence -> dma_fence, step 1: files.

amdgpu: Another pass through.

amdgpu_object.c
amdgpu_pci.c
amdgpu_pm.c
amdgpu_polaris10_smumgr.c
amdgpu_pp_psm.c
amdgpu_ppatomctrl.c
amdgpu_ppatomfwctrl.c
amdgpu_process_pptables_v1_0.o
amdgpu_psp.o
amdgpu_psp_v10_0.c
amdgpu_psp_v11_0.c
amdgpu_psp_v12_0.c
amdgpu_psp_v3_1.c
amdgpu_ras.c
amdgpu_ras_eeprom.c
amdgpu_rc_calc.c
amdgpu_rc_calc_dpi.c
amdgpu_ring.c
amdgpu_rlc.c
amdgpu_rn_clk_mgr.c
amdgpu_sdma_v3_0.c
amdgpu_sdma_v4_0.c
amdgpu_sdma_v5_0.c
amdgpu_si.c
amdgpu_si_dma.c
amdgpu_si_dpm.c
amdgpu_smu.c
amdgpu_smu10_hwmgr.c
amdgpu_smu7_hwmgr.c
amdgpu_smu8_hwmgr.c
amdgpu_smu8_smumgr.c
amdgpu_smu_v11_0.c (and a sprinkling of consts)
amdgpu_smu_v11_0_i2c.c
amdgpu_soc15.c
amdgpu_tonga_smumgr.c
amdgpu_ttm.c
amdgpu_ucode.c
amdgpu_uvd.c
amdgpu_uvd_v5_0.c
amdgpu_uvd_v6_0.c
amdgpu_uvd_v7_0.c
amdgpu_vce.c
amdgpu_vce_v2_0.c
amdgpu_vce_v4_0.c
amdgpu_vcn_v1_0.c
amdgpu_vcn_v2_0.c
amdgpu_vcn_v2_5.c
amdgpu_vega10_hwmgr.c
amdgpu_vega10_powertune.c
amdgpu_vega10_processpptables.c
amdgpu_vega10_reg_init.c
amdgpu_vega12_hwmgr.c
amdgpu_vega20_hwmgr.c
amdgpu_vega20_ppt.c
amdgpu_vega20_reg_init.c
amdgpu_vi.c
amdgpu_virt.c
amdgpu_vram_mgr.c
amdgpu_xgmi.c -- disable (for now?)
amdgpufb.c

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.

Stub dma_addressing_limited, fixup references for caller

This might turn out to be a mistake if we actually need to distinguish
if a device is only capable of 32-bit DMA.


Author: Maya Rashish <maya@NetBSD.org>

Move dma_addr_t to linux/types.h.

Allow maximum number of bits in DMA_BIT_MASK.

Forget about trying to replicate side-loading, will just include
directly


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

begin fixing -Werror build:

Provide missing declarations from older drm
provide more definitions used by code
no idea what might_lock is, but it might be some error-checking
lockdep thing.


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

branches: 1.3.4; 1.3.28; 1.3.30;
Add Linux DMA_BIT_MASK.

branches: 1.2.2; 1.2.4; 1.2.6;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file dma-mapping.h was initially added on branch riastradh-drm2.

amdgpu: Another pass through.

amdgpu_object.c
amdgpu_pci.c
amdgpu_pm.c
amdgpu_polaris10_smumgr.c
amdgpu_pp_psm.c
amdgpu_ppatomctrl.c
amdgpu_ppatomfwctrl.c
amdgpu_process_pptables_v1_0.o
amdgpu_psp.o
amdgpu_psp_v10_0.c
amdgpu_psp_v11_0.c
amdgpu_psp_v12_0.c
amdgpu_psp_v3_1.c
amdgpu_ras.c
amdgpu_ras_eeprom.c
amdgpu_rc_calc.c
amdgpu_rc_calc_dpi.c
amdgpu_ring.c
amdgpu_rlc.c
amdgpu_rn_clk_mgr.c
amdgpu_sdma_v3_0.c
amdgpu_sdma_v4_0.c
amdgpu_sdma_v5_0.c
amdgpu_si.c
amdgpu_si_dma.c
amdgpu_si_dpm.c
amdgpu_smu.c
amdgpu_smu10_hwmgr.c
amdgpu_smu7_hwmgr.c
amdgpu_smu8_hwmgr.c
amdgpu_smu8_smumgr.c
amdgpu_smu_v11_0.c (and a sprinkling of consts)
amdgpu_smu_v11_0_i2c.c
amdgpu_soc15.c
amdgpu_tonga_smumgr.c
amdgpu_ttm.c
amdgpu_ucode.c
amdgpu_uvd.c
amdgpu_uvd_v5_0.c
amdgpu_uvd_v6_0.c
amdgpu_uvd_v7_0.c
amdgpu_vce.c
amdgpu_vce_v2_0.c
amdgpu_vce_v4_0.c
amdgpu_vcn_v1_0.c
amdgpu_vcn_v2_0.c
amdgpu_vcn_v2_5.c
amdgpu_vega10_hwmgr.c
amdgpu_vega10_powertune.c
amdgpu_vega10_processpptables.c
amdgpu_vega10_reg_init.c
amdgpu_vega12_hwmgr.c
amdgpu_vega20_hwmgr.c
amdgpu_vega20_ppt.c
amdgpu_vega20_reg_init.c
amdgpu_vi.c
amdgpu_virt.c
amdgpu_vram_mgr.c
amdgpu_xgmi.c -- disable (for now?)
amdgpufb.c

Implement dma_resv_locking_ctx more carefully.

Provide dma_resv_locking_ctx


Author: Maya Rashish <maya@NetBSD.org>

dma_resv_reserve_shared takes one more argument, num_fences

Fortunately it's ~always 1, and (XXX) I assume our implementation
assumes this, too. But assert it.


Author: Maya Rashish <maya@NetBSD.org>

Add slow path versions of to drm_resv_lock.

Rename reservation_object -> dma_resv.

handle us having a named union in struct ww_mutex


Author: Maya Rashish <maya@NetBSD.org>

Switch to upstream dma-resv.h, it's MIT


Author: Maya Rashish <maya@NetBSD.org>

stub xarray.h and dma-resv.h

Add some upstream side-loads and empty structs.


Author: Maya Rashish <maya@NetBSD.org>

Create stub headers for files included by drm code.


Author: Maya Rashish <maya@NetBSD.org>

branches: 1.2.4; 1.2.6; 1.2.10;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file dma_remapping.h was initially added on branch riastradh-drm2.

linux/dmapool.h: Start filling out stubs.

drm/vmwgfx: First draft. Passes make depend, doesn't build yet.

Need <sys/types.h> for bool.

Namespace Linux symbols.

provide some dummy structs

declaration without implementation of dmi_match (Won't link)
provide div64_u64_rem XXX the naming of the parameters is backwards
ci_send_msg_to_smc_return_parameter isn't unused, keep it


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

branches: 1.3.2; 1.3.4; 1.3.8; 1.3.32; 1.3.34;
Implement Linux dmi_check_system API.

Use it to get a Thinkpad quirk for i915 graphics.

Patch from nonoka@, PR 48708.

branches: 1.2.2;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file dmi.h was initially added on branch riastradh-drm2.

branches: 1.2.4; 1.2.6; 1.2.10;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file efi.h was initially added on branch riastradh-drm2.

Move linux/err.h into common/include.

OK riastradh@

branches: 1.3.2; 1.3.4; 1.3.6; 1.3.10;
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: 1.2.2; 1.2.4; 1.2.6;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file err.h was initially added on branch riastradh-drm2.

Move <linux/errno.h> from drm2 to common.

branches: 1.2.2;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file errno.h was initially added on branch riastradh-drm2.

Include linux/export.h in linux/printk.h, move to common

linux does this through:
printk.h -> linkage.h -> export.h

this will require less patching of drm files.

Author: Maya Rashish <maya@NetBSD.org>

Move IS_REACHABLE to <linux/kernel.h> to avoid double definition.

Define IS_ENABLED and IS_REACHABLE.

branches: 1.2.4; 1.2.6; 1.2.10;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file export.h was initially added on branch riastradh-drm2.

Bunch more Linux header file stubs.

drm: Just disable fb_get_options for now.

Match linux side-loading


Author: Maya Rashish <maya@NetBSD.org>

stub fb_get_options


Author: Maya Rashish <maya@NetBSD.org>

Stub out console_lock/unlock and drm_fb_helper_set_suspend.

Reduce some more diffs.

branches: 1.3.6;
match linux side loading of headers

more bring in OLD drm code to unbreak the build
matching netbsd types etc. to get the build going


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

branches: 1.2.4; 1.2.6; 1.2.10; 1.2.34; 1.2.36;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file fb.h was initially added on branch riastradh-drm2.

branches: 1.1.2; 1.1.6;
Not sure what DRM is doing with the fdtable, but OK...

Create stub headers for files included by drm code.


Author: Maya Rashish <maya@NetBSD.org>

Rename fence -> dma_fence, step 1: files.

FENCE_TRACE is __printflike

branches: 1.15.2; 1.15.6; 1.15.10;
Make extremely noisy fence trace output conditional.

Add fence_is_later and fence_wait_any_timeout.

Add missing namespacing.

Use atomics to manage fence->flags.

Outside users use this field too, and expect it to be managed with
atomics.

Draft implementation of the Linux reservation and fence APIs.

Add fence_enable_sw_signaling prototype.

Add fence_signal prototype.

Define FENCE_TRACE. Need some way to limit it...

Fill out <linux/fence.h>, pending implementing the symbols.

Namespace fence_get.

Add fence_get prototype.

Include <linux/rcupdate.h> in <linux/fence.h>.

Add some <linux/fence.h> prototypes.

provide some dummy structs

declaration without implementation of dmi_match (Won't link)
provide div64_u64_rem XXX the naming of the parameters is backwards
ci_send_msg_to_smc_return_parameter isn't unused, keep it


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

Add stub headers to be filled in as needed.

New fdget/fdput wrappers help to reduce diffs.

missing includes


Author: Maya Rashish <maya@NetBSD.org>

<linux/file.h> oughta be enough to dereference struct file.

fget == fd_getfile

Work out sync file construction API.

branches: 1.2.4; 1.2.6; 1.2.10;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file file.h was initially added on branch riastradh-drm2.

linux: request_firmware_direct

Move Linux firmware shims to their own .c file.

Add firmware_request_nowarn.

correct typo

shows up as null deref in finish_csr_load.


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

Implement Linux asynchronous firmware load API.

just include linux/err.h instead of matching side loading

(it's included by fence.h, for example)

switch void* -> char* because of void pointer arithmetic.


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

branches: 1.6.4; 1.6.28; 1.6.30;
Omit needless firmware handle from Linux struct firmware.

Elide up to one excess driver-named component from firmware pathnames.

Tweak failure case for Linux firmware API.

request_firmware is guaranteed to yield null on failure;
release_firmware on null is guaranteed to be a no-op.

First whack at radeon driver.

No hardware to test yet, but it builds.

branches: 1.2.2; 1.2.4; 1.2.6;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file firmware.h was initially added on branch riastradh-drm2.

linux/frame.h: STACK_FRAME_NON_STANDARD

drm/vmwgfx: First draft. Passes make depend, doesn't build yet.

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

Linux defines struct file in linux/fs.h, match with includes


Author: Maya Rashish <maya@NetBSD.org>

Side-load <linux/wait_bit.h> in <linux/fs.h>.

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: 1.2.4; 1.2.6; 1.2.10; 1.2.36; 1.2.42;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file fs.h was initially added on branch riastradh-drm2.

branches: 1.1.2; 1.1.6;
First whack at radeon driver.

No hardware to test yet, but it builds.

gfp.h and slab.h are now used in kernel.h, so move to common.

Define __GFP_NOFAIL as zero since it's a noop here.

branches: 1.5.4; 1.5.28; 1.5.30;
Remove defunct declarations of alloc_page/free_page.

First whack at radeon driver.

No hardware to test yet, but it builds.

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: 1.2.2; 1.2.4; 1.2.6;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file gfp.h was initially added on branch riastradh-drm2.

i915: hack hack hack

branches: 1.2.2; 1.2.6;
Can't and needn't xcall while cold.

Define synchronize_irq to broadcast xcall noop in <linux/hardirq.h>.

amdgpu_fb.c
amdgpu_fence.c
amdgpu_gart.c
amdgpu_fixpt31_32.c
amdgpu_fiji_smumgr.c
amdgpu_gem.c
amdgpu_gfx_v10_0.c
amdgpu_gfx_v6_0.c through amdgpu_gfx_v9_0.c
amdgpu_gmc.c
amdgpu_gmc_v6_0.c through amdgpu_gmc_v10_0.c
amdgpu_gtt_mgr.c
some amdgpu_h* files
some amdgpu_i* files
some amdgpu_j* files
amdgpu_kms.c
some amdgpu_m* and amdgpu_n* files

sys/param.h for uint32_t


Author: Maya Rashish <maya@NetBSD.org>

Dinky hash_32.

branches: 1.2.4; 1.2.6; 1.2.10;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file hash.h was initially added on branch riastradh-drm2.

amdgpu_fb.c
amdgpu_fence.c
amdgpu_gart.c
amdgpu_fixpt31_32.c
amdgpu_fiji_smumgr.c
amdgpu_gem.c
amdgpu_gfx_v10_0.c
amdgpu_gfx_v6_0.c through amdgpu_gfx_v9_0.c
amdgpu_gmc.c
amdgpu_gmc_v6_0.c through amdgpu_gmc_v10_0.c
amdgpu_gtt_mgr.c
some amdgpu_h* files
some amdgpu_i* files
some amdgpu_j* files
amdgpu_kms.c
some amdgpu_m* and amdgpu_n* files

branches: 1.5.2; 1.5.6;
Durr, declarations don't work like that.

Rewrite hashtable.h locally so it might work here.

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>

something like hashtable.h from openbsd, WIP


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

local versions of headers required


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

linux/hdmi: Use ssize_t for all hdmi_*_infoframe_pack functions.

linux/hdmi: Handle unpacking audio frames too. Sort by number.

linux/hdmi: Make some private functions static.

linux/hdmi: Fix missing hdmi_avi_infoframe_init namespacing.

linux/hdmi.h: Split out logic into .c file.

Shoulda done this ages ago.

linux/hdmi.h: Implement hdmi_infoframe_log.

linux/hdmi.h: Fix spd infoframe header version.

Possibly relevant to PR kern/56740.

linux/hdmi: Update. Add unpack.

linux/hdmi: Fix buffer overrun with index typo.

linux/hdmi: Fix strlcpy abuse -- need strncpy to fill field.

i915: more passes

i915: another whack at it

Fill out some structs. Hopefully order doesn't matter...?

Also stub hdmi_drm_infoframe_init


Author: Maya Rashish <maya@NetBSD.org>

Stub hdmi structs


Author: Maya Rashish <maya@NetBSD.org>

Add some hdmi definitions.

hdmi_audio_infoframe_pack: Initialize the entire output buffer.
hdmi_infoframe_checksum: Fix HDMI infoframe checksum calculations.

branches: 1.5.2;
Fix bounds checking in hdmi_infoframe_header_pack (PR kern/49411).

Back out previous, fix PR kern/49411 correctly.

frame->header.length is not initialized in the vendor packet case.
Instead, make hdmi_infoframe_header_pack take the packet length and
subtract the size of the header when packing the header.

(Could make it take the payload length instead, but for callers it is
more convenient to have a name for the whole packet length.)

Use correct length for hdmi_infoframe_header_pack in <linux/hdmi.h>.

Fixes PR kern/49411.

branches: 1.2.2; 1.2.4; 1.2.6;
First whack at radeon driver.

No hardware to test yet, but it builds.

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.

Side-load <linux/mm.h> in a few places.

Work around `kmap' namespace clash another way.

Sort.

branches: 1.3.4; 1.3.28; 1.3.30;
First whack at radeon driver.

No hardware to test yet, but it builds.

branches: 1.2.2; 1.2.4; 1.2.6;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file highmem.h was initially added on branch riastradh-drm2.

Add some Linux header stubs.

drm: Eliminate hrtimer_destroy and teardown_timer.

These are non-Linux shims added to the Linux API to pair with
hrtimer_setup and setup_timer, but they really only serve to call
callout_destroy, which is not strictly necessary in the NetBSD API.
So although we lose a little bit of diagnostic help this way, we also
reduce diffs a bit and save some trouble that bit us last week.

Add hrtimer_destroy


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

i915: Adapt i915_perf.c.

linux compat: Draft hrtimer shims.

(not actually high-resolution, just a wrapper around callout(9))

Revert "Give up on finding how linux side-loads these"

Found it.

Stub struct hrtimer.

branches: 1.1.2; 1.1.6;
More Linux goo.

branches: 1.1.2; 1.1.6;
First whack at radeon driver.

No hardware to test yet, but it builds.

branches: 1.1.2; 1.1.6;
First whack at radeon driver.

No hardware to test yet, but it builds.

branches: 1.4.4;
Fill in i2c_bit_add_bus.

First whack at radeon driver.

No hardware to test yet, but it builds.

branches: 1.2.2; 1.2.4; 1.2.6;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file i2c-algo-bit.h was initially added on branch riastradh-drm2.

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>

I2C_CLASS_SPD, IORESOURCE_IRQ

Add some sideloads.

match linux side-loading


Author: Maya Rashish <maya@NetBSD.org>

Provide a definition of I2C_M_STOP XXX


Author: Maya Rashish <maya@NetBSD.org>

Support i2c bus lock operations.

Reorganize Linux i2c header file and add a few more shims.

Remove outdated comment. The ->dev.parent stuff is part of the API.

branches: 1.6.2;
Another round of nouveau whack-a-mole.

branches: 1.5.2; 1.5.4;
Whack some nouveau moles.

Add Linux i2c stubs.

First whack at radeon driver.

No hardware to test yet, but it builds.

branches: 1.2.2; 1.2.4; 1.2.6;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file i2c.h was initially added on branch riastradh-drm2.

linux/idr.h: Need <sys/mutex.h> for kmutex_t.

branches: 1.10.2;
linux/idr.h: Update ida_* API.

Teach idr_remove to accept failure.

But teach it to return the data if it succeeds too.

Define idr_init_base.

Nothing actually uses this at the moment, because callers already
specify the base. Not really sure what the point is!

Draft rewrite of idr preload.

Previous idr code assumed every caller would definitely call
idr_preload, idr_alloc, idr_preload_end, but some callers skip
idr_alloc if an intermediate failure happens first, and would
therefore leak idr nodes.

Use a per-lwp single-node cache instead, and print warnings about
leakers.

Add ida_simple_remove.

Same as ida_remove for us. Linux requires caller to handle locking
for ida_remove (so it works with RCU) and ida_simple_remove just uses
a global lock where for users that aren't performance-critical. If
we ever pserialized idr, we'd have to do the same.

Add idr_get_next and idr_for_each_entry.

Sort.

branches: 1.3.4; 1.3.28; 1.3.30;
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.

branches: 1.2.2; 1.2.4; 1.2.6;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file idr.h was initially added on branch riastradh-drm2.

branches: 1.2.4; 1.2.6; 1.2.10;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file ihex.h was initially added on branch riastradh-drm2.

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>

branches: 1.3.2; 1.3.4; 1.3.8;
Implement Linux dmi_check_system API.

Use it to get a Thinkpad quirk for i915 graphics.

Patch from nonoka@, PR 48708.

branches: 1.2.2;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file init.h was initially added on branch riastradh-drm2.

branches: 1.2.4; 1.2.6; 1.2.10;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file input.h was initially added on branch riastradh-drm2.

branches: 1.2.4; 1.2.6; 1.2.10;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file intel-iommu.h was initially added on branch riastradh-drm2.

drm: Sprinkle miscellaneous includes.

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

Match linux side-loading


Author: Maya Rashish <maya@NetBSD.org>

Stub disable_irq, enable_irq.

<linux/interrupt.h> provides the tasklet API.

<linux/interrupt.h> side loads <linux/kernel.h>.

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

From riastradh

branches: 1.3.6;
Define in_interrupt for debugging assertions.

branches: 1.2.4; 1.2.6; 1.2.10; 1.2.34; 1.2.36;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file interrupt.h was initially added on branch riastradh-drm2.

drm: KASSERT(A && B) -> KASSERT(A); KASSERT(B)

branches: 1.13.4;
linux: Actually do post-order tree traversal.

Requires breaking the rbtree(3) abstraction, but this is necessary
because the body of the loop often frees the element, so as is we had
a huge pile of use-after-free going on.

Requires changing struct interval_tree_node's rbnode member to match
the Linux name, since we now use container_of here, and radeon relies
on this.

Touch up the rbtree/intervaltree stubs.

Doesn't work at all yet, but maybe it will help.

Rename struct rb_root_cached field to match Linux as needed.

Move rbtree_postorder_for_each_entry_safe to linux/rbtree.h.

Move struct rb_root from linux interval_tree.h to rbtree.h.

branches: 1.8.4;
interval_tree_iter_next: check the node we return, not the prev one.
Also assert that the interval is intersecting the requested boundary.

branches: 1.7.2;
Return the node we found. From tnn@.

Fill out interval tree a little bit including wacky linux rb stubs.

Fix ordering of interval trees, including using the right member names.

Dereference pointer to use what it points at!

Spell the field names right.

Give the interval trees a chance to compile.

Draft interval tree.

drm: Fix an even number of sign errors in interval_tree_generic.h.

amdgpu: make most of amdgpu_vm.c build

Not quite done yet -- need to adapt the DMA address logic to use
bus_dmamap_t.

Touch up the rbtree/intervaltree stubs.

Doesn't work at all yet, but maybe it will help.

misc drm hacks

Stub linux interval_tree_generic.h, mem_encrypt.h, sync_file.h.

Create stub headers for files included by drm code.


Author: Maya Rashish <maya@NetBSD.org>

drm: Move io_mapping to very own .c file.

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.)

drm: Fix io_mapping_map_wc to map all VAs, not just the first one.

drm: Allow >page-sized io_mapping_map_wc.

i915: Draft gem fault.

i915: ggtt

io_mapping_init/fini

io_mapping_map_wc takes a size argument now. Must be PAGE_SIZE.

<linux/io-mapping.h> needs <sys/mman.h> for PROT_READ/WRITE.

From chs@.

branches: 1.4.2;
A void function should not return a value from another void function.

Make Linux io_mapping actually work.

Can't use bus_space_map/bus_space_unmap in interrupt context, where
Linux uses this io_mapping abstraction. Instead, do uvm_km_alloc up
front and then use bus_space_mmap/pmap_kenter_pa to map it.

XXX Should bus_space_reserve the region (but that's x86-only for now).
XXX Should use direct map if possible.
XXX Should set an MTRR WC too in case of direct map or no PAT.

branches: 1.2.4; 1.2.6; 1.2.8; 1.2.10;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file io-mapping.h was initially added on branch riastradh-drm2.

drm: Make some legacy UMS crap build.

arch_io_reserve_memtype_wc, arch_io_free_memtype_wc stubs.

Rename namespacing for phys_arch_wc_index.

phys_wc_to_mtrr_index got renamed to arch_phys_wc_index upstream.

branches: 1.4.4; 1.4.28; 1.4.30;
First whack at radeon driver.

No hardware to test yet, but it builds.

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: 1.2.2; 1.2.4; 1.2.6;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file io.h was initially added on branch riastradh-drm2.

branches: 1.1.2; 1.1.6;
More Linux header file stubs.

drm: Make resource_size return resource_size_t like Linux.

Because Linux uses resource_size_t for absolute bus addresses, it is
actually bus_addr_t rather than bus_size_t. So using bus_size_t here
causes trouble with the i915 range_overflows macro, which verifies
that types match.

linux: Define IORESOURCE_* to be bits, not indices.

I2C_CLASS_SPD, IORESOURCE_IRQ

DEFINE_RES_MEM, resource_contains

Linux uses inclusive start/end, not start/size. Match.

resource_size

branches: 1.2.4; 1.2.6; 1.2.10;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file ioport.h was initially added on branch riastradh-drm2.

branches: 1.1.2; 1.1.6;
Empty stub <linux/irq.h>.

Not sure why upstream uses it -- the file in Linux exhorts against
its usage in any MI code.

linux: Draft irq_work shims, in terms of a softint.

Doesn't actually run in hardirq context, but who's gonna notice...

linux: Odds and ends.

i915: machete

Match linux which is side-loading

linux/irq_work.h->
asm/irq_work.h->
asm/cpufeature.h

Avoids errors on static_cpu_has


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

Fill in irq_work.h with stubs.

Create stub headers for files included by drm code.


Author: Maya Rashish <maya@NetBSD.org>

<linux/irqdomain.h>

irqs_disabled -- for assertions only.

branches: 1.1.2; 1.1.6;
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.

Stub <linux/irqreturn.h> to just include drm_irq_netbsd.h.

drm: Use getticks(), not hardclock_ticks.

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

From riastradh

branches: 1.12.6;
Define nsecs_to_jiffies.

Define jiffies_to_usecs, badly.

Define get_jiffies_64. Hope a 32-bit approximation is enough!

Call the parameter by its name, not its nickname!

Rework nsecs_to_jiffies64.

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>

branches: 1.6.4; 1.6.28; 1.6.30;
Use tvtohz instead of ad-hacking it myself, suggested by apb@.

Make Linux usecs_to_jiffies work with hz up to 2000.

First whack at radeon driver.

No hardware to test yet, but it builds.

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: 1.2.2; 1.2.4; 1.2.6;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file jiffies.h was initially added on branch riastradh-drm2.

kernel.h and list.h have already been moved to external/bsd/common.

With apologies to future revision control systems for not making it
clear this is effectively a rename.

Make uninitialized_var kludge expand to `x = 0'.

Forgot to commit this the other day.

Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file kernel.h was initially added on branch riastradh-drm2.

amdgpu: more destroy fixes.


Author: phone <mrg@NetBSD.org>
Committer: Taylor R Campbell <riastradh@NetBSD.org>

linux: INIT_KFIFO

statically allocated kfifo

kfifo is used under a spin lock, so its lock must be a spin lock.

From mrg@.

Rewrite <linux/kfifo.h>.

pcq isn't flexible enough for its needs now. Just use a spinlocked
circular buffer.

branches: 1.1.2; 1.1.6; 1.1.30; 1.1.32;
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.

ddb/db_active.h: New home for extern db_active.

This can be included unconditionally, and db_active can then be
queried unconditionally; if DDB is not in the kernel, then db_active
is a constant zero. Reduces need for #include opt_ddb.h, #ifdef DDB.

Need sys/types.h for bool.

branches: 1.2.4; 1.2.6; 1.2.10;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file kgdb.h was initially added on branch riastradh-drm2.

linux: Stub linux/kmemleak.h.

Create stub headers for files included by drm code.


Author: Maya Rashish <maya@NetBSD.org>

branches: 1.2.4; 1.2.6; 1.2.10;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file kobject.h was initially added on branch riastradh-drm2.

drm: Eliminate __HAVE_ATOMIC_AS_MEMBAR conditionals.

Discussed on tech-kern:
https://mail-index.netbsd.org/tech-kern/2023/02/23/msg028729.html

branches: 1.13.4;
linux/kref: Fix memory barriers and use membar_release/acquire.

i915: more work

Provide kref_put_lock


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

Linux refcount API

Omit __insn_barrier where implied or subsumed by volatile.

Implement kref_read.

Insert membars needed for non-x86.

XXX We should really define membar_enter_postatomic and
membar_exit_preatomic to avoid relying on not typoing
__HAVE_ATOMIC_AS_MEMBAR.

Revert "Rename private member of kref, because Linux abuses the abstraction."

This reverts commit d45132a71ee3c2570db2f10eac27283d3ac5b392.

Linux's abuse will be fixed.

Rename private member of kref, because Linux abuses the abstraction.

branches: 1.4.4; 1.4.28; 1.4.30;
First whack at radeon driver.

No hardware to test yet, but it builds.

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: 1.2.2; 1.2.4; 1.2.6;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file kref.h was initially added on branch riastradh-drm2.

drm: Rework Linux `kthread' abstraction to avoid race to sleep.

Requires passing in the caller's lock and condvar to kthread_run, but
for the one user that appears not to be an onerous requirement.

drm: linux kthread stubs

branches: 1.2.4; 1.2.6; 1.2.10;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file kthread.h was initially added on branch riastradh-drm2.

drm: Make via build again.

linux: ktime_get_real_seconds

amdgpu: Another pass through.

amdgpu_object.c
amdgpu_pci.c
amdgpu_pm.c
amdgpu_polaris10_smumgr.c
amdgpu_pp_psm.c
amdgpu_ppatomctrl.c
amdgpu_ppatomfwctrl.c
amdgpu_process_pptables_v1_0.o
amdgpu_psp.o
amdgpu_psp_v10_0.c
amdgpu_psp_v11_0.c
amdgpu_psp_v12_0.c
amdgpu_psp_v3_1.c
amdgpu_ras.c
amdgpu_ras_eeprom.c
amdgpu_rc_calc.c
amdgpu_rc_calc_dpi.c
amdgpu_ring.c
amdgpu_rlc.c
amdgpu_rn_clk_mgr.c
amdgpu_sdma_v3_0.c
amdgpu_sdma_v4_0.c
amdgpu_sdma_v5_0.c
amdgpu_si.c
amdgpu_si_dma.c
amdgpu_si_dpm.c
amdgpu_smu.c
amdgpu_smu10_hwmgr.c
amdgpu_smu7_hwmgr.c
amdgpu_smu8_hwmgr.c
amdgpu_smu8_smumgr.c
amdgpu_smu_v11_0.c (and a sprinkling of consts)
amdgpu_smu_v11_0_i2c.c
amdgpu_soc15.c
amdgpu_tonga_smumgr.c
amdgpu_ttm.c
amdgpu_ucode.c
amdgpu_uvd.c
amdgpu_uvd_v5_0.c
amdgpu_uvd_v6_0.c
amdgpu_uvd_v7_0.c
amdgpu_vce.c
amdgpu_vce_v2_0.c
amdgpu_vce_v4_0.c
amdgpu_vcn_v1_0.c
amdgpu_vcn_v2_0.c
amdgpu_vcn_v2_5.c
amdgpu_vega10_hwmgr.c
amdgpu_vega10_powertune.c
amdgpu_vega10_processpptables.c
amdgpu_vega10_reg_init.c
amdgpu_vega12_hwmgr.c
amdgpu_vega20_hwmgr.c
amdgpu_vega20_ppt.c
amdgpu_vega20_reg_init.c
amdgpu_vi.c
amdgpu_virt.c
amdgpu_vram_mgr.c
amdgpu_xgmi.c -- disable (for now?)
amdgpufb.c

linux: ktime_get_ns

i915: another whack at it

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

provide ktime_ms_delta


Author: Maya Rashish <maya@NetBSD.org>

Stupid ktime_get_raw until I figure out what this should be.

ktime_add

struct timespec serves as a perfectly adequate struct timespec64.

ktime_after

ktime_t is now just int64_t, no longer a union.

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

From riastradh

branches: 1.8.2;
- Eliminate the global "boottime" variable, which was being accessed
without any synchronization against changes by e.g. clock_settime().
- Replace with new getbinboottime() / getnanoboottime() / getmicroboottime()
functions (naming mirrors that of other time access functions in kern_tc.c).
It returns the (maybe-converted) value of timebasebin, which also tracks
our estimate of when the system was booted (i.e. the legacy "boottime" was
redundant).

XXX There needs to be a lockless synchronization mechanism for reading
timebasebin, but this is a problem in kern_tc.c that pre-existed these
"boottime" changes. At least now the problem is centralized in one location.

Need <sys/kernel.h> for boottime.

Rework Linux ktime.

- Remove union crap which has been removed upstream.
- Define everything in terms of ns_to_ktime and ktime_to_ns.
- Add ktime_to_us.
- Add ktime_us_delta.

Define ktime_get_raw_ns.

New ktime utilities. Fix monotonic offset business.

branches: 1.3.4; 1.3.28; 1.3.30;
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.

branches: 1.2.2; 1.2.4; 1.2.6;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file ktime.h was initially added on branch riastradh-drm2.

kernel.h and list.h have already been moved to external/bsd/common.

With apologies to future revision control systems for not making it
clear this is effectively a rename.

Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file list.h was initially added on branch riastradh-drm2.

branches: 1.2.4; 1.2.6; 1.2.10;
Merge riastradh-drm2 to HEAD.

branches: 1.1.2;
file list_sort.h was initially added on branch riastradh-drm2.

linux/llist: Use membar_release and membar_datadep_consumer.

No need for membar_acquire here! Loads are all data-dependent.

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

i915: more work

linux/llist.h: Various fixups.

Linux wants llist_node::next.

Fix , -> ; typo in `for' header.

Reimplement GPL <linux/llist.h> using only the upstream comments.

Hope I gathered what this is trying to do...!

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>