drm/i915/ttm: Embed a ttm buffer object in the i915 gem object

Embed a struct ttm_buffer_object into the i915 gem object, making sure
we alias the gem object part. It's a bit unfortunate that the
struct ttm_buffer_ojbect embeds a gem object since we otherwise could
make the TTM part private to the TTM backend, and use the usual
i915 gem object for the other backends.
To make this a bit more storage efficient for the other backends,
we'd have to use a pointer for the gem object which would require
a lot of changes in the driver. We postpone that for later.

Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Acked-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210602083818.241793-3-thomas.hellstrom@linux.intel.com

drm/i915/ttm Initialize the ttm device and memory managers

Temporarily remove the buddy allocator and related selftests
and hook up the TTM range manager for i915 regions.

Also modify the mock region selftests somewhat to account for a
fragmenting manager.

Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210602083818.241793-2-thomas.hellstrom@linux.intel.com

Revert "i915: use io_mapping_map_user"

This reverts commit b739f125e4ebd73d10ed30a856574e13649119ed.

We are unfortunately seeing more issues like we did in 293837b9ac8d
("Revert "i915: fix remap_io_sg to verify the pgprot""), except this is
now for the vm_fault_gtt path, where we are now hitting the same
BUG_ON(!pte_none(*pte)):

[10887.466150] kernel BUG at mm/memory.c:2183!
[10887.466162] invalid opcode: 0000 [#1] PREEMPT SMP PTI
[10887.466168] CPU: 0 PID: 7775 Comm: ffmpeg Tainted: G     U            5.13.0-rc3-CI-Nightly #1
[10887.466174] Hardware name: To Be Filled By O.E.M. To Be Filled By O.E.M./J4205-ITX, BIOS P1.40 07/14/2017
[10887.466177] RIP: 0010:remap_pfn_range_notrack+0x30f/0x440
[10887.466188] Code: e8 96 d7 e0 ff 84 c0 0f 84 27 01 00 00 48 ba 00 f0 ff ff ff ff 0f 00 4c 89 e0 48 c1 e0 0c 4d 85 ed 75 96 48 21 d0 31 f6 eb a9 <0f> 0b 48 39 37 0f 85 0e 01 00 00 48 8b 0c 24 48 39 4f 08 0f 85 00
[10887.466193] RSP: 0018:ffffc90006e33c50 EFLAGS: 00010286
[10887.466198] RAX: 800000000000002f RBX: 00007f5e01800000 RCX: 0000000000000028
[10887.466201] RDX: 0000000000000001 RSI: ffffea0000000000 RDI: 0000000000000000
[10887.466204] RBP: ffffea000033fea8 R08: 800000000000002f R09: ffff8881072256e0
[10887.466207] R10: ffffc9000b84fff8 R11: 0000000017dab000 R12: 0000000000089f9f
[10887.466210] R13: 800000000000002f R14: 00007f5e017e4000 R15: ffff88800cffaf20
[10887.466213] FS:  00007f5e04849640(0000) GS:ffff888278000000(0000) knlGS:0000000000000000
[10887.466216] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[10887.466220] CR2: 00007fd9b191a2ac CR3: 00000001829ac000 CR4: 00000000003506f0
[10887.466223] Call Trace:
[10887.466233]  vm_fault_gtt+0x1ca/0x5d0 [i915]
[10887.466381]  ? ktime_get+0x38/0x90
[10887.466389]  __do_fault+0x37/0x90
[10887.466395]  __handle_mm_fault+0xc46/0x1200
[10887.466402]  handle_mm_fault+0xce/0x2a0
[10887.466407]  do_user_addr_fault+0x1c5/0x660

Reverting this commit is reported to fix the issue.

Reported-by: Eero Tamminen <eero.t.tamminen@intel.com>
References: https://gitlab.freedesktop.org/drm/intel/-/issues/3519
Fixes: b739f125e4eb ("i915: use io_mapping_map_user")
Cc: Christoph Hellwig <hch@lst.de>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20210527185145.458021-1-matthew.auld@intel.com

drm/i915/selftests: Fix return value check in live_breadcrumbs_smoketest()

In case of error, the function live_context() returns ERR_PTR() and never
returns NULL. The NULL test in the return value check should be replaced
with IS_ERR().

Fixes: 52c0fdb25c7c ("drm/i915: Replace global breadcrumbs with per-context interrupt tracking")
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/33c46ef24cd547d0ad21dc106441491a@intel.com
[tursulin: Wrap commit text, fix Fixes: tag.]
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>

drm/i915: use DEVICE_ATTR_RO macro

Use DEVICE_ATTR_RO() helper instead of plain DEVICE_ATTR(),
which makes the code a bit shorter and easier to read.

Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210528100403.21548-1-yuehaibing@huawei.com

Merge drm/drm-next into drm-intel-gt-next

Pulling in -rc2 fixes and TTM changes that next upcoming patches depend
on.

Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>

Merge tag 'drm-intel-gt-next-2021-05-28' of git://anongit.freedesktop.org/drm/drm-intel into drm-next

UAPI Changes:
- Add reworked uAPI for DG1 behind CONFIG_BROKEN (Matt A, Abdiel)

Driver Changes:

- Fix for Gitlab issues #3293 and #3450:
  Avoid kernel crash on older L-shape memory machines

- Add Wa_14010733141 (VDBox SFC reset) for Gen11+ (Aditya)
- Fix crash in auto_retire active retire callback due to
  misalignment (Stephane)
- Fix overlay active retire callback alignment (Tvrtko)
- Eliminate need to align active retire callbacks (Matt A, Ville,
  Daniel)
- Program FF_MODE2 tuning value for all Gen12 platforms (Caz)
- Add Wa_14011060649 for TGL,RKL,DG1 and ADLS (Swathi)
- Create stolen memory region from local memory on DG1 (CQ)
- Place PD in LMEM on dGFX (Matt A)
- Use WC when default state object is allocated in LMEM (Venkata)
- Determine the coherent map type based on object location (Venkata)
- Use lmem physical addresses for fb_mmap() on discrete (Mohammed)
- Bypass aperture on fbdev when LMEM is available (Anusha)
- Return error value when displayable BO not in LMEM for dGFX (Mohammed)
- Do release kernel context if breadcrumb measure fails (Janusz)
- Hide modparams for compiled-out features (Tvrtko)
- Apply Wa_22010271021 for all Gen11 platforms (Caz)
- Fix unlikely ref count race in arming the watchdog timer (Tvrtko)
- Check actual RC6 enable status in PMU (Tvrtko)
- Fix a double free in gen8_preallocate_top_level_pdp (Lv)
- Use trylock in shrinker for GGTT on BSW VT-d and BXT (Maarten)
- Remove erroneous i915_is_ggtt check for
  I915_GEM_OBJECT_UNBIND_VM_TRYLOCK (Maarten)

- Convert uAPI headers to real kerneldoc (Matt A)
- Clean up kerneldoc warnings headers (Matt A, Maarten)
- Fail driver if LMEM training failed (Matt R)
- Avoid div-by-zero on Gen2 (Ville)
- Read C0DRB3/C1DRB3 as 16 bits again and add _BW suffix (Ville)
- Remove reference to struct drm_device.pdev (Thomas)
- Increase separation between GuC and execlists code (Chris, Matt B)

- Use might_alloc() (Bernard)
- Split DGFX_FEATURES from GEN12_FEATURES (Lucas)
- Deduplicate Wa_22010271021 programming on (Jose)
- Drop duplicate WaDisable4x2SubspanOptimization:hsw (Tvrtko)
- Selftest improvements (Chris, Hsin-Yi, Tvrtko)
- Shuffle around init_memory_region for stolen (Matt)
- Typo fixes (wengjianfeng)

[airlied: fix conflict with fixes in i915_active.c]
Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/YLCbBR22BsQ/dpJB@jlahtine-mobl.ger.corp.intel.com

Merge tag 'drm-misc-next-2021-06-01' of git://anongit.freedesktop.org/drm/drm-misc into drm-next

drm-misc-next for 5.14:

UAPI Changes:

 * Use DRM driver names for fbdev

Cross-subsystem Changes:

Core Changes:

 * Fix leaked DMA handles

 * Improve documentation around DRM_CLIENT_CAP_*

 * Cleanups

 * dp_mst: Use kHz as link-rate unit during init

 * fourcc: Remove drm_gem_format_name() and drm_format_name_buf

 * gem-cma: Fix mmap for buffers with write combining

 * ttm: Don't override pre-set vm_ops; ttm_bo_mmap() removal and cleanups

Driver Changes:

 * drm/amdgpu: Fix hot unplug during suspend; Implement mmap as GEM object
   function; Use %p4cc format-string modifier; Cleanups

 * drm/bridge: Cdns: Fix PM reference leak, Cleanups; Lt8912b: Fix Coccinelle
   warnings; Fix Kconfig dependencies; Fixes and cleanups

 * drm/hisilicon/kirin: Cleanups

 * drm/nouveau: Implement mmap as GEM object function

 * drm/radeon: Implement mmap as GEM object function

 * drm/rockchip: Remove generic drivers during init; Add scaling for RK3036
   win1; Fix missing registers for RK3066 and 3188; Add alpha support for
   RK3036, RK3066, RK3126 and RK3188; Fixes and cleanups

 * drm/simpledrm: Use %p4cc: format-string modifier

 * drm/vmwgfx: Cleanups

 * fbdev/matrox: Use modern module_init()

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/YLZOKiYE6XFmE/MH@linux-uq9g.fritz.box

drm/i915: Disable mmap ioctl for gen12+

The platform should exclusively use mmap_offset, one less path to worry
about for discrete.

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210601074654.3103-14-thomas.hellstrom@linux.intel.com

drm/i915: Fix i915_sg_page_sizes to record dma segments rather than physical pages

All users of this function actually want the dma segment sizes, but that's
not what's calculated. Fix that and rename the function to
i915_sg_dma_sizes to reflect what's calculated.

Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210601074654.3103-4-thomas.hellstrom@linux.intel.com

drm/i915: Don't free shared locks while shared

We are currently sharing the VM reservation locks across a number of
gem objects with page-table memory. Since TTM will individiualize the
reservation locks when freeing objects, including accessing the shared
locks, make sure that the shared locks are not freed until that is done.
For PPGTT we add an additional refcount, for GGTT we take additional
measures to make sure objects sharing the GGTT reservation lock are
freed at GGTT takedown

Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210601074654.3103-3-thomas.hellstrom@linux.intel.com

drm/i915: Untangle the vma pages_mutex

Any sleeping dma_resv lock taken while the vma pages_mutex is held
will cause a lockdep splat.
Move the i915_gem_object_pin_pages() call out of the pages_mutex
critical section.

Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210601074654.3103-2-thomas.hellstrom@linux.intel.com

drm: document minimum kernel version for DRM_CLIENT_CAP_*

The kernel versions including the following commits are referenced:

DRM_CLIENT_CAP_STEREO_3D
61d8e3282541 ("drm: Add a STEREO_3D capability to the SET_CLIENT_CAP ioctl")

DRM_CLIENT_CAP_UNIVERSAL_PLANES
681e7ec73044 ("drm: Allow userspace to ask for universal plane list (v2)")
c7dbc6c9ae5c ("drm: Remove command line guard for universal planes")

DRM_CLIENT_CAP_ATOMIC
88a48e297b3a ("drm: add atomic properties")
8b72ce158cf0 ("drm: Always enable atomic API")

DRM_CLIENT_CAP_ASPECT_RATIO
7595bda2fb43 ("drm: Add DRM client cap for aspect-ratio")

DRM_CLIENT_CAP_WRITEBACK_CONNECTORS
d67b6a206507 ("drm: writeback: Add client capability for exposing writeback connectors")

Signed-off-by: Simon Ser <contact@emersion.fr>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Acked-by: Daniel Stone <daniels@collabora.com>
Acked-by: Pekka Paalanen <pekka.paalanen@collabora.com>
Link: https://patchwork.freedesktop.org/patch/434202/

drm: clarify and linkify DRM_CLIENT_CAP_WRITEBACK_CONNECTORS docs

Make it clear that the client is responsible for enabling ATOMIC
prior to enabling WRITEBACK_CONNECTORS. Linkify the reference to
ATOMIC.

Signed-off-by: Simon Ser <contact@emersion.fr>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Acked-by: Daniel Stone <daniels@collabora.com>
Acked-by: Pekka Paalanen <pekka.paalanen@collabora.com>
Link: https://patchwork.freedesktop.org/patch/434200/

drm: reference mode flags in DRM_CLIENT_CAP_* docs

In the docs for DRM_CLIENT_CAP_STEREO_3D and
DRM_CLIENT_CAP_ASPECT_RATIO, reference the DRM_MODE_FLAG_* defines
that get set when the cap is enabled.

Signed-off-by: Simon Ser <contact@emersion.fr>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Acked-by: Daniel Stone <daniels@collabora.com>
Acked-by: Pekka Paalanen <pekka.paalanen@collabora.com>
Link: https://patchwork.freedesktop.org/patch/434201/

drm: bridge: cdns-mhdp8546: Fix PM reference leak in

pm_runtime_get_sync will increment pm usage counter even it failed.
Forgetting to putting operation will result in reference leak here.
Fix it by replacing it with pm_runtime_resume_and_get to keep usage
counter balanced.

Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Reviewed-by: Robert Foss <robert.foss@linaro.org>
Signed-off-by: Robert Foss <robert.foss@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20210531135622.3348252-1-yukuai3@huawei.com

drm/bridge: lt8912b: Drop unused includes

The Lontium bridge is including legacy header files for GPIO
but not using them. Delete the includes.

Cc: Adrien Grassein <adrien.grassein@gmail.com>
Cc: Robert Foss <robert.foss@linaro.org>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Reviewed-by: Adren Grassein <adrien.grassein@gmail.com>
Signed-off-by: Robert Foss <robert.foss@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20210529002759.468964-1-linus.walleij@linaro.org

drm/rockchip: vop: add PX30 version info

To reduce memory various Rockchip VOP versions share
common reg structures. However more recent added SoCs not
always have to same futures as the old ones.
Add PX30 missing version info, so all VOP version checks
work correct if needed in the future.

Signed-off-by: Johan Jonker <jbx6244@gmail.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20210131125016.10837-1-jbx6244@gmail.com

drm/rockchip: cdn-dp: fix sign extension on an int multiply for a u64 result

The variable bit_per_pix is a u8 and is promoted in the multiplication
to an int type and then sign extended to a u64. If the result of the
int multiplication is greater than 0x7fffffff then the upper 32 bits will
be set to 1 as a result of the sign extension. Avoid this by casting
tu_size_reg to u64 to avoid sign extension and also a potential overflow.

Fixes: 1a0f7ed3abe2 ("drm/rockchip: cdn-dp: add cdn DP support for rk3399")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Reviewed-by: Guenter Roeck <groeck@chromium.org>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20200915162049.36434-1-colin.king@canonical.com

drm/rockchip: lvds: Fix an error handling path

'ret' is know to be 0 a this point. Checking the return value of
'phy_init()' and 'phy_set_mode()' was intended instead.

So add the missing assignments.

Fixes: cca1705c3d89 ("drm/rockchip: lvds: Add PX30 support")
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/248220d4815dc8c8088cebfab7d6df5f70518438.1619881852.git.christophe.jaillet@wanadoo.fr

drm/rockchip: remove unused function

Fix the following clang warning:

drivers/gpu/drm/rockchip/dw-mipi-dsi-rockchip.c:320:20: warning: unused
function 'dsi_set' [-Wunused-function].

Reported-by: Abaci Robot <abaci@linux.alibaba.com>
Signed-off-by: Jiapeng Chong <jiapeng.chong@linux.alibaba.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/1618476421-114429-1-git-send-email-jiapeng.chong@linux.alibaba.com

drm/rockchip: dsi: remove extra component_del() call

commit cf6d100dd238 ("drm/rockchip: dsi: add dual mipi support") added
this devcnt field and call to component_del(). However, these both
appear to be erroneous changes left over from an earlier version of the
patch. In the version merged, nothing ever modifies devcnt, meaning
component_del() runs unconditionally and in addition to the
component_del() calls in dw_mipi_dsi_rockchip_host_detach(). The second
call fails to delete anything and produces a warning in dmesg.

If we look at the previous version of the patch[1], however, we see that
it had logic to calculate devcnt and call component_add() in certain
situations. This was removed in v6, and the fact that the deletion code
was not appears to have been an oversight.

[1] https://patchwork.kernel.org/project/dri-devel/patch/20180821140515.22246-8-heiko@sntech.de/

Fixes: cf6d100dd238 ("drm/rockchip: dsi: add dual mipi support")
Cc: stable@vger.kernel.org
Signed-off-by: Thomas Hebb <tommyhebb@gmail.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/201385acb0eeb5dfb037afdc6a94bfbcdab97f99.1618797778.git.tommyhebb@gmail.com

drm/rockchip: dsi: move all lane config except LCDC mux to bind()

When we first enable the DSI encoder, we currently program some per-chip
configuration that we look up in rk3399_chip_data based on the device
tree compatible we match. This data configures various parameters of the
MIPI lanes, including on RK3399 whether DSI1 is slaved to DSI0 in a
dual-mode configuration. It also selects which LCDC (i.e. VOP) to scan
out from.

This causes a problem in RK3399 dual-mode configurations, though: panel
prepare() callbacks run before the encoder gets enabled and expect to be
able to write commands to the DSI bus, but the bus isn't fully
functional until the lane and master/slave configuration have been
programmed. As a result, dual-mode panels (and possibly others too) fail
to turn on when the rockchipdrm driver is initially loaded.

Because the LCDC mux is the only thing we don't know until enable time
(and is the only thing that can ever change), we can actually move most
of the initialization to bind() and get it out of the way early. That's
what this change does. (Rockchip's 4.4 BSP kernel does it in mode_set(),
which also avoids the issue, but bind() seems like the more correct
place to me.)

Tested on a Google Scarlet board (Acer Chromebook Tab 10), which has a
Kingdisplay KD097D04 dual-mode panel. Prior to this change, the panel's
backlight would turn on but no image would appear when initially loading
rockchipdrm. If I kept rockchipdrm loaded and reloaded the panel driver,
it would come on. With this change, the panel successfully turns on
during initial rockchipdrm load as expected.

Fixes: 2d4f7bdafd70 ("drm/rockchip: dsi: migrate to use dw-mipi-dsi bridge driver")
Signed-off-by: Thomas Hebb <tommyhebb@gmail.com>
Tested-by: Jonathan Liu <net147@gmail.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/55fe7f3454d8c91dc3837ba5aa741d4a0e67378f.1618797813.git.tommyhebb@gmail.com

drm/rockchip: cdn-dp-core: add MODULE_FIRMWARE macro

The CDP DP component of the rockchip GPU driver requires firmware
so define MODULE_FIRMWARE for rockchip/dptx.bin so the details
are available in the module info.

Signed-off-by: Peter Robinson <pbrobinson@gmail.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20210506092951.515813-1-pbrobinson@gmail.com

drm/rockchip: cdn-dp-core: add missing clk_disable_unprepare() on error in cdn_dp_grf_write()

After calling clk_prepare_enable(), clk_disable_unprepare() need
be called when calling regmap_write() failed.

Fixes: 1a0f7ed3abe2 ("drm/rockchip: cdn-dp: add cdn DP support for rk3399")
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Yang Yingliang <yangyingliang@huawei.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20210519134928.2696617-1-yangyingliang@huawei.com

drm: rockchip: set alpha_en to 0 if it is not used

alpha_en should be set to 0 if it is not used, i.e. to disable alpha
blending if it was enabled before and should be disabled now.

Fixes: 2aae8ed1f390 ("drm/rockchip: Add per-pixel alpha support for the PX30 VOP")
Signed-off-by: Alex Bee <knaerzche@gmail.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20210528130554.72191-6-knaerzche@gmail.com

drm: rockchip: add alpha support for RK3036, RK3066, RK3126 and RK3188

With
commit 2aae8ed1f390 ("drm/rockchip: Add per-pixel alpha support for the PX30 VOP")
alpha support was introduced for PX30's VOP.
RK3036, RK3066, RK3126 and RK3188 VOPs support alpha blending in the
same manner.
With the exception of RK3066 all of them support pre-multiplied alpha.

Signed-off-by: Alex Bee <knaerzche@gmail.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20210528130554.72191-5-knaerzche@gmail.com

drm: rockchip: add missing registers for RK3066

Add dither_up, dsp_lut_en and data_blank registers to enable their
respective functionality for RK3066's VOP.

While at that also fix .rb_swap and .format registers for all windows,
which have to be set though RK3066_SYS_CTRL1 register.
Also remove .scl from win1: Scaling is only supported on the primary
plane.

Signed-off-by: Alex Bee <knaerzche@gmail.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20210528130554.72191-4-knaerzche@gmail.com

drm: rockchip: add missing registers for RK3188

Add dither_up, dsp_lut_en and data_blank registers to enable their
respective functionality for RK3188's VOP.
While at that also fix .dsp_blank register which is (only) set with
BIT24 (same as RK3066)

Signed-off-by: Alex Bee <knaerzche@gmail.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20210528130554.72191-3-knaerzche@gmail.com

drm: rockchip: add scaling for RK3036 win1

Add the registers needed to make scaling work on RK3036's win1.

Signed-off-by: Alex Bee <knaerzche@gmail.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20210528130554.72191-2-knaerzche@gmail.com

drm/rockchip: remove existing generic drivers to take over the device

There are drivers that register framebuffer devices very early in the boot
process and make use of the existing framebuffer as setup by the firmware.

If one of those drivers has registered a fbdev, then the fallback fbdev of
the DRM driver won't be bound to the framebuffer console. To avoid that,
remove any existing generic driver and take over the graphics device.

By doing that, the fb mapped to the console is switched correctly from the
early fbdev to the one registered by the rockchip DRM driver:

    [   40.752420] fb0: switching to rockchip-drm-fb from EFI VGA

Signed-off-by: Javier Martinez Canillas <javierm@redhat.com>
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Tested-by: Peter Robinson <pbrobinson@gmail.com>
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Heiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20210516074833.451643-1-javierm@redhat.com

drm: Fix for GEM buffers with write-combine memory

The previous commit wrongly assumed that dma_mmap_wc() could be replaced
by pgprot_writecombine() + dma_mmap_pages(). It did work on my setup,
but did not work everywhere.

Use dma_mmap_wc() when the buffer has the write-combine cache attribute,
and dma_mmap_pages() when it has the non-coherent cache attribute.

Signed-off-by: Paul Cercueil <paul@crapouillou.net>
Reported-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Reviewed-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Tested-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Fixes: cf8ccbc72d61 ("drm: Add support for GEM buffers backed by non-coherent memory")
Link: https://patchwork.freedesktop.org/patch/msgid/20210527230334.151947-1-paul@crapouillou.net

drm/dp_mst: Use kHz as link rate units when settig source max link caps at init

[why]
Link rate in kHz is what is eventually required to calculate the link
bandwidth, which makes kHz a more generic unit. This should also make
forward-compatibility with new DP standards easier.

[how]
- Replace 'link rate DPCD code' with 'link rate in kHz' when used with
drm_dp_mst_topology_mgr_init()
- Add/remove related DPCD code conversion from/to kHz where applicable

Signed-off-by: Nikola Cornij <nikola.cornij@amd.com>
Acked-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Lyude Paul <lyude@redhat.com>
Signed-off-by: Lyude Paul <lyude@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210512210011.8425-2-nikola.cornij@amd.com

drm/i915: Add Wa_14010733141

The WA requires the following procedure for VDBox SFC reset:

If (MFX-SFC usage is 1) {
1.Issue a MFX-SFC forced lock
2.Wait for MFX-SFC forced lock ack
3.Check the MFX-SFC usage bit
If (MFX-SFC usage bit is 1)
Reset VDBOX and SFC
else
Reset VDBOX
Release the force lock MFX-SFC
}
else if(HCP+SFC usage is 1) {
1.Issue a VE-SFC forced lock
2.Wait for SFC forced lock ack
3.Check the VE-SFC usage bit
If (VE-SFC usage bit is 1)
Reset VDBOX
else
Reset VDBOX and SFC
Release the force lock VE-SFC.
}
else
Reset VDBOX

- Restructure: the changes to the original code flow should stay
  relatively minimal; we only need to do an extra HCP check after the
  usual VD-MFX check and, if true, switch the register/bit we're
  performing the lock on.(MattR)

v2:
- Assign unlock mask using paired_engine->mask instead of using
  BIT(paired_vecs->id). (Daniele)

Bspec: 52890, 53509

Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Matt Roper <matthew.d.roper@intel.com>
Cc: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Aditya Swarup <aditya.swarup@intel.com>
Co-developed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210526094852.286424-2-aditya.swarup@intel.com

Merge drm/drm-next into drm-misc-next

i915 is broken without -rc3, let's bring that tag in to fix it.

Signed-off-by: Maxime Ripard <maxime@cerno.tech>

drm/bridge: DRM_CROS_EC_ANX7688 should depend on I2C_CROS_EC_TUNNEL

The ChromeOS EC ANX7688 bridge is connected to a ChromeOS Embedded
Controller, and is accessed using I2C tunneling through the Embedded
Controller.  Hence add a dependency on I2C_CROS_EC_TUNNEL, to prevent
asking the user about this driver when configuring a kernel without
support for the ChromeOS EC tunnel I2C bus.

Fixes: 44602b10d7f2a5f7 ("drm/bridge: Add ChromeOS EC ANX7688 bridge driver support")
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Reviewed-by: Enric Balletbo i Serra <enric.balletbo@collabora.com>
Signed-off-by: Robert Foss <robert.foss@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/d107d1840b83607baee8571cc5d88973fc32b519.1622015323.git.geert+renesas@glider.be

drm/bridge: lt8912b: fix platform_no_drv_owner.cocci warnings

./drivers/gpu/drm/bridge/lontium-lt8912b.c:758:3-8: No need to set .owner here. The core will do it.

 Remove .owner field if calls are used which set it automatically

Generated by: scripts/coccinelle/api/platform_no_drv_owner.cocci

Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Zou Wei <zou_wei@huawei.com>
Reviewed-by: Robert Foss <robert.foss@linaro.org>
Signed-off-by: Robert Foss <robert.foss@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/1622109018-54648-1-git-send-email-zou_wei@huawei.com

drm: bridge: cdns-mhdp8546: Fix inconsistent indenting

Eliminate the follow smatch warning:

drivers/gpu/drm/bridge/cadence/cdns-mhdp8546-core.c:2143
cdns_mhdp_bridge_atomic_reset() warn: inconsistent indenting.

Reported-by: Abaci Robot <abaci@linux.alibaba.com>
Signed-off-by: Jiapeng Chong <jiapeng.chong@linux.alibaba.com>
Reviewed-by: Robert Foss <robert.foss@linaro.org>
Signed-off-by: Robert Foss <robert.foss@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/1621852953-51325-1-git-send-email-jiapeng.chong@linux.alibaba.com

drm/bridge: cdns: Fix PM reference leak in cdns_dsi_transfer()

pm_runtime_get_sync will increment pm usage counter even it failed.
Forgetting to putting operation will result in reference leak here.
Fix it by replacing it with pm_runtime_resume_and_get to keep usage
counter balanced.

Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Zou Wei <zou_wei@huawei.com>
Reviewed-by: Robert Foss <robert.foss@linaro.org>
Signed-off-by: Robert Foss <robert.foss@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/1621840862-106024-1-git-send-email-zou_wei@huawei.com

Merge v5.13-rc3 into drm-next

drm/i915 is extremely on fire without the below revert from -rc3:

commit 293837b9ac8d3021657f44c9d7a14948ec01c5d0
Author: Linus Torvalds <torvalds@linux-foundation.org>
Date:   Wed May 19 05:55:57 2021 -1000

    Revert "i915: fix remap_io_sg to verify the pgprot"

Backmerge so we don't have a too wide bisect window for anything
that's a more involved workload than booting the driver.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915/params: Align visibility of device level and global modparams

We have a few modparams which get conditionally exposed based on a Kconfig
options and in most cases this also means portions of the driver
implementing the respective feature are also left out.

Align the visibility of device level and global modparams to make them
consistent in this respect.

v2:
 * Fix misplaced parentheses.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Jani Nikula <jani.nikula@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210526141946.2347085-1-tvrtko.ursulin@linux.intel.com

drm/fourcc: Remove struct drm_format_buf_name

The structure is unused. Remove it.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210516121315.30321-5-tzimmermann@suse.de

drm: Remove drm_get_format_name()

The %p4cc printk format modifier was recently added to print fourcc codes,
replacing drm_get_format_name(). The function is no longer needed, so
remove it.

Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Petr Mladek <pmladek@suse.com>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210516121315.30321-4-tzimmermann@suse.de

drm/simpledrm: Use %p4cc to print 4CC format

Replace use of struct drm_format_name_buf with %p4cc for printing
4CC formats.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210516121315.30321-3-tzimmermann@suse.de

drm/amdgpu: Use %p4cc to print 4CC format

Replace use of struct drm_format_name_buf with %p4cc for printing
4CC formats.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Acked-by: Christian König <christian.koenig@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210516121315.30321-2-tzimmermann@suse.de

fbdev: matrox: use modern module_init()

This is one of the last drivers with a global init_module() function
instead of the modern module_init() annotation. Convert it over.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20210514213316.635070-1-arnd@kernel.org

drm/fb-helper: improve DRM fbdev emulation device names

Framebuffer devices that are registered by DRM drivers for fbdev emulation
have a "drmfb" suffix in their name. But makes them to be quite confusing
for drivers that already have "drm" in their name:

$ cat /proc/fb
0 rockchipdrmdrmfb

$ cat /proc/fb
0 simpledrmdrmfb

Also, there isn't a lot of value in adding these "drmfb" suffices to their
names, since users shouldn't really care if the FB devices were registered
by a real fbdev driver or a DRM driver using the fbdev emulation.

What programs should be interested about is if there's a DRM device, and
there are better ways to query that info than reading this procfs entry.

So let's just remove the suffix, which leads to much better device names:

$ cat /proc/fb
0 rockchipdrm

$ cat /proc/fb
0 simpledrm

Suggested-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Javier Martinez Canillas <javierm@redhat.com>
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20210525151313.3379622-1-javierm@redhat.com

drm/hisilicon/kirin: Use the correct HiSilicon copyright

s/Hisilicon/HiSilicon/.
It should use capital S, according to
https://www.hisilicon.com/en.

Signed-off-by: Hao Fang <fanghao11@huawei.com>
Acked-by: Tian Tao <tiantao6@hisilicon.com>
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/1621678529-14389-1-git-send-email-fanghao11@huawei.com

drm: fix leaked dma handles after removing drm_pci_free

After removing drm_pci_alloc/free, some instances where drm_pci_free()
would have kfreed the dma handle were skipped.

Ensure these handles are freed properly.

Signed-off-by: Joseph Kogut <joseph.kogut@gmail.com>
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20210518212859.4148903-1-joseph.kogut@gmail.com

drm/kconfig: Remove unused select of DRM_KMS_FB_HELPER

The option DRM_KMS_FB_HELPER has been removed. Also remove the last
remaining select statement for it.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Fixes: 91185d55b32e ("drm: Remove DRM_KMS_FB_HELPER Kconfig option")
Acked-by: Maxime Ripard <maxime@cerno.tech>
Reported-by: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Maxime Ripard <mripard@kernel.org>
Cc: David Airlie <airlied@linux.ie>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: dri-devel@lists.freedesktop.org
Link: https://patchwork.freedesktop.org/patch/msgid/20210526100825.29450-1-tzimmermann@suse.de

drm/ttm: Remove ttm_bo_mmap() and friends

The function ttm_bo_mmap is unused. Remove it and it's helpers; including
the verify_access callback in struct ttm_device_funcs.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Christian König <christian.koenig@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210525151055.8174-8-tzimmermann@suse.de

drm/vmwgfx: Inline vmw_verify_access()

Vmwgfx is the only user of the TTM's verify_access callback. Inline
the call and avoid the indirection through the function pointer.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Zack Rusin <zackr@vmware.com>
Acked-by: Christian König <christian.koenig@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210525151055.8174-7-tzimmermann@suse.de

drm/vmwgfx: Inline ttm_bo_mmap() into vmwgfx driver

The vmwgfx driver is the only remaining user of ttm_bo_mmap(). Inline
the code. The internal helper ttm_bo_vm_lookup() is now also part of
vmwgfx as vmw_bo_vm_lookup().

v2:
* replace pr_err() with drm_err() (Zack)

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Zack Rusin <zackr@vmware.com>
Acked-by: Christian König <christian.koenig@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210525151055.8174-6-tzimmermann@suse.de

drm/nouveau: Implement mmap as GEM object function

Moving the driver-specific mmap code into a GEM object function allows
for using DRM helpers for various mmap callbacks.

The GEM object function is provided by GEM TTM helpers. Nouveau's
implementation of verify_access is unused and has been removed. Access
permissions are validated by the DRM helpers.

As a side effect, nouveau_ttm_vm_ops and nouveau_ttm_fault() are now
implemented in nouveau's GEM code.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Christian König <christian.koenig@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210525151055.8174-5-tzimmermann@suse.de

drm/radeon: Implement mmap as GEM object function

Moving the driver-specific mmap code into a GEM object function allows
for using DRM helpers for various mmap callbacks.

This change also allows to support prime-based mmap via DRM's helper
drm_gem_prime_mmap().

Permission checks are implemented by drm_gem_mmap(), with an additional
check for radeon_ttm_tt_has_userptr() in the GEM object function. The
function radeon_verify_access() is now unused and has thus been removed.

As a side effect, radeon_ttm_vm_ops and radeon_ttm_fault() are now
implemented in amdgpu's GEM code.

v3:
* remove unnecessary checks from mmap (Christian)
v2:
* rename radeon_ttm_vm_ops and radeon_ttm_fault() to
  radeon_gem_vm_ops and radeon_gem_fault() (Christian)
* fix commit description (Alex)

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Christian König <christian.koenig@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210525151055.8174-4-tzimmermann@suse.de

drm/amdgpu: Implement mmap as GEM object function

Moving the driver-specific mmap code into a GEM object function allows
for using DRM helpers for various mmap callbacks.

This change resolves several inconsistencies between regular mmap and
prime-based mmap. The vm_ops field in vma is now set for all mmap'ed
areas. Previously it way only set for regular mmap calls, prime-based
mmap used TTM's default vm_ops. The function amdgpu_verify_access() is
no longer being called and therefore removed by this patch.

As a side effect, amdgpu_ttm_vm_ops and amdgpu_ttm_fault() are now
implemented in amdgpu's GEM code.

v4:
* rebased
v3:
* rename mmap function to amdgpu_gem_object_mmap() (Christian)
* remove unnecessary checks from mmap (Christian)
v2:
* rename amdgpu_ttm_vm_ops and amdgpu_ttm_fault() to
  amdgpu_gem_vm_ops and amdgpu_gem_fault() (Christian)
* the check for kfd_bo has meanwhile been removed

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Christian König <christian.koenig@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210525151055.8174-3-tzimmermann@suse.de

drm/ttm: Don't override vm_ops callbacks, if set

Drivers may want to set their own callbacks for a VM area. Only set
TTM's callbacks if the vm_ops field is clear.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Christian König <christian.koenig@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210525151055.8174-2-tzimmermann@suse.de

drm/amdgpu: Fix clang warning: unused label 'exit'

Problem:
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c:332:1: warning: unused label 'exit' [-Wunused-label]
exit:
^~~~~

Fix: Put #ifdef CONFIG_64BIT around exit

Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210525184431.1170373-1-andrey.grodzovsky@amd.com

drm/amdgpu: Fix crash when hot unplug in BACO

Problem:
When device goes into runtime suspend due to prolonged
inactivity (e.g. BACO sleep) and then hot unplugged,
PCI core will try to wake up the device as part of
unplug process. Since the device is gone all HW
programming during rpm resume fails leading
to a bad SW state later during pci remove handling.

Fix:
Use a flag we use for PCIe error recovery to avoid
accessing registres. This allows to successfully complete
rpm resume sequence and finish pci remove.

v2: Renamed HW access block flag

Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Link: https://gitlab.freedesktop.org/drm/amd/-/issues/1081
Link: https://patchwork.freedesktop.org/patch/msgid/20210521204122.762288-2-andrey.grodzovsky@amd.com

drm/amdgpu: Rename flag which prevents HW access

Make it's name not feature but function descriptive.

Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210521204122.762288-1-andrey.grodzovsky@amd.com

drm/i915/gt: Move CS interrupt handler to the backend

The different submission backends each have their own preferred
behaviour and interrupt setup. Let each handle their own interrupts.

This becomes more useful later as we to extract the use of auxiliary
state in the interrupt handler that is backend specific.

Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20210521183215.65451-4-matthew.brost@intel.com

drm/i915/gt: Move submission_method into intel_gt

Since we setup the submission method for the engines once, it is easy to
assign an enum and use that instead of probing into the backends.

Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20210521183215.65451-3-matthew.brost@intel.com

drm/i915/gt: Move engine setup out of set_default_submission

Now that we no longer switch back and forth between guc and execlists,
we no longer need to restore the backend's vfunc and can leave them set
after initialisation. The only catch is that we lose the submission on
wedging and still need to reset the submit_request vfunc on unwedging.

Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20210521183215.65451-2-matthew.brost@intel.com

drm/ingenic: Add option to alloc cached GEM buffers

Alloc GEM buffers backed by noncoherent memory on SoCs where it is
actually faster than write-combine.

This dramatically speeds up software rendering on these SoCs, even for
tasks where write-combine memory should in theory be faster (e.g. simple
blits).

v3: The option is now selected per-SoC instead of being a module
    parameter.

v5: - Fix drm_atomic_get_new_plane_state() used to retrieve the old
      state
    - Use custom drm_gem_fb_create()
    - Only check damage clips and sync DMA buffers if non-coherent
      buffers are used

Signed-off-by: Paul Cercueil <paul@crapouillou.net>
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20210523170415.90410-4-paul@crapouillou.net

drm: Add and export function drm_fb_cma_sync_non_coherent

This function can be used by drivers that use damage clips and have
CMA GEM objects backed by non-coherent memory. Calling this function
in a plane's .atomic_update ensures that all the data in the backing
memory have been written to RAM.

v3: - Only sync data if using GEM objects backed by non-coherent memory.
    - Use a drm_device pointer instead of device pointer in prototype

v5: - Rename to drm_fb_cma_sync_non_coherent
    - Invert loops for better cache locality
    - Only sync BOs that have the non-coherent flag
    - Move to drm_fb_cma_helper.c to avoid circular dependency

Signed-off-by: Paul Cercueil <paul@crapouillou.net>
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20210523170415.90410-3-paul@crapouillou.net

drm: Add support for GEM buffers backed by non-coherent memory

Having GEM buffers backed by non-coherent memory is interesting in the
particular case where it is faster to render to a non-coherent buffer
then sync the data cache, than to render to a write-combine buffer, and
(by extension) much faster than using a shadow buffer. This is true for
instance on some Ingenic SoCs, where even simple blits (e.g. memcpy)
are about three times faster using this method.

Add a 'map_noncoherent' flag to the drm_gem_cma_object structure, which
can be set by the drivers when they create the dumb buffer.

Since this really only applies to software rendering, disable this flag
as soon as the CMA objects are exported via PRIME.

v3: New patch. Now uses a simple 'map_noncoherent' flag to control how
    the objects are mapped, and use the new dma_mmap_pages function.

v4: Make sure map_noncoherent is always disabled when creating GEM
    objects meant to be used with dma-buf.

Signed-off-by: Paul Cercueil <paul@crapouillou.net>
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20210523170415.90410-2-paul@crapouillou.net

drm/vc4: hdmi: Fix PM reference leak in vc4_hdmi_encoder_pre_crtc_co()

pm_runtime_get_sync will increment pm usage counter even it failed.
Forgetting to putting operation will result in reference leak here.
Fix it by replacing it with pm_runtime_resume_and_get to keep usage
counter balanced.

Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Zou Wei <zou_wei@huawei.com>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://patchwork.freedesktop.org/patch/msgid/1621840854-105978-1-git-send-email-zou_wei@huawei.com

drm/panel: panel-simple: Add missing pm_runtime_dont_use_autosuspend() calls

The PM Runtime docs specifically call out the need to call
pm_runtime_dont_use_autosuspend() in the remove() callback if
pm_runtime_use_autosuspend() was called in probe():

> Drivers in ->remove() callback should undo the runtime PM changes done
> in ->probe(). Usually this means calling pm_runtime_disable(),
> pm_runtime_dont_use_autosuspend() etc.

We should do this. This fixes a warning splat that I saw when I was
testing out the panel-simple's remove().

Fixes: 3235b0f20a0a ("drm/panel: panel-simple: Use runtime pm to avoid excessive unprepare / prepare")
Signed-off-by: Douglas Anderson <dianders@chromium.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210517130450.v7.1.I9e947183e95c9bd067c9c1d51208ac6a96385139@changeid

drm/vc4: hdmi: Add a workqueue to set scrambling

It looks like some displays (like the LG 27UL850-W) don't enable the
scrambling when the HDMI driver enables it. However, if we set later the
scrambler enable bit, the display will work as expected.

Let's create delayed work queue to periodically look at the display
scrambling status, and if it's not set yet try to enable it again.

Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Reviewed-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210507150515.257424-12-maxime@cerno.tech

drm/vc4: hdmi: Enable the scrambler

The HDMI controller on the BCM2711 includes a scrambler in order to
reach the HDMI 2.0 modes that require it. Let's add the support for it.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Reviewed-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210507150515.257424-11-maxime@cerno.tech

drm/vc4: hdmi: Check and warn if we can't reach 4kp60 frequencies

In order to reach the frequencies needed to output at 594MHz, the
firmware needs to be configured with the appropriate parameters in the
config.txt file (enable_hdmi_4kp60 and force_turbo).

Let's detect it at bind time, warn the user if we can't, and filter out
the relevant modes.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://patchwork.freedesktop.org/patch/msgid/20210507150515.257424-10-maxime@cerno.tech

drm/vc4: hdmi: Properly compute the BVB clock rate

The BVB clock rate computation doesn't take into account a mode clock of
594MHz that we're going to need to support 4k60.

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://patchwork.freedesktop.org/patch/msgid/20210507150515.257424-9-maxime@cerno.tech

drm/vc4: hvs: Make the HVS bind first

We'll need to have the HVS binding before the HDMI controllers so that
we can check whether the firmware allows to run in 4kp60. Reorder a bit
the component list, and document the current constraints we're aware of.

Acked-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://patchwork.freedesktop.org/patch/msgid/20210507150515.257424-8-maxime@cerno.tech

drm/vc4: hdmi: Prevent clock unbalance

Since we fixed the hooks to disable the encoder at boot, we now have an
unbalanced clk_disable call at boot since we never enabled them in the
first place.

Let's mimic the state of the hardware and enable the clocks at boot if
the controller is enabled to get the use-count right.

Cc: <stable@vger.kernel.org> # v5.10+
Fixes: 09c438139b8f ("drm/vc4: hdmi: Implement finer-grained hooks")
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Reviewed-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210507150515.257424-7-maxime@cerno.tech

drm/vc4: crtc: Lookup the encoder from the register at boot

At boot, we can't rely on the vc4_get_crtc_encoder since we don't have a
state yet and thus will not be able to figure out which connector is
attached to our CRTC.

However, we have a muxing bit in the CRTC register we can use to get the
encoder currently connected to the pixelvalve. We can thus read that
register, lookup the associated register through the vc4_pv_data
structure, and then pass it to vc4_crtc_disable so that we can perform
the proper operations.

Fixes: 875a4d536842 ("drm/vc4: drv: Disable the CRTC at boot time")
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Reviewed-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210507150515.257424-6-maxime@cerno.tech

drm/vc4: crtc: Fix vc4_get_crtc_encoder logic

The vc4_get_crtc_encoder function currently only works when the
connector->state->crtc pointer is set, which is only true when the
connector is currently enabled.

However, we use it as part of the disable path as well, and our lookup
will fail in that case, resulting in it returning a null pointer we
can't act on.

We can access the connector that used to be connected to that crtc
though using the old connector state in the disable path.

Since we want to support both the enable and disable path, we can
support it by passing the state accessor variant as a function pointer,
together with the atomic state.

Fixes: 792c3132bc1b ("drm/vc4: encoder: Add finer-grained encoder callbacks")
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Reviewed-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210507150515.257424-5-maxime@cerno.tech

drm/vc4: crtc: Pass the drm_atomic_state to config_pv

The vc4_crtc_config_pv will need to access the drm_atomic_state
structure and its only parent function, vc4_crtc_atomic_enable already
has access to it. Let's pass it as a parameter.

Fixes: 792c3132bc1b ("drm/vc4: encoder: Add finer-grained encoder callbacks")
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Reviewed-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210507150515.257424-4-maxime@cerno.tech

drm/vc4: crtc: Skip the TXP

The vc4_set_crtc_possible_masks is meant to run over all the encoders
and then set their possible_crtcs mask to their associated pixelvalve.

However, since the commit 39fcb2808376 ("drm/vc4: txp: Turn the TXP into
a CRTC of its own"), the TXP has been turned to a CRTC and encoder of
its own, and while it does indeed register an encoder, it no longer has
an associated pixelvalve. The code will thus run over the TXP encoder
and set a bogus possible_crtcs mask, overriding the one set in the TXP
bind function.

In order to fix this, let's skip any virtual encoder.

Cc: <stable@vger.kernel.org> # v5.9+
Fixes: 39fcb2808376 ("drm/vc4: txp: Turn the TXP into a CRTC of its own")
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://patchwork.freedesktop.org/patch/msgid/20210507150515.257424-3-maxime@cerno.tech

drm/vc4: txp: Properly set the possible_crtcs mask

The current code does a binary OR on the possible_crtcs variable of the
TXP encoder, while we want to set it to that value instead.

Cc: <stable@vger.kernel.org> # v5.9+
Fixes: 39fcb2808376 ("drm/vc4: txp: Turn the TXP into a CRTC of its own")
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://patchwork.freedesktop.org/patch/msgid/20210507150515.257424-2-maxime@cerno.tech

drm/vc4: Separate VEC compatible variants

The VEC's DAC on BCM2711 is slightly different compared to the one on
BCM283x and needs different configuration. In particular, bit 3
(mask 0x8) switches the BCM2711 DAC input to "self-test input data",
which makes the output unusable. Separating two compatible variants in
devicetrees and the DRM driver was therefore necessary.

The configurations used for both variants have been borrowed from
Raspberry Pi (model 3B for BCM283x, 4B for BCM2711) firmware defaults.

Signed-off-by: Mateusz Kwiatkowski <kfyatek+publicgit@gmail.com>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Reviewed-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210520150344.273900-4-maxime@cerno.tech

dt-bindings: display: bcm2835-vec: Add BCM2711 compatible

The BCM2711 VEC uses a slightly different, incompatible, setup than the
one used for the earlier SoC. Add a new compatible for it.

Signed-off-by: Mateusz Kwiatkowski <kfyatek+publicgit@gmail.com>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Acked-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210520150344.273900-3-maxime@cerno.tech

drm/vc4: Fix clock source for VEC PixelValve on BCM2711

On the BCM2711 (Raspberry Pi 4), the VEC is actually connected to
output 2 of pixelvalve3.

NOTE: This contradicts the Broadcom docs, but has been empirically
tested and confirmed by Raspberry Pi firmware devs.

Signed-off-by: Mateusz Kwiatkowski <kfyatek+publicgit@gmail.com>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Reviewed-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210520150344.273900-2-maxime@cerno.tech

Linux 5.13-rc3

Merge tag 'perf-urgent-2021-05-23' of git://git./linux/kernel/git/tip/tip

Pull perf fixes from Thomas Gleixner:
 "Two perf fixes:

   - Do not check the LBR_TOS MSR when setting up unrelated LBR MSRs as
     this can cause malfunction when TOS is not supported

   - Allocate the LBR XSAVE buffers along with the DS buffers upfront
     because allocating them when adding an event can deadlock"

* tag 'perf-urgent-2021-05-23' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  perf/x86/lbr: Remove cpuc->lbr_xsave allocation from atomic context
  perf/x86: Avoid touching LBR_TOS MSR for Arch LBR

Merge tag 'locking-urgent-2021-05-23' of git://git./linux/kernel/git/tip/tip

Pull locking fixes from Thomas Gleixner:
 "Two locking fixes:

   - Invoke the lockdep tracepoints in the correct place so the ordering
     is correct again

   - Don't leave the mutex WAITER bit stale when the last waiter is
     dropping out early due to a signal as that forces all subsequent
     lock operations needlessly into the slowpath until it's cleaned up
     again"

* tag 'locking-urgent-2021-05-23' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  locking/mutex: clear MUTEX_FLAGS if wait_list is empty due to signal
  locking/lockdep: Correct calling tracepoints

Merge tag 'irq-urgent-2021-05-23' of git://git./linux/kernel/git/tip/tip

Pull irq fixes from Thomas Gleixner:
 "A few fixes for irqchip drivers:

   - Allocate interrupt descriptors correctly on Mainstone PXA when
     SPARSE_IRQ is enabled; otherwise the interrupt association fails

   - Make the APPLE AIC chip driver depend on APPLE

   - Remove redundant error output on devm_ioremap_resource() failure"

* tag 'irq-urgent-2021-05-23' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  irqchip: Remove redundant error printing
  irqchip/apple-aic: APPLE_AIC should depend on ARCH_APPLE
  ARM: PXA: Fix cplds irqdesc allocation when using legacy mode

Merge tag 'x86_urgent_for_v5.13_rc3' of git://git./linux/kernel/git/tip/tip

Pull x86 fixes from Borislav Petkov:

 - Fix how SEV handles MMIO accesses by forwarding potential page faults
   instead of killing the machine and by using the accessors with the
   exact functionality needed when accessing memory.

 - Fix a confusion with Clang LTO compiler switches passed to the it

 - Handle the case gracefully when VMGEXIT has been executed in
   userspace

* tag 'x86_urgent_for_v5.13_rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/sev-es: Use __put_user()/__get_user() for data accesses
  x86/sev-es: Forward page-faults which happen during emulation
  x86/sev-es: Don't return NULL from sev_es_get_ghcb()
  x86/build: Fix location of '-plugin-opt=' flags
  x86/sev-es: Invalidate the GHCB after completing VMGEXIT
  x86/sev-es: Move sev_es_put_ghcb() in prep for follow on patch

Merge tag 'powerpc-5.13-4' of git://git./linux/kernel/git/powerpc/linux

Pull powerpc fixes from Michael Ellerman:

 - Fix breakage of strace (and other ptracers etc.) when using the new
   scv ABI (Power9 or later with glibc >= 2.33).

 - Fix early_ioremap() on 64-bit, which broke booting on some machines.

Thanks to Dmitry V. Levin, Nicholas Piggin, Alexey Kardashevskiy, and
Christophe Leroy.

* tag 'powerpc-5.13-4' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux:
  powerpc/64s/syscall: Fix ptrace syscall info with scv syscalls
  powerpc/64s/syscall: Use pt_regs.trap to distinguish syscall ABI difference between sc and scv syscalls
  powerpc: Fix early setup to make early_ioremap() work

Merge tag 'kbuild-fixes-v5.13' of git://git./linux/kernel/git/masahiroy/linux-kbuild

Pull Kbuild fixes from Masahiro Yamada:

 - Fix short log indentation for tools builds

 - Fix dummy-tools to adjust to the latest stackprotector check

* tag 'kbuild-fixes-v5.13' of git://git.kernel.org/pub/scm/linux/kernel/git/masahiroy/linux-kbuild:
  kbuild: dummy-tools: adjust to stricter stackprotector check
  scripts/jobserver-exec: Fix a typo ("envirnoment")
  tools build: Fix quiet cmd indentation

Merge branch 'akpm' (patches from Andrew)

Merge misc fixes from Andrew Morton:
 "10 patches.

  Subsystems affected by this patch series: mm (pagealloc, gup, kasan,
  and userfaultfd), ipc, selftests, watchdog, bitmap, procfs, and lib"

* emailed patches from Andrew Morton <akpm@linux-foundation.org>:
  userfaultfd: hugetlbfs: fix new flag usage in error path
  lib: kunit: suppress a compilation warning of frame size
  proc: remove Alexey from MAINTAINERS
  linux/bits.h: fix compilation error with GENMASK
  watchdog: reliable handling of timestamps
  kasan: slab: always reset the tag in get_freepointer_safe()
  tools/testing/selftests/exec: fix link error
  ipc/mqueue, msg, sem: avoid relying on a stack reference past its expiry
  Revert "mm/gup: check page posion status for coredump."
  mm/shuffle: fix section mismatch warning

userfaultfd: hugetlbfs: fix new flag usage in error path

In commit d6995da31122 ("hugetlb: use page.private for hugetlb specific
page flags") the use of PagePrivate to indicate a reservation count
should be restored at free time was changed to the hugetlb specific flag
HPageRestoreReserve.  Changes to a userfaultfd error path as well as a
VM_BUG_ON() in remove_inode_hugepages() were overlooked.

Users could see incorrect hugetlb reserve counts if they experience an
error with a UFFDIO_COPY operation.  Specifically, this would be the
result of an unlikely copy_huge_page_from_user error.  There is not an
increased chance of hitting the VM_BUG_ON.

Link: https://lkml.kernel.org/r/20210521233952.236434-1-mike.kravetz@oracle.com
Fixes: d6995da31122 ("hugetlb: use page.private for hugetlb specific page flags")
Signed-off-by: Mike Kravetz <mike.kravetz@oracle.com>
Reviewed-by: Mina Almasry <almasry.mina@google.com>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Muchun Song <songmuchun@bytedance.com>
Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Miaohe Lin <linmiaohe@huawei.com>
Cc: Mina Almasry <almasrymina@google.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

lib: kunit: suppress a compilation warning of frame size

lib/bitfield_kunit.c: In function `test_bitfields_constants':
lib/bitfield_kunit.c:93:1: warning: the frame size of 7456 bytes is larger than 2048 bytes [-Wframe-larger-than=]
 }
 ^

As the description of BITFIELD_KUNIT in lib/Kconfig.debug, it "Only useful
for kernel devs running the KUnit test harness, and not intended for
inclusion into a production build".  Therefore, it is not worth modifying
variable 'test_bitfields_constants' to clear this warning.  Just suppress
it.

Link: https://lkml.kernel.org/r/20210518094533.7652-1-thunder.leizhen@huawei.com
Signed-off-by: Zhen Lei <thunder.leizhen@huawei.com>
Cc: Shuah Khan <skhan@linuxfoundation.org>
Cc: Vitor Massaru Iha <vitor@massaru.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

proc: remove Alexey from MAINTAINERS

People Cc me and I don't have time.

Link: https://lkml.kernel.org/r/YKarMxHJBIhMHQIh@localhost.localdomain
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

linux/bits.h: fix compilation error with GENMASK

GENMASK() has an input check which uses __builtin_choose_expr() to
enable a compile time sanity check of its inputs if they are known at
compile time.

However, it turns out that __builtin_constant_p() does not always return
a compile time constant [0].  It was thought this problem was fixed with
gcc 4.9 [1], but apparently this is not the case [2].

Switch to use __is_constexpr() instead which always returns a compile time
constant, regardless of its inputs.

Link: https://lore.kernel.org/lkml/42b4342b-aefc-a16a-0d43-9f9c0d63ba7a@rasmusvillemoes.dk
Link: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=19449
Link: https://lore.kernel.org/lkml/1ac7bbc2-45d9-26ed-0b33-bf382b8d858b@I-love.SAKURA.ne.jp
Link: https://lkml.kernel.org/r/20210511203716.117010-1-rikard.falkeborn@gmail.com
Signed-off-by: Rikard Falkeborn <rikard.falkeborn@gmail.com>
Reported-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Cc: Ard Biesheuvel <ardb@kernel.org>
Cc: Yury Norov <yury.norov@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

watchdog: reliable handling of timestamps

Commit 9bf3bc949f8a ("watchdog: cleanup handling of false positives")
tried to handle a virtual host stopped by the host a more
straightforward and cleaner way.

But it introduced a risk of false softlockup reports.  The virtual host
might be stopped at any time, for example between
kvm_check_and_clear_guest_paused() and is_softlockup().  As a result,
is_softlockup() might read the updated jiffies and detects a softlockup.

A solution might be to put back kvm_check_and_clear_guest_paused() after
is_softlockup() and detect it.  But it would put back the cycle that
complicates the logic.

In fact, the handling of all the timestamps is not reliable.  The code
does not guarantee when and how many times the timestamps are read.  For
example, "period_ts" might be touched anytime also from NMI and re-read in
is_softlockup().  It works just by chance.

Fix all the problems by making the code even more explicit.

1. Make sure that "now" and "period_ts" timestamps are read only once.
   They might be changed at anytime by NMI or when the virtual guest is
   stopped by the host.  Note that "now" timestamp does this implicitly
   because "jiffies" is marked volatile.

2. "now" time must be read first.  The state of "period_ts" will
   decide whether it will be used or the period will get restarted.

3. kvm_check_and_clear_guest_paused() must be called before reading
   "period_ts".  It touches the variable when the guest was stopped.

As a result, "now" timestamp is used only when the watchdog was not
touched and the guest not stopped in the meantime.  "period_ts" is
restarted in all other situations.

Link: https://lkml.kernel.org/r/YKT55gw+RZfyoFf7@alley
Fixes: 9bf3bc949f8aeefeacea4b ("watchdog: cleanup handling of false positives")
Signed-off-by: Petr Mladek <pmladek@suse.com>
Reported-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Reviewed-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

kasan: slab: always reset the tag in get_freepointer_safe()

With CONFIG_DEBUG_PAGEALLOC enabled, the kernel should also untag the
object pointer, as done in get_freepointer().

Failing to do so reportedly leads to SLUB freelist corruptions that
manifest as boot-time crashes.

Link: https://lkml.kernel.org/r/20210514072228.534418-1-glider@google.com
Signed-off-by: Alexander Potapenko <glider@google.com>
Cc: Marco Elver <elver@google.com>
Cc: Vincenzo Frascino <vincenzo.frascino@arm.com>
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Andrey Konovalov <andreyknvl@gmail.com>
Cc: Elliot Berman <eberman@codeaurora.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

tools/testing/selftests/exec: fix link error

Fix the link error by adding '-static':

  gcc -Wall  -Wl,-z,max-page-size=0x1000 -pie load_address.c -o /home/yang/linux/tools/testing/selftests/exec/load_address_4096
  /usr/bin/ld: /tmp/ccopEGun.o: relocation R_AARCH64_ADR_PREL_PG_HI21 against symbol `stderr@@GLIBC_2.17' which may bind externally can not be used when making a shared object; recompile with -fPIC
  /usr/bin/ld: /tmp/ccopEGun.o(.text+0x158): unresolvable R_AARCH64_ADR_PREL_PG_HI21 relocation against symbol `stderr@@GLIBC_2.17'
  /usr/bin/ld: final link failed: bad value
  collect2: error: ld returned 1 exit status
  make: *** [Makefile:25: tools/testing/selftests/exec/load_address_4096] Error 1

Link: https://lkml.kernel.org/r/20210514092422.2367367-1-yangyingliang@huawei.com
Fixes: 206e22f01941 ("tools/testing/selftests: add self-test for verifying load alignment")
Signed-off-by: Yang Yingliang <yangyingliang@huawei.com>
Cc: Chris Kennelly <ckennelly@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

ipc/mqueue, msg, sem: avoid relying on a stack reference past its expiry

do_mq_timedreceive calls wq_sleep with a stack local address.  The
sender (do_mq_timedsend) uses this address to later call pipelined_send.

This leads to a very hard to trigger race where a do_mq_timedreceive
call might return and leave do_mq_timedsend to rely on an invalid
address, causing the following crash:

  RIP: 0010:wake_q_add_safe+0x13/0x60
  Call Trace:
   __x64_sys_mq_timedsend+0x2a9/0x490
   do_syscall_64+0x80/0x680
   entry_SYSCALL_64_after_hwframe+0x44/0xa9
  RIP: 0033:0x7f5928e40343

The race occurs as:

1. do_mq_timedreceive calls wq_sleep with the address of `struct
   ext_wait_queue` on function stack (aliased as `ewq_addr` here) - it
   holds a valid `struct ext_wait_queue *` as long as the stack has not
   been overwritten.

2. `ewq_addr` gets added to info->e_wait_q[RECV].list in wq_add, and
   do_mq_timedsend receives it via wq_get_first_waiter(info, RECV) to call
   __pipelined_op.

3. Sender calls __pipelined_op::smp_store_release(&this->state,
   STATE_READY).  Here is where the race window begins.  (`this` is
   `ewq_addr`.)

4. If the receiver wakes up now in do_mq_timedreceive::wq_sleep, it
   will see `state == STATE_READY` and break.

5. do_mq_timedreceive returns, and `ewq_addr` is no longer guaranteed
   to be a `struct ext_wait_queue *` since it was on do_mq_timedreceive's
   stack.  (Although the address may not get overwritten until another
   function happens to touch it, which means it can persist around for an
   indefinite time.)

6. do_mq_timedsend::__pipelined_op() still believes `ewq_addr` is a
   `struct ext_wait_queue *`, and uses it to find a task_struct to pass to
   the wake_q_add_safe call.  In the lucky case where nothing has
   overwritten `ewq_addr` yet, `ewq_addr->task` is the right task_struct.
   In the unlucky case, __pipelined_op::wake_q_add_safe gets handed a
   bogus address as the receiver's task_struct causing the crash.

do_mq_timedsend::__pipelined_op() should not dereference `this` after
setting STATE_READY, as the receiver counterpart is now free to return.
Change __pipelined_op to call wake_q_add_safe on the receiver's
task_struct returned by get_task_struct, instead of dereferencing `this`
which sits on the receiver's stack.

As Manfred pointed out, the race potentially also exists in
ipc/msg.c::expunge_all and ipc/sem.c::wake_up_sem_queue_prepare.  Fix
those in the same way.

Link: https://lkml.kernel.org/r/20210510102950.12551-1-varad.gautam@suse.com
Fixes: c5b2cbdbdac563 ("ipc/mqueue.c: update/document memory barriers")
Fixes: 8116b54e7e23ef ("ipc/sem.c: document and update memory barriers")
Fixes: 0d97a82ba830d8 ("ipc/msg.c: update and document memory barriers")
Signed-off-by: Varad Gautam <varad.gautam@suse.com>
Reported-by: Matthias von Faber <matthias.vonfaber@aox-tech.de>
Acked-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Christian Brauner <christian.brauner@ubuntu.com>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Revert "mm/gup: check page posion status for coredump."

While reviewing [1] I came across commit d3378e86d182 ("mm/gup: check
page posion status for coredump.") and noticed that this patch is broken
in two ways.  First it doesn't really prevent hwpoison pages from being
dumped because hwpoison pages can be marked asynchornously at any time
after the check.  Secondly, and more importantly, the patch introduces a
ref count leak because get_dump_page takes a reference on the page which
is not released.

It also seems that the patch was merged incorrectly because there were
follow up changes not included as well as discussions on how to address
the underlying problem [2]

Therefore revert the original patch.

Link: http://lkml.kernel.org/r/20210429122519.15183-4-david@redhat.com
Link: http://lkml.kernel.org/r/57ac524c-b49a-99ec-c1e4-ef5027bfb61b@redhat.com
Link: https://lkml.kernel.org/r/20210505135407.31590-1-mhocko@kernel.org
Fixes: d3378e86d182 ("mm/gup: check page posion status for coredump.")
Signed-off-by: Michal Hocko <mhocko@suse.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Cc: Aili Yao <yaoaili@kingsoft.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/shuffle: fix section mismatch warning

clang sometimes decides not to inline shuffle_zone(), but it calls a
__meminit function.  Without the extra __meminit annotation we get this
warning:

  WARNING: modpost: vmlinux.o(.text+0x2a86d4): Section mismatch in reference from the function shuffle_zone() to the function .meminit.text:__shuffle_zone()
  The function shuffle_zone() references
  the function __meminit __shuffle_zone().
  This is often because shuffle_zone lacks a __meminit
  annotation or the annotation of __shuffle_zone is wrong.

shuffle_free_memory() did not show the same problem in my tests, but it
could happen in theory as well, so mark both as __meminit.

Link: https://lkml.kernel.org/r/20210514135952.2928094-1-arnd@kernel.org
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Nathan Chancellor <nathan@kernel.org>
Cc: Nick Desaulniers <ndesaulniers@google.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Wei Yang <richard.weiyang@linux.alibaba.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>