Merge tag 'drm-habanalabs-next-2023-10-10' of https://git./linux/kernel/git/ogabbay/linux into drm-next

This tag contains habanalabs driver changes for v6.7.

The notable changes are:

- uAPI changes:
  - Expose tsc clock sampling to better sync clock information in profiler.
  - Enhance engine error reporting in the info ioctl.
  - Block access to the eventfd operations through the control device.
  - Disable the option of the user to register multiple times with the same
    offset for timestamp dump by the driver. If a user wants to use the same
    offset in the timestamp buffer for different interrupt, it needs to first
    de-register the offset.
  - When exporting dma-buf (for p2p), force the user to specify size/offset
    in multiples of PAGE_SIZE. This is instead of the driver doing the
    rounding to PAGE_SIZE, which has caused the driver to map more memory
    than was intended by the user.

- New features and improvements:
  - Complete the move of the driver to the accel subsystem by removing the
    custom habanalabs class and major and registering to accel subsystem.
  - Move the firmware interface files to include/linux/habanalabs. This is
    a pre-requisite for upstreaming the NIC drivers of Gaudi (as they need to
    include those files).
  - Perform device hard-reset upon PCIe AXI drain event to prevent the failure
    from cascading to different IP blocks in the SoC. In secured environments,
    this is done automatically by the firmware.
  - Print device name when it is removed for better debuggability.
  - Add support for trace of dma map sgtable operations.
  - Optimize handling of user interrupts by splitting the interrupts to two
    lists. One list for fast handling and second list for handling with
    timestamp recording, which is slower.
  - Prevent double device hard-reset due to 2 adjacent H/W events.
  - Set device status 'malfunction' while in rmmod.

- Firmware related fixes:
  - Extend preboot timeout because preboot loading might take longer than
    expected in certain cases.
  - Add a protection mechanism for the Event Queue. In case it is full, the
    firmware will be able to notify about it through a dedicated interrupt.
  - Perform device hard-reset in case scrubbing of memory has failed.

- Bug fixes and code cleanups:
  - Small fixes of dma-buf handling in Gaudi2, such as handling an offset != 0,
    using the correct exported size, creation of sg table.
  - Fix spmu mask creation.
  - Fix bug in wait for cs completion for decoder workloads.
  - Cleanup Greco name from documentation.
  - Fix bug in recording timestamp during cs completion interrupt handling.
  - Fix CoreSight ETF configuration and flush logic.
  - Fix small bug in hpriv_list handling (the list that contains the private
    data per process that opens our device).

Signed-off-by: Dave Airlie <airlied@redhat.com>
# -----BEGIN PGP SIGNATURE-----
#
# iQEzBAABCgAdFiEE7TEboABC71LctBLFZR1NuKta54AFAmUlHoQACgkQZR1NuKta
# 54DsXQf8CW+W4iWJf5UDTj/E/giu9rVRrsUsU0hhCcXbecIxRsLObYXtulENu5/u
# VuEAo/tAvo0LUKi8pdIv6ernDKaxZ1+fimlfXMCzllAA/ts3yp1NgunprsIsx3tv
# YgcJ2GNR8UlVZ1qYuZl+4dOTyD0yfRMROUXBe7wqKnUXOEepOiLBxq6W15tZiJnx
# L+V0yGkNk6pAoADIXLW9EgEXiN/bJZCXGPWp06i/Nz7cHIHJGoV59wAqftqllCtk
# 8ZMkLByjlQKPhc5AgWBtKE8EGVip3sm7b/Q2Gq0ZXdZiebyVJ+AjuuDOdtq1UCIw
# Rcp2576E7rByIBu3RAFlrioWhuR5Zw==
# =2ien
# -----END PGP SIGNATURE-----
# gpg: Signature made Tue 10 Oct 2023 19:51:00 AEST
# gpg:                using RSA key ED311BA00042EF52DCB412C5651D4DB8AB5AE780
# gpg: Can't check signature: No public key
From: Oded Gabbay <ogabbay@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/ZSUfiX4J7v4Wn0cU@ogabbay-vm-u22.habana-labs.com

Merge tag 'drm-intel-gt-next-2023-10-12' of git://anongit.freedesktop.org/drm/drm-intel into drm-next

Driver Changes:

Fixes/improvements/new stuff:

- Register engines early to avoid type confusion (Mathias Krause)
- Suppress 'ignoring reset notification' message [guc] (John Harrison)
- Update 'recommended' version to 70.12.1 for DG2/ADL-S/ADL-P/MTL [guc] (John Harrison)
- Enable WA 14018913170 [guc, dg2] (Daniele Ceraolo Spurio)

Future platform enablement:

- Clean steer semaphore on resume (Nirmoy Das)
- Skip MCR ops for ring fault register [mtl] (Nirmoy Das)
- Make i915_gem_shrinker multi-gt aware [gem] (Jonathan Cavitt)
- Enable GGTT updates with binder in MTL (Nirmoy Das, Chris Wilson)
- Invalidate the TLBs on each GT (Chris Wilson)

Miscellaneous:

- Clarify type evolution of uabi_node/uabi_engines (Mathias Krause)
- Annotate struct ct_incoming_msg with __counted_by [guc] (Kees Cook)
- More use of GT specific print helpers [gt] (John Harrison)

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/ZSfKotZVdypU6NaX@tursulin-desk

Merge tag 'drm-intel-next-2023-10-12' of git://anongit.freedesktop.org/drm/drm-intel into drm-next

drm/i915 feature pull #2 for v6.7:

Features and functionality:
- Preparation for i915 display code reuse in upcoming Xe driver (Jani)
- Drop the fastboot module parameter and use the platform defaults (Arun)
- Enable new LNL FBC features (Vinod)
- Add LNL display feature capability reads (Vinod)

Refactoring and cleanups:
- Locally enable W=1 warnings by default in i915 (Jani)
- Move HDCP GSC message code to a separate file (Suraj)
- GVT include cleanups (Jani)
- Move more display init under display/ (Jani)
- DPLL ID refactoring (Ville)
- Better abstraction of GT0 (Jani)
- Move VGA decode function to GMCH code (Uma)
- Use local64_try_cmpxchg() to optimize PMU event read (Uros Bizjak)
- Clean up FBC checks (Ville)
- Constify and unify state checker calling conventions (Ville)
- Add display step name helper (Chaitanya)

Documentation:
- Update CCS and GSC CS documentation (Rodrigo)
- Fix a number of documentation typos (Randy Dunlap)

Fixes:
- VLV DSI fixes and quirks (Hans)
- Fix crtc state memory leaks (Suraj)
- Increase LSPCON mode settle timeout (Niko Tsirakis)
- Stop clobbering old crtc state during state check (Ville)
- Fix VLV color state readout (Ville)
- Fix cx0 PHY pipe reset to allow S0iX (Khaled)
- Ensure DP MST pbn_div is up-to-date after sink reconnect (Imre)
- Drop an unnecessary NULL check to fix static analyzer warning (Suraj)
- Use an explicit rather than implicit include for frontbuffer tracking (Jouni)

Merges:
- Backmerge drm-next to fix a conflict (Jani)

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/87r0m00xew.fsf@intel.com

Merge tag 'drm-misc-next-2023-10-12' of git://anongit.freedesktop.org/drm/drm-misc into drm-next

drm-misc-next for v6.7-rc1:

Contains the previous pull request drm-misc-next-2023-10-06 + following:

Cross-subsystem Changes:
- Rename fb_pgprot to pgprot_framebuffer and remove file argument/
- Update iosys-map documentation typos.

Core Changes:
- Assorted fixes to drm/panel.
- Add HPD state to drm_connector_oob_hotplug_event(), and implement
  oob hotplug events in bridge connector.
- Replace drm_framebuffer_plane_width/height  with calls to
  drm_format_info_plane_width/height.

Driver Changes:
- Clock and debug fixes for bridge/samsung-dsim.
- More btree -> maple tree conversions.
- Assorted bugfixes in rockchip, panel-tpo-tpg110,
- Add LTK050H3148W-CTA6 panel support.
- Assorted small fixes in host1x, tegra, simpledrm.
- Suspend fixes for host1x.

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/3812345e-b086-4d72-8504-f58d84e8feab@linux.intel.com

drm/i915/dsi: Add some debug logging to mipi_exec_i2c (v2)

Add some debug logging to mipi_exec_i2c, to make debugging various
issues seen with it easier.

Changes in v2:
- Drop unnecessary __func__ drm_dbg_kms() argument

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230920195613.304091-5-hdegoede@redhat.com

drm/i915/vlv_dsi: Add DMI quirk for backlight control issues on Lenovo Yoga Tab 3 (v2)

On the Lenovo Yoga Tab 3 Pro YT3-X90F there are 2 issues with the backlight
on/off MIPI sequences:

1. The backlight on sequence has an I2C MIPI sequence element which uses
   bus 0, but there is a bogus I2cSerialBus resource under the GPU in
   the DSDT which causes i2c_acpi_find_adapter() to pick the wrong bus.

2. There is no backlight off sequence, causing the backlight to stay on.

Add a DMI quirk fixing both issues.

v2:
- Add Closes tag to gitlab issue with drm.debug=0xe, VBT info

Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/9380
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230920195613.304091-4-hdegoede@redhat.com

drm/i915/vlv_dsi: Add DMI quirk for wrong I2C bus and panel size on Lenovo Yoga Tablet 2 series (v3)

On the Lenovo Yoga Tablet 2 830 / 1050 there are 2 problems:

1. The I2C MIPI sequence elements reference bus 3. ACPI has I2C1 - I2C7
   which under Linux become bus 0 - 6. And the MIPI sequence reference
   to bus 3 is indented for I2C3 which is bus 2 under Linux.

   This leads to errors like these:
   [  178.244049] i2c_designware 80860F41:03: controller timed out
   [  178.245703] i915 0000:00:02.0: [drm] *ERROR* Failed to xfer payload of size (1) to reg (169)
   There are 3 timeouts when the panel is on, delaying
   waking up the screen on a key press by 3 seconds.

   Note mipi_exec_i2c() cannot just subtract 1 from the bus
   given in the I2C MIPI sequence element. Since on other
   devices the I2C bus-numbers used in the MIPI sequences do
   actually start at 0.

2. width_/height_mm contain a bogus 192mm x 120mm size. This is
   especially a problem on the 8" 830 version which uses a 10:16
   portrait screen where as the bogus size is 16:10.

Add a DMI quirk to override the I2C bus and the panel size with
the correct values.

Note both the 10" 1050 models as well as the 8" 830 models use the same
mainboard and thus the same DMI strings. The 10" 1050 uses a 1920x1200
landscape screen, where as the 8" 830 uses a 1200x1920 portrait screen,
so the quirk handling uses the display resolution to detect the model.

v2:
- Also override i2c_bus_num to fix mipi_exec_i2c() timeouts

v3:
- Add Closes tag to gitlab issue with drm.debug=0xe, VBT info

Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/9379
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230920195613.304091-3-hdegoede@redhat.com

drm/i915/vlv_dsi: Add DMI quirk for wrong panel modeline in BIOS on Asus TF103C (v3)

Vtotal is wrong in the BIOS supplied modeline for the DSI panel on
the Asus TF103C leading to the last line of the display being shown
as the first line.

Original: "1280x800": 60 67700 1280 1312 1328 1376 800 808 812 820 0x8 0xa
Fixed:    "1280x800": 60 67700 1280 1312 1328 1376 800 808 812 816 0x8 0xa

The factory installed Android has a hardcoded modeline in its kernel,
causing it to not suffer from this BIOS bug;
and the Android boot-splash which uses the EFI FB which does have this bug
has the last line all black causing the bug to not be visible.

This commit introduces a generic DMI based quirk mechanism to vlv_dsi for
doing various fixups, and uses this to correct the modeline.

v2:
- s/mode_fixup/dmi_quirk/ to make the new DMI quirk mechanism more generic
- Add a comment with the old and new modelines to the patch and commit msg

v3:
- Add Closes tag to gitlab issue with drm.debug=0xe, VBT info

Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/9381
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230920195613.304091-2-hdegoede@redhat.com

drm/simpledrm: Fix power domain device link validity check

We need to check if a link is non-NULL before trying to delete it.

Fixes: 61df9ca23107 ("drm/simpledrm: Add support for multiple "power-domains"")
Signed-off-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Thomas Zimmermann <tzimmermann@suse.de>
Cc: Janne Grunau <j@jannau.net>
Cc: Eric Curtin <ecurtin@redhat.com>
Cc: Neal Gompa <neal@gompa.dev>
Cc: Sven Peter <sven@svenpeter.dev>
Cc: Javier Martinez Canillas <javierm@redhat.com>
Cc: dri-devel@lists.freedesktop.org
Link: https://patchwork.freedesktop.org/patch/msgid/20231011143230.1107731-1-thierry.reding@gmail.com

iosys-map: fix kernel-doc typos

Correct spelling of "beginning".

Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Cc: Thomas Zimmermann <tzimmermann@suse.de>
Cc: dri-devel@lists.freedesktop.org
Reviewed-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20230930221428.18463-2-rdunlap@infradead.org

drm: Replace drm_framebuffer plane size functions with its equivalents

The functions drm_framebuffer_plane_{width,height} and
fb_plane_{width,height} do exactly the same job of its
equivalents drm_format_info_plane_{width,height} from drm_fourcc.

The only reason to have these functions on drm_framebuffer
would be if they would added a abstraction layer to call it just
passing a drm_framebuffer pointer and the desired plane index,
which is not the case, where these functions actually implements
just part of it. In the actual implementation, every call to both
drm_framebuffer_plane_{width,height} and fb_plane_{width,height} should
pass some drm_framebuffer attribute, which is the same as calling the
drm_format_info_plane_{width,height} functions.

The drm_format_info_pane_{width,height} functions are much more
consistent in both its implementation and its location on code. The
kind of calculation that they do is intrinsically derivated from the
drm_format_info struct and has not to do with drm_framebuffer, except
by the potential motivation described above, which is still not a good
justification to have drm_framebuffer functions to calculate it.

So, replace each drm_framebuffer_plane_{width,height} and
fb_plane_{width,height} call to drm_format_info_plane_{width,height}
and remove them.

Signed-off-by: Carlos Eduardo Gallo Filho <gcarlos@disroot.org>
Reviewed-by: André Almeida <andrealmeid@igalia.com>
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20230926141519.9315-3-gcarlos@disroot.org

drm: Remove plane hsub/vsub alignment requirement for core helpers

The drm_format_info_plane_{height,width} functions was implemented using
regular division for the plane size calculation, which cause issues [1][2]
when used on contexts where the dimensions are misaligned with relation
to the subsampling factors. So, replace the regular division by the
DIV_ROUND_UP macro.

This allows these functions to be used in more drivers, making further
work to bring more core presence on them possible.

[1] http://patchwork.freedesktop.org/patch/msgid/20170321181218.10042-3-ville.syrjala@linux.intel.com
[2] https://patchwork.freedesktop.org/patch/msgid/20211026225105.2783797-2-imre.deak@intel.com

Signed-off-by: Carlos Eduardo Gallo Filho <gcarlos@disroot.org>
Reviewed-by: André Almeida <andrealmeid@igalia.com>
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20230926141519.9315-2-gcarlos@disroot.org

fbdev: Replace fb_pgprotect() with pgprot_framebuffer()

Rename the fbdev mmap helper fb_pgprotect() to pgprot_framebuffer().
The helper sets VMA page-access flags for framebuffers in device I/O
memory.

Also clean up the helper's parameters and return value. Instead of
the VMA instance, pass the individial parameters separately: existing
page-access flags, the VMAs start and end addresses and the offset
in the underlying device memory rsp file. Return the new page-access
flags. These changes align pgprot_framebuffer() with other pgprot_()
functions.

v4:
* fix commit message (Christophe)
v3:
* rename fb_pgprotect() to pgprot_framebuffer() (Arnd)

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Geert Uytterhoeven <geert@linux-m68k.org> # m68k
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230922080636.26762-3-tzimmermann@suse.de

fbdev: Avoid file argument in fb_pgprotect()

Only PowerPC's fb_pgprotect() needs the file argument, although
the implementation in either phys_mem_access_prot() or
pci_phys_mem_access_prot() does not use it. Pass NULL to the internal
helper in preparation of further updates. A later patch will remove
the file parameter from fb_pgprotect().

While at it, replace the shift operation with PHYS_PFN().

v5:
* state function names in commit description (Javier)

Suggested-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230922080636.26762-2-tzimmermann@suse.de

drm/i915: Remove the module parameter 'fastboot'

By default fastboot is enabled on all Display 9+ platforms and disabled
on older platforms. Its not necessary to retain this as a module
parameter.

Signed-off-by: Arun R Murthy <arun.r.murthy@intel.com>
Acked-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Uma Shankar <uma.shankar@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230926091157.635438-1-arun.r.murthy@intel.com

drm/i915/hdcp: Move common message filling function to its own file

Create a new file intel_hdcp_gsc_message that contain functions
which fill the hdcp messages we send to gsc cs this refactor will
help us reuse code for Xe later on

--v2
-add the missed file for proper build

--v3
-use forward declarations instead of #includes [Jani]

--v4
-move linux/err.h to intel_hdcp_gsc_message.c from
intel_hdcp_gsc_message.h [Jani]

--v5
-move linux include on top of drm includes [Uma]

Signed-off-by: Suraj Kandpal <suraj.kandpal@intel.com>
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Acked-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Uma Shankar <uma.shankar@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231009095537.653619-3-suraj.kandpal@intel.com

drm/i915/hdcp: Move checks for gsc health status

Move checks for gsc components required for HDCP 2.2
to work into intel_hdcp_gsc.c. This will also help
with XE refactor on HDCP's side.

Signed-off-by: Suraj Kandpal <suraj.kandpal@intel.com>
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Signed-off-by: Uma Shankar <uma.shankar@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231009095537.653619-2-suraj.kandpal@intel.com

drm/i915/display: Use correct method to free crtc_state

Even though there is no leaking of resource here lets
just use the correct method to free crtc_state

Signed-off-by: Suraj Kandpal <suraj.kandpal@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231010183101.704439-3-suraj.kandpal@intel.com

drm/i915/display: Use intel_crtc_destroy_state instead kfree

intel_encoder_current_mode() seems to leak some resource because
it uses kfree instead of intel_crtc_destroy_state let us fix that.

Signed-off-by: Suraj Kandpal <suraj.kandpal@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231010183101.704439-2-suraj.kandpal@intel.com

drm/tegra: gem: Do not return NULL in tegra_bo_mmap()

It's confusing for a function to return NULL and ERR_PTR()-encoded error
codes on failure. Make sure we only ever return the latter since that's
what callers already expect.

Reported-by: Sui Jingfeng <suijingfeng@loongson.cn>
Reviewed-by: Sui Jingfeng <suijingfeng@loongson.cn>
Signed-off-by: Thierry Reding <treding@nvidia.com>
Link: https://patchwork.freedesktop.org/patch/msgid/ZSVuVcqdGfGtQIQj@orome.fritz.box

drm/tegra: Zero-initialize iosys_map

UBSAN reports an invalid load for bool, as the iosys_map is read
later without being initialized. Zero-initialize it to avoid this.

Reported-by: Ashish Mhetre <amhetre@nvidia.com>
Signed-off-by: Mikko Perttunen <mperttunen@nvidia.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230901115910.701518-2-cyndis@kapsi.fi

gpu: host1x: Correct allocated size for contexts

Original implementation over allocates the memory size for the
contexts list. The size of memory for the contexts list is based
on the number of iommu groups specified in the device tree.

Fixes: 8aa5bcb61612 ("gpu: host1x: Add context device management code")
Signed-off-by: Johnny Liu <johnliu@nvidia.com>
Signed-off-by: Mikko Perttunen <mperttunen@nvidia.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230901115910.701518-1-cyndis@kapsi.fi

gpu: host1x: Syncpoint interrupt sharding

Support sharded syncpoint interrupts on Tegra234+. This feature
allows specifying one of eight interrupt lines for each syncpoint
to lower processing latency of syncpoint threshold
interrupts.

Signed-off-by: Mikko Perttunen <mperttunen@nvidia.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230901114008.672433-1-cyndis@kapsi.fi

gpu: host1x: Enable system suspend callbacks

With the previous CDMA stop fix, executing runtime PM ops around
system suspend now makes channel submissions work after system
suspend, so do that.

Signed-off-by: Mikko Perttunen <mperttunen@nvidia.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230901111510.663401-3-cyndis@kapsi.fi

gpu: host1x: Stop CDMA before suspending

Before going into suspend, wait all CDMA to go idle and stop it.
This will ensure no channel is still active while we enter
suspend, and ensures the driver doesn't think that CDMA is still
active when coming back from suspend (as HW state has been reset).

Signed-off-by: Mikko Perttunen <mperttunen@nvidia.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230901111510.663401-2-cyndis@kapsi.fi

gpu: host1x: Add locking in channel allocation

Add locking around channel allocation to avoid race conditions.

Signed-off-by: Mikko Perttunen <mperttunen@nvidia.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230901111510.663401-1-cyndis@kapsi.fi

drm/tegra: hub: Increase buffer size to ensure all possible values can be stored

When converting from int to string, we must allow for up to 10-chars (2147483647).

Fixes the following W=1 kernel build warning(s):

 drivers/gpu/drm/tegra/hub.c: In function ‘tegra_display_hub_probe’:
 drivers/gpu/drm/tegra/hub.c:1106:47: warning: ‘%u’ directive output may be truncated writing between 1 and 10 bytes into a region of size 4 [-Wformat-truncation=]
 drivers/gpu/drm/tegra/hub.c:1106:42: note: directive argument in the range [0, 4294967294]
 drivers/gpu/drm/tegra/hub.c:1106:17: note: ‘snprintf’ output between 6 and 15 bytes into a destination of size 8

Signed-off-by: Lee Jones <lee@kernel.org>
Signed-off-by: Thierry Reding <treding@nvidia.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230824073710.2677348-16-lee@kernel.org

drm/tegra: Remove two unused function declarations

Commit 776dc3840367 ("drm/tegra: Move subdevice infrastructure to host1x")
removed the implementation but not the declaration.

Signed-off-by: Yue Haibing <yuehaibing@huawei.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230809030226.3412-1-yuehaibing@huawei.com

drm/tegra: Remove surplus else after return

else is not generally useful after return

Signed-off-by: Sui Jingfeng <suijingfeng@loongson.cn>
Signed-off-by: Thierry Reding <treding@nvidia.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230626143331.640454-2-suijingfeng@loongson.cn

drm/i915: enable W=1 warnings by default

We enable a bunch more compiler warnings than the kernel
defaults. However, they've drifted to become a unique set of warnings,
and have increasingly fallen behind from the W=1 set.

Align with the W=1 warnings from scripts/Makefile.extrawarn for clarity,
by copy-pasting them with s/KBUILD_CFLAGS/subdir-ccflags-y/ to make it
easier to compare in the future.

Some of the -Wextra warnings do need to be disabled, just like in
Makefile.extrawarn, but take care to not disable them for W=2 or W=3
builds, depending on the warning.

v2: Add back some -Wextra warning disables (Nathan)

Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Nick Desaulniers <ndesaulniers@google.com>
Cc: Nathan Chancellor <nathan@kernel.org>
Cc: Masahiro Yamada <masahiroy@kernel.org>
Reviewed-by: Nathan Chancellor <nathan@kernel.org>
Reviewed-by: Masahiro Yamada <masahiroy@kernel.org>
[Final s/KBUILD_CFLAGS/subdir-ccflags-y/ fix while applying]
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/69a812273091b6535ddc7f9346289d71bb30f43d.1697009258.git.jani.nikula@intel.com

drm/i915: drop -Wall and related disables from cflags as redundant

The kernel top level Makefile, and recently scripts/Makefile.extrawarn,
have included -Wall, and the disables -Wno-format-security and
$(call cc-disable-warning,frame-address,) for a very long time. They're
redundant in our local subdir-ccflags-y and can be dropped.

Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Nick Desaulniers <ndesaulniers@google.com>
Cc: Nathan Chancellor <nathan@kernel.org>
Cc: Masahiro Yamada <masahiroy@kernel.org>
Reviewed-by: Nathan Chancellor <nathan@kernel.org>
Acked-by: Nick Desaulniers <ndesaulniers@google.com>
Reviewed-by: Masahiro Yamada <masahiroy@kernel.org>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/5ab754ddc2e342c75deb8476275984918e573beb.1697009258.git.jani.nikula@intel.com

drm/i915/cx0: Only clear/set the Pipe Reset bit of the PHY Lanes Owned

Currently, with MFD/pin assignment D, the driver clears the pipe reset bit
of lane 1 which is not owned by display. This causes the display
to block S0iX.

By not clearing this bit for lane 1 and keeping whatever default, S0ix
started to work. This is already what the driver does at the end
of the phy lane reset sequence (Step#8)

Bspec: 65451
Fixes: 619a06dba6fa ("drm/i915/mtl: Reset only one lane in case of MFD")
Cc: Mika Kahola <mika.kahola@intel.com>
Cc: Gustavo Sousa <gustavo.sousa@intel.com>
Signed-off-by: Khaled Almahallawy <khaled.almahallawy@intel.com>
Reviewed-by: Gustavo Sousa <gustavo.sousa@intel.com>
Signed-off-by: Radhakrishna Sripada <radhakrishna.sripada@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231005001310.154396-1-khaled.almahallawy@intel.com

drm/i915/gvt: move structs intel_gvt_irq_info and intel_gvt_irq_map to interrupt.c

Structs intel_gvt_irq_info and intel_gvt_irq_map are not used outside of
interrupt.c. Hide them, and reduce includes.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Zhenyu Wang <zhenyuw@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230926121904.499888-4-jani.nikula@intel.com

drm/i915/gvt: move struct engine_mmio to mmio_context.c

struct engine_mmio is not used outside of mmio_context.c. Hide it, and
reduce includes.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Zhenyu Wang <zhenyuw@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230926121904.499888-3-jani.nikula@intel.com

drm/i915/gvt: don't include gvt.h from intel_gvt_mmio_table.h

intel_gvt_mmio_table.c has no need to include the massive
gvt.h. Simplify.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Zhenyu Wang <zhenyuw@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230926121904.499888-2-jani.nikula@intel.com

drm/i915/gvt: remove unused to_gvt() and reduce includes

gvt.h has no need to include i915_drv.h once the unused to_gvt() has
been removed.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Zhenyu Wang <zhenyuw@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230926121904.499888-1-jani.nikula@intel.com

drm/i915/xe2lpd: update the scaler feature capability

Update the number of scalers per pipe based on the display
capabilities reported.

v1: define the field values instead of the magic number (JaniN)

Bspec: 71161
Signed-off-by: Vinod Govindapillai <vinod.govindapillai@intel.com>
Reviewed-by: Jouni Högander <jouni.hogander@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231001113155.80659-4-vinod.govindapillai@intel.com
Signed-off-by: Jouni Högander <jouni.hogander@intel.com>

drm/i915/xe2lpd: update the dsc feature capability

Update the global dsc flag based on the display capabilities
reported.

v1: define the field values instead of the magic number (JaniN)

Bspec: 71161
Signed-off-by: Vinod Govindapillai <vinod.govindapillai@intel.com>
Reviewed-by: Jouni Högander <jouni.hogander@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231001113155.80659-3-vinod.govindapillai@intel.com
Signed-off-by: Jouni Högander <jouni.hogander@intel.com>

drm/i915/xe2lpd: display capability register definitions

Register definitions to track the reported scalable display
feature configurations

Bspec: 71161
Signed-off-by: Vinod Govindapillai <vinod.govindapillai@intel.com>
Reviewed-by: Jouni Högander <jouni.hogander@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231001113155.80659-2-vinod.govindapillai@intel.com
Signed-off-by: Jouni Högander <jouni.hogander@intel.com>

drm/bridge_connector: implement oob_hotplug_event

Implement the oob_hotplug_event() callback. Translate it to the HPD
notification sent to the HPD bridge in the chain.

Reviewed-by: Janne Grunau <j@jannau.net>
Link: https://lore.kernel.org/r/20231009174048.2695981-4-dmitry.baryshkov@linaro.org
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20231009174048.2695981-4-dmitry.baryshkov@linaro.org

drm/bridge_connector: stop filtering events in drm_bridge_connector_hpd_cb()

In some cases the bridge drivers would like to receive hotplug events
even in the case new status is equal to the old status. In the DP case
this is used to deliver "attention" messages to the DP host. Stop
filtering the events in the drm_bridge_connector_hpd_cb() and let
drivers decide whether they would like to receive the event or not.

Reviewed-by: Janne Grunau <j@jannau.net>
Link: https://lore.kernel.org/r/20231009174048.2695981-3-dmitry.baryshkov@linaro.org
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20231009174048.2695981-3-dmitry.baryshkov@linaro.org

drm: Add HPD state to drm_connector_oob_hotplug_event()

In some implementations, such as the Qualcomm platforms, the display
driver has no way to query the current HPD state and as such it's
impossible to distinguish between disconnect and attention events.

Add a parameter to drm_connector_oob_hotplug_event() to pass the HPD
state.

Also push the test for unchanged state in the displayport altmode driver
into the i915 driver, to allow other drivers to act upon each update.

Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Heikki Krogerus <heikki.krogerus@linux.intel.com>
Link: https://lore.kernel.org/r/20231009174048.2695981-2-dmitry.baryshkov@linaro.org
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20231009174048.2695981-2-dmitry.baryshkov@linaro.org

drm/i915: Add wrapper for getting display step

Add a wrapper around intel_step_name that takes in driver data as an
argument. This wrapper will help maintain compatibility with the
proposed xe driver.

Signed-off-by: Chaitanya Kumar Borah <chaitanya.kumar.borah@intel.com>
Signed-off-by: Animesh Manna <animesh.manna@intel.com>
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Signed-off-by: Uma Shankar <uma.shankar@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231003065211.1052385-1-chaitanya.kumar.borah@intel.com

Merge drm/drm-next into drm-misc-next

Updating drm-misc-next to the state of Linux v6.6-rc2.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>

drm/i915: More use of GT specific print helpers

Update a bunch of GT related print messages in non-GT files to use the
GT specific helpers.

Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
Reviewed-by: Andi Shyti <andi.shyti@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231009183802.673882-3-John.C.Harrison@Intel.com

drm/i915/gt: More use of GT specific print helpers

A bunch of print messages got missed in the update to using sub-system
specific helpers. So update those.

Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
Reviewed-by: Andi Shyti <andi.shyti@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231009183802.673882-2-John.C.Harrison@Intel.com

drm/i915/display: Free crtc_state in verify_crtc_state

Free hw_crtc_state in verify_crtc_state after we are done using
this or else it's just a resource leak.

Fixes: 2745bdda2095 ("drm/i915: Stop clobbering old crtc state during state check")
Signed-off-by: Suraj Kandpal <suraj.kandpal@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231010053208.691260-1-suraj.kandpal@intel.com

drm/panel: ltk050h3146w: add support for Leadtek LTK050H3148W-CTA6 variant

The LTK050H3148W-CTA6 is a 5.0" 720x1280 DSI display, whose driving
controller is a Himax HX8394-F, slightly different from LTK050H3146W by
its init sequence, mode details and mode flags.

Cc: Quentin Schulz <foss+kernel@0leil.net>
Signed-off-by: Klaus Goger <klaus.goger@theobroma-systems.com>
Signed-off-by: Quentin Schulz <quentin.schulz@theobroma-systems.com>
Reviewed-by: Paul Kocialkowski <paul.kocialkowski@bootlin.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20220131164723.714836-2-quentin.schulz@theobroma-systems.com

dt-bindings: ltk050h3146w: add compatible for LTK050H3148W-CTA6 variant

The LTK050H3148W-CTA6 is a 5.0" 720x1280 DSI display, whose driving
controller is a Himax HX8394-F, slightly different from LTK050H3146W by
its init sequence, mode details and mode flags.

Cc: Quentin Schulz <foss+kernel@0leil.net>
Signed-off-by: Quentin Schulz <quentin.schulz@theobroma-systems.com>
Reviewed-by: Paul Kocialkowski <paul.kocialkowski@bootlin.com>
Acked-by: Rob Herring <robh@kernel.org>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20220131164723.714836-3-quentin.schulz@theobroma-systems.com

drm/panel: ltk050h3146w: add mipi_dsi_device.mode_flags to of_match_data

To prepare for a new display to be supported by this driver which has a
slightly different set of DSI mode related flags, let's move the
currently hardcoded mode flags to the .data field of of_device_id
structure.

Cc: Quentin Schulz <foss+kernel@0leil.net>
Signed-off-by: Quentin Schulz <quentin.schulz@theobroma-systems.com>
Reviewed-by: Paul Kocialkowski <paul.kocialkowski@bootlin.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20220131164723.714836-1-quentin.schulz@theobroma-systems.com

drm/panel/panel-tpo-tpg110: fix a possible null pointer dereference

In tpg110_get_modes(), the return value of drm_mode_duplicate() is
assigned to mode, which will lead to a NULL pointer dereference on
failure of drm_mode_duplicate(). Add a check to avoid npd.

Signed-off-by: Ma Ke <make_ruc2021@163.com>
Reviewed-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://lore.kernel.org/r/20231009090446.4043798-1-make_ruc2021@163.com
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20231009090446.4043798-1-make_ruc2021@163.com

drm/panel: fix a possible null pointer dereference

In versatile_panel_get_modes(), the return value of drm_mode_duplicate()
is assigned to mode, which will lead to a NULL pointer dereference
on failure of drm_mode_duplicate(). Add a check to avoid npd.

Signed-off-by: Ma Ke <make_ruc2021@163.com>
Reviewed-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://lore.kernel.org/r/20231007033105.3997998-1-make_ruc2021@163.com
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20231007033105.3997998-1-make_ruc2021@163.com

drm/i915: Fix VLV color state readout

VLV was missed when the color.get_config() hook was added.
Remedy that.

Not really sure what the final plan here was since a bunch of
color related readout was left in intel_display.c anyway,
but that's for anothr day to figure out...

Cc: Jani Nikula <jani.nikula@intel.com>
Fixes: 9af09dfcdfa1 ("drm/i915/color: move pre-SKL gamma and CSC enable read to intel_color")
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231009145828.12960-1-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>

drm/i915/guc: Enable WA 14018913170

The GuC handles the WA, the KMD just needs to set the flag to enable
it on the appropriate platforms.

Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Reviewed-by: Vinay Belgaumkar <vinay.belgaumkar@intel.com>
Reviewed-by: Andi Shyti <andi.shyti@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231006013553.1339418-1-John.C.Harrison@Intel.com

drm/rockchip: dsi: Use devm_platform_get_and_ioremap_resource()

Convert platform_get_resource(), devm_ioremap_resource() to a single
call to devm_platform_get_and_ioremap_resource(), as this is exactly
what this function does.

Signed-off-by: Yang Li <yang.lee@linux.alibaba.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20230421081303.122452-1-yang.lee@linux.alibaba.com

drm/rockchip: remove redundant of_match_ptr

The driver depends on CONFIG_OF, so it is not necessary to use
of_match_ptr here.

Even for drivers that do not depend on CONFIG_OF, it's almost always
better to leave out the of_match_ptr(), since the only thing it can
possibly do is to save a few bytes of .text if a driver can be used both
with and without it. Hence we remove of_match_ptr.

Signed-off-by: Zhu Wang <wangzhu9@huawei.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20230731125304.87059-1-wangzhu9@huawei.com

drm/rockchip: cdn-dp: Fix some error handling paths in cdn_dp_probe()

cdn_dp_audio_codec_init() can fail. So add some error handling.

If component_add() fails, the previous cdn_dp_audio_codec_init() call
should be undone, as already done in the remove function.

Fixes: 88582f564692 ("drm/rockchip: cdn-dp: Don't unregister audio dev when unbinding")
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/8494a41602fadb7439630921a9779640698f2f9f.1693676045.git.christophe.jaillet@wanadoo.fr

drm/rockchip: vop2: Convert to use maple tree register cache

The maple tree register cache is based on a much more modern data structure
than the rbtree cache and makes optimisation choices which are probably
more appropriate for modern systems than those made by the rbtree cache.

Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20231001-drm-rockchip-maple-v1-1-ca396ab75be7@kernel.org

drm/rockchip: vop2: Demote message in mod_supported to drm_dbg_kms

Checking if a modifier is supported by a plane is normal behavior. It is
normal that a plane may not support certain modifiers. Failing the check
doesn't justify an error message in the kernel log and may mislead
users.

Demote the error message to drm_dbg_kms to only print the message if the
respective debug messages are enabled. This is similar to the behavior
in rockchip_drm_vop.c.

Signed-off-by: Michael Tretter <m.tretter@pengutronix.de>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20231009103753.830458-1-m.tretter@pengutronix.de

drm/i915/uapi: fix doc typos

Correct typo of "its".
Add commas for clarity.
Capitalize L3.

Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Cc: intel-gfx@lists.freedesktop.org
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Maxime Ripard <mripard@kernel.org>
Cc: Thomas Zimmermann <tzimmermann@suse.de>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: dri-devel@lists.freedesktop.org
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231008214942.28439-1-rdunlap@infradead.org

Documentation/gpu: fix Panfrost documentation build warnings

Fix issues revealed by `make htmldocs` after adding Panfrost DRM
documentation file.

Signed-off-by: Adrián Larumbe <adrian.larumbe@collabora.com>
Fixes: f11b0417eec2 ("drm/panfrost: Add fdinfo support GPU load metrics")
Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202310030917.Txzlpoeq-lkp@intel.com
Reviewed-by: Steven Price <steven.price@arm.com>
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231005141239.132783-1-adrian.larumbe@collabora.com

drm/i915/guc: Annotate struct ct_incoming_msg with __counted_by

Prepare for the coming implementation by GCC and Clang of the __counted_by
attribute. Flexible array members annotated with __counted_by can have
their accesses bounds-checked at run-time via CONFIG_UBSAN_BOUNDS (for
array indexing) and CONFIG_FORTIFY_SOURCE (for strcpy/memcpy-family
functions).

As found with Coccinelle[1], add __counted_by for struct ct_incoming_msg.

Cc: Jani Nikula <jani.nikula@linux.intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Cc: David Airlie <airlied@gmail.com>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: "Gustavo A. R. Silva" <gustavoars@kernel.org>
Cc: John Harrison <John.C.Harrison@Intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: Michal Wajdeczko <michal.wajdeczko@intel.com>
Cc: Matt Roper <matthew.d.roper@intel.com>
Cc: intel-gfx@lists.freedesktop.org
Cc: dri-devel@lists.freedesktop.org
Cc: linux-hardening@vger.kernel.org
Link: https://github.com/kees/kernel-tools/blob/trunk/coccinelle/examples/counted_by.cocci
Signed-off-by: Kees Cook <keescook@chromium.org>
Reviewed-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Reviewed-by: Andi Shyti <andi.shyti@linux.intel.com>
Signed-off-by: Andi Shyti <andi.shyti@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231006201744.work.135-kees@kernel.org

accel/habanalabs/gaudi2: fix spmu mask creation

event_types_num received from the user can be 0. In that case, the
event_mask should be 0.

In addition, to create a correct mask we need to match the number
of event types to the bit location such that bit 0 represents a single
event type, bit 1 represents 2 types and so on.

Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
Reviewed-by: Ofir Bitton <obitton@habana.ai>

accel/habanalabs/gaudi2: perform hard-reset upon PCIe AXI drain event

Non-completed transactions from PCIe towards the device are handled by
the AXI drain mechanism. This handling is in the PCIe level, but the
transactions are still there in the device consuming some queues
entries, and therefore the device must be reset.
Modify to perform hard-reset upon PCIe AXI drain events.

Signed-off-by: Tomer Tayar <ttayar@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: fix bug in decoder wait for cs completion

The decoder interrupts are handled in the interrupt context
same as all user interrupts.
In such case, the wait list should be protected by
spin_lock_irqsave in order to avoid deadlock that might happen
with the user submission flow.

Signed-off-by: farah kassabri <fkassabri@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: remove wrong doc for init_phys_pg_pack_from_userptr

The function does not pin the pages so remove that from the inline doc.

Signed-off-by: Dafna Hirschfeld <dhirschfeld@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: add missing debugfs function stubs

Two function stubs were removed in an earlier commit but are now needed
again:

drivers/accel/habanalabs/common/device.c: In function 'hl_device_init':
drivers/accel/habanalabs/common/device.c:2231:14: error: implicit declaration of function 'hl_debugfs_device_init'; did you mean 'drm_debugfs_dev_init'? [-Werror=implicit-function-declaration]
 2231 |         rc = hl_debugfs_device_init(hdev);
drivers/accel/habanalabs/common/device.c:2367:9: error: implicit declaration of function 'hl_debugfs_device_fini'; did you mean 'hl_debugfs_remove_file'? [-Werror=implicit-function-declaration]
 2367 |         hl_debugfs_device_fini(hdev);
      |         ^~~~~~~~~~~~~~~~~~~~~~

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: minor cosmetic update to habanalabs.h

- Update copyright years
- Align fields in struct hl_userptr
- Fix comments

Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
Reviewed-by: Ofir Bitton <obitton@habana.ai>

accel/habanalabs/gaudi: remove define used for simulator

We don't support simulator in upstream.

Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
Reviewed-by: Ofir Bitton <obitton@habana.ai>

accel/habanalabs: remove leftover code

This code was added as part of a bigger feature which was never
upstreamed, so remove this code.

Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
Reviewed-by: Ofir Bitton <obitton@habana.ai>

accel/habanalabs: print device name when it is removed

Notifies the user which device was removed. It is important in
a server with multiple devices.

Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
Reviewed-by: Ofir Bitton <obitton@habana.ai>

accel/habanalabs: remove unused field

flags in struct wait_interrupt_data is not used anywhere so remove it.

Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
Reviewed-by: Ofir Bitton <obitton@habana.ai>

accel/habanalabs/gaudi: remove unused structure definition

struct gaudi_nic_status is not used anywhere in the code.

Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
Reviewed-by: Ofir Bitton <obitton@habana.ai>

accel/habanalabs: change Greco to Gaudi2

Greco was not upstreamed so no point of mentioning it here.

Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
Reviewed-by: Ofir Bitton <obitton@habana.ai>

accel/habanalabs: minor cosmetics update to trace file

- Update copyright years
- Add missing newline at end of file

Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
Reviewed-by: Ofir Bitton <obitton@habana.ai>

accel/habanalabs: minor cosmetics update to cpucp_if.h

- Update copyright years
- Align comments

Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
Reviewed-by: Ofir Bitton <obitton@habana.ai>

accel/habanalabs: trace dma map sgtable

Traces the DMA [un]map_sgtable using the new traces we added.

Signed-off-by: Ohad Sharabi <osharabi@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: add traces for dma mappings

In order to get a full picture of DMA mappings (e.g. to track DMAR
errors), DMA mappings APIs should be covered.

Signed-off-by: Ohad Sharabi <osharabi@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: remove unused asic functions

asic_dma_{un}map_single() asic-specific functions are no longer called
from the common code, so delete these functions.

In addition, delete the gaudi2 implementation as they are also not
called.

Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
Reviewed-by: Ofir Bitton <obitton@habana.ai>

accel/habanalabs: update boot status print

FW shutdown preparation status was added to spec.

Signed-off-by: Ariel Suller <asuller@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: extend preboot timeout when preboot might take longer

There are cases such when FW runs MBIST, that preboot is expected to take
longer than the usual. In such cases the firmware reports status
SECURITY_READY/IN_PREBOOT and we extend the timeout waiting for it.
This is currently implemented for Gaudi2 only.

Signed-off-by: Dafna Hirschfeld <dhirschfeld@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: add fw status SHUTDOWN_PREP

update hl_boot_if.h from specs to include
CPU_BOOT_STATUS_FW_SHUTDOWN_PREP

Signed-off-by: Dafna Hirschfeld <dhirschfeld@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: add debug prints to dump content of SG table for dma-buf

Add debug prints to dump the content of the SG table which is prepared
when the dma-buf map op is called.

Signed-off-by: Tomer Tayar <ttayar@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: add missing offset handling for dma-buf

On devices with virtual device memory (Gaudi2 onwards), user can provide
an offset within an allocated device memory from which he wants to
export a dma-buf object.
The offset value is verified by driver, but it is not taken into
consideration when the importer driver maps the dma-buf and the SG table
it prepared.
Add the missing offset handling.

Signed-off-by: Tomer Tayar <ttayar@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: set hl_dmabuf_priv.device_address only when needed

The device_address member of 'struct hl_dmabuf_priv' is used only when
virtual device memory is not supported and dma-buf is exported from
address.
Set the value of this field only when it is relevant, and add "phys" to
its name so it would be clearer that it can't be a device virtual
address.

Signed-off-by: Tomer Tayar <ttayar@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: fix SG table creation for dma-buf mapping

In some cases the calculated number of required entries for the dma-buf
SG table is wrong. For example, if the page size is larger than both the
dma max segment size of the importer device and from the exported side,
or if the exported size is part of a phys_pg_pack that is composed of
several pages.
In these cases, redundant entries will be added to the SG table.

Modify the method that the number of entries is calculated, and the way
they are prepared.

Signed-off-by: Tomer Tayar <ttayar@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: split user interrupts pending list

Currently driver maintain one list for both pending user interrupts
which seeks to wait till CQ reaches it's target value and also the ones
that seeks to get timestamp records when the CQ reaches it's target
value.
This causes delay in handling the waiters which gets higher priority
than the timestamp records.
In order to solve this, let's split the list into two,
one for each case and each one is protected by it's own spinlock.
Waiters will be handled within the interrupt context first,
then the timestamp records will be set.
Freeing the timestamp related memory will be handled in a workqueue.

Signed-off-by: farah kassabri <fkassabri@habana.ai>
Reviewed-by: Tomer Tayar <ttayar@habana.ai>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: optimize timestamp registration handler

Currently we use dynamic allocation inside the irq handler
in order to allocate free node to be used for the free jobs.

This operation is expensive, especially when we deal with large
burst of events records that get released at the same time.

The alternative is to have pre allocated pool of free nodes
and just fetch nodes from this pool at irq handling time instead
of allocating them.

In case the pool becomes full, then the driver will fallback to
dynamic allocations.

As part of the optimization also update the unregister flow
upon re-using a timestamp record, by making the operation much
simpler and quicker. We already have the record in the registration
flow and now we just seek to re-use with different interrupt.
Therefore, no need to look for buffer according to the user handle.

Signed-off-by: farah kassabri <fkassabri@habana.ai>
Reviewed-by: Tomer Tayar <ttayar@habana.ai>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: fix bug in timestamp interrupt handling

There is a potential race between user thread seeking to re-use
a timestamp record with new interrupt id, while this record is still
in the middle of interrupt handling and it is about to be freed.
Imagine the driver set the record in_use to 0 and only then fill the
free_node information. This might lead to unpleasant scenario where
the new registration thread detects the record as free to use, and
change the cq buff address. That will cause the free_node to get
the wrong buffer address to put refcount to.

Signed-off-by: farah kassabri <fkassabri@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: tiny refactor of hl_map_dmabuf()

alloc_sgt_from_device_pages() includes relatively many parameters, and
in a subsequent change another offset parameter is going to be added.
Using structure fields directly when calling this function, and in
hl_map_dmabuf() it is done twice, makes it a little bit difficult to
understand the meaning of the parameters.
To make it clearer, assign the required values into local variables with
explicit names, and use the variables when calling the function.

Signed-off-by: Tomer Tayar <ttayar@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: export dma-buf only if size/offset multiples of PAGE_SIZE

It is currently allowed for a user to export dma-buf with size and
offset that are not multiples of PAGE_SIZE.
The exported memory is mapped for the importer device, and there it will
be rounded to PAGE_SIZE, leading to actually exporting more than the
user intended to.
To make the user be aware of it, accept only size and offset which are
multiple of PAGE_SIZE.

Signed-off-by: Tomer Tayar <ttayar@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: use exported size from dma_buf and not from phys_pg_pack

The 'exported_size' member in 'struct hl_vm_phys_pg_pack' is used to
keep the exported dma-buf size, to be later used when the buffer is
mapped.
However it is possible that the same phys_pg_pack will be exported more
than once, and independently of when the mapping takes place.
Remove this member from the phys_pg_pack structure, and simply use the
size in the dma-buf object as the exported size when mapping.

Signed-off-by: Tomer Tayar <ttayar@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: always pass exported size to alloc_sgt_from_device_pages()

For Gaudi1 the exported dma-buf is always composed of a single page, and
therefore the exported size is equal to this page's size.
When calling alloc_sgt_from_device_pages(), we pass 0 as the exported
size and internally calculate it as "number of pages * page size".
This makes alloc_sgt_from_device_pages() less clear, because the
exported size parameter is not understood as a restriction on the pages'
size.
Modify to always pass the exported size explicitly.

Signed-off-by: Tomer Tayar <ttayar@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: prevent sending heartbeat before events are enabled

After the heartbeat mechanism is now expanded to be used also
for EQ health check, we shouldn't send heartbeat messages
to FW before driver allow events to be received from FW.

Because if the driver will send two heartbeats before it enables
events to be received from FW, then the EQ health check
will fail and reset the device.

Signed-off-by: farah kassabri <fkassabri@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs/gaudi2: add eq health check using irq

This is the second patch for applying the eq health check mechanism
which will add support for the interrupt flow for gaudi2 asic.

More info about the interrupt mechanism:
set a dedicated msix for the eq error interrupt, and add
interrupt handler for it.
when FW detects some issue with EQ like EQ_FULL, it'll
raise that interrupt and driver should reset the device.
Driver will inform the FW which msix index to use through
the already existing handshake mechanism which will
send msix info message to fw.

Signed-off-by: farah kassabri <fkassabri@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs/gaudi2: handle eq health heartbeat check

Add mechanism for fw eq health check. this will be done using two flows:
using the heartbeat mechanism and raising a dedicated interrupt to
indicate an eq failure like EQ full.
This patch will add implementation for the eq heartbeat for gaudi2 asic.

More info about the heartbeat mechanism:
Expand the heartbeat mechanism to monitor a new event that
will be sent from FW upon receiving heartbeat message.
that way driver can know that the eq is working or not.

Signed-off-by: farah kassabri <fkassabri@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs/gaudi2: print power-mode changes

Print to kernel log any device power mode changes events reported by
the FW.

Signed-off-by: Moti Haimovski <mhaimovski@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: add tsc clock sampling to clock sync info

Add tsc clock to clock sync info, to enable using this clock for
sampling and sync it with device time.

Signed-off-by: Hen Alon <halon@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: fix inline doc typos

Fix two typos

Signed-off-by: Dafna Hirschfeld <dhirschfeld@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

accel/habanalabs: disable events ioctls on control device

Because it is not used and also, for graceful reset to work
those ioctls should run on the compute device.

Signed-off-by: Dafna Hirschfeld <dhirschfeld@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>