Matthew Brost [Wed, 18 Jan 2023 22:43:56 +0000 (14:43 -0800)]
drm/xe: Invalidate TLB after unbind is complete
This gets tricky as we can't do the TLB invalidation until the unbind
operation is done on the hardware and we can't signal the unbind as
complete until the TLB invalidation is done. To work around this we
create an unbind fence which does a TLB invalidation after unbind is
done on the hardware, signals on TLB invalidation completion, and this
fence is installed in the BO dma-resv slot and installed in out-syncs
for the unbind operation.
Signed-off-by: Matthew Brost <matthew.brost@intel.com> Suggested-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com Suggested-by: Thomas Hellström <thomas.hellstrom@linux.intel.com> Reviewed-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com> Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Matthew Brost [Fri, 20 Jan 2023 17:17:50 +0000 (09:17 -0800)]
drm/xe: Don't process TLB invalidation done in CT fast-path
We can't currently do this due to TLB invalidation done handler
expecting the seqno being received in-order, with the fast-path a TLB
invalidation done could pass one being processed in the slow-path in an
extreme corner case. Remove TLB invalidation done from the fast-path for
now and in a follow up reenable this once the TLB invalidation done
handler can deal with out of order seqno.
Matthew Auld [Mon, 16 Jan 2023 10:46:21 +0000 (10:46 +0000)]
drm/xe/ggtt: fix GGTT scratch usage for DG2
Scratch page is in VRAM, and therefore requires 64K GTT layout. In GGTT
world this just means having 16 consecutive entries, with 64K GTT
alignment for the GTT address of the first entry (also matching physical
alignment). However to keep things simple just dump it into system
memory, like we already do for ppGTT. While we are here, also give it
known default value.
Matthew Auld [Thu, 26 Jan 2023 11:31:34 +0000 (11:31 +0000)]
drm/xe/ggtt: fix alignment usage for DG2
Spec says we need to use 64K VRAM pages for GGTT on platforms like DG2.
In GGTT this just means aligning the GTT address to 64K and ensuring
that we have 16 consecutive entries each pointing to the respective 4K
entry. We already ensure we have 64K pages underneath, so it's just a
case of forcing the GTT alignment.
Matthew Auld [Thu, 19 Jan 2023 12:16:51 +0000 (12:16 +0000)]
drm/xe/ppgtt: fix scratch page usage on DG2
On DG2 when running the xe_vm IGT, the kernel generates loads of CAT
errors and GT resets (sometimes at least). On small-bar systems seems
to trigger a lot more easily (maybe due to difference in allocation
strategy). Appears to be related to scratch, since we seem to use the
64K TLB hint on scratch entries, even though the scratch page is a 4K
vram page. Bumping the scratch page size and physical alignment seems
to fix it. Or at least we no longer hit:
However to keep things simple, so we don't have to deal with 64K TLB
hints, just move the scratch page into system memory on platforms that
require 64K VRAM pages.
Signed-off-by: Matthew Auld <matthew.auld@intel.com> Reviewed-by: Matthew Brost <matthew.brost@intel.com> Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com> Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
If we consider dpt it looks like we can already in theory hit this, if
we run out of vram and stolen vram. It at least seems reasonable to
allow calling this on any object which supports CPU access.
Note this also changes the behaviour with stolen VRAM and suspend, such
that we no longer attempt to migrate stolen objects into system memory.
However nothing in stolen should ever need to be restored (same on
integrated), so should be fine. Also on small-bar systems the stolen
portion is pretty much always non-CPU accessible, and currently pinned
objects use plain memcpy when being moved, which doesn't play nicely.
Signed-off-by: Matthew Auld <matthew.auld@intel.com> Reviewed-by: Matthew Brost <matthew.brost@intel.com> Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com> Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
drm/xe/gt: Enable interrupt while initializing root gt
At present the interrupts are enabled while initializing the last GT.
But this is incorrect for a Multi-GT platform, as root GT initialization
will fail with interrupt disabled. Interrupts are required for
the GuC submission triggered during initialization.
Enable the interrupt during the root GT initialization.
Lucas De Marchi [Fri, 13 Jan 2023 23:09:07 +0000 (15:09 -0800)]
drm/xe: Add documentation for mem_type
mem_type field was added in commit d8b52a02cb40 ("drm/xe: Implement
stolen memory.") to designate the TTM memory type for that mgr. Add
kernel-doc with its description.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com> Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com> Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Lucas De Marchi [Thu, 22 Dec 2022 21:58:12 +0000 (13:58 -0800)]
drm/xe/gt: Fix min() with u32 and u64
Fix the following error while building for 32b:
In file included from ../drivers/gpu/drm/xe/xe_gt.c:6:
../drivers/gpu/drm/xe/xe_gt.c: In function ‘gt_ttm_mgr_init’:
../include/linux/minmax.h:20:35: error: comparison of distinct pointer types lacks a cast [-Werror]
20 | (!!(sizeof((typeof(x) *)1 == (typeof(y) *)1)))
| ^~
Cast it to u64 so size of the second operand matches the first one when
building it for 32 bits.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com> Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com> Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Lucas De Marchi [Thu, 22 Dec 2022 20:18:26 +0000 (12:18 -0800)]
drm/xe: Fix tracepoints on 32b
Leave the types as u64, but cast the pointers to unsigned long before
assigning so the compiler doesn't throw warning about casting a pointer
to integer of different size.
Also, size_t should use %zu, not %ld.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com> Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com> Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Lucas De Marchi [Thu, 22 Dec 2022 20:15:24 +0000 (12:15 -0800)]
drm/xe/mmio: Use non-atomic writeq/readq variant for 32b
writeq() and readq() and other functions working on 64 bit variables
are not provided by 32b arch. For that it's needed to choose between
linux/io-64-nonatomic-hi-lo.h and linux/io-64-nonatomic-lo-hi.h,
spliting the read/write in 2 accesses. For xe driver, it doesn't matter
much, so just choose one and include in xe_mmio.h.
This also removes some ifdef CONFIG_64BIT we had around because of the
missing 64bit functions.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com> Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com> Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Matthew Brost [Thu, 12 Jan 2023 22:25:31 +0000 (17:25 -0500)]
drm/xe/guc: Report submission version of GuC firmware
Starting in 70.6.* GuC firmware the CSS header includes the submission
version, pull this from the CSS header. Prior 70.* versions accidentally
omitted this informatio so hard code to the correct values. This
information will be used by VFs when communicating with the PF.
Signed-off-by: Matthew Brost <matthew.brost@intel.com> Reviewed-by: Philippe Lecluse <philippe.lecluse1@gmail.com> Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
This adds support for stolen memory, with the same allocator as
vram_mgr. This allows us to skip a whole lot of copy-paste,
by re-using parts of xe_ttm_vram_mgr.
The stolen memory may be bound using VM_BIND, so it performs like any
other memory region.
We should be able to map a stolen BO directly using the physical memory
location instead of through GGTT even on old platforms, but I don't know
what the effects are on coherency.
Matthew Brost [Thu, 12 Jan 2023 22:25:14 +0000 (17:25 -0500)]
drm/xe: Take memory ref on kernel job creation
When a job is inflight we may access memory to read the hardware seqno.
All user jobs have VM open which has a ref but kernel jobs do not
require VM so it is possible to not have memory ref. To avoid this, take
a memory ref on kernel job creation.
Rodrigo Vivi [Thu, 12 Jan 2023 22:25:03 +0000 (17:25 -0500)]
drm/xe: Implement a local xe_mmio_wait32
Then, move the i915_utils.h include to its user.
The overall goal is to kill all the usages of the i915_utils
stuff.
Yes, wait_for also depends on <linux/delay.h>, so they go
together to where it is needed. It will be likely needed
anyway directly for udelay or usleep_range.
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com> Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Matthew Brost [Thu, 30 Mar 2023 21:31:57 +0000 (17:31 -0400)]
drm/xe: Introduce a new DRM driver for Intel GPUs
Xe, is a new driver for Intel GPUs that supports both integrated and
discrete platforms starting with Tiger Lake (first Intel Xe Architecture).
The code is at a stage where it is already functional and has experimental
support for multiple platforms starting from Tiger Lake, with initial
support implemented in Mesa (for Iris and Anv, our OpenGL and Vulkan
drivers), as well as in NEO (for OpenCL and Level0).
The new Xe driver leverages a lot from i915.
As for display, the intent is to share the display code with the i915
driver so that there is maximum reuse there. But it is not added
in this patch.
This initial work is a collaboration of many people and unfortunately
the big squashed patch won't fully honor the proper credits. But let's
get some git quick stats so we can at least try to preserve some of the
credits:
Co-developed-by: Matthew Brost <matthew.brost@intel.com> Co-developed-by: Matthew Auld <matthew.auld@intel.com> Co-developed-by: Matt Roper <matthew.d.roper@intel.com> Co-developed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com> Co-developed-by: Francois Dugast <francois.dugast@intel.com> Co-developed-by: Lucas De Marchi <lucas.demarchi@intel.com> Co-developed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Co-developed-by: Philippe Lecluse <philippe.lecluse@intel.com> Co-developed-by: Nirmoy Das <nirmoy.das@intel.com> Co-developed-by: Jani Nikula <jani.nikula@intel.com> Co-developed-by: José Roberto de Souza <jose.souza@intel.com> Co-developed-by: Rodrigo Vivi <rodrigo.vivi@intel.com> Co-developed-by: Dave Airlie <airlied@redhat.com> Co-developed-by: Faith Ekstrand <faith.ekstrand@collabora.com> Co-developed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Co-developed-by: Mauro Carvalho Chehab <mchehab@kernel.org> Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com> Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Dave Airlie [Fri, 8 Dec 2023 06:26:20 +0000 (16:26 +1000)]
Merge tag 'drm-misc-next-2023-12-07' of git://anongit.freedesktop.org/drm/drm-misc into drm-next
drm-misc-next for 6.8:
UAPI Changes:
- Remove Userspace Mode-Setting ioctls
- v3d: New uapi to handle jobs involving the CPU
Cross-subsystem Changes:
Core Changes:
- atomic: Add support for FB-less planes which got reverted a bit
later for lack of IGT tests and userspace code, Dump private objects
state in drm_state_dump.
- dma-buf: Add fence deadline support
- encoder: Create per-encoder debugfs directory, move the bridge chain
file to that directory
Driver Changes:
- Include drm_auth.h in driver that use it but don't include it, Drop
drm_plane_helper.h from drivers that include it but don't use it
- imagination: Plenty of small fixes
- panfrost: Improve interrupt handling at poweroff
- qaic: Convert to persistent DRM devices
- tidss: Support for the AM62A7, a few probe improvements, some cleanups
- v3d: Support for jobs involving the CPU
- bridge:
- Create transparent aux-bridge for DP/USB-C
- lt8912b: Add suspend/resume support and power regulator support
- panel:
- himax-hx8394: Drop prepare, unprepare and shutdown logic, Support
panel rotation
- New panels: BOE BP101WX1-100, Powkiddy X55, Ampire AM8001280G,
Evervision VGG644804, SDC ATNA45AF01
Dave Airlie [Fri, 8 Dec 2023 05:06:04 +0000 (15:06 +1000)]
Merge tag 'drm-intel-next-2023-12-07' of git://anongit.freedesktop.org/drm/drm-intel into drm-next
- Improve display debug msgs and other general clean-ups (Ville, Rahuul)
- PSR fixes and improvements around selective fetch (Jouni, Ville)
- Remove FBC restrictions for Xe2LPD displays (Vinod)
- Skip some timing checks on BXT/GLK DSI transcoders (Ville)
- DP MST Fixes (Ville)
- Correct the input parameter on _intel_dsb_commit (heminhong)
- Fix IP version of the display WAs (Bala)
- DGFX uses direct VBT pin mapping (Clint)
- Proper handling of bool on PIPE_CONF_CHECK macros (Jani)
- Skip state verification with TBT-ALT mod (Mika Kahona)
- General organization of display code for reusage with Xe
(Jouni, Luca, Jani, Maarten)
- Squelch a sparse warning (Jani)
- Don't use "proxy" headers (Andy Shevchenko)
- Use devm_gpiod_get() for all GPIOs (Hans)
- Fix ADL+ tiled plane stride (Ville)
- Use octal permissions in display debugfs (Jani)
Luca Coelho [Fri, 1 Dec 2023 10:00:32 +0000 (12:00 +0200)]
drm/i915: handle uncore spinlock when not available
The uncore code may not always be available (e.g. when we build the
display code with Xe), so we can't always rely on having the uncore's
spinlock.
To handle this, split the spin_lock/unlock_irqsave/restore() into
spin_lock/unlock() followed by a call to local_irq_save/restore() and
create wrapper functions for locking and unlocking the uncore's
spinlock. In these functions, we have a condition check and only
actually try to lock/unlock the spinlock when I915 is defined, and
thus uncore is available.
This keeps the ifdefs contained in these new functions and all such
logic inside the display code.
After reading the code, I think this is what happens:
We have a DSI host defined in the device tree and a DSI peripheral under
that host (i.e. an i2c device using the DSI as data bus doesn't exhibit
this behavior).
The host driver calls mipi_dsi_host_register(), which causes (via a few
functions) mipi_dsi_device_add() to be called for the DSI peripheral. So
now we have a DSI device under the host, but attach hasn't been called.
Normally the probing of the devices continues, and eventually the DSI
peripheral's driver will call mipi_dsi_attach(), attaching the
peripheral.
However, if the host driver's probe encounters an error after calling
mipi_dsi_host_register(), and before the peripheral has called
mipi_dsi_attach(), the host driver will do cleanups and return an error
from its probe function. The cleanups include calling
mipi_dsi_host_unregister().
mipi_dsi_host_unregister() will call two functions for all its DSI
peripheral devices: mipi_dsi_detach() and mipi_dsi_device_unregister().
The latter makes sense, as the device exists, but the former may be
wrong as attach has not necessarily been done.
To fix this, track the attached state of the peripheral, and only detach
from mipi_dsi_host_unregister() if the peripheral was attached.
Note that I have only tested this with a board with an i2c DSI
peripheral, not with a "pure" DSI peripheral.
However, slightly related, the unregister machinery still seems broken.
E.g. if the DSI host driver is unbound, it'll detach and unregister the
DSI peripherals. After that, when the DSI peripheral driver unbound
it'll call detach either directly or using the devm variant, leading to
a crash. And probably the driver will crash if it happens, for some
reason, to try to send a message via the DSI bus.
Tomi Valkeinen [Fri, 3 Nov 2023 13:14:04 +0000 (15:14 +0200)]
drm/framebuffer: Fix use of uninitialized variable
smatch reports:
drivers/gpu/drm/drm_framebuffer.c:654 drm_mode_getfb2_ioctl() error: uninitialized symbol 'ret'.
'ret' is possibly not set when there are no errors, causing the error
above. I can't say if that ever happens in real-life, but in any case I
think it is good to initialize 'ret' to 0.
Tomi Valkeinen [Fri, 3 Nov 2023 13:14:03 +0000 (15:14 +0200)]
drm/drm_file: fix use of uninitialized variable
smatch reports:
drivers/gpu/drm/drm_file.c:967 drm_show_memory_stats() error: uninitialized symbol 'supported_status'.
'supported_status' is only set in one code path. I'm not familiar with
the code to say if that path will always be ran in real life, but
whether that is the case or not, I think it is good to initialize
'supported_status' to 0 to silence the warning (and possibly fix a bug).
Dan Carpenter [Wed, 6 Dec 2023 15:05:15 +0000 (18:05 +0300)]
drm/bridge: nxp-ptn3460: simplify some error checking
The i2c_master_send/recv() functions return negative error codes or
they return "len" on success. So the error handling here can be written
as just normal checks for "if (ret < 0) return ret;". No need to
complicate things.
Btw, in this code the "len" parameter can never be zero, but even if
it were, then I feel like this would still be the best way to write it.
Fixes: 914437992876 ("drm/bridge: nxp-ptn3460: fix i2c_master_send() error checking") Suggested-by: Neil Armstrong <neil.armstrong@linaro.org> Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org> Reviewed-by: Robert Foss <rfoss@kernel.org> Signed-off-by: Robert Foss <rfoss@kernel.org> Link: https://patchwork.freedesktop.org/patch/msgid/04242630-42d8-4920-8c67-24ac9db6b3c9@moroto.mountain
Donald Robson [Mon, 4 Dec 2023 15:13:37 +0000 (15:13 +0000)]
drm/imagination: Removed unused functions in pvr_fw_trace
Fixing the warning below due to an unused file level vtable. Removing
only this causes additional warnings for the now unused functions, so
I've removed those too.
>> drivers/gpu/drm/imagination/pvr_fw_trace.c:205:37: warning: 'pvr_fw_trace_group_mask_fops' defined but not used [-Wunused-const-variable=]
205 | static const struct file_operations pvr_fw_trace_group_mask_fops = {
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~
Changes since v1:
- Corrected hash in Fixes tag.
Reported-by: kernel test robot <lkp@intel.com> Closes: https://lore.kernel.org/oe-kbuild-all/202311302054.MVYPxFCE-lkp@intel.com/ Fixes: cb56cd610866 ("drm/imagination: Add firmware trace to debugfs") Signed-off-by: Donald Robson <donald.robson@imgtec.com> Signed-off-by: Maxime Ripard <mripard@kernel.org> Link: https://patchwork.freedesktop.org/patch/msgid/20231204151337.60930-1-donald.robson@imgtec.com
Remove unnecessary include statements for <drm/drm_plane_helper.h>.
The file contains helpers for non-atomic code and should not be
required by most drivers. No functional changes.
drm/simpledrm: Do not include <drm/drm_plane_helper.h>
Remove unnecessary include statements for <drm/drm_plane_helper.h>.
The file contains helpers for non-atomic code and should not be
required by most drivers. No functional changes.
drm/ofdrm: Do not include <drm/drm_plane_helper.h>
Remove unnecessary include statements for <drm/drm_plane_helper.h>.
The file contains helpers for non-atomic code and should not be
required by most drivers. No functional changes.
drm/solomon: Do not include <drm/drm_plane_helper.h>
Remove unnecessary include statements for <drm/drm_plane_helper.h>.
The file contains helpers for non-atomic code and should not be
required by most drivers. No functional changes.
drm/shmobile: Do not include <drm/drm_plane_helper.h>
Remove unnecessary include statements for <drm/drm_plane_helper.h>.
The file contains helpers for non-atomic code and should not be
required by most drivers. No functional changes.
drm/loongson: Do not include <drm/drm_plane_helper.h>
Remove unnecessary include statements for <drm/drm_plane_helper.h>.
The file contains helpers for non-atomic code and should not be
required by most drivers. No functional changes.
drm/amdgpu: Do not include <drm/drm_plane_helper.h>
Remove unnecessary include statements for <drm/drm_plane_helper.h>.
The file contains helpers for non-atomic code and should not be
required by most drivers. No functional changes.
The AGP subsystem supports a user-space interface via /dev/agpgart. It
is only enabled with DRM support for mode setting in user space. (i.e.,
CONFIG_DRM_LEGACY). All of that DRM code has been removed and the option
will go away. Hence remove the AGP frontend.
Modern DRM drivers with kernel mode setting handle AGP support internally.
drm: Remove locking for legacy ioctls and DRM_UNLOCKED
Modern DRM drivers acquire ioctl locks by themselves. Legacy ioctls
for user-space mode setting used to acquire drm_global_mutex. After
removing the ioctl entry points, also remove the locking code. The only
legacy ioctl without global locking was VBLANK_WAIT, which has been
removed as well. Hence remove the related DRM_UNLOCKED flag.
Remove all hooks and calls into code for user-space mode setting from
the DRM core. Without the drivers and ioctl entry points, none of this
is required any longer.
drm: Remove the legacy DRM_IOCTL_MODESET_CTL ioctl
DRM drivers with user-space mode setting have been removed in Linux
v6.3. [1] Now remove the ioctl entry points for these drivers. Invoking
any of the ioctl ops will unconditionally return -EINVAL to user space.
Invoking DRM_IOCTL_MODESET_CTL is different from the other legacy
ioctl ops as it returns 0 even without CONFIG_DRM_LEGACY set. From the
original commit 29935554b384 ("drm: Disallow DRM_IOCTL_MODESET_CTL for
KMS drivers") it is not apparent how or why the operation differs from
the others. It is likely just an oversight in commit 61ae227032e7
("drm: allow removal of legacy codepaths (v4.1)"), which allowed
disabling leagacy ioctls in the first place. Still keep this removal
separate from the other ioctls to allow an easy revert, if necessary.
DRM drivers with user-space mode setting have been removed in Linux
v6.3. [1] Now remove the ioctl entry points for these drivers. Invoking
any of the ioctl ops will unconditionally return -EINVAL to user space.
This has already been the behavior for kernels without CONFIG_DRM_LEGACY
set.
Ville Syrjälä [Mon, 4 Dec 2023 20:24:43 +0000 (22:24 +0200)]
drm/i915: Fix ADL+ tiled plane stride when the POT stride is smaller than the original
plane_view_scanout_stride() currently assumes that we had to pad the
mapping stride with dummy pages in order to align it. But that is not
the case if the original fb stride exceeds the aligned stride used
to populate the remapped view, which is calculated from the user
specified framebuffer width rather than the user specified framebuffer
stride.
Ignore the original fb stride in this case and just stick to the POT
aligned stride. Getting this wrong will cause the plane to fetch the
wrong data, and can lead to fault errors if the page tables at the
bogus location aren't even populated.
TODO: figure out if this is OK for CCS, or if we should instead increase
the width of the view to cover the entire user specified fb stride
instead...
The drm_atomic_print_new_state() already prints private object state via
drm_atomic_private_obj_print_state(). Add private object state dumping
to __drm_state_dump(), so that it is also included into drm_state_dump()
output and into debugfs/dri/N/state file.
projects / mirror_ubuntu-kernels.git / log