Most platforms with a Mali-400 or Mali-450 GPU also have support for
changing the GPU clock frequency. Add devfreq support so the GPU clock
rate is updated based on the actual GPU usage when the
"operating-points-v2" property is present in the board.dts.
The actual devfreq code is taken from panfrost_devfreq.c and modified so
it matches what the lima hardware needs:
- a call to dev_pm_opp_set_clkname() during initialization because there
are two clocks on Mali-4x0 IPs. "core" is the one that actually clocks
the GPU so we need to control it using devfreq.
- locking when reading or writing the devfreq statistics because (unlike
than panfrost) we have multiple PP and GP IRQs which may finish jobs
concurrently.
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200319203427.2259891-3-martin.blumenstingl@googlemail.com
heap buffer is used as output of GP and input of PP for
Mali Utgard GPU. Size of heap buffer depends on the task
so is a runtime variable.
Previously we just create a large enough buffer as heap
buffer. Now we add a heap buffer type to be able to
increase the backup memory dynamically when GP fail due
to lack of heap memory.
Reviewed-by: Vasily Khoruzhick <anarsoul@gmail.com>
Tested-by: Andreas Baierl <ichgeh@imkreisrum.de>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200116131157.13346-4-yuq825@gmail.com
Do not need to maintain our own shmem memory management
code as drm_gem_shmem_helpers provides it. And we can
also benifit from the work of others with shared code.
This is also a preparation for implementing buffer madv.
Reviewed-by: Vasily Khoruzhick <anarsoul@gmail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20191010140152.17747-2-yuq825@gmail.com
The authentication can be circumvented, by design, by using the render
node.
From the driver POV there is no distinction between primary and render
nodes, thus we can drop the token.
Cc: Qiang Yu <yuq825@gmail.com>
Cc: lima@lists.freedesktop.org
Cc: David Airlie <airlied@linux.ie>
Cc: Daniel Vetter <daniel@ffwll.ch>
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Qiang Yu <yuq825@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190527081741.14235-6-emil.l.velikov@gmail.com
There is no point to print deferred probe (and its failures to get
resources) as an error. For example getting a regulator causes three
unneeded error messages:
lima 13000000.gpu: failed to get regulator: -517
lima 13000000.gpu: regulator init fail -517
lima 13000000.gpu: Fatal error during GPU init
Also do not print clock rates before the initialization finishes
because they will be duplicated after deferral. Each probe step already
prints error so remove the final error message "Fatal error during GPU
init".
In case of multiple probe tries this would pollute the dmesg.
Fixes: a1d2a63399 ("drm/lima: driver for ARM Mali4xx GPUs")
Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190621162117.22533-2-krzk@kernel.org
After "5918045c4ed4 drm/scheduler: rework job destruction", lima started
to leak memory due to buffers not being destroyed after job execution in
the drm scheduler.
This started happening because the drm scheduler only destroyed buffers
after cancelling the job timeout handler, and for lima this handler was
never started as lima specified a MAX_SCHEDULE_TIMEOUT timeout.
Lima seems to run well in its current state with a real timeout, so to
make it more aligned with the other drivers from now on, let's use a
real default timeout.
This also fixes the observed memory leaks.
The 500ms value was chosen as it is the current value for all other
embedded gpu drivers using drm sched.
Signed-off-by: Erico Nunes <nunes.erico@gmail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190520224229.21111-1-nunes.erico@gmail.com
- Mali 4xx GPUs have two kinds of processors GP and PP. GP is for
OpenGL vertex shader processing and PP is for fragment shader
processing. Each processor has its own MMU so prcessors work in
virtual address space.
- There's only one GP but multiple PP (max 4 for mali 400 and 8
for mali 450) in the same mali 4xx GPU. All PPs are grouped
togather to handle a single fragment shader task divided by
FB output tiled pixels. Mali 400 user space driver is
responsible for assign target tiled pixels to each PP, but mali
450 has a HW module called DLBU to dynamically balance each
PP's load.
- User space driver allocate buffer object and map into GPU
virtual address space, upload command stream and draw data with
CPU mmap of the buffer object, then submit task to GP/PP with
a register frame indicating where is the command stream and misc
settings.
- There's no command stream validation/relocation due to each user
process has its own GPU virtual address space. GP/PP's MMU switch
virtual address space before running two tasks from different
user process. Error or evil user space code just get MMU fault
or GP/PP error IRQ, then the HW/SW will be recovered.
- Use GEM+shmem for MM. Currently just alloc and pin memory when
gem object creation. GPU vm map of the buffer is also done in
the alloc stage in kernel space. We may delay the memory
allocation and real GPU vm map to command submission stage in the
furture as improvement.
- Use drm_sched for GPU task schedule. Each OpenGL context should
have a lima context object in the kernel to distinguish tasks
from different user. drm_sched gets task from each lima context
in a fair way.
mesa driver can be found here before upstreamed:
https://gitlab.freedesktop.org/lima/mesa
v8:
- add comments for in_sync
- fix ctx free miss mutex unlock
v7:
- remove lima_fence_ops with default value
- move fence slab create to device probe
- check pad ioctl args to be zero
- add comments for user/kernel interface
v6:
- fix comments by checkpatch.pl
v5:
- export gp/pp version to userspace
- rebase on drm-misc-next
v4:
- use get param interface to get info
- separate context create/free ioctl
- remove unused max sched task param
- update copyright time
- use xarray instead of idr
- stop using drmP.h
v3:
- fix comments from kbuild robot
- restrict supported arch to tested ones
v2:
- fix syscall argument check
- fix job finish fence leak since kernel 5.0
- use drm syncobj to replace native fence
- move buffer object GPU va map into kernel
- reserve syscall argument space for future info
- remove kernel gem modifier
- switch TTM back to GEM+shmem MM
- use time based io poll
- use whole register name
- adopt gem reservation obj integration
- use drm_timeout_abs_to_jiffies
Cc: Eric Anholt <eric@anholt.net>
Cc: Rob Herring <robh@kernel.org>
Cc: Christian König <ckoenig.leichtzumerken@gmail.com>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: Alex Deucher <alexdeucher@gmail.com>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Rob Clark <robdclark@gmail.com>
Cc: Dave Airlie <airlied@gmail.com>
Signed-off-by: Andreas Baierl <ichgeh@imkreisrum.de>
Signed-off-by: Erico Nunes <nunes.erico@gmail.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Signed-off-by: Marek Vasut <marex@denx.de>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: Simon Shields <simon@lineageos.org>
Signed-off-by: Vasily Khoruzhick <anarsoul@gmail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Rob Herring <robh@kerrnel.org>
Signed-off-by: Eric Anholt <eric@anholt.net>
Link: https://patchwork.freedesktop.org/patch/291200/
