linux-stable.git/drivers/accel, branch linux-6.3.y

accel/ivpu: Fix sporadic VPU boot failure

2023-06-14T09:16:57+00:00

commit a3efabee5878b8d7b1863debb78cb7129d07a346 upstream.

Wait for AON bit in HOST_SS_CPR_RST_CLR to return 0 before
starting VPUIP power up sequence, otherwise the VPU device
may sporadically fail to boot.

An error in power up sequence is propagated to the runtime
power management - the device will be in an error state
until the VPU driver is reloaded.

Fixes: 35b137630f08 ("accel/ivpu: Introduce a new DRM driver for Intel VPU")
Cc: stable@vger.kernel.org # 6.3.x
Signed-off-by: Andrzej Kacprowski 
Reviewed-by: Krystian Pradzynski 
Signed-off-by: Stanislaw Gruszka 
Link: https://patchwork.freedesktop.org/patch/msgid/20230607094502.388489-1-stanislaw.gruszka@linux.intel.com
Signed-off-by: Greg Kroah-Hartman

accel/ivpu: Do not trigger extra VPU reset if the VPU is idle

2023-06-14T09:16:57+00:00

commit 9f7e3611f6c828fcb6001c39d8e7a523a4f31525 upstream.

Turning off the PLL and entering D0i3 will reset the VPU so
an explicit IP reset is redundant.
But if the VPU is active, it may interfere with PLL disabling
and to avoid that, we have to issue an additional IP reset
to silence the VPU before turning off the PLL.

Fixes: a8fed6d1e0b9 ("accel/ivpu: Fix power down sequence")
Cc: stable@vger.kernel.org # 6.3.x
Signed-off-by: Andrzej Kacprowski 
Reviewed-by: Stanislaw Gruszka 
Reviewed-by: Jeffrey Hugo 
Signed-off-by: Stanislaw Gruszka 
Link: https://patchwork.freedesktop.org/patch/msgid/20230525103818.877590-1-stanislaw.gruszka@linux.intel.com
Signed-off-by: Greg Kroah-Hartman

accel/ivpu: Do not use mutex_lock_interruptible

2023-06-14T09:16:51+00:00

[ Upstream commit b563e47957af4ff71736c5cc4120a59b055ab583 ]

If we get signal when waiting for the mmu->lock we do not invalidate
current MMU configuration that might result in undefined behavior.

Additionally there is little or no benefit on break waiting for
ipc->lock. In current code base, we keep this lock for short periods.

Fixes: 263b2ba5fc93 ("accel/ivpu: Add Intel VPU MMU support")
Reviewed-by: Krystian Pradzynski 
Reviewed-by: Jeffrey Hugo 
Signed-off-by: Stanislaw Gruszka 
Link: https://patchwork.freedesktop.org/patch/msgid/20230525103818.877590-2-stanislaw.gruszka@linux.intel.com
Signed-off-by: Sasha Levin

accel/ivpu: Reserve all non-command bo's using DMA_RESV_USAGE_BOOKKEEP

2023-06-14T09:16:48+00:00

[ Upstream commit 411360257c1f4fccaa20143098b6d3fcc9d4e4dc ]

Use DMA_RESV_USAGE_BOOKKEEP reservation for buffer objects, except for
command buffers for which we use DMA_RESV_USAGE_WRITE (since VPU can
write to command buffer context save area).

Fixes: 0ec8671837a6 ("accel/ivpu: Fix S3 system suspend when not idle")
Reviewed-by: Jeffrey Hugo 
Signed-off-by: Stanislaw Gruszka 
Link: https://patchwork.freedesktop.org/patch/msgid/20230413063810.3167511-1-stanislaw.gruszka@linux.intel.com
Signed-off-by: Sasha Levin

accel/ivpu: ivpu_ipc needs GENERIC_ALLOCATOR

2023-06-14T09:16:48+00:00

[ Upstream commit 50d30040eb856ff6b0382b4d9dc332dc15597729 ]

Drivers that use the gen_pool*() family of functions should
select GENERIC_ALLOCATOR to prevent build errors like these:

ld: drivers/accel/ivpu/ivpu_ipc.o: in function `gen_pool_free':
include/linux/genalloc.h:172: undefined reference to `gen_pool_free_owner'
ld: drivers/accel/ivpu/ivpu_ipc.o: in function `gen_pool_alloc_algo':
include/linux/genalloc.h:138: undefined reference to `gen_pool_alloc_algo_owner'
ld: drivers/accel/ivpu/ivpu_ipc.o: in function `gen_pool_free':
include/linux/genalloc.h:172: undefined reference to `gen_pool_free_owner'
ld: drivers/accel/ivpu/ivpu_ipc.o: in function `ivpu_ipc_init':
drivers/accel/ivpu/ivpu_ipc.c:441: undefined reference to `devm_gen_pool_create'
ld: drivers/accel/ivpu/ivpu_ipc.o: in function `gen_pool_add_virt':
include/linux/genalloc.h:104: undefined reference to `gen_pool_add_owner'

Fixes: 5d7422cfb498 ("accel/ivpu: Add IPC driver and JSM messages")
Signed-off-by: Randy Dunlap 
Reported-by: kernel test robot 
Link: https://lore.kernel.org/all/202305221206.1TaugDKP-lkp@intel.com/
Cc: Oded Gabbay 
Cc: dri-devel@lists.freedesktop.org
Cc: Jacek Lawrynowicz 
Cc: Stanislaw Gruszka 
Cc: Andrzej Kacprowski 
Cc: Krystian Pradzynski 
Cc: Jeffrey Hugo 
Cc: Daniel Vetter 
Reviewed-by: Jeffrey Hugo 
Signed-off-by: Stanislaw Gruszka 
Link: https://patchwork.freedesktop.org/patch/msgid/20230526044519.13441-1-rdunlap@infradead.org
Signed-off-by: Sasha Levin

accel/ivpu: Remove D3hot delay for Meteorlake

2023-05-24T16:30:02+00:00

[ Upstream commit cb949ce504e829193234e26cb3042bb448465d52 ]

VPU on MTL has hardware optimizations and does not require 10ms
D0 - D3hot transition delay imposed by PCI specification (PCIe
r6.0, sec 5.9.) .

The delay removal is traditionally done by adding PCI ID to
quirk_remove_d3hot_delay() in drivers/pci/quirks.c . But since
we do not need that optimization before driver probe and we
can better specify in the ivpu driver on what (future) hardware
use the optimization, we do not use quirk_remove_d3hot_delay()
for that.

Signed-off-by: Karol Wachowski 
Signed-off-by: Stanislaw Gruszka 
Reviewed-by: Jeffrey Hugo 
Signed-off-by: Jacek Lawrynowicz 
Link: https://patchwork.freedesktop.org/patch/msgid/20230403121545.2995279-1-stanislaw.gruszka@linux.intel.com
Signed-off-by: Sasha Levin

accel/habanalabs: postpone mem_mgr IDR destruction to hpriv_release()

2023-05-24T16:30:01+00:00

[ Upstream commit 2e8e9a895c4589f124a37fc84d123b5114406e94 ]

The memory manager IDR is currently destroyed when user releases the
file descriptor.
However, at this point the user context might be still held, and memory
buffers might be still in use.
Later on, calls to release those buffers will fail due to not finding
their handles in the IDR, leading to a memory leak.
To avoid this leak, split the IDR destruction from the memory manager
fini, and postpone it to hpriv_release() when there is no user context
and no buffers are used.

Signed-off-by: Tomer Tayar 
Reviewed-by: Oded Gabbay 
Signed-off-by: Oded Gabbay 
Signed-off-by: Sasha Levin

accel/ivpu: PM: remove broken ivpu_dbg() statements

2023-05-11T14:16:57+00:00

[ Upstream commit 17ab1ea679be48d905559d968a7622f5f212de6e ]

When CONFIG_PM is disabled, the driver fails to build:

drivers/accel/ivpu/ivpu_pm.c: In function 'ivpu_rpm_get':
drivers/accel/ivpu/ivpu_pm.c:240:84: error: 'struct dev_pm_info' has no member named 'usage_count'
  240 |         ivpu_dbg(vdev, RPM, "rpm_get count %d\n", atomic_read(&vdev->drm.dev->power.usage_count));
      |                                                                                    ^
include/linux/dynamic_debug.h:223:29: note: in definition of macro '__dynamic_func_call_cls'
  223 |                 func(&id, ##__VA_ARGS__);                       \
      |                             ^~~~~~~~~~~
include/linux/dynamic_debug.h:249:9: note: in expansion of macro '_dynamic_func_call_cls'
  249 |         _dynamic_func_call_cls(_DPRINTK_CLASS_DFLT, fmt, func, ##__VA_ARGS__)
      |         ^~~~~~~~~~~~~~~~~~~~~~
include/linux/dynamic_debug.h:272:9: note: in expansion of macro '_dynamic_func_call'
  272 |         _dynamic_func_call(fmt, __dynamic_dev_dbg,              \
      |         ^~~~~~~~~~~~~~~~~~
include/linux/dev_printk.h:155:9: note: in expansion of macro 'dynamic_dev_dbg'
  155 |         dynamic_dev_dbg(dev, dev_fmt(fmt), ##__VA_ARGS__)
      |         ^~~~~~~~~~~~~~~
drivers/accel/ivpu/ivpu_drv.h:65:17: note: in expansion of macro 'dev_dbg'
   65 |                 dev_dbg((vdev)->drm.dev, "[%s] " fmt, #type, ##args);          \
      |                 ^~~~~~~
drivers/accel/ivpu/ivpu_pm.c:240:9: note: in expansion of macro 'ivpu_dbg'
  240 |         ivpu_dbg(vdev, RPM, "rpm_get count %d\n", atomic_read(&vdev->drm.dev->power.usage_count));
      |         ^~~~~~~~

It would be possible to rework these statements to only conditionally print
the reference counter, or to make the driver depend on CONFIG_PM, but my
impression is that these are not actually needed at all if the driver generally
works, or they could be put back when required. Just remove all four of these
to make the driver build in all configurations.

Fixes: 852be13f3bd3 ("accel/ivpu: Add PM support")
Signed-off-by: Arnd Bergmann 
Reviewed-by: Stanislaw Gruszka 
Signed-off-by: Jacek Lawrynowicz 
Link: https://patchwork.freedesktop.org/patch/msgid/20230126163804.3648051-1-arnd@kernel.org
Signed-off-by: Sasha Levin

accel/ivpu: Fix S3 system suspend when not idle

2023-04-05T07:07:26+00:00

Wait for VPU to be idle in ivpu_pm_suspend_cb() before powering off
the device, so jobs are not lost and TDRs are not triggered after
resume.

Fixes: 852be13f3bd3 ("accel/ivpu: Add PM support")
Signed-off-by: Stanislaw Gruszka 
Reviewed-by: Jeffrey Hugo 
Signed-off-by: Jacek Lawrynowicz 
Link: https://patchwork.freedesktop.org/patch/msgid/20230331113603.2802515-3-stanislaw.gruszka@linux.intel.com

accel/ivpu: Add dma fence to command buffers only

2023-04-05T07:06:38+00:00

Currently job->done_fence is added to every BO handle within a job. If job
handle (command buffer) is shared between multiple submits, KMD will add
the fence in each of them. Then bo_wait_ioctl() executed on command buffer
will exit only when all jobs containing that handle are done.

This creates deadlock scenario for user mode driver in case when job handle
is added as dependency of another job, because bo_wait_ioctl() of first job
will wait until second job finishes, and second job can not finish before
first one.

Having fences added only to job buffer handle allows user space to execute
bo_wait_ioctl() on the job even if it's handle is submitted with other job.

Fixes: cd7272215c44 ("accel/ivpu: Add command buffer submission logic")
Signed-off-by: Karol Wachowski 
Signed-off-by: Stanislaw Gruszka 
Reviewed-by: Jeffrey Hugo 
Signed-off-by: Jacek Lawrynowicz 
Link: https://patchwork.freedesktop.org/patch/msgid/20230331113603.2802515-2-stanislaw.gruszka@linux.intel.com