Add support for powering off IGPU for switching between
legacy to SMC mode/vice-versa or changing SMC configuration.
The power off can be issued as follows
echo 0 > /dev/nvgpu/igpu0/power
The following steps are done during a poweroff.
1) Deterministic channel idle
2) Acquire write_lock on l->busy semaphore.
3) Wait till power_usage decrements to indicate 0 active jobs.
4) Invoke pm_runtime_put_sync_suspend()
5) Invoke nvgpu_gr_remove_support() to clear existing GR memory.
6) Release write_lock on l->busy
7) Deterministic channel unidle.
Part of the sequence matches that of the gk20a_do_idle code.
The common parts are extracted into new functions
gk20a_block_new_jobs_and_idle() and gk20a_unblock_jobs()
For joint-rail case, the current implementation, does a railgate
and then sets pm_runtime_set_autosuspend_delay(-1) to disable
regular runtime resume/suspend.
Remove clearing of NVGPU_SUPPORT_MIG status during state change
ias it leads to inconsistencies.
Jira NVGPU-6920
Change-Id: I0b3eb3278176122ac061c1e8a94ebfb3c17c3925
Signed-off-by: Debarshi Dutta <ddutta@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nvgpu/+/2578501
Reviewed-by: svc_kernel_abi <svc_kernel_abi@nvidia.com>
Reviewed-by: Vijayakumar Subbu <vsubbu@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: Antony Clince Alex <aalex@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
GVS: Gerrit_Virtual_Submit
Remove static class definition and registration for iGPU and dGPU.
Create the class dynamically in gk20a_user_init() and setup the callback
function to create devnode name based on GPU type.
For now add nvgpu_pci_devnode() callback for dGPU that sets correct
dev node path for dGPUs. For iGPU, Android apparently does not honor dev
node path set in callback and hence override the device name for iGPU
with function nvgpu_devnode().
Destroy the class in gk20a_user_deinit().
This will overall be helpful in adding multiple classes and dev nodes
for each GPU instance in MIG mode.
Set GPU device pointer as the parent of new devices created with
device_create(). This is helpful in getting GPU device name in
callback function nvgpu_pci_devnode().
Update functions to not pass class structure and interface names :
nvgpu_probe()
gk20a_user_init()
gk20a_user_deinit()
nvgpu_remove()
Remove static interface name format like INTERFACE_NAME since it is no
longer needed.
Update GK20A_NUM_CDEVS to 10 since there are 10 dev nodes per GPU right
now.
Jira NVGPU-5648
Change-Id: I5d41db5a0f87fa4a558297fb4135a9fbfcd51080
Signed-off-by: Deepak Nibade <dnibade@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nvgpu/+/2423492
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
ifdef function prototypes with CONFIG_* defines. This fixes MISRA rule
8.6 violations which complain about undefined functions.
Also moved nvgpu_channel_get_from_file prototype to ioctl_channel.h &
nvgpu_probe to driver_common.h as those are linux specific. Define
nvgpu_init_soc_vars in posix/soc.c as it is implemented in QNX.
JIRA NVGPU-3873
Change-Id: I5d2b238e1b5d1318867cd2416ac5f03cc6ab7c6a
Signed-off-by: Sagar Kamble <skamble@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/2196794
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
Reviewed-by: Deepak Nibade <dnibade@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Alex Waterman <alexw@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
For safety build, nvgpu driver should enter SW quiesce state
in case an uncorrectable error has occurred. In this state, any
activity on the GPU should be prevented, without powering off the GPU.
Also, a minimal set of operations should be used to enter SW quiesce
state.
Entering SW quiesce state does the following:
- set sw_quiesce_pending: when this flag is set, interrupt
handlers exit after masking interrupts. This should help mitigate
an interrupt storm.
- wake up thread to complete quiescing.
The thread performs the following:
- set NVGPU_DRIVER_IS_DYING to prevent allocation of new resources
- disable interrupts
- disable fifo scheduling
- preempt all runlists
- set error notifier for all active channels
Note: for channels with usermode submit enabled, userspace can
still ring doorbell, but this will not trigger any work on
engines since fifo scheduling is disabled.
Jira NVGPU-3493
Change-Id: I639a32da754d8833f54dcec1fa23135721d8d89a
Signed-off-by: Thomas Fleury <tfleury@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/2172391
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Usermode submit needs to access the usermode region of registers from
userspace. Store the start address of register resource in struct
nvgpu_os_linux to be used in remap to userspace.
Bug 200145225
Change-Id: I3796b6bf67942af0cc16c86accb82a013032bfc8
Signed-off-by: Konsta Holtta <kholtta@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1811838
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
The power rail of dGPU is managed with help of a set of
GPIOs. Using those GPIOs add an interface to power off and
power on dGPU.
Before dGPU is powered off, new work is blocked by setting
NVGPU_DRIVER_IS_DYING and current jobs are allowed to finish
by waiting for gpu to be idle.
The tegra PCIe controller driver provided APIs
tegra_pcie_attach_controller() and tegra_pcie_detach_controller()
are used to manage PCIe link shutdown, PCIe refclk management
and PCIe rescan.
JIRA NVGPU-1100
Change-Id: Ifae5b81535f40dceca5292a987d3daf6984f3210
Signed-off-by: Nitin Kumbhar <nkumbhar@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1749847
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
While removing nvgpu driver, devm mapped reg mappings
are released on driver_unregister. For iGPU, these
regs are explicitly unmapped with iounmap(). This
results in "Trying to vfree() nonexistent vm area"
warnings on driver removal.
Address this by using devm* variants to map all IO regions
of both iGPU and dGPU and let the driver unregister
release these mappings.
Also, lock out GPU regs in driver removal path.
Bug 1987855
Change-Id: I0388daf90bea3eaf8752255059cfd3ceabf66e7d
Signed-off-by: Nitin Kumbhar <nkumbhar@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1730539
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
