This is adding code to select MIG mode and boot
the GPU with selected mig config.
For testing MIG, after system boots
1. write mig_mode_config by
echo x > /sys/devices/gpu.0/mig_mode_config for igpu
echo x > /sys/devices/./platform/14100000.pcie/pci0001:00/0001:00:00.0/0001:01:00.0/ for dgpu
2. Then run any nvgpu* tests or nvrm_gpu_info.
If the mig_mode need to be changed , note down the supported
configs by "cat mig_mode_config_list" and reboot the system
3. Follow steps 1 and 2.
example output:
"cat mig_mode_config" 2
"cat mig_mode_config_list"
+++++++++ Config list Start ++++++++++
CONFIG_ID : 0 for CONFIG NAME : 2 GPU instances each with 4 GPCs
CONFIG_ID : 1 for CONFIG NAME : 4 GPU instances each with 2 GPCs
CONFIG_ID : 2 for CONFIG NAME : 7 GPU instances - 1 GPU instance with 2
GPCs + 6 GPU instances each with 1 GPC
CONFIG_ID : 3 for CONFIG NAME : 5 GPU instances - 1 GPU instance with 4
GPCs + 4 GPU instances each with 1 GPC
CONFIG_ID : 4 for CONFIG NAME : 4 GPU instances - 1 GPU instance with 2
GPCs + 2 GPU instances each with 1 GPC + 1 GPU instance with 4 GPCs
CONFIG_ID : 5 for CONFIG NAME : 6 GPU instances - 2 GPU instances each
with 2 GPCs + 4 GPU instances each with 1 GPC
CONFIG_ID : 6 for CONFIG NAME : 5 GPU instances - 1 GPU instance with
2 GPCs + 2 GPU instances each with 1 GPC + 2 GPU instances with 2 GPCs
CONFIG_ID : 7 for CONFIG NAME : 5 GPU instances - 2 GPU instances each
with 2 GPCs + 1 GPC instance with 2 GPCs + 2 GPU instances with 1 GPC
CONFIG_ID : 8 for CONFIG NAME : 5 GPU instances - 1 GPC instance with 2
GPCs + 2 GPU instances each with 1 GPC + 2 GPU instances each with 2
GPCs
CONFIG_ID : 9 for CONFIG NAME : 1 GPU instance with 8 GPCs
++++++++++ Config list End +++++++++++
JIRA NVGPU-6633
Change-Id: I3e56f8c836e1ced8753a60f328da63916faa7696
Signed-off-by: dt <dt@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nvgpu/+/2522821
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Lazy bootstrap is a secure iGPU feature where LS falcons(FECS and
GPCCS) are bootstrapped by LSPMU in both cold boot and recovery boot.
As there is no ACR running after boot, we need LSPMU to bootstrap LS
falcons to support recovery.
In absence of LSPMU, ACR will bootstrap LS falcons but recovery is
not supported.
This CL will enable Low secure falcon manager(lsfm) to support
Lazy bootstrap feature. This will allow nvgpu to send cmds
to lspmu to bootstrap LS falcons.
Bug 200709761
Signed-off-by: Ramesh Mylavarapu <rmylavarapu@nvidia.com>
Change-Id: I65d17cf5e07a45c040a9bb75f75cf18eb509cd4f
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nvgpu/+/2506162
Reviewed-by: Seema Khowala <seemaj@nvidia.com>
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
Reviewed-by: svc-mobile-cert <svc-mobile-cert@nvidia.com>
Reviewed-by: Vijayakumar Subbu <vsubbu@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
GVS: Gerrit_Virtual_Submit
1) nvgpu poweron sequence split into two stages:
- nvgpu_early_init() - Initializes the sub units
which are required to be initialized before the grgmr init.
For creating dev node, grmgr init and its dependency unit
needs to move to early stage of GPU power on.
After successful nvgpu_early_init() sequence,
NvGpu can indetify the number of MIG instance required
for each physical GPU.
- nvgpu_finalize_poweron() - Initializes the sub units which
can be initialized at the later stage of GPU power on sequence.
- grmgr init depends on the following HAL sub units,
* device - To get the device caps.
* priv_ring - To get the gpc count and other
MIG config programming.
* fb - MIG config programming.
* ltc - MIG config programming.
* bios, bus, ecc and clk - dependent module of
priv_ring/fb/ltc.
2) g->ops.xve.reset_gpu() should be called before GPU sub unit
initialization. Hence, added g->ops.xve.reset_gpu() HAL in the
early stage of dGPU power on sequence.
3) Increased xve_reset timeout from 100ms to 200ms.
4) Added nvgpu_assert() for gpc_count, gpc_mask and
max_veid_count_per_tsg for identify the GPU boot
device probe failure during nvgpu_init_gr_manager().
JIRA NVGPU-6633
Change-Id: I5d43bf711198e6b3f8eebcec3027ba17c15fc692
Signed-off-by: Lakshmanan M <lm@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nvgpu/+/2521894
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Reviewed-by: Debarshi Dutta <ddutta@nvidia.com>
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
Reviewed-by: svc-mobile-cert <svc-mobile-cert@nvidia.com>
Reviewed-by: Vaibhav Kachore <vkachore@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
GVS: Gerrit_Virtual_Submit
Currently, there are few chip specific erratas present in nvgpu code.
For better traceability of the erratas and corresponding fixes,
introduce flags to indicate existing erratas on a chip. These flags
decide if a corresponding solution is applied to the chip(s).
This patch introduces below functions to handle errata flags:
- nvgpu_init_errata_flags
- nvgpu_set_errata
- nvgpu_is_errata_present
- nvgpu_print_errata_flags
- nvgpu_free_errata_flags
nvgpu_print_errata_flags: print below details of erratas present in chip
1. errata flag name
2. chip where the errata was first discovered
3. short description of the errata
Flags corresponding to erratas present in a chip are set during chip hal
init sequence.
JIRA NVGPU-6510
Change-Id: Id5a8fb627222ac0a585aba071af052950f4de965
Signed-off-by: Vedashree Vidwans <vvidwans@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nvgpu/+/2498095
Reviewed-by: Seema Khowala <seemaj@nvidia.com>
Reviewed-by: Vaibhav Kachore <vkachore@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
GVS: Gerrit_Virtual_Submit
1) NvGpu dev node needs to be created in gpu power on
early stage to avoid latency introduced by udevd.
For creating dev node, device and grmgr init
needs to move to early stage of GPU power on.
After grmgr init, NvGpu can identify the number of MIG
instance required for each physical GPU.
For that, added a new API nvgpu_early_poweron() to handle
early init which is required for before dev node creation.
2) Removed fifo dependency in nvgpu_init_gr_manager()
3) Used get_max_subctx_count() directly to query
the veid/subctx count.
JIRA NVGPU-6633
Change-Id: Ib9d7c3e184c71237b0da9305515ccd8ceda1d5ad
Signed-off-by: Lakshmanan M <lm@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nvgpu/+/2517173
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
All graphics code is under CONFIG_NVGPU_GRAPHICS and all the HALs are
in non-fusa files. In order to support graphics in safety,
CONFIG_NVGPU_GRAPHICS needs to be enabled. But since most of the HALs
are in non-fusa files, this causes huge compilation problem.
Fix this by moving all graphics specific HALs used on gv11b to fusa
files. Graphics specific HALs not used on gv11b remain in non-fusa files
and need not be protected with GRAPHICS config.
Protect call to nvgpu_pmu_save_zbc() also with config
CONFIG_NVGPU_POWER_PG, since it is implemented under that config.
Delete hal/ltc/ltc_gv11b.c since sole function in this file is moved to
fusa file.
Enable nvgpu_writel_loop() in safety build since it is needed for now.
This will be revisited later once requirements are clearer.
Move below CTXSW methods under CONFIG_NVGPU_NON_FUSA for now. Safety
CTXSW ucode does not support these methods. These too will be revisited
later once requirements are clearer.
NVGPU_GR_FALCON_METHOD_PREEMPT_IMAGE_SIZE
NVGPU_GR_FALCON_METHOD_CTXSW_DISCOVER_ZCULL_IMAGE_SIZE
Jira NVGPU-6460
Change-Id: Ia095a04a9ba67126068aa7193f491ea27477f882
Signed-off-by: Deepak Nibade <dnibade@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nvgpu/+/2513675
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Separated set_atomic_mode functionality from
init_fs_state/enable_nvlink and created new
fb gops for set_atomic_mode.
In gpu init sequence, set_atomic_mode is
called after acr_construct_execute to take care
of design changes required for nvgpu-next
architectures.
Updated fb_gv11b_init_test to use set_atomic_mode
gops along with init_fs_state.
Bug 3268664
Change-Id: I1ab9eb21cc4cce77f3325c4e8821a75b6e85fba2
Signed-off-by: Seshendra Gadagottu <sgadagottu@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nvgpu/+/2508095
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
This patch is to disable the clock arbitor for TU104.
TU104 is not a POR for Drive 6.0 so disabling it to easy migration
of clk arb for GA100.
As a first step all the NVRM Clock tests will be skipped by setting
NVGPU_SUPPORT_CLOCK_CONTROLS to false for TU104.
Then clk arbitor will be rewritten for GA100 and enabled back.
This patch implements by adding a new flag NVGPU_CLK_ARB_ENABLED which
holds the status of clk arbitor for each platform and disables them for
TU104
Bug 200699763
Change-Id: I51cd5c7821bdc0b48080c17a70735925b278ddf5
Signed-off-by: absalam <absalam@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nvgpu/+/2515086
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
Reviewed-by: svc-mobile-misra <svc-mobile-misra@nvidia.com>
Reviewed-by: svc-mobile-cert <svc-mobile-cert@nvidia.com>
Reviewed-by: Deepak Nibade <dnibade@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
GVS: Gerrit_Virtual_Submit
- moved reg fields to gk20a
- added os abstract register accessor in nvgpu/io.h
- defined linux register access abstract implementation
- hook up with posix. posix implementation of the register accessor uses
the high 4 bit of address to identify register apertures then call the
according callbacks.
It helps to unify code across OSes.
Bug 2999617
Signed-off-by: Richard Zhao <rizhao@nvidia.com>
Change-Id: Ifcb737e4b4d5b1d8bae310ae50b1ce0aa04f750c
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nvgpu/+/2497937
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
Reviewed-by: Alex Waterman <alexw@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
GVS: Gerrit_Virtual_Submit
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
When force closing the channel, NEXT and CTX_RELOAD bits might be set.
Currently CTX_RELOAD bit is ignored. However, due to this, the channel
created after the erroneous unbind encounters FECS fault.
If the channel is unbound while it is running, fifo unbind error
happens and can lead to unspecified behavior.
By moving CTX_RELOAD to other channel in the TSG, the channel can be
unbound safely. In other cases, if the channel is truly running
something when it is being unbound it should either get
preempted or be handled through engine reset.
Bug 200701444
Change-Id: Iba956544dcaa1144c6064247257c64cbe9a29ae6
Signed-off-by: Sagar Kamble <skamble@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nvgpu/+/2515083
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
Reviewed-by: svc-mobile-misra <svc-mobile-misra@nvidia.com>
Reviewed-by: svc-mobile-cert <svc-mobile-cert@nvidia.com>
Reviewed-by: Deepak Nibade <dnibade@nvidia.com>
Reviewed-by: Alex Waterman <alexw@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
GVS: Gerrit_Virtual_Submit
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
This is one of the steps in restructuring of interrupt code.
- Move ISR logic to common code. This will allow us to add mixed ASIL
error handling levels.
- Modify nonstall ISR to use threaded interrupts. Bottom half of
nonstall ISR will run nonstall operations instead of adding work to
workqueues.
- Remove nonstall workqueue implementation.
JIRA NVGPU-6351
Change-Id: I5f891b0de4b0c34f6ac05522a5da08dc36221aa6
Signed-off-by: Vedashree Vidwans <vvidwans@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nvgpu/+/2467713
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Add below updates to common.gr doxygen:
- Add doxygen comments for APIs that are mentioned in RM SWAD and in
RM-common.gr traceability document.
- Comment about valid ranges for input parameters of bunch of functions.
- Add nvgpu_assert() to ensure correct value is passed as input
parameter to number of functions.
- Add references to relevant functions with @see.
- Update Targets field for unit tests to cover newly doxygenated
functions.
- Update unit test test_gr_init_hal_pd_skip_table_gpc to take care of
new asserts added into some APIs.
Jira NVGPU-6180
Change-Id: Ie889bed96b6428b1fd86dcf30b322944464e9d12
Signed-off-by: Deepak Nibade <dnibade@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nvgpu/+/2469397
(cherry picked from commit 5d7d7e9ce1c4efe836ab842d7962a3aee4e8972f)
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nvgpu/+/2469394
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
Reviewed-by: svc-mobile-misra <svc-mobile-misra@nvidia.com>
Reviewed-by: svc-mobile-cert <svc-mobile-cert@nvidia.com>
Reviewed-by: Alex Waterman <alexw@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
GVS: Gerrit_Virtual_Submit
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
