For easy of extensibility to future chips, below data and
functions applicable to T23x are moved to dev-t18x.c
- chip data for supported chips
- set_boot_vec: set up EVP
- set_boot_freqs: set ADSP freq as per DT property
- check_wfi_status: check WFI status from AMISC
- dump_core_state: ADSP core and state dump
Bug 3682950
Change-Id: Iffba56addc9a68ec33de390a379fe725e727cfad
Signed-off-by: Viswanath L <viswanathl@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3107450
GVS: Gerrit_Virtual_Submit <buildbot_gerritrpt@nvidia.com>
Reviewed-by: Dara Ramesh <dramesh@nvidia.com>
- With multiple ADSPs in use, only one of them may configure AMC,
so use amc_not_avlbl flag to skip reg parsing, if not needed
- Unify reg enums for ADSP and AON; this means that reg DT prop
for AON should have a dummy first entry (against AMC)
- Remove adsp_prid from chip data, as AMISC base in reg DT prop
will be offset as per the ADSP instance
Bug 3682950
Bug 4165898
Change-Id: Ic6068aeeb7bbb3540cad36371a6715f8b43ee8b6
Signed-off-by: Viswanath L <viswanathl@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3106899
Reviewed-by: svcacv <svcacv@nvidia.com>
Reviewed-by: Dara Ramesh <dramesh@nvidia.com>
GVS: Gerrit_Virtual_Submit <buildbot_gerritrpt@nvidia.com>
ADSP host driver is enhanced to be multi-instance
capable and reentrant:
- Trailing unique identifier string in compatible DT property, like
"adsp", "adsp1" or "aon", is used to identify the driver instances
- Each probed driver instance is inserted into a global list, from
which the handle can be fetched using 'nvadsp_get_handle' API
(passing the above unique identifier as argument)
- Above unique identifier is also used as name for the DBFS
directory (containing files like adsp_console, adsp_logger, etc.)
- 'nvadsp_get_handle' is the only exported API; all other APIs are
accessible via function pointers within 'struct nvadsp_handle'
- APIs above maintain one-is-to-one correspondence with all
legacy APIs, with the addition of a new argument
'struct nvadsp_handle *' at the beginning
- Legacy APIs continue to be supported, but they are hardwired to
work only if the kernel probes just one driver instance
- All driver files are cleaned up to not use any global state
variables (necessary for reentrancy)
Bug 3682950
Change-Id: Id5db49e861b2f81716ae8352b36b406654da2bbd
Signed-off-by: Viswanath L <viswanathl@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3092701
GVS: Gerrit_Virtual_Submit <buildbot_gerritrpt@nvidia.com>
Reviewed-by: Dara Ramesh <dramesh@nvidia.com>
- Added num_irqs to chip data structure and modified enum to
suit the driver to both AON and ADSP
- Added num_regs to chip data and modified reg bank enum
to suit both AON and ADSP
- Added chipdata variable to indicate if amc is not available or
relevant
- Modified aram_exit to execute only if aram size is not null
- Avoid passing fw name for registration as it currently passing "APE"
- WFI status check logic is dsp specific. Hence separated out into
new API
- Mapping mbx interrupts is specific to AON. Hence added platform device
specific API to map the shared interrupts
- Avoided empty interrupt spamming for level triggered interrupts.
This fix does no harm even for edge triggered interrupts
Bug 4165898
Change-Id: I99cf03c06ed5d290e067b4fd3569febadbda607a
Signed-off-by: Asha Talambedu <atalambedu@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3035268
Reviewed-by: Viswanath L <viswanathl@nvidia.com>
Reviewed-by: Dara Ramesh <dramesh@nvidia.com>
GVS: Gerrit_Virtual_Submit <buildbot_gerritrpt@nvidia.com>
Introduce a new field 'adsp_boot_config_hwmbox' in chip data
for passing boot config to ADSP, replacing the existing
method of writing chip_id into hwmbox1.
The new field presently encodes adsp_os_secload flag. This
may be useful for firmware to do specific settings based on
boot mode, e.g. do AST settings if boot is in non-secure mode.
Bug 3682950
Change-Id: I51e5e0c8f15a8160247a7febb50998defec13d42
Signed-off-by: Viswanath L <viswanathl@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3052180
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Below changes are made to support future
extensibility for new chips:
- Chip data field 'adsp_prid' is added to specify DSP ID
- Chip data field 'adsp_elf' is added to specify ELF file name
- DT property "nvidia,cluster_mem" is added to specify cluster
internal memories; ELF segments will be loaded into these
using intermediate shadow buffers so that unaligned access
is avoided
- "nvidia,adsp_mem" entries are made optional (0x0 in the size
field will indicate bypass)
- Chip extension functions 'set_boot_vec' and 'set_boot_freqs'
are added to support chip specific settings
Bug 4164138
Bug 3682950
Change-Id: I538ae0df57fa199a6eef75441167a9a5ba6924e8
Signed-off-by: Viswanath L <viswanathl@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nvidia/+/2894321
(cherry picked from commit fa7fc9d08af5fab9d0b2c5eb1b9b9a78446ade0e)
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/2971980
Reviewed-by: Sharad Gupta <sharadg@nvidia.com>
GVS: Gerrit_Virtual_Submit <buildbot_gerritrpt@nvidia.com>
There is a decompress_done flag in the adsp OS firmware
that was earlier used to indicate that decompress is done
at the adsp os cold boot
However this flag cannot be part of firmware
as fw needs to authenticated at the sc7 resume as well and
hence the fw should not have modifications
On fw side, the flag is removed. Instead the same information
is preserved across sc7 cycles via HWMBOX5 and this driver
enables the decompress bit to indicate the same to adsp at
cold start
Bug 3491011
Change-Id: I77d1ff6defdf1228f4cd4278cf5bb667df51fad1
Signed-off-by: Uday Gupta <udayg@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nvidia/+/2660706
Tested-by: Asha Talambedu <atalambedu@nvidia.com>
Reviewed-by: svcacv <svcacv@nvidia.com>
Reviewed-by: Asha Talambedu <atalambedu@nvidia.com>
Reviewed-by: svc_kernel_abi <svc_kernel_abi@nvidia.com>
Reviewed-by: Sharad Gupta <sharadg@nvidia.com>
Reviewed-by: Viswanath L <viswanathl@nvidia.com>
GVS: Gerrit_Virtual_Submit
ACAST settings are expected to be done by the driver in backdoor
boot. Provision is added to read the ACAST address ranges and
the SMMU stream ID from DT.
ACAST settings are done either for physical memory, or for SMMU
mapped memory, as per the backdoor boot configuration.
Also, call to ACAST init function is moved from runtime resume
to device probe.
Bug 200745795
Bug 200746669
Bug 200773359
Change-Id: Ia53820968ce4d48966c39d72b4c2106be99815fa
Signed-off-by: Viswanath L <viswanathl@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nvidia/+/2644989
Reviewed-by: svc_kernel_abi <svc_kernel_abi@nvidia.com>
Reviewed-by: Mohan Kumar D <mkumard@nvidia.com>
Reviewed-by: Sharad Gupta <sharadg@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
GVS: Gerrit_Virtual_Submit
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Provision is made to pass the Tegra platform information to ADSP
("nvidia,tegra_platform" DT property, as per enum tegra_platform);
this will be useful to convey whether the platform is CMODEL, in
which case the ADSP may take CMODEL specific paths, like using
HSP shared interrupts.
Upper 16 bits of nvadsp_os_args.chip_id is used to convey the
platform information; this may be improved in future as a
separate structure member.
Bug 200745795
Bug 200746669
Change-Id: Ibd9423eebb06fc4d243c132ab1cb7c6dc558ba5e
Signed-off-by: Viswanath L <viswanathl@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nvidia/+/2622156
Reviewed-by: svcacv <svcacv@nvidia.com>
Reviewed-by: svc_kernel_abi <svc_kernel_abi@nvidia.com>
Reviewed-by: Mohan Kumar D <mkumard@nvidia.com>
Reviewed-by: Asha Talambedu <atalambedu@nvidia.com>
Reviewed-by: Sharad Gupta <sharadg@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
GVS: Gerrit_Virtual_Submit
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Add Interconnect API support for ADSP Memory bandwidth requirement
handling.
From Kernel 4.14 and beyond
- On T23X and newer platforms, the interconnects driver will be initialized
and calls to BWMGR will be stubbed out.
- On T194 and earlier platforms, BWMGR driver will be initialized and calls
to interconnect framework will be stubbed out.
Added code check with K5.9 as mc_utils.h header is present from K5.9 build
For Kernel 4.9 and earlier
- T23x will not be supported.
- T194 and earlier platforms will use BWMGR. Interconnect framework is not
available.
JIRA TAS-1060
Change-Id: Ie0b4040f374a3c6102cb7aff2506d8cf2f0eab69
Signed-off-by: Mohan Kumar <mkumard@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nvidia/+/2414369
Reviewed-by: Viswanath L <viswanathl@nvidia.com>
Reviewed-by: Sameer Pujar <spujar@nvidia.com>
Reviewed-by: Sharad Gupta <sharadg@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
GVS: Gerrit_Virtual_Submit
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Change exposes lpthread/adsp usage trigger via
amixer. This is needed to make sound card aware
of lpthread state. Without this STR adsp feature
breaks as currently lpthread is written in such
a manner which does not allow adsp to suspend.
Bug 200552183
Change-Id: I2cc1e0fd805a982686bbd034b2c6e094b56df23b
Signed-off-by: Dipesh Gandhi <dipeshg@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/2198693
(cherry picked from commit e9bd14fd6eb0305cc583092752e6bdccc9c5e76a)
Reviewed-on: https://git-master.nvidia.com/r/2250202
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Swati Sachdeva <ssachdeva@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Uday Gupta <udayg@nvidia.com>
Reviewed-by: Nitin Pai <npai@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
- Workqueue scheduling latency causes issues for
adsp apps requiring real time behaviour. Hence
use RT kthread to schedule the file io functions
- file open with O_TRUNC flag may take long time
Hence we open the file just once and just reset
the read/write pointers in a file close call
- Expose sysfs entry to actually close all files
opened by kernel
Bug 1846449
Jira EMA-847
Change-Id: I3c130b36e5f5f5a0e65ddb4c44efc8754fb006d2
Signed-off-by: Gaurav Tendolkar <gtendolkar@nvidia.com>
Reviewed-on: http://git-master/r/1314984
(cherry picked from commit 00f5d0b48aa0df9a720c6fd15b4cffd827408375)
Signed-off-by: Hariharan Sivaraman <hariharans@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/2029203
(cherry picked from commit 6760f7d61a589b90a8c156a38e4453b518a20b05)
Reviewed-on: https://git-master.nvidia.com/r/2030551
(cherry picked from commit c136844cd4147c25478fe39a15122a1f43553079)
Reviewed-on: https://git-master.nvidia.com/r/2150778
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Niranjan Dighe <ndighe@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Nitin Pai <npai@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
NVADSP driver tries to allocate shared memory for ADSP
by requesting the DMA APIs to allocate at address hardcoded
in DT. This poses an issue where if kernel has already allocated
that memory to a different driver in same iommu group, this
memory becomes unavailable and ADSP does not boot
Adding support to allocate shared memory for ADSP dynamically
to avoid dependencies on other drivers
Jira EMA-1213
Change-Id: I1bc8d49f17ec8226d34f3c943cccabef97b2afb6
Signed-off-by: Hariharan Sivaraman <hariharans@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/2134341
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Uday Gupta <udayg@nvidia.com>
GVS: Gerrit_Virtual_Submit
Tested-by: Niranjan Dighe <ndighe@nvidia.com>
Reviewed-by: Nitin Pai <npai@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
There is no point in managing iovad that's not binded to
device's IOMMU domain. So this patch just removes iovad
managing code and does a correct iommu_dma_alloc_iova()
to reserve the iova.
A difficulty of using iommu_dma_alloc_iova() is that we
mostly do not use aligned addresses for the target iova,
but IOMMU core aligns the iova address with the size by
power of two due to cache friendly. So the trick here is
to align the start address and to allocate a larger iova
region that would cover the target iova -- the excessive
part will be freed during iommu_dma_free_iova() by IOMMU
core.
Note that the dma_free_coherent() does both iommu_unmap
and iommu_dma_free_iova, so there's no need of an extra
free_iova() in the exit routine. However, for the remap
part, the tmp_iova is allocated and mapped, so it should
be unmapped and freed. So this patch also adds a free()
of tmp_iova in the nvadsp_dma_alloc_at_map() function.
Bug 200444660
Change-Id: I54126c2bd9ff3dd688aaa5c6f9befaad620ab548
Signed-off-by: Nicolin Chen <nicolinc@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/2096704
Reviewed-by: Automatic_Commit_Validation_User
GVS: Gerrit_Virtual_Submit
Reviewed-by: Krishna Reddy <vdumpa@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
The dma_alloc_at_coherent is implemented with a hack in the
iommu core code, which is unlikely able to get upstream. So
this patch removes its callers in the nvadsp code, and then
adds an alternative helper function to do something similar.
The dma_alloc_at_coherent was used to allocate a DMA buffer
and to map it to a specified iova address. And this can be
done by remapping a normal DMA buffer to the iova address.
Bug 200444660
Change-Id: Ic90aec550daa29df69d97c4b230c8f9cc1519166
Signed-off-by: Nicolin Chen <nicolinc@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/2083356
Reviewed-by: Krishna Reddy <vdumpa@nvidia.com>
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Dump name (first 4 chars) of current thread and irq
number of last interrupt handled from hwmbox reg 4 and 7
respectively.
Bug 200295526
Change-Id: I4b53521cbfece2177e3cf6ed80b559b58e1d1c0a
Signed-off-by: Nitin Kumbhar <nkumbhar@nvidia.com>
Reviewed-on: http://git-master/r/1475487
(cherry picked from commit 64a799ae7d8e8f8783212fb4c28209cf760b38a7)
Reviewed-on: http://git-master/r/1478825
(cherry picked from commit cab60f6977fc8070e4206db01e2a2d444860c90c)
Reviewed-on: https://git-master/r/1513585
Reviewed-on: https://git-master.nvidia.com/r/1537329
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: svccoveritychecker <svccoveritychecker@nvidia.com>
GVS: Gerrit_Virtual_Submit
The state of ADSP is logged by writing to a mailbox
register. Adding support to kernel driver to dump ADSP
state when crash happens.
Jira EMA-404
Bug 1901511
Bug 1893324
Bug 200239577
Change-Id: Ia2beb84a7d5bc871f7062bd6a661333af0b0e24d
Signed-off-by: Hariharan Sivaraman <hariharans@nvidia.com>
Reviewed-on: http://git-master/r/1229864
(cherry picked from commit 0027e7a51f6134b4a7b332c17f7c7c26ac727ebc)
Signed-off-by: Nitin Kumbhar <nkumbhar@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1537327
Reviewed-by: svccoveritychecker <svccoveritychecker@nvidia.com>
GVS: Gerrit_Virtual_Submit
The request irqs are moved just before starting ADSP and freed when
adsp is suspended/stopped.
This is since the new agic driver is based on a device-driver model
and requires all interrupts to be freed before it could suspend.
Bug 200270956
Change-Id: I8ecd05ebe52020f11be79b9a1da37a85fed432ac
Signed-off-by: Ajay Nandakumar <anandakumarm@nvidia.com>
Reviewed-on: http://git-master/r/1478838
Signed-off-by: Nitin Kumbhar <nkumbhar@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1537326
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: svccoveritychecker <svccoveritychecker@nvidia.com>
GVS: Gerrit_Virtual_Submit
When ADSP OS is loaded from adsp.elf, it loads two sections : adsp os
and the vector table. The vector table is copied to a local buffer
before writing to the EVP registers.
There is a probability that, if the adsp.elf is hacked / modified by
an attacker, it can write to other APE register spaces. Hence, the
destination address obtained from the elf is checked with evp base
before copying.
Also, this fixes out of array-bound writing to evp buffer.
Bug 1684844
Change-Id: I8981dc5a5db8e0c0653ff46a1576df2c82e420be
Signed-off-by: Ajay Nandakumar <anandakumarm@nvidia.com>
Reviewed-on: http://git-master/r/1118936
(cherry picked from commit 9f7120e03e66b5f6e2bf67f09063da20945be238)
Reviewed-on: http://git-master/r/1458894
Signed-off-by: Nitin Kumbhar <nkumbhar@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1537317
Reviewed-by: svccoveritychecker <svccoveritychecker@nvidia.com>
GVS: Gerrit_Virtual_Submit
