For upstream dma engine slave_id in <struct dma_slave_config> is obsolete.
The Correct way to specify slave_id is
to call dma_request_chan API along with device name defined in dts node.
mods_tegra_dma: mods_tegra_dma {
compatible = "nvidia,mods_tegra_dma";
dmas = <&gpcdma 15>, <&gpcdma 16>;
dma-names = "spi1_rx", "spi2_tx";
status = "okay";
};
To keep kernel ABI compatibility,
add a new ioctl cmd MODS_ESC_DMA_REQUEST_HANDLE_2.
A new struct MODS_DMA_HANDLE_2 is used together with it.
A new function esc_mods_dma_request_channel_2 is implemented accordingly.
JIRA:TM-728
Change-Id: Ic84f294e2248d800b9aacd597a7ce0a84fc94a1d
Signed-off-by: andersm <andersm@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nvidia/+/2711537
Signed-off-by: xueyuanb <xueyuanb@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nvidia/+/2753148
Reviewed-by: Chris Dragan <kdragan@nvidia.com>
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
Reviewed-by: svc-mobile-cert <svc-mobile-cert@nvidia.com>
Reviewed-by: Bitan Biswas <bbiswas@nvidia.com>
GVS: Gerrit_Virtual_Submit <buildbot_gerritrpt@nvidia.com>
The perforce version of the MODS kernel driver
does not pull in any of the Tegra specific driver
functions. This prevents the perforce driver
from being compiled on any Tegra system or non
Tegra system with CONFIG_ARCH_TEGRA=y.
To allow the perforce driver to be compiled
replace CONFIG_ARCH_TEGRA with MODS_HAS_TEGRA
which is set based on CONFIG_ARCH_TEGRA in git
but left unset in perforce.
Bug 3397113
Signed-off-by: Lael Jones <lajones@nvidia.com>
Change-Id: Ie113d632c4dcc372058b9a1e3a549a70b8f7c03f
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nvidia/+/2607859
Reviewed-by: svcacv <svcacv@nvidia.com>
Reviewed-by: svc_kernel_abi <svc_kernel_abi@nvidia.com>
Reviewed-by: Chris Dragan <kdragan@nvidia.com>
Reviewed-by: Sachin Nikam <snikam@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
GVS: Gerrit_Virtual_Submit
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
To allocate larger buffer, MODS code is allocating noncontiguous pages
and then calls dma buffer api to do smmu mapping to get a continuous
iova address. This change will fix the error when the scatterlist's
dma address is 0xffffffffffffffff, skip checking the sg since it's
already merged to the head node.
JIRA: TM-129
Signed-off-by: Carl Dong <carld@nvidia.com>
Change-Id: I45aa3b0afcf5c86179fb293cd48326b14c2f3efc
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nvidia/+/2552672
Reviewed-by: svc_kernel_abi <svc_kernel_abi@nvidia.com>
Reviewed-by: svcacv <svcacv@nvidia.com>
Reviewed-by: Sachin Nikam <snikam@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
GVS: Gerrit_Virtual_Submit
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Fix build linux-5.7-rc5 errors including following:
- change timespec to timespec64. replace getnstimeofday
with ktime_get_ts64
- replace usage of macro FIELD_SIZEOF with sizeof_field in ethtool.c
nvethernet and eqos files.
- support 2 arguments for of_get_phy_mode call
bug 200617764
Change-Id: I46067d7d36d08ee9556b2722e9ccec707b8853d4
Signed-off-by: Bitan Biswas <bbiswas@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nvidia/+/2347244
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Summary: Previously mods stored handles for both
clocks and resets in the same list. This meant
reset signals that didn't have a corresponding
clock signal could not be toggled. This change
adds a separate list of structures to handle
resets and decouple it from the clock handling.
Change-Id: I2a83dd5cb7b7fe412acf7ae1815b18b747ad2cac
Signed-off-by: Ellis Roberts <ellisr@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nvidia/+/2225415
Reviewed-by: automaticguardword <automaticguardword@nvidia.com>
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Bharat Nihalani <bnihalani@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
GVS: Gerrit_Virtual_Submit
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
In git, we have moved this header to uapi directory, so that it can be
installed with the headers_install target. It now has to follow
guidelines of well-formed headers usable in userspace programs.
* Update SPDX license header.
* Don't use kernel CONFIG options.
Bug 2657961
Change-Id: Iee2bbfc6c931b5e8c5ee71cef9dfdfbb7f50d114
Signed-off-by: Chris Dragan <kdragan@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nvidia/+/2310801
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Bharat Nihalani <bnihalani@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Bug 200582229
This change will add smmu mapping/unmapping support for MODS
allocated memory buffer by using kernel smmu api dma_map_sg_attrs()
and dma_unmap_sg_attrs()
1. new IOCTL MODS_ESC_IOMMU_DMA_MAP_MEMORY and
MODS_ESC_IOMMU_DMA_UNMAP_MEMORY is added to do smmu mapping/unmapping
and get smmu mapped address for MODS allocated buffers;
2. a platform device and driver is added to match mods smmu sid in dtb,
the device is used to do smmu mapping/unampping; and mods smmu devices
are added in dtb to match with the driver with smmu config
Signed-off-by: Dong Qiang <carld@nvidia.com>
Change-Id: I47d4d1c43f14fc6b51f0c00cee127e1d51377f64
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nvidia/+/2180897
Reviewed-by: Chris Dragan <kdragan@nvidia.com>
Reviewed-by: Bharat Nihalani <bnihalani@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
GVS: Gerrit_Virtual_Submit
* Add new cl_* macros for printing client-specific messages.
This adds client index to messages in code which is called from
user space. This makes it easier to diagnose errors when multiple
clients (processes) open the driver.
* Use kernel print levels appropriate to messages being printed.
* Print process pid when opening the driver.
* Print information about access token being set or released.
* Print information about PCI devices being enabled or disabled.
PCI device enablement is tied to the driver getting ownership
of the device. It also prevents multiple MODS instances from
owning the same PCI device.
* Print info about device DMA mask, print device DMA mask when DMA
mapping fails.
* Print more info about SRIOV actions.
* Disable SRIOV when the driver is being closed by the process
which enables SRIOV. Leaving SRIOV enabled when disabling PCI
device is an error and leads to stale functions, unusable on
subsequent MODS runs.
* Clean up log messages in a few places.
Change-Id: I51953e984a55c0990e90f89d8260f215c8c58498
Signed-off-by: Chris Dragan <kdragan@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nvidia/+/2279439
Tested-by: Ellis Roberts <ellisr@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Bharat Nihalani <bnihalani@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Summary: Support to hook mods interrupts
for tegra mods devices was broken due to
gpu-mods changes that went uncaught in sanity.
This change modifies the kernel to allow the tegra
code paths to work again. This change also adds
support to map irq numbers to gpio input pins
to expand functionality in the irq framework.
Change-Id: Iee65f9bcaf09453c1740cd5cf7f64b76c87e00c3
Signed-off-by: Ellis Roberts <ellisr@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/2259509
Reviewed-by: Automatic_Commit_Validation_User
GVS: Gerrit_Virtual_Submit
Reviewed-by: Chris Dragan <kdragan@nvidia.com>
Reviewed-by: Yu-Huan Hsu <yhsu@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Summary: Add debug fs to allow user to program
the error control registers in RAS so mods can
selectively disable errors if so desired. A
current example of this usecase is disabling
ras from throttling, which is thrown during
the mods Soctherm OC throttling test
Bug 200533168
Change-Id: Id9007e9e13ae9563ad2aae107b130bcd951f0bc4
Signed-off-by: Ellis Roberts <ellisr@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/2224479
GVS: Gerrit_Virtual_Submit
Reviewed-by: Kiran Kasamsetty <kkasamsetty@nvidia.com>
Reviewed-by: Rohan Sreeram <rsreeram@nvidia.com>
Reviewed-by: Bharat Nihalani <bnihalani@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
The sg->length may happen to be a large size, and this
will overflow the local segment_size when accumulating
sg->length to it at line "segment_size += sg->length".
Since segment_size of MODS_DMABUF_GET_PHYSICAL_ADDRESS
structure is defined in u64 type, this patch fixes the
bug by changing the local varible to u64 as well.
Bug 2521255
Change-Id: Id1fb6d149f55f4fd9238a399393ed2efae77c84a
Signed-off-by: Nicolin Chen <nicolinc@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/2105445
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Ashish Mhetre <amhetre@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Sachin Nikam <snikam@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
