Account NvMap allocated memory into both RSS and CG tracking to make
efficient OOM kill decisions during memory pressure.
NvMap allocates memory via kernel APIs like alloc_pages, the kernel
memory is not accounted on behalf of process who requests the
allocation. Hence in case OOM, the OOM killer never kills the process
who has allocated memory via NvMap even though this process might be
holding most of the memory.
Solve this issue using following approach:
- Use __GFP_ACCOUNT and __GFP_NORETRY flag
- __GFP_NORETRY will not let the current allocation flow to go into OOM
path, so that it will never trigger OOM.
- __GFP_ACCOUNT causes the allocation to be accounted to kmemcg. So any
allocation done by NvMap will be definitely accounted to kmemcg and
cgroups can be used to define memory limits.
- Add RSS counting for the process which allocates by NvMap, so that OOM
score for that process will get updated and OOM killer can pick this
process based upon the OOM score.
- Every process that has a reference to NvMap Handle would have the
memory size accounted into its RSS. On releasing the reference to
handle, the RSS would be reduced.
Bug 5222690
Change-Id: I3fa9b76ec9fc8d7f805111cb96e11e2ab1db42ce
Signed-off-by: Ketan Patil <ketanp@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3447072
(cherry picked from commit 858d73775a)
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3458546
GVS: buildbot_gerritrpt <buildbot_gerritrpt@nvidia.com>
Tested-by: Amulya Yarlagadda <ayarlagadda@nvidia.com>
Reviewed-by: Amulya Yarlagadda <ayarlagadda@nvidia.com>
OpenRM maps the buffer with remap_pfn_range and then it's user VA is
passed to libnvrm_mem to create a handle out of it. NvMap uses
get_user_pages to get user pages from the VA. It fails for the buffer
mapped with remap_pfn_range. Then it fallbacks to follow_pfn or
follow_pfnmap_start functions to obtain pfn from the VA and then obtain
page pointer from it. But as get_user_pages fails, the corresponding
error prints are getting generated even when not required. Hence reduce
the log level to debug to avoid these unnecessary spews.
Bug 5383624
Change-Id: Idbd3cfe93ce3fac6e27efc5c3bb7a200fc183d26
Signed-off-by: Ketan Patil <ketanp@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3425552
GVS: buildbot_gerritrpt <buildbot_gerritrpt@nvidia.com>
Reviewed-by: Pritesh Raithatha <praithatha@nvidia.com>
This change adds the support for programming streamids to
allow tsec fw on t264 to access PA at a low privilege level.
It also includes the synchronization logic to communicate
with the fw regarding completion of stream id programming
so that the fw can go ahead and initialize itself.
In addition to this, the mailbox used for communicating init done
from tsec fw to ccplex is changed from NV_PTSEC_FALCON_MAILBOX0 to
NV_PTSEC_MAILBOX1 since CCPLEX does not have access to the former from
t26x onwards. Hence falcon based mailboxes are used for tsec-psc comms
and non-falcon ones for tsec-ccplex comms (stream id comms and init done).
Jira TSEC-14
Change-Id: I2871a52222cd69786a8cc3f53162a80486611bb5
Signed-off-by: Sahil Patki <spatki@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3366343
Reviewed-by: Bharat Nihalani <bnihalani@nvidia.com>
GVS: buildbot_gerritrpt <buildbot_gerritrpt@nvidia.com>
(cherry picked from commit db54fde9c4d786b22b7f8694753de3ec80649b17)
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3400219
Android builds don't have CONFIG_NUMA enabled hence
/sys/devices/system/node/node0/meminfo is not present on android.
While nvscibuf calls the QueryHeapParams to check presence of the
hugetlbfs based carveout, the error prints will be seen due to absence
of the above sysfs file. Hence first check whethere there are multiple
numa nodes are not. If not, then use /proc/meminfo file to retrieve the
hugetlbfs size otherwise use the meminfo sysfs node from the
corresponding numa node.
Bug 5200644
Change-Id: I5495de91726d323210807e86f22757b798226fca
Signed-off-by: Ketan Patil <ketanp@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3338255
Reviewed-by: Pritesh Raithatha <praithatha@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Reviewed-by: svcacv <svcacv@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Reviewed-by: Jian-Min Liu <jianminl@nvidia.com>
GVS: buildbot_gerritrpt <buildbot_gerritrpt@nvidia.com>
[1] BUG_ON -> WARN_ON as per the recommendation from svcacv.
[2] Updated the power management interface
- Setting delay for rail, power and clock gating is easier to
manage with separate API.
- delay_us has been removed from the gating.
- New stat interface has been exposed for any stats related
to the power management operations.
[3] Supports autosuspend feature.
[4] Supports lpwr telemetries for the rail, power, and clock.
- idle time in us
- active time in us
- idle count
- active count
- entry and exit latencies in us (max, min and total)
Bug 5185298
Jira DLA-7905
Change-Id: I81267d7c3e9c35c81d29a19efe9192d436322ee9
Signed-off-by: Arvind M <am@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3318099
(cherry picked from commit c7cf5bb3b24dd9ed58e55a3fec56c8b1503c7e01)
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3338010
Reviewed-by: Vishal Thoke <vthoke@nvidia.com>
Reviewed-by: Mitch Harwell <mharwell@nvidia.com>
Tested-by: Amit Sharma (SW-TEGRA) <amisharma@nvidia.com>
Reviewed-by: Amit Sharma (SW-TEGRA) <amisharma@nvidia.com>
GVS: buildbot_gerritrpt <buildbot_gerritrpt@nvidia.com>
Reviewed-by: Ayush Kumar <ayushk@nvidia.com>
Gitlab commit 52e76e5680447 ("Map only PVA's syncpoints as R...")
Changes since last deployment:
- Map only PVA's syncpoints as RW
- fw: treat unregister resource ID 0 as noop
- Fix tegrastats issue that mixes two reads together
- Delpoy VPU ELF files to tegra tree
- kmd: bugfix: check if ctx inited before async ops
Change-Id: Ife641b1a27580e6f74f2b6e806eebc6750f3f846
Signed-off-by: nanwa <nanwa@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3354736
Reviewed-by: Mohnish Jain <mohnishj@nvidia.com>
GVS: buildbot_gerritrpt <buildbot_gerritrpt@nvidia.com>
VPR feature is needed for N1-auto on embedded, while this feature is not
required on 7.0. Hence enable this feature to unblock the VPR testing.
Later on in 7.0.4.0 branch create a patch to disable this feature. Once
we have a uniform solution for handling this from program team, we will
switch to that approach.
Bug 5240456
Change-Id: Ib3773bff8b44525f4a1c3b68cc134d96f5c45ee3
Signed-off-by: Ketan Patil <ketanp@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3348574
GVS: buildbot_gerritrpt <buildbot_gerritrpt@nvidia.com>
Reviewed-by: Sachin Nikam <snikam@nvidia.com>
Reviewed-by: Pritesh Raithatha <praithatha@nvidia.com>
Remove power management ops from the host1x bus-side DLA driver. The host1x
bus-side driver is a virtual device and should not directly access hardware.
PM ops should only be used in the platform device driver.
This fixes suspend/resume test failure when tegra-drm is loaded without any
enabled engines under host1x node.
Bug 5055946
Change-Id: I2e07f759e1d92b4c5a92739e1b5439d39800773f
Signed-off-by: Vamsee Vardhan Thummala <vthummala@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3337661
Reviewed-by: Sourab Gupta <sourabg@nvidia.com>
Reviewed-by: Mikko Perttunen <mperttunen@nvidia.com>
Reviewed-by: Mainak Sen <msen@nvidia.com>
GVS: buildbot_gerritrpt <buildbot_gerritrpt@nvidia.com>
Deprecate the usage of nvhost_module_pm_ops as part of the wider
nvhost API deprecation while maintaining equivalent functionality.
The NVCSI driver does not register prepare_poweroff, poweron_reset,
finalize_poweron callbacks nor set the clock gating regs with nvhost,
so the runtime operations will only call clk APIs.
Bug 4921620
Change-Id: I22a75c9f58ec0980aa97997b674c26a9bb62d47b
Signed-off-by: Akihiro Mizusawa <amizusawa@nvidia.com>
Signed-off-by: Mainak Sen <msen@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3321086
GVS: buildbot_gerritrpt <buildbot_gerritrpt@nvidia.com>
Reviewed-by: Frank Chen <frankc@nvidia.com>
Reviewed-by: Chinniah Poosapadi <cpoosapadi@nvidia.com>
Reviewed-by: FNU Raunak <fraunak@nvidia.com>
Tested-by: FNU Raunak <fraunak@nvidia.com>
Deprecate the usage of nvhost_module_pm_ops as part of the wider
nvhost API deprecation while maintaining equivalent functionality.
The capture-support driver does not register prepare_poweroff,
poweron_reset, finalize_poweron callbacks nor set the clock gating
regs with nvhost, so the runtime operations will only call clk APIs.
Bug 4921620
Change-Id: I438621258378bb6e81fc7a82e3334dcbfa180dc9
Signed-off-by: Akihiro Mizusawa <amizusawa@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3321085
Reviewed-by: Chinniah Poosapadi <cpoosapadi@nvidia.com>
Reviewed-by: Frank Chen <frankc@nvidia.com>
Tested-by: FNU Raunak <fraunak@nvidia.com>
GVS: buildbot_gerritrpt <buildbot_gerritrpt@nvidia.com>
Reviewed-by: FNU Raunak <fraunak@nvidia.com>
Deprecate the usage of nvhost_module_pm_ops as part of the wider
nvhost API deprecation while maintaining equivalent functionality.
The ISP driver does not register prepare_poweroff, poweron_reset,
finalize_poweron callbacks nor set the clock gating regs with nvhost,
so the runtime operations will only call clk APIs.
Bug 4921620
Change-Id: Iadf7b4104f622507e8b2e82fe054151e6318d747
Signed-off-by: Akihiro Mizusawa <amizusawa@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3321084
GVS: buildbot_gerritrpt <buildbot_gerritrpt@nvidia.com>
Tested-by: FNU Raunak <fraunak@nvidia.com>
Reviewed-by: Frank Chen <frankc@nvidia.com>
Reviewed-by: FNU Raunak <fraunak@nvidia.com>
Reviewed-by: Chinniah Poosapadi <cpoosapadi@nvidia.com>
Reviewed-by: svcacv <svcacv@nvidia.com>
Deprecate the usage of nvhost_module_pm_ops as part of the wider
nvhost API deprecation while maintaining equivalent functionality.
The VI driver does not register prepare_poweroff, poweron_reset,
finalize_poweron callbacks nor set the clock gating regs with nvhost,
so the runtime operations will only call clk APIs.
Bug 4921620
Change-Id: I639d428f5ea1ed1071d4502a877841b7049b9cd2
Signed-off-by: Akihiro Mizusawa <amizusawa@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3320983
Reviewed-by: svcacv <svcacv@nvidia.com>
Reviewed-by: Frank Chen <frankc@nvidia.com>
Tested-by: FNU Raunak <fraunak@nvidia.com>
GVS: buildbot_gerritrpt <buildbot_gerritrpt@nvidia.com>
Reviewed-by: FNU Raunak <fraunak@nvidia.com>
Reviewed-by: Chinniah Poosapadi <cpoosapadi@nvidia.com>
- The VI is allowed to access the whole of the Guest VM currently as it
is a stage-2 SMMU device. Hence it can read and write to any memory.
This may lead to breach of confidentiality and integrity. In order to
restrict the VI from accessing the whole VM, a mempool is being created
by HV that would restrict access to VI to just the mempool memory.
- In nvmap, we have some special rules for IVM carveout like if user
specify IVM carveout to allocate from, then if first IVM carveout does
not have sufficient memory then nvmap tries to allocate from next IVM
carveout. We don't want these rules applicable for VI mempool heap,
hence only carveout initialization part would remain similar to IVM
carveouts but rest of the operations would be performed normally like
any other carveout.
- Add DT binding doc for VI-carveout.
Bug 4648721
Change-Id: Ib40415a4c80da908654c86162c1cd4b50b33ef31
Signed-off-by: Ketan Patil <ketanp@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3196238
Reviewed-by: Pritesh Raithatha <praithatha@nvidia.com>
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
When building the PVA driver with the compiler option
'-Werror=declaration-after-statement' the build fails with the following
error ...
drivers/video/tegra/host/pva/./src/kmd/common/pva_kmd_shared_buffer.c:
In function ‘shared_buffer_process_msg’:
drivers/video/tegra/host/pva/./src/kmd/common/pva_kmd_shared_buffer.c:
error: ISO C90 forbids mixed declarations and code
[-Werror=declaration-after-statement]
187 | uint32_t msg_size = safe_subu32(header.size, sizeof(header));
| ^~~~~~~~
Fix this build issue by ensuring the variable 'msg_size' is defined at
the start of the shared_buffer_process_msg() function.
Change-Id: Ibd4a51caae06bfda2e11dfc6e3f1cf2f26d618c7
Signed-off-by: Jon Hunter <jonathanh@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3330348
GVS: buildbot_gerritrpt <buildbot_gerritrpt@nvidia.com>
Reviewed-by: svcacv <svcacv@nvidia.com>
Reviewed-by: Mohnish Jain <mohnishj@nvidia.com>
Reviewed-by: Nan Wang <nanwa@nvidia.com>
After a deeper analysis, it was found that multithreaded cache flush is
not required. Only gathering the pages and then doing cache flush is
enough to get the perf improvement, the cache flush time of each page is
almost half in this case as compared with not gathering the pages. So it
means there is continuous cache eviction is happening. So, while
allocating big pages from allocator, instead of allocating one big page
and flushing it, allocate all required big pages and then flush pages
at last.
Bug 4628529
Change-Id: I4d8113010a8c82a1cab124e26b14e8a3f5791cd1
Signed-off-by: Ketan Patil <ketanp@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3325002
Reviewed-by: Pritesh Raithatha <praithatha@nvidia.com>
GVS: buildbot_gerritrpt <buildbot_gerritrpt@nvidia.com>
Reviewed-by: Ajay Nandakumar Mannargudi <anandakumarm@nvidia.com>
This reverts commit b90900fdece2eb002b5061681aead4a74401b484.
After a deeper analysis, it was found that multithreaded cache flush is
not required. Only gathering the pages and then doing cache flush is
enough to get the perf improvement, the cache flush time of each page is
almost half in this case as compared with not gathering the pages. So it
means there is continuous cache eviction is happening. Hence revert the
multithreaded cache flush support to simply the code and patch 3325002
adds gathering of pages support.
Bug 4628529
Change-Id: Ieb9237c6c2935cbf88036bc3ab83c33dc4d5e38d
Signed-off-by: Ketan Patil <ketanp@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3325001
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Reviewed-by: Sachin Nikam <snikam@nvidia.com>
GVS: buildbot_gerritrpt <buildbot_gerritrpt@nvidia.com>
Reviewed-by: Pritesh Raithatha <praithatha@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
- Added support for booting from GSC on L4T and updated the VPU allowlist.
- Enabled debug logs by default for debug builds and added a DebugFS node for VPU.
- Addressed data corruption issues in QNX devctls and fixed various KMD data handling bugs.
- Updated DMA configurations and handling.
- Implemented nsight trace logging support and updated VPU stats and NVLOG implementations.
- Other minor bug fixes and documentation updates.
Gitlab commit dc34d709b865 ("intf: Enable following intf tests")
Change-Id: Ide42e0390af9eece5e9ed88ac3eba57b9272d6a1
Signed-off-by: nanwa <nanwa@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3323529
Reviewed-by: Mohnish Jain <mohnishj@nvidia.com>
GVS: buildbot_gerritrpt <buildbot_gerritrpt@nvidia.com>
Remove dependencies on the following deprecated
nvhost API functions by implementing their functionality
directly in the NVDLA driver:
1. nvhost_module_init:
Replaced with direct implementation in nvdla_module_init:
- Get clock resources and set clock rates
- Initialize reset control
- Set up power management runtime
- Create debugfs entries
2. nvhost_module_deinit:
Replaced with direct implementation in nvdla_module_deinit:
- Disable power management runtime
- Free falcon firmware if present
- Clean up debugfs entries
3. nvhost_module_reset:
Replaced with direct implementation in nvdla_module_reset:
- Handle prepare_poweroff callback
- Acquire and release reset control
- Load clock gating registers
- Handle finalize_poweron callback
4. nvhost_module_busy:
Replaced with direct implementation in nvdla_module_busy:
- Get PM runtime sync
- Handle PM runtime errors
5. nvhost_module_idle and nvhost_module_idle_mult:
Replaced with direct implementations:
- Mark last busy for PM runtime
- Handle autosuspend or regular PM runtime put
6. nvhost_module_pm_ops:
Replaced with custom nvdla_pm_ops structure:
- Implement nvdla_runtime_suspend and nvdla_runtime_resume
- Implement nvdla_suspend and nvdla_resume
- Configure SET_RUNTIME_PM_OPS and SET_SYSTEM_SLEEP_PM_OPS
7. nvhost_module_load_regs:
Replaced with nvdla_module_load_regs:
- Implement register manipulation with direct memory mapped I/O
- Use writel instead of host1x_writel
Bug 4921620
Jira HOSTX-5963
Change-Id: I3006a92a05966da138da31730e97f74f786c9d0c
Signed-off-by: Mainak Sen <msen@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3321246
Reviewed-by: Arvind M <am@nvidia.com>
Reviewed-by: Mitch Harwell <mharwell@nvidia.com>
GVS: buildbot_gerritrpt <buildbot_gerritrpt@nvidia.com>
Replace deprecated nvhost client device functions with
their direct implementations to maintain compatibility
with newer host1x drivers:
1. Replace nvhost_client_device_get_resources with
direct implementation:
- Get host1x data from parent device
- Map device resources directly
2. Replace nvhost_client_device_init with
nvdla_client_device_init:
- Allocate character device region
- Create device node using nvdla_client_device_create
3. Replace nvhost_client_device_release with
nvdla_client_device_release:
- Clean up device node
- Unregister character device region
This change ensures the NVDLA driver continues to work
correctly with the host1x driver while removing
dependencies on deprecated functions.The device node
"nvhost-ctrl-dla" is still created with the same
functionality as before.
Bug 4921461
Jira HOSTX-5963
Change-Id: Ib27000fa47fdd72f80ca4030232d31e100c01ea4
Signed-off-by: Mainak Sen <msen@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3320695
Reviewed-by: svcacv <svcacv@nvidia.com>
GVS: buildbot_gerritrpt <buildbot_gerritrpt@nvidia.com>
Reviewed-by: Arvind M <am@nvidia.com>
Reviewed-by: Mitch Harwell <mharwell@nvidia.com>
Replace the deprecated nvhost_syncpt_unit_interface_init()
function with host1x_syncpt_get_shim_info() in
the NVDLA driver. This change:
1. Adds a local def of struct nvhost_syncpt_interface
2. Implements the functionality in nvdla_sync_syncpt.c
3. Adds necessary include files for IOMMU and DMA mapping
This change ensures the NVDLA driver continues to work
correctly with the host1x driver while removing the
dependency on deprecated functions.
Bug 4921461
Jira HOSTX-5963
Change-Id: I085a24045758a2ecbb026654c3ee57a8dfda9271
Signed-off-by: Mainak Sen <msen@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3320585
GVS: buildbot_gerritrpt <buildbot_gerritrpt@nvidia.com>
Reviewed-by: svcacv <svcacv@nvidia.com>
Reviewed-by: Mitch Harwell <mharwell@nvidia.com>
Reviewed-by: Arvind M <am@nvidia.com>
