Modify the GK20A_TRACE_PRINTK config such that it depends on TRACING
instead of FTRACE_PRINTK. The latter is not in upstream Linux nor in our
downstream 4.9, and this option is default n anyway so this is a pretty
safe change.
Change-Id: If4ce5a041c8392d0bc54a60730c6ab3115b0062a
Signed-off-by: Konsta Holtta <kholtta@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1674114
GVS: Gerrit_Virtual_Submit
Reviewed-by: Deepak Nibade <dnibade@nvidia.com>
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Add new characteristics flag NVGPU_GPU_FLAGS_SUPPORT_USER_SYNCPOINT to indicate
support for this new API
Add new flag NVGPU_SUPPORT_USER_SYNCPOINT for use of core driver.
Set this flag for VGPU-GV11B
Bug 200326065
Jira NVGPU-179
Change-Id: I6c992b13268b688a2bbc93a3331e987ea2f7dd0c
Signed-off-by: Lakshmanan M <lm@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1670452
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
Reviewed-by: Deepak Nibade <dnibade@nvidia.com>
Tested-by: Jitendra Pratap Singh Chauhan <jchauhan@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Vijayakumar Subbu <vsubbu@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
struct nvgpu_semaphore represents (mainly) a threshold value that a sema
at some index will get and struct nvgpu_semaphore_int (aka "hw_sema")
represents the allocation (and write access) of a semaphore index and
the next value that the sema at that index can have. The threshold
object doesn't need a pointer to the sema allocation that is not even
guaranteed to exist for the whole threshold lifetime, so replace the
pointer by the position of the sema in the sema pool.
This requires some modifications to pass a hw sema around explicitly
because it now represents write access more explicitly.
Delete also the index field of semaphore_int because it can be directly
derived from the offset in the sema location and is thus unnecessary.
Jira NVGPU-512
Change-Id: I40be523fd68327e2f9928f10de4f771fe24d49ee
Signed-off-by: Konsta Holtta <kholtta@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1658102
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Add IPA to PA translation for GV100 nvlink / pass-through mode
- define platform->phys_addr(g, ipa) method
- call nvgpu_init_soc_vars from nvgpu_tegra_pci_probe
- in nvgpu_init_soc_vars, define set platform->phys_addr to
nvgpu_tegra_hv_ipa_pa, if hypervisor is present.
- in __nvgpu_sgl_phys, use sg_phys, then apply platform->phys_addr
if defined.
- implement IPA to PA translation in nvgpu_tegra_hv_ipa_pa
Bug 200392719
Change-Id: I622049ddc62c2a57a665dd259c1bb4ed3843a537
Signed-off-by: Thomas Fleury <tfleury@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1673582
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
Reviewed-by: Richard Zhao <rizhao@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Vijayakumar Subbu <vsubbu@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Add __nvgpu_sgl_phys function that can be used to implement IPA
to PA translation in a subsequent change.
Adapt existing function prototypes to add pointer to gpu context,
as we will need to check if IPA to PA translation is needed.
JIRA EVLR-2442
Bug 200392719
Change-Id: I5a734c958c8277d1bf673c020dafb31263f142d6
Signed-off-by: Thomas Fleury <tfleury@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1673142
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Default semaphore payload size is 16byte. Set it to 4 byte
to avoid double increment of associated sync point with
semaphore release.
Also removed extra 0 op function from syncpoint increment
command.
Bug 2066025
Change-Id: Ia282cc5625827d356b5ba963adb7b1b3c703a931
Signed-off-by: seshendra Gadagottu <sgadagottu@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1669714
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Replace the padding in nvgpu_channel_wdt_args with a timeout value in
milliseconds, and add NVGPU_IOCTL_CHANNEL_WDT_FLAG_SET_TIMEOUT to
signify the existence of this new field. When the new flag is included
in the value of wdt_status, the field is used to set a per-channel
timeout to override the per-GPU default.
Add NVGPU_IOCTL_CHANNEL_WDT_FLAG_DISABLE_DUMP to disable the long debug
dump when a timed out channel gets recovered by the watchdog. Printing
the dump to serial console takes easily several seconds. (Note that
there is NVGPU_TIMEOUT_FLAG_DISABLE_DUMP about ctxsw timeout separately
for NVGPU_IOCTL_CHANNEL_SET_TIMEOUT_EX as well.)
The behaviour of NVGPU_IOCTL_CHANNEL_WDT is changed so that either
NVGPU_IOCTL_CHANNEL_ENABLE_WDT or NVGPU_IOCTL_CHANNEL_DISABLE_WDT has to
be set. The old behaviour was that other values were silently ignored.
The usage of the global default debugfs-controlled ch_wdt_timeout_ms is
changed so that its value takes effect only for newly opened channels
instead of in realtime. Also, zero value no longer means that the
watchdog is disabled; there is a separate flag for that after all.
gk20a_fifo_recover_tsg used to ignore the value of "verbose" when no
engines were found. Correct this.
Bug 1982826
Bug 1985845
Jira NVGPU-73
Change-Id: Iea6213a646a66cb7c631ed7d7c91d8c2ba8a92a4
Signed-off-by: Konsta Holtta <kholtta@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1510898
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
With proper wrap-handling comparisons now supported, it's safe to not
reset a kernel-managed semaphore to 0 when initializing it to be used by
some channel; the value can be left unchanged, so that any pending waits
on other channels for this sema can't get corrupted anymore. This makes
semaphore values very similar to syncpoints, i.e., just monotonically
increasing counters.
Also clear the semaphore sea to values of 0xfffffff0 when allocating it.
This way it takes 16 increments on each sema to wrap over the 32-bit
integer range; such wrapping would eventually happen if the memory was
initialized to zeros, so this way any bugs possibly caused by wrapping
not taken into account would uncover quickly after boot.
Jira NVGPU-514
Change-Id: I93f9b1d32d020a4c23824f5856bc463b1895b99d
Signed-off-by: Konsta Holtta <kholtta@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1652087
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
Reviewed-by: Alex Waterman <alexw@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
The value of NV_PERF_PMASYS_MEM_BUMP is different for Volta
and NVGPU_IOCTL_CHANNEL_CYCLE_STATS_SNAPSHOT_CMD_FLUSH did not
have correct behavior on GV11B due to that.
The patch adds an instance of css_hw_set_handled_snapshots
for Volta to fix that.
Bug 1960846
Bug 2068936
Change-Id: Ic057338d3b1b951a66d070267e69a90f136598b9
Signed-off-by: Martin Radev <mradev@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1668568
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Compare gpu semaphores in the kernel in the same way as the hardware
does: released if value is over threshold, but at most half of u32's
range. This makes it possible to skip zeroing the sema values when semas
are allocated, so that they'd be just monotonically increasing numbers
like syncpoints are.
Jira NVGPU-514
Change-Id: I3bae352fbacfe9690666765b9ecdeae6f0813ea1
Signed-off-by: Konsta Holtta <kholtta@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1652086
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Alex Waterman <alexw@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>
- Created ops for below boardobj methods to support gp10x & gv10x
branch boardobj changes, and defined methods for gv10x with
postfix _v1 with below names
boardobjgrp_pmucmd_construct_impl
boardobjgrp_pmuset_impl
boardobjgrp_pmugetstatus_impl
is_boardobjgrp_pmucmd_id_valid
- These ops are assigned based on PMU version to respective
chip.
- Modified BOARDOBJGRP_PMU_CMD_GRP_SET_CONSTRUCT &
BOARDOBJGRP_PMU_CMD_GRP_GET_STATUS_CONSTRUCT to support
gp10x & gv10x branch changes
- Updated struct boardobjgrp_pmu_cmd to include members
needed for gv10x boardobj changes
- Created "struct nv_pmu_rpc_struct_board_obj_grp_cmd"
to execute BOARD_OBJ_GRP_CMD using RPC.
- Defined method boardobjgrp_pmucmdsend_rpc() to
send BOARD_OBJ_GRP_CMD to PMU.
Change-Id: If2551bdda80e897e7b21d2966881586f3bbc7a9b
Signed-off-by: Mahantesh Kumbar <mkumbar@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1656511
GVS: Gerrit_Virtual_Submit
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
- Added ops "pmu.alloc_super_surface" to create
memory space for pmu super surface
- Defined method nvgpu_pmu_sysmem_surface_alloc()
to allocate pmu super surface memory & assigned
to "pmu.alloc_super_surface" for gv100
- "pmu.alloc_super_surface" set to NULL for gp106
- Memory space of size "struct nv_pmu_super_surface"
is allocated during pmu sw init setup if
"pmu.alloc_super_surface" is not NULL &
free if error occur.
- Added ops "pmu_ver.config_pmu_cmdline_args_super_surface"
to describe PMU super surface details to PMU ucode
as part of pmu command line args command if
"pmu.alloc_super_surface" is not NULL.
- Updated pmu_cmdline_args_v6 to include member
"struct flcn_mem_desc_v0 super_surface"
- Free allocated memory for PMU super surface in
nvgpu_remove_pmu_support() method
- Added "struct nvgpu_mem super_surface_buf" to "nvgpu_pmu" struct
- Created header file "gpmu_super_surf_if.h" to include interface
about pmu super surface, added "struct nv_pmu_super_surface"
to hold super surface members along with rsvd[x] dummy space
to sync members offset with PMU super surface members.
Change-Id: I2b28912bf4d86a8cc72884e3b023f21c73fb3503
Signed-off-by: Mahantesh Kumbar <mkumbar@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1656571
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
When adding a sema wait to a pushbuf, verify that the sema threshold has
been incremented from the original value by reading the incremented
field instead of value (which is set to nonzero by
nvgpu_semaphore_incr()). Value could be 0 even after an increment if new
semas weren't reset to 0.
Jira NVGPU-514
Change-Id: I295451fbc7eb9e597aea12d73074e99f74a6a899
Signed-off-by: Konsta Holtta <kholtta@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1658100
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
/sys/devices/gpu.0/gfxp_wfi_timeout_unit
usec - microseconds
sysclk - gpu clock count
Treat gr_fe_gfxp_wfi_timeout_r as context-switched
register on gv11b.
Set default gfxp_wfi_timeout to 100 usec to match
gp10b at 1GHz.
bug 1888344
Signed-off-by: Kirill Artamonov <kartamonov@nvidia.com>
Change-Id: I7fa64ce6912ae861244856807543b17bd7a26bed
Reviewed-on: https://git-master.nvidia.com/r/1651517
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Check these conditions for fecs arbiter idle:
1. Wait for gr_fecs_arb_ctx_cmd to idle
2. Wait for gr_fecs_ctxsw_status_1:arb_busy to idle
Just waiting for condition 1, only guarantees dispatching
of command but not completion of work.
Bug 2056266
Change-Id: I3fcbdb9cda91fee0ff345a24bf23ba7dabf268d0
Signed-off-by: seshendra Gadagottu <sgadagottu@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1667550
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
PMU instance block layout is not getting
populated with subctx pdb info as max_subctx_count is
0x0 and is getting initialized after PMU instance block
initialization gets completed.
Therefore initializing max_subctx_count before using it.
For Volta, FECS can bind itself with the PMU instance
only if SC PDB info is populated.
Bug 2051863
Bug 200392620
Change-Id: Id4fc26502e189c15cb57cb36cc09387dad773dc5
Signed-off-by: Deepak Goyal <dgoyal@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1666585
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
PMU ucode records supported feature list for a
particular chip as support mask sent
via PMU_PG_PARAM_CMD_GR_INIT_PARAM.
It then enables selective feature list through
enable mask sent via
PMU_PG_PARAM_CMD_SUB_FEATURE_MASK_UPDATE cmd.
Right now only ELPG state machine mask was enabled.
Only ELPG state machine was getting executed
but other crucial steps in ELPG entry/exit sequence
were getting skipped.
Bug 200392620.
Bug 200296076.
Change-Id: I5e1800980990c146c731537290cb7d4c07e937c3
Signed-off-by: Deepak Goyal <dgoyal@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1665767
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Seshendra Gadagottu <sgadagottu@nvidia.com>
Tested-by: Seshendra Gadagottu <sgadagottu@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>
QNX is going to reuse the 'struct gk20a_event_id_data' data structure
for its event handling. Contrary to linux which works on fd polling,
QNX needs the information about the pid to find out which process to
deliver the event to.
The patch adds the requisite field.
Change-Id: I466e9cfaacb921c6e1c41bd92c66194b81fc6774
Signed-off-by: Sourab Gupta <sourabg@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1657994
Reviewed-by: Deepak Nibade <dnibade@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
In case deferred_reset_pending is set in gk20a_fifo_handle_mmu_fault() and in
gv11b_fifo_teardown_ch_tsg(), we skip resetting the engines and skip setting
the error notifier
Then we call gk20a_channel_abort()/gk20a_fifo_abort_tsg() which aborts the
channels, and resets the syncpoint values to release all the waiters
But since we don't set error notifier this could lead User to assume a
successful submission without any error
To fix this disable channel/TSG in case deferred_reset_pending is set and skip
calls to gk20a_channel_abort()/gk20a_fifo_abort_tsg()
Note that we finally abort the channel when channel is being closed
Bug 200363077
Change-Id: Ia48ca369701c14d1913d8f7b66ed466b7b840224
Signed-off-by: Deepak Nibade <dnibade@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1664319
(cherry picked from commit ac40d082e8)
Reviewed-on: https://git-master.nvidia.com/r/1666445
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
Reviewed-by: Seema Khowala <seemaj@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
We right now allocate a nvgpu managed syncpoint in c->sync and share
that with user space
But to avoid conflicts between user space and kernel space increments
allocate a separate "client managed" syncpoint for User space in c->user_sync
Add new API nvgpu_nvhost_get_syncpt_client_managed() to request a client managed
syncpoint from nvhost.
Note that nvhost/nvgpu do not keep track of MAX/threshold value of this syncpoint
Update gk20a_channel_syncpt_create() to receive a flag to indicate whether a
User space syncpoint is required or not
Unset NVGPU_SUPPORT_USER_SYNCPOINT for gp10b since we don't want to allocate
double syncpoints per channel on that platform
For gv11b, once we move to use user space submits, support for c->sync will be
dropped so we keep using only one syncpoint per channel
Bug 200326065
Jira NVGPU-179
Change-Id: I78d94de4276db1c897ea2a4fe4c2db8b2a179722
Signed-off-by: Deepak Nibade <dnibade@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1665828
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
The operations in struct nvgpu_sgt_ops have a scatter-gather list (sgl)
argument which is a void pointer. Change the type signatures to take
struct nvgpu_sgl * which is an opaque marker type that makes it more
difficult to pass around wrong arguments, as anything goes for void *.
Explicit types add also self-documentation to the code.
For some added safety, some explicit type casts are now required in
implementors of the nvgpu_sgt_ops interface when converting between the
general nvgpu_sgl type and implementation-specific types. This is not
purely a bad thing because the casts explain clearly where type
conversions are happening.
Jira NVGPU-30
Jira NVGPU-52
Jira NVGPU-305
Change-Id: Ic64eed6d2d39ca5786e62b172ddb7133af16817a
Signed-off-by: Konsta Holtta <kholtta@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1643555
GVS: Gerrit_Virtual_Submit
Reviewed-by: Vijayakumar Subbu <vsubbu@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Also revert other changes related to IO coherence. This may be the
culprit in a recent dev-kernel lockdown.
Bug 2070609
Change-Id: Ida178aef161fadbc6db9512521ea51c702c1564b
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1665914
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Srikar Srimath Tirumala <srikars@nvidia.com>
For some reason the GPU does not like the mappings created by the
DMA API for coherent sysmem buffers. But a plain vmap() does seem
to work. To work around this, when we are using coherent sysmem,
force the NO_KERNEL_MAPPING flag to on and then make a vmap() in
the nvgpu DMA API wrapper. The rest of the driver will be none the
wiser but will work as expected.
This problem is not understood yet but it is being tracked in bug
2040115. Once this bug is understood this WAR should either be
determined as necessary or reverted with an appropriate fix.
Bug 2040115
JIRA EVLR-2333
Change-Id: Idae7a0c92441f0309df572ac18697af49bb6ff2b
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1657568
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
