FECS ucode does a priv holdoff around the assertion of context
reset. So, priv transactions (e.g. mailbox1 register write) might
fail due to this. Hence, do write with ack i.e. write and read
it back to make sure write happened for mailbox1.
Bug 200417403
Change-Id: I463be1cb8fdd477106b87786cb0603327a22cebe
Signed-off-by: Vaibhav Kachore <vkachore@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/2023494
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Create new gr/fecs_trace unit with common/gr/fecs_trace/fecs_trace.c
as common source file and include/nvgpu/gr/fecs_trace.h as common
header file
This unit will be common between Linux and QNX
Corresponding HAL files will be added under common/gr/fecs_trace/
as more functionality is moved to new unit
For now move struct gk20a_fecs_trace_hash_ent to new unit and
rename it as struct nvgpu_fecs_trace_context_entry
Add vmid field to this struct since it is required for QNX
Remove use of hashtables and simply use linked list to simplify
the code. FECS tracing is not a performance sensitive use case
so perf hit could be ignored
Rename hash_lock mutex to list_lock
struct gk20a_fecs_trace and mutex list_lock are still declared in
gk20a/fecs_trace_gk20a.c, hence they cannot be used in new unit yet
Rename and update all gk20a_fecs_trace_hash_*() APIs to appropriate
nvgpu_gr_fecs_trace_*() APIs
Remove gk20a_fecs_trace_hash_dump() since it is not being used
Jira NVGPU-1880
Change-Id: I89c2715baa770dbbd864ea70ab43d83d98ba693c
Signed-off-by: Deepak Nibade <dnibade@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/2022903
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
gk20a_fecs_trace_poll() right now calls gk20a_fecs_trace_ring_read()
to read the trace ring buffer written by FECS
gk20a_fecs_trace_ring_read() returns number of trace entries written
to local buffer if successful, otherwise returns error
In case there is really an invalid entry, gk20a_fecs_trace_poll()
will just stop reading more entries, write current read pointer to
h/w and return
When gk20a_fecs_trace_poll() is called next time, we again read that
invalid entry, and again skip it, and again return
This keeps happening, and we never move on to read new entries
Fix this by always continuing to read next entry irrespective
of current entry is valid or not
gk20a_fecs_trace_poll() now just prints a debug message instead of
breaking the loop
Bug 200491708
Change-Id: I8cf823d7dc5f72ebd70cab2f115af32f3d20ac78
Signed-off-by: Deepak Nibade <dnibade@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/2020167
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Debarshi Dutta <ddutta@nvidia.com>
Reviewed-by: Alex Waterman <alexw@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Add new unit common/gr/subctx.c to manage GR subcontext
This unit provides interfaces to allocate/free/load GR subcontext
Add new header file include/nvgpu/gr/subctx.h to declare all the
interfaces.
Right now channel_gk20a structure directly includes a nvgpu_mem
for context header.
Declare a new structure nvgpu_gr_subctx for subcontext and include
this from channel_gk20a
Make all necessary changes to refer ctx_header from subctx instead
of directly referencing it from channel
Jira NVGPU-1613
Change-Id: I9eb1ee8f26fa88d2881f9b294935b65e9cbcc9b4
Signed-off-by: Deepak Nibade <dnibade@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1990129
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
- Remove handling for channels that are no more bound to tsg
as channel could be referenceable but no more part of a tsg
- Use tsg_gk20a_from_ch to get pointer to tsg for a given channel
- Clear unhandled gr interrupts
Bug 2429295
JIRA NVGPU-1580
Change-Id: I9da43a2bc9a0282c793b9f301eaf8e8604f91d70
Signed-off-by: Seema Khowala <seemaj@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1972492
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Rename __nvgpu_set_enabled() to nvgpu_set_enabled(). The original
double underscore was present to indicate that this function is a
function with potentially unintended side effects (enabling a feature
has wide ranging impact).
To not lose this documentation a comment was added to convey that this
function must be used with care.
JIRA NVGPU-1029
Change-Id: I8bfc6fa4c17743f9f8056cb6a7a0f66229ca2583
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1989434
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Currently all the global contex buffers are mapped into each graphics
context. Move all the mapping/unmapping support to gr/ctx unit since
all the mappings are owned by context itself
Add nvgpu_gr_ctx_map_global_ctx_buffers() that maps all the global
context buffers into given gr_ctx
Add nvgpu_gr_ctx_get_global_ctx_va() that returns VA of the mapping
for requested index
Remove g->ops.gr.map_global_ctx_buffers() since it is no longer
required. Also remove below APIs
gr_gk20a_map_global_ctx_buffers()
gr_gk20a_unmap_global_ctx_buffers()
gr_tu104_map_global_ctx_buffers()
Remove global_ctx_buffer_size from nvgpu_gr_ctx since it is no
longer used
Jira NVGPU-1527
Change-Id: Ic185c03757706171db0f5a925e13a118ebbdeb48
Signed-off-by: Deepak Nibade <dnibade@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1987739
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Add new unit common/gr/ctx.c to manage GR context
This unit provides interfaces to allocate/free/map/unmap GR context,
patch context, pm context, ctxsw {preempt/spill/betacb/pagepool/rtvcb}
buffers.
It also provides APIs to set size of above buffers
Add new header file include/nvgpu/gr/ctx.h to declare all the interfaces.
Move nvgpu_gr_ctx, patch_desc, pm_ctx_desc, zcull_ctx_desc structures
to this unit
Add new structure nvgpu_gr_ctx_desc to hold context description
parameters. For now we add sizes of all the buffers here.
Add this structure to gr_gk20a for global reference
Remove gr_gp10b_alloc_buffer() since it is no longer used
Rename g->ops.gr.alloc_gfxp_rtv_cb() to g->ops.gr.init_gfxp_rtv_cb()
since this HAL now only sets the size of rtvcb ctxsw buffer
Remove gr->ctx_vars.buffer_size and gr->ctx_vars.buffer_total_size
since they were redundant. We already have gr->ctx_vars.golden_image_size
to denote golden image size
Jira NVGPU-1527
Change-Id: I8847b347f80235209dd5e28d979e79984ab85408
Signed-off-by: Deepak Nibade <dnibade@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1987702
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
OS specific implementation of fecs trace bind_channel function
needs to handle special case for vserver to retrieve vmid from
channel id. Native code should be independent of server code.
Modify struct fecs_trace member function bind_channel to pass
vmid parameter enabling retrieving and passing vmid from server
code.
Jira GVSCI-44
Change-Id: I96223376f2068e2cbf60a9c9b35ff564a65e5dc3
Signed-off-by: Aparna Das <aparnad@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1970693
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Add new unit common/gr/global_ctx.c to manage GR global context buffers
This unit provides interfaces to allocate/free/map/unmap all the global
context buffers. It also provides APIs to get/set size of the buffers,
and to get memory handle of the buffers
Use interfaces exposed by this unit instead of directly accessing global
context buffers in common code
Add new header file include/nvgpu/gr/global_ctx.h to declare all the
interfaces.
Rename "struct gr_ctx_buffer_desc" to "struct nvgpu_gr_global_ctx_buffer_desc"
which holds all data for each global context
Remove void *priv since it is no longer used
Add size to the desc structure to store the requested size
Remove global_ctx_buffer_size from struct nvgpu_gr_ctx since it is no longer
used for any real purpose
Jira NVGPU-1625
Change-Id: I3feaf47bc2fdf192f36b136f2ef80a49d1782c5d
Signed-off-by: Deepak Nibade <dnibade@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1977884
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>
In gk20a_gr_handle_fecs_error(), we right now check the error code in
mailbox to identify if we hit timestamp buffer full error interrupt
This error code right now is hard coded to 0x26
But on Turing ucode this error code is set to 0x32
Add new HAL g->ops.fecs_trace.get_buffer_full_mailbox_val() to get
correct error code per platform and use this in
gk20a_gr_handle_fecs_error()
Bug 200471541
Bug 2469604
Change-Id: I7325354b39d35b1c8b218e554814316d22950469
Signed-off-by: Deepak Nibade <dnibade@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1978144
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
Reviewed-by: svc-mobile-misra <svc-mobile-misra@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 separate new unit gr/ctxsw_prog that provides interface to access
h/w header files hw_ctxsw_prog_*.h
Add below chip specific files that access above h/w unit and provide
interface through g->ops.gr.ctxsw_prog.*() HAL for rest of the units
common/gr/ctxsw_prog/ctxsw_prog_gm20b.c
common/gr/ctxsw_prog/ctxsw_prog_gp10b.c
common/gr/ctxsw_prog/ctxsw_prog_gv11b.c
Remove all the h/w header includes from rest of the units and code.
Remove direct calls to h/w headers ctxsw_prog_*() and use HALs
g->ops.gr.ctxsw_prog.*() instead
In gr_gk20a_find_priv_offset_in_ext_buffer(), h/w header
ctxsw_prog_extended_num_smpc_quadrants_v() is only defined on gk20a
And since we don't support gk20a remove corresponding code
Add missing h/w header ctxsw_prog_main_image_pm_mode_ctxsw_f() for
some chips
Add new h/w header ctxsw_prog_gpccs_header_stride_v()
Jira NVGPU-1526
Change-Id: I170f5c0da26ada833f94f5479ff299c0db56a732
Signed-off-by: Deepak Nibade <dnibade@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1966111
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
If enabled, fecs trace updating happens from ucode
side even when there is no fecs trace dumper application
to consume it. Due to this, trace buffer will get
eventually full and ucode will trigger ctxsw_intr0.
Reset fecs_trace buffer to handle timestamp buffer full
ctxsw_intr0.
Bug 2361571
Bug 200472922
Change-Id: Ia26a17635fc6bd6e8663b8af983acc91839ecfcd
Signed-off-by: Seema Khowala <seemaj@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1965370
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
- Remove the usage of NVGPU_CTXSW_FILTER_ISSET splattered
across nvgpu, and replace with a MACRO defined in common code.
The usage is still inside Linux, but this helps the
subsequent unification efforts, e.g. to unify the fecs trace
path.
- Remove "uapi/linux/nvgpu.h" from common code.
EVLR-3078
Change-Id: I60b0e1627576a4b255671d58530d8c773ea6154c
Signed-off-by: Vaibhav Kachore <vkachore@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1803210
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
- Remove the usage of nvgpu_ctxsw_trace_entry splattered
across nvgpu, and replace with a struct defined in common code.
The usage is still inside Linux, but this helps the
subsequent unification efforts, e.g. to unify the fecs trace
path.
- Remove "asm/barrier.h" as "nvgpu/barrier.h" is already
included.
EVLR-3078
Change-Id: Iabfb105b891b0078ed326a8047ef14ebe1888cf2
Signed-off-by: Vaibhav Kachore <vkachore@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1803208
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
The graphics subctx header object is nothing but memory. Drop the
dependency to gr header file in the channel header file and substitute
struct nvgpu_mem for struct ctx_header_desc.
Jira NVGPU-967
Change-Id: Ic3976391016c42d2ada4aac3e0851a1222244ce9
Signed-off-by: Konsta Holtta <kholtta@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1807370
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
In the current code, gk20a.h includes io.h which gets directly included
in a lot of other files. io.h contains methods which uses a struct
gk20a as a parameter leading to a circular dependency between io.h
and gk20a.h. This can be mitigated by removing io.h from gk20a.h as
part of larger effort to moving gk20a.h to nvgpu/gk20a.h
JIRA NVGPU-597
Change-Id: I93e504fa9371b88152737b342a75580c65e8f712
Signed-off-by: Debarshi Dutta <ddutta@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1787316
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Now that GR buffers always have a kernel mapping, remove the unnecessary
calls to nvgpu_mem_begin() and nvgpu_mem_end() on these buffers:
- global ctx buffer mem in gr
- gr ctx mem in a tsg
- patch ctx mem in a gr ctx
- pm ctx mem in a gr ctx
- ctx_header mem in a channel (subctx header)
Change-Id: Id2a8ad108aef8db8b16dce5bae8003bbcd3b23e4
Signed-off-by: Konsta Holtta <kholtta@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1760599
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Alex Waterman <alexw@nvidia.com>
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
- On t186, ucode expects physical address to be
programmed for FECS trace buffer.
- On t194, ucode expects GPU VA to be programmed
for FECS trace buffer. This patch adds extra
support to handle this change for linux native.
- Increase the size of FECS trace buffer (as few
entries were getting dropped due to overflow of
FECS trace buffer.)
- This moves FECS trace buffer handling in global
context buffer.
- This adds extra check for updation of mailbox1
register. (Bug 200417403)
EVLR-2077
Change-Id: I7c3324ce9341976a1375e0afe6c53c424a053723
Signed-off-by: Vaibhav Kachore <vkachore@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1536028
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Nirav Patel <nipatel@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Switch all logging to nvgpu_log*(). gk20a_dbg* macros are
intentionally left there because of use from other repositories.
Because the new functions do not work without a pointer to struct
gk20a, and piping it just for logging is excessive, some log messages
are deleted.
Change-Id: I00e22e75fe4596a330bb0282ab4774b3639ee31e
Signed-off-by: Terje Bergstrom <tbergstrom@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1704148
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>
When using a coherent DMA API wee must make sure to program
any aperture fields with the coherent aperture setting. To
do this the nvgpu_aperture_mask() function was modified to
take a third aperture mask argument, a coherent setting, so
that code can use this function to generate coherent aperture
settings.
The aperture choice is some what tricky: the default version
of this function uses the state of the DMA API to determine
what aperture to use for SYSMEM: either coherent or
non-coherent internally. Thus a kernel user need only specify
the normal nvgpu_mem struct and the correct mask should be
chosen. Due to many uses of nvgpu_mem structs not created
directly from the DMA API wrapper it's easier to translate
SYSMEM to SYSMEM_COH after creation.
However, the GMMU mapping code, will encounter buffers from
userspace with difference coerency attributes than the DMA
API. Thus the __nvgpu_aperture_mask() really respects the
aperture setting passed in regardless of the DMA API state.
This aperture setting is pulled from NVGPU_VM_MAP_IO_COHERENT
since this is either passed in from userspace or set by the
kernel when using coherent DMA. The aperture field in attrs
is upgraded to coh if this flag is set.
This change also adds a coherent sysmem mask everywhere that
it can. There's a couple places that do not have a coherent
register field defined yet. These need to eventually be
defined and added.
Lastly the aperture mask code has been mvoed from the Linux
vm.c code to the general vm.c code since this function has
no Linux dependencies.
Note: depends on https://git-master.nvidia.com/r/1664536 for
new register fields.
JIRA EVLR-2333
Change-Id: I4b347911ecb7c511738563fe6c34d0e6aa380d71
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1655220
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Split the ctxsw trace "core" API code into <nvgpu/ctxsw_trace.h>. This
is not perect though since there's some Linuxisms present in the HAL
and as such that code has to be hidden by the ctxsw tracing CONFIG. But
this patch should work for QNX such that it will allow the code to
build as long as CONFIG_GK20A_CTXSW_TRACE is not set.
Also fix the copywrite notice in the ctxsw code present under
common/linux to be GPL.
JIRA NVGPU-287
Change-Id: I94715864caf335b7220185492e4629d713b025e0
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1589429
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Move much of the remaining generic MM code to a new common location:
common/mm/mm.c. Also add a corresponding <nvgpu/mm.h> header. This
mostly consists of init and cleanup code to handle the common MM
data structures like the VIDMEM code, address spaces for various
engines, etc.
A few more indepth changes were made as well.
1. alloc_inst_block() has been added to the MM HAL. This used to be
defined directly in the gk20a code but it used a register. As a
result, if this register hypothetically changes in the future,
it would need to become a HAL anyway. This path preempts that
and for now just defines all HALs to use the gk20a version.
2. Rename as much as possible: global functions are, for the most
part, prepended with nvgpu (there are a few exceptions which I
have yet to decide what to do with). Functions that are static
are renamed to be as consistent with their functionality as
possible since in some cases function effect and function name
have diverged.
JIRA NVGPU-30
Change-Id: Ic948f1ecc2f7976eba4bb7169a44b7226bb7c0b5
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1574499
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
This change solves crashes during bind that were introduced in the driver
during the OS unification refactoring due to lack of coverage of the remove()
function.
The fixes during remove are:
(1) Prevent NULL dereference on GPUs with secure boot
(2) Prevent NULL dereferences when fecs_trace is not enabled
(3) Added PRAMIN blocker during driver removal if HW is no longer accesible
(4) Prevent double free of debugfs nodes as they are handled on the
debugfs_remove_recursive() call
(5) quiesce() can now be called without checking is HW accesible flag is set
(6) added function to free irq so no IRQ association is left on the driver after
it is removed
(7) prevent NULL dereference on nvgpu_thread_stop() if the thread is already
stopped
JIRA: EVLR-1739
Change-Id: I787d38f202d5267a6b34815f23e1bc88110e8455
Signed-off-by: David Nieto <dmartineznie@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1563005
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Reorganize HAL initialization to remove inheritance and construct
the gpu_ops struct at compile time. This patch only covers the
fecs_trace sub-module of the gpu_ops struct.
Perform HAL function assignments in hal_gxxxx.c through the
population of a chip-specific copy of gpu_ops.
Jira NVGPU-74
Change-Id: I84485ad64997270c6a0fce3c95dc9ceb0094cf6c
Signed-off-by: Sunny He <suhe@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1527419
Reviewed-by: Automatic_Commit_Validation_User
GVS: Gerrit_Virtual_Submit
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
