The problem here, and the solution, requires some background
so let's start there.
During page table programming page directories (PDs) are
allocated as needed. Each PD can range in size, depending on
chip, from 256 bytes all the way up to 32KB (gk20a 2-level
page tables).
In HW, two distinct PTE sizes are supported: large and small.
The HW supports mixing these at will. The second to last level
PDE has pointers to both a small and large PD with
corresponding PTEs. Nvgpu doesn't handle that well and as a
result historically we split the GPU virtual address space
up into a small page region and a large page region. This
makes the GMMU programming logic easier since we now only have
to worry about one type of PD for any given region.
But this presents issues for CUDA and UVM. They want to be
able to mix PTE sizes in the same GPU virtual memory range.
In general we still don't support true dual page directories.
That is page directories with both the small and large next
level PD populated. However, we will allow adjecent PDs to
have different sized next-level PDs.
Each last level PD maps the same amount. On Pascal+ that's
2MB. This is true regardless of the PTE coverage (large or
small). That means the last level PD will be different in
size depending on the PTE size.
So - going back to the SW we allocate PDs as needed when
programming the page tables. When we do this allocation we
allocate just enough space for the PD to contain the
necessary number of PTEs for the page size. The problem
manifests when a PD flips in size from large to small PTEs.
Consider the following mapping operations:
map(gpu_va -> phys) [large-pages]
unmap(gpu_va)
map(gpu_va -> phys) [small-pages]
In the first map/unmap we go and allocate all the necessary
PDs and PTEs to build this translation. We do so assuming a
large page size. When unmapping, as an optimzation/quirk of
nvgpu, we leave the PDs around. We know they may well be used
again in the future.
But if we swap the size of the mapping from large to small
then we now need more space in the PD for PTEs. But the logic
in the GMMU coding assumes if the PD has memory allocated then
that memory is sufficient. This worked back when there was no
potential for a PD to swap in page size. But now that there is
we have to re-allocate the PD doesn't have enough space for
the required PTEs.
So that's the fix - reallocate PDs when they require more
space than they currently have.
Change-Id: I9de70da6acfd20c13d7bdd54232e4d4657840394
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1933076
Reviewed-by: svc-misra-checker <svc-misra-checker@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Nicolas Benech <nbenech@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>
Add two new sub-directories under MM: gmmu and allocators.
The allocators directory is for all the allocator code we have.
There's a fair amount and as such could be considered a component
with a bunch of sub-units.
The new GMMU directory will contain the GMMU component (which used to
be a single unit). The new GMMU component is comprised of the
page_table and pd_cache units. Also when we migrate the chip specific
GMMU code out of mm_gk20a.c and mm_gp10b.c it will be placed in this
new GMMU directory.
JIRA NVGPU-1390
Change-Id: I7aa47ea2a32612b7d69972671fccb72770e1ae09
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1944385
Reviewed-by: Nicolas Benech <nbenech@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>
This A) clears a dangling pointer which should make the code
more robust, and B) allows for easier unit testing of the
nvgpu_pd_cache_do_free() functions since there's now a
tangible change to the nvgpu_gmmu_pd after this function runs.
JIRA NVGPU-1323
Change-Id: I57db02c9e74324b8e3c3fd4a2c14565dfd0048aa
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1949936
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Nicolas Benech <nbenech@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 following error checking code in nvgpu_pd_cache_alloc() cannot
hit the greater than PAGE_SIZE check:
if ((bytes & (bytes - 1U)) != 0U ||
(bytes >= PAGE_SIZE ||
bytes < NVGPU_PD_CACHE_MIN)) {
/* ... */
This is because the nvgpu_pd_cache_alloc() function is only called,
specifically when, bytes is less than PAGE_SIZE! As such we would
only see this case when there's a bug.
So change the error condition to now check only for bytes being a
power of 2 and being greater than NVGPU_PD_CACHE_MIN. The greater
than page size check has been turned into an assert since this
should really never happen in practice unless there's a bug.
JIRA NVGPU-1323
Change-Id: I97cd701aa4f045345606b90c97a8478b4a06e189
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1946731
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
MISRA rule 14.4 doesn't allow the usage of integer types as booleans
in the controlling expression of an if statement or an iteration
statement.
Fix violations where the result of a bitwise operation is used as a
boolean in the controlling expression of if and loop statements.
JIRA NVGPU-1020
Change-Id: I6a756ee1bbb45d43f424d2251eebbc26278db417
Signed-off-by: Amurthyreddy <amurthyreddy@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1936334
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
MISRA rule 14.4 doesn't allow the usage of non-boolean variable as
boolean in the controlling expression of an if statement or an
iteration statement.
Fix violations where a non-boolean variable is used as a boolean in the
controlling expression of if and loop statements.
JIRA NVGPU-1022
Change-Id: Ia96f3bc6ca645ba8538faf7a9fa3a9ccf9df40d3
Signed-off-by: Amurthyreddy <amurthyreddy@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1943168
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Fixed the erroneous timeout message in sync-unmap and corrected the
condition for returning ETIMEDOUT.
In the original codeflow, after waiting for mapped_buffer release,
nvgpu_timeout_expired is called to check whether to return ETIMEDOUT.
However, if there is a delay between the end of the waiting and the
nvgpu_timeout_expired check, ETIMEDOUT is returned regardless if the
mapped_buffer is released with the timeout message printed. This is
an incorrect behavior. This patch fixes it by letting the refcount of
the mapped_buffer be the only source to determine the return value.
Bug 200434475
Change-Id: I8ca170c811da415c24045ab643da26476bc7463c
Signed-off-by: Kyle Guo <kyleg@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1945388
Reviewed-by: svc-misra-checker <svc-misra-checker@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Alex Waterman <alexw@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
This makes unit testing easier because it hides a difficult to
test branch under POSIX. This branch is not functional to the
rest of the code in the pd_cache so moving it to the POSIX
header should avoid headache of testing it in the pd_cache
code.
JIRA NVGPU-1323
Change-Id: Id5ca2627c83cf6dbbe68dd8ad7bfe9def71761cc
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1945145
Reviewed-by: svc-misra-checker <svc-misra-checker@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>
MISRA rule 14.4 doesn't allow the usage of non-boolean variable as
boolean in the controlling expression of an if statement or an
iteration statement.
Fix violations where a non-boolean variable is used as a boolean in the
controlling expression of if and loop statements.
JIRA NVGPU-1022
Change-Id: I957f8ca1fa0eb00928c476960da1e6e420781c09
Signed-off-by: Amurthyreddy <amurthyreddy@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1941002
GVS: Gerrit_Virtual_Submit
Reviewed-by: Adeel Raza <araza@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
MISRA Rule-15.6 requires that all if-else blocks and loop blocks
be enclosed in braces, including single statement blocks. Fix errors
due to single statement if-else and loop blocks without braces
by introducing the braces.
JIRA NVGPU-775
Change-Id: Ib70621d39735abae3fd2eb7ccf77f36125e2d7b7
Signed-off-by: Srirangan Madhavan <smadhavan@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1928745
GVS: Gerrit_Virtual_Submit
Reviewed-by: Adeel Raza <araza@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
MISRA Rule 21.15 prohibits use of memcpy() with incompatible ptrs
to qualified/unqualified types.
To circumvent this issue we've introduced a new MISRA-compliant
nvgpu_memcpy() function.
This change switches all offending uses of memcpy() in mm/*
over to use nvgpu_memcpy() with appropriate casts applied.
JIRA NVGPU-849
Change-Id: I17be87475fde62e969b014d4d0fa455dae5d4373
Signed-off-by: Scott Long <scottl@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1941257
Reviewed-by: Philip Elcan <pelcan@nvidia.com>
Reviewed-by: svc-misra-checker <svc-misra-checker@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Alex Waterman <alexw@nvidia.com>
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
MISRA Rule-17.7 requires the return value of all functions to be used.
Fix is either to use the return value or change the function to return
void. This patch contains fix for all 17.7 violations instandard C functions
in common code.
JIRA NVGPU-1036
Change-Id: Id6dea92df371e71b22b54cd7a521fc22812f9b69
Signed-off-by: Nicolas Benech <nbenech@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1929899
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Sometimes longer 64-bit values are composed of pairs of low and high
parts in a same nvgpu_mem. Add a function to read the two and to do the
bit ops given the low and high word offsets.
Jira NVGPU-886
Change-Id: I4fe698d7c5a891ef5740ea0fb31549195e2e6c43
Signed-off-by: Konsta Holtta <kholtta@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1928206
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Update the PD cache code to handle 64KB pages. To do this the
number of partial/full lists is expanded for when we have 64K
pages. Currently we only explicitly support 4K and 64K page
sizes. Other pages sizes (16K for example) will fail compilation
during preprocessing.
This change also cleans up the definitions for some of the
internal structs. They have been moved into pd_cache.c since
they are not used outside of pd_cache.c.
This allows the following functions to be removed from the global
context:
__nvgpu_pd_cache_alloc_direct()
__nvgpu_pd_cache_free_direct()
They have been replaced by calls to nvgpu_pd_{alloc,free}().
The nvgpu_pd_mem_entry alloc_map also had to be expanded to a
real bitmap. 32 or 64 bits is not sufficient for packing 256
byte PDs into a 64K page (there's 256 PDs per nvgpu_pd_mem_entry
in that case). To prevent doing too many find_first_zero
operations on the bitmap an 'allocs' field was also added which
tracks how many allocs are done. We can use this instead of
comparing a mask against the bitmap to determine if an
nvgpu_pd_mem_entry is full.
Note: there's still a limitation with the TLB invalidate code:
it simply assumes an nvgpu_mem is a 1 to 1 with a PDB. This means
we can't invalidate a PDB allocated at an offset greater than 0
in a nvgpu_pd_mem_entry. This in turn means we must always use a
full page size for a context's PDB.
Bug 1977822
Change-Id: I6a7a3a95b7c902bc6487cba05fde58fbc4a25da5
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1718755
Reviewed-by: Konsta Holtta <kholtta@nvidia.com>
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Background:
In Hypervisor mode dGPU device is configured in pass through mode for
the Guest (QNX/Linux). GMMU programming is handled by the guest which
converts a mapped buffer's GVA into SGLes in IPA (Intermediate/Guest
Physical address) which is then translated into PA (Acutual Physical
address) and programs the GMMU PTEes with correct GVA to PA mapping.
Incase of the vgpu this work is delegated to the RM server which takes care
of the GMMU programming and IPA to PA conversion.
Problem:
The current GMMU mapping logic in the guest assumes that PA range is
continuous over a given IPA range. Hence, it doesn't account for holes being
present in the PA range. But this is not the case, a continous IPA range
can be mapped to dis-contiguous PA ranges. In this situation the mapping
logic sets up GMMU PTEes ignoring the holes in physical memory and
creates GVA => PA mapping which intrudes into the PA ranges which are
reserved. This results in memory being corrupted.
This change takes into account holes being present in a given PA range and
for a given IPA range it also identifies the discontiguous PA ranges and
sets up the PTE's appropriately.
Bug 200451447
Jira VQRM-5069
Change-Id: I354d984f6c44482e4576a173fce1e90ab52283ac
Signed-off-by: aalex <aalex@nvidia.com>
Signed-off-by: Antony Clince Alex <aalex@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1850972
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
MISRA rule-14.4 doesn't allow the usage of function pointers & integer
types as booleans in the controlling expression of an if statement or
an iteration statement.
Fix violations where a function pointer or a function whose return
value is an integer, is used as a boolean in the controlling expression
of if and loop statements.
JIRA NVGPU-1021
Change-Id: Ic5336268394ba4396ce80744c25930d2fb44dc42
Signed-off-by: Amurthyreddy <amurthyreddy@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1932147
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Address MISRA Rule 20.7 violation: Macro parameter expands into an
expression without being wrapped by parentheses.
Some of the exception the coverity is not able to catch are:
1. Macro parameters passed as parameter to another macro.
i.e NVGPU_ACCESS_ONCE. Fixing these by additional parantheses.
2. Macro parameter used as type. i.e. type parameter in container_of.
While at it, update copyright date rage for list.h and types.h.
JIRA NVGPU-841
Change-Id: I4b793981d671069289720e8c041bad8125961c0c
Signed-off-by: Sagar Kamble <skamble@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1929823
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
Reviewed-by: Automatic_Commit_Validation_User
GVS: Gerrit_Virtual_Submit
Reviewed-by: Konsta Holtta <kholtta@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>
Fixed timeout checking flow in sync-unmap where timeout message could
be printed while there is not a timeout.
The original codeflow doesn't check refcount of a mapped_buffer again
after 10ms waiting. So if the buffer is released during the last
round of waiting, the timeout message will still be printed. The new
codeflow combines refcount and timeout checking. So that there won't
be an inconsistency between the two.
Bug 200434475
Change-Id: Id0a0aebcb24906a1ec7113e669227788e729564b
Signed-off-by: Kyle Guo <kyleg@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1930236
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>
struct gk20a from gk20a.h needs defination of struct gk20a_ce_app
and ce2_gk20a.h needs defination of struct gk20a. This creates
a circular dependency.
Fix this by making gk20a_ce_app a pointer to skip knowing the
complete type details and using forward declarations for struct
gk20a_ce_app and struct gk20a.
The gk20a_ce_app pointer is alloc'ed in gk20a_init_ce_support()
and free'ed in gk20a_ce_destroy.
JIRA NVGPU-611
Change-Id: I4d62d5f2b2d1492db73bae69f90a1fe5586fba76
Signed-off-by: Nitin Kumbhar <nkumbhar@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1917945
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
This patch moves the increment and decrement of the user mapped
buffer count to the insert/remove mapped buffer functions since
this value should only ever change when these functions are called.
Bug 200105199
Change-Id: I5b0a86d00e9e948c48e313153a668eb2e10fca49
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1917791
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Nvgpu uses many ways to check if sync points are enabled. The four
ways used to be:
platform->has_syncpoints
g->has_syncpoints
nvgpu_is_enabled(g, NVPGU_HAS_SYNCPOINTS)
gk20a_platform_has_syncpoints()
This patch standardizes all usage to now be nvgpu_has_syncpoints()
which is based on gk20a_platform_has_syncpoints() - just renamed to
be general to nvgpu.
All usage of the other forms have now been consolidated. However,
under the hood nvgpu_has_syncpoints() does check the is_enabled
flag. This flag is now set where g->has_syncpoints used to be set
based on the platform data.
The basic dependency chain is this:
nvgpu_has_syncpoints -> NVGPU_HAS_SYNCPOINTS ->
platform->has_syncpoints
However, note: there are several places where syncpoints can be
disabled if some other driver initialization fails (for ex. host1x).
Also note that nvgpu_has_syncpoints() also considers a disable
variable set by debugfs.
Bug 2327574
Change-Id: Ia2375a80f5f2e27285e6175568dd13e6bb25fd33
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1803975
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Input paran VM Will be used to get PTE size for GVA spaces and it can be
NULL if buddy_allocator initialization is not for GVA space. As part of
recent changes VM was being accessed for non GVA space case and for that
VM was NULL, causing the crash.
In this fix added a prior check and VM will be accessed only if input
flag is set to GPU_ALLOC_GVA_SPACE.
Bug 200452730
Change-Id: I9dcb1f2bd5e639c37a7152b99453bb5b46fe4087
Signed-off-by: Amurthyreddy <amurthyreddy@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1842530
Reviewed-by: svc-misra-checker <svc-misra-checker@nvidia.com>
Reviewed-by: Automatic_Commit_Validation_User
GVS: Gerrit_Virtual_Submit
Tested-by: Prateek Sethi <prsethi@nvidia.com>
Reviewed-by: Vaibhav Kachore <vkachore@nvidia.com>
Reviewed-by: Vijayakumar Subbu <vsubbu@nvidia.com>
Reviewed-by: Deepak Nibade <dnibade@nvidia.com>
Reviewed-by: Alex Waterman <alexw@nvidia.com>
MISRA rule 21.2 doesn't allow the use of macro names which start with
an underscore. These leading underscores are to be removed from the
macro names. This patch will fix such violations in common directory
by renaming them to follow the convention,
'NVGPU_PARENT-DIR_HEADER-NAME' when there is no keyword repetition
between file name and directory or 'NVGPU_HEADER-NAME' when there
is repetition.
JIRA NVGPU-1028
Change-Id: Idf10f6b179cfd96bfb8ab8e9e2bf79c26591905d
Signed-off-by: smadhavan <smadhavan@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1809086
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
We have a h/w bug on some chips and we need to support below additional
HALs to implement a s/w WAR
gops.fifo.init_pdb_cache_war()
gops.fifo.deinit_pdb_cache_war()
gops.fb.apply_pdb_cache_war()
Add new API nvgpu_init_mm_pdb_cache_war() to initialize WAR sequence
and call this from MM initialization and before setting up rest of the
memory management units
Deinitialize WAR while cleaning up MM support
Add pdb_cache_war_mem member to gk20a to hold all the memory needed
for the WAR
Bug 200449545
Change-Id: Id2ac0d940c7881c7b0cf396413273c0f329a1a1f
Signed-off-by: Deepak Nibade <dnibade@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1834901
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
-Created struct nvgpu_acr to hold acr module related member
within single struct which are currently spread across multiple structs
like nvgpu_pmu, pmu_ops & gk20a.
-Created struct hs_flcn_bl struct to hold ACR HS bootloader specific members
-Created struct hs_acr to hold ACR ucode specific members like bootloader data
using struct hs_flcn_bl, acr type & falcon info on which ACR ucode need to run.
-Created acr ops under struct nvgpu_acr to perform ACR specific operation,
currently ACR ops were part PMU which caused to have always dependence
on PMU even though ACR was not executing on PMU.
-Added acr_remove_support ops which will be called as part of
gk20a_remove_support() method, earlier acr cleanup was part of
pmu remove_support method.
-Created define for ACR types,
-Ops acr_sw_init() function helps to set ACR properties
statically for chip currently in execution & assign ops to point to
needed functions as per chip.
-Ops acr_sw_init execute at early as nvgpu_init_mm_support calls acr
function to alloc blob space.
-Created ops to fill bootloader descriptor & to patch WPR info to ACR uocde
based on interfaces used to bootstrap ACR ucode.
-Created function gm20b_bootstrap_hs_acr() function which is now common
HAL for all chips to bootstrap ACR, earlier had 3 different function for
gm20b/gp10b, gv11b & for all dgpu based on interface needed.
-Removed duplicate code for falcon engine wherever common falcon code can be used.
-Removed ACR code dependent on PMU & made changes to use from nvgpu_acr.
JIRA NVGPU-1148
Change-Id: I39951d2fc9a0bb7ee6057e0fa06da78045d47590
Signed-off-by: Mahantesh Kumbar <mkumbar@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1813231
GVS: Gerrit_Virtual_Submit
Reviewed-by: svc-misra-checker <svc-misra-checker@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>
MISRA 21.2 states that we may not use reserved identifiers; since
all identifiers beginning with '_' are reserved by libc, the usage
of '__' as a prefix is disallowed.
Fixes for all the pd_cache functions that use '__' prefixes. This
was trivial: the '__' prefix was simply deleted.
JIRA NVGPU-1029
Change-Id: Ia91dabe3ef97fb17a2a85105935fb3a72d7c2c5e
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1813643
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
MISRA 21.2 states that we may not use reserved identifiers; since
all identifiers beginning with '_' are reserved by libc, the usage
of '__' as a prefix is disallowed.
Handle the 21.2 fixes for nvgpu_mem.c and mm.c; this deletes the
'__' prefixes and slightly renames the __nvgpu_aperture_mask()
function since there's a coherent version and a general version.
Change-Id: Iee871ad90db3f2622f9099bd9992eb994e0fbf34
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1813623
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
MISRA 21.2 states that we may not use reserved identifiers; since
all identifiers beginning with '_' are reserved by libc, the usage
of '__' as a prefix is disallowed.
Fixes for the buddy allocator. These are all pretty trivial; mostly
the '__' prefix is simply deleted since the functions are all static.
The one exception is the list accessors were also renamed slightly to
include a 'do' component to show more clearly that the other list
accessors are convenience functions.
JIRA NVGPU-1029
Change-Id: Ia39cdde9a81e740537fc9b4d70d87eadc12f8d4d
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1813601
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
MISRA 21.2 states that we may not use reserved identifiers; since
all identifiers beginning with '_' are reserved by libc, the usage
of '__' as a prefix is disallowed.
This fixes places in the public allocator APIs. This consists of
the various init routines which are used to create an allocator
and the debug macro used within the allocator code.
The buddy allocator was handled by collapsing the internal
'__' prepended version with the non-prefixed version. The only
required change was in the page_allocator code which now had to
pass in a NULL vm pointer (since the VM is not needed for managing
VIDMEM).
JIRA NVGPU-1029
Change-Id: I484a144e61789bf594c525c1ca307b96d120830f
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1813578
Reviewed-by: svc-misra-checker <svc-misra-checker@nvidia.com>
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Debug page was allocated and programmed to HUB MMU in GR code. This
introduces a dependency from GR to FB and is anyway the wrong place.
Move the code to allocate memory to generic MM code, and the code
to program the addresses to FB.
Change-Id: Ib6d3c96efde6794cf5e8cd4c908525c85b57c233
Signed-off-by: Terje Bergstrom <tbergstrom@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1801423
Reviewed-by: svc-misra-checker <svc-misra-checker@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
The issues are:
1. Non-fixed allocs must take into account explicit PTE size
requests. Previously the PTE size was determines from the
allocation size which was incorect. To do this, the PTE size
is now plumbed through all GPU VA allocations. This is what
the new alloc_pte() op does.
2. Fix buddy PTE size assignment. This changes a '<=' into a
'<' in the buddy allocation logic. Effectively this is now
leaving the PTE size for buddy blocks equal to the PDE block
size as 'ANY'.
This prevents a buddy block of PDE size which has yet to be
allocated from having a specific PDE size. Without this its
possible to do a fixed alloc that fails unexpectedly due to
mismatching PDE sizes.
Consider two PDE block sized fixed allocs that are contained
in one buddy twice the size of a PDE block. Let's call these
fixed allocs S and B (small and big). Let's assume that two
fixed allocs are done, each targeting S and B, in that order.
With the current logic the first alloc, when we create the
two buddies S and B, causes both S and B to have a PTE size of
SMALL. Now when the second alloc happens we attempt to find
a buddy B with a PTE size of either BIG or ANY. But we cannot
becasue B already has size SMALL. This casues us to appear
like we have a conflicting fixed alloc despite this not being
the case.
3. Misc cleanups & bug fixes:
- Clean up some MISRA issues
- Delete an extraneous unlock that could have caused a
deadlock.
Bug 200105199
Change-Id: Ib5447ec6705a5a289ac0cf3d5e90c79b5d67582d
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1768582
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
We use bootstrap vidmem allocator for all the vidmem allocations that happen
boot time
And we need to program physical address for all the potential vidmem buffers
that we program into h/w and are needed during boot
So force the allocator to allocate contigous memory
We otherwise see a warning dump when we program physical address of memory which
is allocated in multiple pages
Bug 2180284
Jira NVGPUT-12
Change-Id: Ib9c2d42ea463bc424c2cb4da8ffd8ebae436e0f6
Signed-off-by: Deepak Nibade <dnibade@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1805467
Reviewed-by: Automatic_Commit_Validation_User
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>
We right now have a bootstrap carveout in vidmem of size 16M and having base
address at {total_vidmem_size - 256M}
So this design divides rest of the vidmem into two chunks
And the size of bootstrap carveout is also small and insufficient for vidmem
allocations during boot
Hence increase the bootstrap vidmem carevout to 256M and move it to the end
of entire vidmem
Rename the carevout name for wpr_co to bootstrap_co as it is more appropriate
Also update __nvgpu_vidmem_do_clear_all() to clear only one chunk of vidmem
instead of two
Bug 2180284
Jira NVGPUT-12
Change-Id: I9c8d62bcd705c7112385df3d4f714e0190b48e17
Signed-off-by: Deepak Nibade <dnibade@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1805466
GVS: Gerrit_Virtual_Submit
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
Reviewed-by: Seshendra Gadagottu <sgadagottu@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
MISRA 21.2 states that we may not use reserved identifiers; since
all identifiers beginning with '_' are reserved by libc, the usage
of '__' as a prefix is disallowed.
This change fixes uses of '__' prepended to static functions defined
in page_allocator.c. Since these funcitons were static and had no
naming conflicts the '__' was simply removed. For free_pages() this
also needed to have a nvgpu specific prefix since free_pages()
conflicts with a Linux kernel function name.
JIRA NVGPU-1029
Change-Id: I8b96e55244bd2de166f7dcc64d2b30427757ea8f
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1812826
Reviewed-by: Vinod Gopalakrishnakurup <vinodg@nvidia.com>
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
MISRA Rule-17.7 requires the return value of all functions to be used.
Fix is either to use the return value or change the function to return
void. This patch contains fix for calls to nvgpu_mutex_init and
improves related error handling.
JIRA NVGPU-677
Change-Id: I609fa138520cc7ccfdd5aa0e7fd28c8ca0b3a21c
Signed-off-by: Nicolas Benech <nbenech@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1805598
Reviewed-by: svc-misra-checker <svc-misra-checker@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Alex Waterman <alexw@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
This migrates the common DMA code (os agnostic) to the
common directory. This new unit will be the common DMA
allocator that lets users allocate SYSMEM, VIDMEM, or
either. Other units will be responsible for actually
handling the mechanics of allocating VIDMEM or SYSMEM.
Also update the names of the DMA related files so that
tmake doesn't complain about duplicate C file names. To
do this call the common DMA file dma.c and prepend the
OS to the other DMA files. So now we have:
common/mm/dma.c
os/posix/posix-dma.c
os/linux/linux-dma.c
JIRA NVGPU-990
Change-Id: I22d2d41803ad89be7d9c28f87864ce4fedf10836
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1799807
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
