This patch does a couple of things. First it renames
NVGPU_DMA_COHERENT to NVGPU_USE_COHERENT_SYSMEM since the former
is somewhat ambiguous in meaning. The latter clearly states what
must happen: nvgpu needs to treat sysmem as coherent. This flag
does simply follow the state of the DMA API but there's no reason
to expect a casual reader of the code to know that when the DMA
API is coherent nvgpu must treat sysmem as coherent.
One thing to note though: when the dGPU is using PCIe and the
PCIe controller is coherent, it doesn't actually matter what we
do. However, we use this flag for determining how to make CPU
mappings in nvgpu_mem_begin() so this flag is still relevant for
the CPU side of things.
Next this patch adds a check in the core kernel GMMU mapping
routine to make sure that when the NVGPU_USE_COHERENT_SYSMEM flag
is set that the IO coherent flag is passed into the mapping code.
This is the primary fix that made NVLINK start working.
Finally the setting of the USE_COHERENT_SYSMEM flag and the
NVGPU_SUPPORT_IO_COHERENCE flag were set both for PCIe and for
iGPUs. The iGPU also must correctly match it's CPU mappings and
GPU mappings for proper operation.
JIRA EVLR-2333
Change-Id: Icd5f07167c9f48a0a2e8493e34c9cc6238e56907
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1654519
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Add User space API NVGPU_AS_IOCTL_GET_SYNC_RO_MAP to get read-only syncpoint
address map in user space
We already map whole syncpoint shim to each address space with base address
being vm->syncpt_ro_map_gpu_va
This new API exposes this base GPU_VA address of syncpoint map, and unit size
of each syncpoint to user space.
User space can then calculate address of each syncpoint as
syncpoint_address = base_gpu_va + (syncpoint_id * syncpoint_unit_size)
Note that this syncpoint address is read_only, and should be only used for
inserting semaphore acquires.
Adding semaphore release with this address would result in MMU_FAULT
Define new HAL g->ops.fifo.get_sync_ro_map and set this for all GPUs supported
on Xavier SoC
Bug 200327559
Change-Id: Ica0db48fc28fdd0ff2a5eb09574dac843dc5e4fd
Signed-off-by: Deepak Nibade <dnibade@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1649365
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 GPU has multiple different operating modes in respect to IOMMU'ability.
As such there needs to be a clean way to tell the driver whether it is
IOMMU'able or not. This state also does not always reflect what is possible:
all becasue the GPU can generate IOMMU'ed memory requests doesn't mean it
wants to.
The nvgpu_iommuable() API has now existed for a little while which is a
useful way to convey whether nvgpu should consider the GPU as IOMMU'able.
However, there is also the g->mm.bypass_smmu flag which used to be able to
override what the GPU decided it should do. Typically it was assigned
the same value as nvgpu_iommuable() but that was not necessarily a
requirment.
This patch removes all the usages of g->mm.bypass_smmu and instead uses the
nvgpu_iommuable() function. All places where the check against
g->mm.bypass_smmu have been replaced with nvgpu_iommuable(). The code
should now be much cleaner.
Subsequently other checks can also be placed in the nvgpu_iommuable()
function. For example, when NVLINK comes online and the GPU should no
longer consider DMA addresses and instead use scatter-gather lists
directly the ngpu_iommuable() function will be able to check the state of
NVLINK and then act accordingly.
Change-Id: I0da6262386de15709decac89d63d3eecfec20cd7
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1648332
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
These '\n' were leftover from the previous debugging macro usage
which did no add the '\n' automagically. However, once swapped over
to the nvgpu logging system the '\n' is added and no longer needs
to be present in the code.
This did require one extra modification though to keep things
consistent. The __alloc_pstat() macro, used for sending output
either to a seq_file or the terminal, needed to add the '\n' for
seq_printf() calls and the '\n' had to be deleted in the C files.
Change-Id: I4d56317fe2a87bd00033cfe79d06ffc048d91049
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1613641
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Lots of code paths were split to T19x specific code paths and structs
due to split repository. Now that repositories are merged, fold all of
them back to main code paths and structs and remove the T19x specific
Kconfig flag.
Change-Id: Id0d17a5f0610fc0b49f51ab6664e716dc8b222b6
Signed-off-by: Terje Bergstrom <tbergstrom@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1640606
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
In the buddy allocator use the actual size of the PDE to determine
the pte_blk_order field which is used to determine what page size a
buddy has (or doesn't). Previously this was just set as the large
page size times 1024 which would over allocate PDE ranges for Pascal+
chips. This caused userspace, which was using the real PDE size, to
sometime allocate small and large pages in what the buddy allocator
mistakenly thought was one PDE.
Bug 200105199
Change-Id: I7ab7db7962015fc268bad61b558a18704133e1cb
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1639731
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>
Add a field in vm_gk20a to identify guest managed VM, with the
corresponding checks to ensure that there's no kernel section for
guest managed VMs.
Also make the __nvgpu_vm_init function available globally, so that
the vm can be allocated elsewhere, requisite fields set, and passed
to the function to initialize the vm.
Change-Id: Iad841d1b8ff9c894fe9d350dc43d74247e9c5512
Signed-off-by: Sourab Gupta <sourabg@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1617171
Reviewed-by: Alex Waterman <alexw@nvidia.com>
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@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>
In __nvgpu_gmmu_do_update_page_table(), and in case of non-IOMMU mappings,
we call nvgpu_sgt_get_phys() to get physical address
But this API ignores mapping attributes including l3_alloc attribute
specified by user space, and this breaks L3 cache allocations
Fix this by using g->ops.mm.gpu_phys_addr() which also considers the
mapping attributes and returns appropriate physical address
Jira GPUT19X-10
Bug 200279508
Change-Id: Ibc0d29f7cb576a9d6893a97b1912d9ff4bc78e02
Signed-off-by: Deepak Nibade <dnibade@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1621245
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Seshendra Gadagottu <sgadagottu@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
Tested-by: Alex Waterman <alexw@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
The current code does not properly calculate the indexes within the PDE
to access the proper entry, and it has a bug in assignement of the big
page entries. This change fixes the issue by:
(1) Passing a pointer to the level structure and dereferencing the
index offset to the next level.
(2) Changing the format of the address.
(3) Ensuring big pages are only selected if their address is set.
Bug 200364599
Change-Id: I46e32560ee341d8cfc08c077282dcb5549d2a140
Signed-off-by: David Nieto <dmartineznie@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1610562
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Deepak Bhosale <dbhosale@nvidia.com>
Increase the ctagline that is programmed in the page tables when the
buffer offset crosses the compression page boundaries. This fixes
compressible-kind fixed-address mapping with 4k pages when the GPU VA
is not aligned by the compression page size.
Bug 1995897
Bug 2011640
Bug 2011668
Change-Id: I1f1f9750635a20a916527c9d18fda7f8aa6b1b1f
Signed-off-by: Sami Kiminki <skiminki@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1608465
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Add an alignment check for compressible-kind fixed-address
mappings. If we're using page size smaller than the comptag line
coverage window, the GPU VA and the physical buffer offset must be
aligned in respect to that window.
Bug 1995897
Bug 2011640
Bug 2011668
Change-Id: If68043ee2828d54b9398d77553d10d35cc319236
Signed-off-by: Sami Kiminki <skiminki@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1606439
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Add a translation layer to convert from the NVGPU_AS_* flags to
to new set of NVGPU_VM_MAP_* and NVGPU_VM_AREA_ALLOC_* flags.
This allows the common MM code to not depend on the UAPI header
defined for Linux.
In addition to this change a couple of other small changes were
made:
1. Deprecate, print a warning, and ignore usage of the
NVGPU_AS_MAP_BUFFER_FLAGS_MAPPABLE_COMPBITS flag.
2. Move the t19x IO coherence flag from the t19x UAPI header
to the regular UAPI header.
JIRA NVGPU-293
Change-Id: I146402b0e8617294374e63e78f8826c57cd3b291
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1599802
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Commit 81868a187f updated barrier
usage to use the nvgpu wrappers and in doing so downgraded many
plain barriers {mb(), wmb(), rmb()} to the SMP versions of these
barriers.
The SMP version of the barriers in question are only issued
when running on an SMP machine. In most of the cases mentioned
above this is fine since the barriers are present to faciliate
proper ordering across CPUs. A single CPU is always coherent
with itself, so on a non-SMP case we don't need those barriers.
However, there are a few places where the barriers in use (GMMU
page table programming, IO accessors, userd) where the barrier
usage is for communicating and establishing ordering for the
GPU. We need these barriers for both SMP machines and non-SMP
machines. Therefor we must use the plain barrier versions.
Change-Id: I376129840b7dc64af8f3f23f88057e4e81360f89
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1599744
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Simplify compbits alloc by making the alloc function re-callable for
the buffer, and making it return the comptags info. This simplifies
the calling code: alloc_or_get vs. get + alloc + get again.
Add tracking whether the allocated compbits need clearing before they
can be used in PTEs. We do this, since clearing is part of the gmmu
map call on vgpu, which can fail.
Bug 1902982
Change-Id: Ic4ab8d326910443b128e82491d302a1f49120f5b
Signed-off-by: Sami Kiminki <skiminki@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1597130
GVS: Gerrit_Virtual_Submit
Reviewed-by: Alex Waterman <alexw@nvidia.com>
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@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>
Clean up the comptag-related data structures and allocation logic. The
most important change is that we only ever try comptag allocation once
to prevent incorrect map aliasing.
If we were to retry the allocation on further map calls, the following
situation would become possible:
(1) Request compressible kind mapping for a buffer. Comptag alloc failed
and we proceed with incompressible kind fallback.
(2) Request another compressible kind mapping for a buffer. Comptag alloc
retry succeeded and now we use the compressible kind.
(3) After writes through the compressible kind mapping, the buffer is no
longer legible via the fallback incompressible kind mapping.
The other changes are about removing the unused comptag-related fields
in gk20a_comptags and nvgpu_mapped_buf, and retrieving comptags info
only for compressible buffers. We also make nvgpu_ctag_buffer_info and
nvgpu_vm_compute_compression as private mm/vm.c definitions, since
they're not used elsewhere.
Bug 1902982
Change-Id: I0c9fe48ccc585a80dd2c05ec606a079c1c1d41f1
Signed-off-by: Sami Kiminki <skiminki@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1595153
GVS: Gerrit_Virtual_Submit
Reviewed-by: Alex Waterman <alexw@nvidia.com>
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@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>
TSG/CHANNEL_SET_PRIORITY IOCTLs are deprecated and user space should be using
combination of timeslice and interleave levels to decide the priority
Hence remove the IOCTLs and all corresponding APIs
Jira NVGPU-393
Change-Id: I7cf0785689269536eca0c278c774b0e9e74f8c2f
Signed-off-by: Deepak Nibade <dnibade@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1598581
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Final VM mapping refactoring. Move most of the logic in the VM
map path to the common/mm/vm.c code and use the generic APIs
previously implemented to deal with comptags and map caching.
This also updates the mapped_buffer struct to finally be free
of the Linux dma_buf and scatter gather table pointers. This
is replaced with the nvgpu_os_buffer struct.
JIRA NVGPU-30
JIRA NVGPU-71
JIRA NVGPU-224
Change-Id: If5b32886221c3e5af2f3d7ddd4fa51dd487bb981
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1583987
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>
Drastically simplify and move the aligment computation for buffers
getting mapped into the SGT code. An SGT is all that is needed for
computing the alignment.
However, this did require that a new SGT op was added:
nvgpu_sgt_iommuable()
This function returns true if the passed SGT is IOMMU'able and must
be implemented by an SGT implementation that has IOMMU'able buffers.
If this function is left as NULL then it is assumed that the buffer
is not IOMMU'able.
Also cleanup the parameter ordering convention among all nvgpu_sgt
functions. Previously there was a mishmash of different parameter
orderings. This patch now standardizes on the gk20a first approach
seen everywhere else in the driver.
JIRA NVGPU-30
JIRA NVGPU-246
JIRA NVGPU-71
Change-Id: Ic4ab7b752847cf795c7cfafed5a07818217bba86
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1583985
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>
Since NVGPU_AS_MAP_BUFFER_FLAGS_DIRECT_KIND_CTRL was made mandatory,
kernel does not need to know the details about the PTE kinds
anymore. Thus, we can remove the kind_gk20a.h header and the code
related to kind table setup, as well as simplify buffer mapping code
a bit.
Bug 1902982
Change-Id: Iaf798023c219a64fb0a84da09431c5ce4bc046eb
Signed-off-by: Sami Kiminki <skiminki@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1560933
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Request explicitly contiguous DMA memory for large page directory
allocations. Large in this case means greater than PAGE_SIZE. This
is necessary if the GPU's DMA allocator is set to, by default,
allocate discontiguous memory.
Bug 2015747
Change-Id: I3afe9c2990522058f6aa45f28030bc82a369ca69
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1593093
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: Terje Bergstrom <tbergstrom@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Make NVGPU_AS_MAP_BUFFER_FLAGS_DIRECT_KIND_CTRL mandatory for all map
IOCTLs. We'll clean up the legacy kernel code in subsequent patches.
Remove support for NVGPU_AS_IOCTL_MAP_BUFFER. It has been superseded
by NVGPU_AS_IOCTL_MAP_BUFFER_EX.
Remove legacy definitions to nvgpu_map_buffer_args and the related
flags, and update the in-kernel map calls accordingly by switching to
the newer definitions.
Bug 1902982
Change-Id: Ie9a7f02b8d5d0ec7c3722c4481afab6d39b4fbd0
Signed-off-by: Sami Kiminki <skiminki@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1560932
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Simplify the copyengine code by deleting support for the
ce_event_callback feature that has never been used. Similarly, create a
channel without the finish callback to get rid of that Linux dependency,
and delete the finish callback function as it now serves no purpose.
Delete also the submitted_seq_number and completed_seq_number fields
that are only written to.
Jira NVGPU-259
Change-Id: I02d15bdcb546f4dd8895a6bfb5130caf88a104e2
Signed-off-by: Konsta Holtta <kholtta@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1589320
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Deepak Nibade <dnibade@nvidia.com>
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>
Remove the buffer_attrs struct and replace it with a more
streamlined nvgpu_ctag_buffer_info struct. This struct allows
several different fields to all be passed by pointer to the
various kind/compression functions in the VM map process.
This path also moves several comptag/kind related functions
to the core vm.c code since these functions can be reused by
other OSes.
Change-Id: I2a0f0a1c4b554ce4c8f2acdbe3161392e717d3bf
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1583984
GVS: Gerrit_Virtual_Submit
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Remove the always true field 'user_mapped' from the mapped_buf
struct. All mapped_bufs are mapped by a user request since they
always originate from a dma_buf (for Linux, that is). As such
there is a fair amount of logic that could be deleted.
Linux specific: the own_mem_ref field was also be deleted. The
logic of only storing a dma_buf ref when the buffer is mapped
for the first time by a user is easy: when the mapped buffer is
found in the map cache release the outstanding dma_buf ref taken
earlier on in the map path. If the map cache does not have the
buffer simply let the higher level map code keep the dma_buf ref.
The dma_buf ref is released when the nvgpu_vm_unmap_system()
call-back is called by the unmap path.
JIRA NVGPU-30
JIRA NVGPU-71
Change-Id: I229d136713812a7332bdadd5ebacd85d983bbbf0
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1583983
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>
Re-organize the unmap code to be better split between OS specific
requirements and common core requirements. The new code flow works
as follows:
nvgpu_vm_unmap()
Is the primary entrance to the unmap path. It takes a VM and a GPU
virtual address to unmap. There's also an optional batch mapping
struct.
This function is responsible for making sure there is a real buffer
and that if it's being called on a fixed mapping then the mapping
will definitely be freed (since buffers are ref-counted). Then this
function decrements the ref-count and returns.
If the ref-count hits zero then __nvgpu_vm_unmap_ref() is called
which just calls __nvgpu_vm_unmap() with the relevant batch struct
if present. This is where the real work is done. __nvgpu_vm_unmap()
clears the GMMU mapping, removes the mapped buffer from the various
lists and trees it may be in and then calls the
nvgpu_vm_unmap_system() function. This function handles any OS
specific stuff and must be defined by all VM OS implementations.
There's a a short cut used by some other core VM code to free
mappings without going through nvgpu_vm_map(). Mostly they just
directly decrement the mapping ref-count which can then call
__nvgpu_vm_unmap_ref() if the ref-count hits zero.
JIRA NVGPU-30
JIRA NVGPU-71
Change-Id: Ic626d37ab936819841bab45214f027b40ffa4e5a
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1583982
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>
Move vm_priv.h to <nvgpu/linux/vm.h> and rename nvgpu_vm_map()
to nvgpu_vm_map_linux(). Also remove a redundant unmap function
from the unmap path. These changes are the beginning of reworking
the nvgpu Linux mapping and unmapping code.
The rest of this patch is just the necessary changes to use the
new map function naming and the new path to the Linux vm header.
Patch Series Goal
-----------------
There's two major goals for this patch series. Note that these
goals are not achieved in this patch. There will be subsequent
patches.
1. Remove all last vestiges of Linux code from common/mm/vm.c
2. Implement map caching in the common/mm/vm.c code
To accomplish this firstly the VM mapping code needs to have the
struct nvgpu_mapped_buf data struct be completely Linux free. That
means implementing an abstraction for this to hold the Linux stuff
that mapped buffers carry about (SGT, dma_buf). This is why the
vm_priv.h code has been moved: it will need to be included by the
<nvgpu/vm.h> header so that the OS specific struct can be pulled
into struct nvgpu_mapped_buf.
Next renaming the nvgpu_vm_map() to nvgpu_vm_map_linux() is in
preparation for adding a new nvgpu_vm_map() that handles the
map caching with nvgpu_mapped_buf. The mapping code is fairly
straight forward: nvgpu_vm_map does OS generic stuff; each OS
then calls this function from an nvgpu_vm_map_<OS>() or the like
that does any OS specific adjustments/management.
Freeing buffers is much more tricky however. The maps are all
reference counted since userspace does not track buffers and
expects us to handle this instead. Ugh! Since there's ref-counts
the free code will require a callback into the OS specific code
since the OS specific code cannot free a buffer directly. THis
make's the path for freeing a buffer quite convoluted.
JIRA NVGPU-30
JIRA NVGPU-71
Change-Id: I5e0975f60663a0d6cf0a6bd90e099f51e02c2395
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1578896
GVS: Gerrit_Virtual_Submit
Reviewed-by: David Martinez Nieto <dmartineznie@nvidia.com>
Reviewed-by: Terje Bergstrom <tbergstrom@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>
