This patch deals with cleanups meant to make things simpler for the
upcoming os abstraction patches for the sync framework. This patch
causes some substantial changes which are listed out as follows.
1) sync_timeline is moved out of gk20a_fence into struct
nvgpu_channel_linux. New function pointers are created to facilitate os
independent methods for enabling/disabling timeline and are now named
as os_fence_framework. These function pointers are located in the struct
os_channel under struct gk20a.
2) construction of the channel_sync require nvgpu_finalize_poweron_linux()
to be invoked before invocations to nvgpu_init_mm_ce_context(). Hence,
these methods are now moved away from gk20a_finalize_poweron() and
invoked after nvgpu_finalize_poweron_linux().
3) sync_fence creation is now delinked from fence construction and move
to the channel_sync_gk20a's channel_incr methods. These sync_fences are
mainly associated with post_fences.
4) In case userspace requires the sync_fences to be constructed, we
try to obtain an fd before the gk20a_channel_submit_gpfifo() instead of
trying to do that later. This is used to avoid potential after effects
of duplicate work submission due to failure to obtain an unused fd.
JIRA NVGPU-66
Change-Id: I42a3e4e2e692a113b1b36d2b48ab107ae4444dfa
Signed-off-by: Debarshi Dutta <ddutta@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1678400
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
When runtime pm is disabled, then gpu rail will be on as soon as
nvgpu module is loaded. If pm suspend/resume called before gpu
hw initialization(g->poweron = false) then pm suspend is skipping
gpu railgate, which is causing issues with SC7 entry/exit.
To fix this issue:
1. During pm suspend, if g->poweron is false, check for runtime pm
disable to railgate gpu rail.
2. On pm resume, check for runtime pm disable to enable gpu rail,
though gpu driver not initialized.
Bug 2073029
Change-Id: I7631109d79cda5882d2864557f1b7b3d2d89c9f6
Signed-off-by: seshendra Gadagottu <sgadagottu@nvidia.com>
Signed-off-by: Seema Khowala <seemaj@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1679010
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Delete the proxy waiter for non-semaphore-backed syncfds in sema wait
path to simplify code, to remove dependencies to the sync framework (and
thus Linux) and to support upcoming refactorings. This feature has never
been used for actually foreign fences.
Jira NVGPU-43
Jira NVGPU-66
Change-Id: I2b539aefd2d096a7bf5f40e61d48de7a9b3dccae
Signed-off-by: Konsta Holtta <kholtta@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1665119
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>
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>
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>
Enabling gpu scaling driver after finalize poweron,
will make gpu booting happen at initially set
frequency(1GHz).
Also doing platform specific init scale after enabling
scaling driver.
Bug 2049965
Bug 2039013
Bug 200377508
Change-Id: I633f8f5a25d9de18cbb3a022913b8b725ccd87e5
Signed-off-by: seshendra Gadagottu <sgadagottu@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1644703
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>
Fix a race condition where we'd still be booting up the gpu and/or
initializing the driver but elsewhere assume that all is done already.
Some userspace APIs to make sure that we're ready by testing
g->gr.sw_ready, but this flag is set in the middle of bootup; there are
other things after gr initialization. Add a new flag that is enabled
after bootup is fully complete at the end of finalize_poweron, and
change the checks in user API paths to test the new flag only.
These checks are only in the ioctl paths for ctrl, dbg and tsg, and in
the ctrl device's opening path.
The gr.sw_ready flag is still left there to signify whether just gr has
had its bookkeeping initialized.
Bug 200370011
Change-Id: I2995500e06de46430d9b835de1e9d60b3f01744e
Signed-off-by: Konsta Holtta <kholtta@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1640124
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Add cleanup to the gk20a_probe() function since it will often fail
due to probe deferal. These "failures" cause this function to be
called multiple times and potentially allocate many resources over
and over again, leaking the old allocations.
Bug 200369627
Change-Id: Ic0bba0ae6542485135d9cb7393086e4460cd271d
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1640628
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
This patch removes linux dependencies from sim_gk20a.h under
gk20a/sim_gk20a.h. The following changes are made in this patch.
1) Created a linux based structure sim_gk20a_linux that contains a
common sim_gk20a struct inside it. The common struct sim_gk20a doesn't
contain any linux specific structs.
2) The common struct sim_gk20a contains an added function pointer which
is used to invoke gk20a_sim_esc_readl() method.
3) sim_gk20a.c is moved to nvgpu/common/linux along with a new header
sim_gk20a.h that contains the definition of struct sim_gk20a_linux.
4) struct gk20a now contains a pointer of sim_gk20a instead of the
entire object. The memory for this struct is allocated and initialized during
gk20a_init_support() and freed during invocation of
gk20_remove_support().
5) We first obtain the pointer for struct sim_gk20a_linux from the
pointer of sim_gk20a using the container_of method in order to work on
the struct.
JIRA NVGPU-386
Change-Id: Ic82b8702642377f82694577a53c3ca0b9c1bb2ab
Signed-off-by: Debarshi Dutta <ddutta@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1603073
GVS: Gerrit_Virtual_Submit
Reviewed-by: Deepak Nibade <dnibade@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>
Move the implementation for channel job update callbacks that is based
on Linux specific work_struct usage to Linux-specific code.
This requires a bit of extra work for allocating OS-specific priv data
for channels which is also done in this patch. The priv data will be
used more when more OS-specific features are moved.
Jira NVGPU-259
Change-Id: I24bc0148a827f375b56a1c96044685affc2d1e8c
Signed-off-by: Konsta Holtta <kholtta@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1589321
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
GPU hardware block needs tegra fuse clock to mirror
gpu fuses from tegra fuses to gpu domain.
Tegra fuse driver provided following APIs to
enable/disable tegra fuse clock:
int tegra_fuse_clock_enable(void);
int tegra_fuse_clock_disable(void);
To ensure that tegra fuse clock is disabled by nvgpu
driver when gpu hardware block is not in use by:
Calling tegra_fuse_clock_enable() while doing
gk20a_pm_unrailgate() and calling
tegra_fuse_clock_disable() while doing
gk20a_pm_railgate().
Bug 2019897
Change-Id: I61688829fd9a8b0c1ffa9d34db6393550f333866
Signed-off-by: seshendra Gadagottu <sgadagottu@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1595297
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Move fuse override DT handling to Linux code. All the chip specific
fuse override functions did the same thing, so delete the HAL and
call the same function to read the DT overrides on all chips.
Also remove the fuse override functionality from dGPU. There are no
DT entries for PCIe devices, so it would've failed anyway.
JIRA NVGPU-259
Change-Id: Iba64a5d53bf4eb94198c0408a462620efc2ddde4
Signed-off-by: Terje Bergstrom <tbergstrom@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1593687
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
Reviewed-by: Deepak Nibade <dnibade@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>
ioctl_channel.h and cde.h referred to multiple structures that were
not forward declared or explitly #included in. Add several forward
declarations and #includes. Also add #include for
<uapi/linux/nvgpu.h> to multiple Linux .c files that were missing it.
Change-Id: Iefd52e71224d5810b5abbcc765f92bc535d7a28b
Signed-off-by: Terje Bergstrom <tbergstrom@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1591634
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
struct gk20a is now part of nvgpu_os_linux in Linux builds. gk20a.c
still frees struct gk20a by kfree(struct gk20a *), which is wrong.
Create a new function pointer in struct gk20a for freeing the
structure and call kfree(struct nvgpu_os_linux *) instead.
JIRA NVGPU-259
Change-Id: I412ee993002cb2a42f0db015fc676de43418ec2f
Signed-off-by: Terje Bergstrom <tbergstrom@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1591012
GVS: Gerrit_Virtual_Submit
Reviewed-by: Deepak Nibade <dnibade@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Move gk20a/platform_gk20a.h to linux specific directory as
common/linux/platform_gk20a.h since this file includes all linux specific
stuff
Fix #includes in all the files to include this file with correct path
Remove #include of this file where it is no more needed
Fix gk20a_init_sim_support() to receive struct gk20a as parameter
instead of receiving linux specific struct platform_device
NVGPU-316
Change-Id: I5ec08e776b753af4d39d11c11f6f068be2ac236f
Signed-off-by: Deepak Nibade <dnibade@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1589938
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>
Migrate ctxsw_trace_gk20a.c to common/linux/ctxsw_trace.c. This
has been done becasue the ctxsw tracing code is currently too
tightly tied to the Linux OS due to usage of a couple system calls:
- poll()
- mmap()
And general Linux driver framework code. As a result pulling the
logic out of the FECS tracing code is simply too large a scope for
time time being.
Instead the code was just copied as much as possible. The HAL ops
for the FECS code was hidden behind the FECS tracing config so
that the vm_area_struct is not used when QNX does not define said
config. All other non-HAL functions called by the FECS ctxsw
tracing code ha now also been hidden by this config. This is not
pretty but for the time being it seems like the way to go.
JIRA NVGPU-287
Change-Id: Ib880ab237f4abd330dc66998692c86c4507149c2
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1586547
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
We currently check that no job is pending before proceeding with
suspend. This prevents suspend, when we could simply disable and
preempt all channels. Moreover, pending jobs accounting is done
using pm_runtime usage count, which is not updated for GPUs with
pm_runtime disabled (e.g. vgpu).
Replaced the check on pm_runtime usage count, with a check on
gk20a handle usage count. Suspend is allowed when there is no
task inside a busy/idle sequence.
JIRA EVLR-1852
Change-Id: I79e71c8112182622dbd1c7c46cd84befa61a5c4d
Signed-off-by: Thomas Fleury <tfleury@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1552348
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 currently check that engine is idle before proceeding with
suspend. This prevents suspend when we could simply disable and
preempt all channels. Moreover, doing such a check in virtualization
case, would require to query engine status from RM server, before
proceeding with suspend.
Removed check on engine idle for system suspend.
JIRA EVLR-1852
Change-Id: Ic6dc65af14b00f236db20038dfc04fa0928c1fe2
Signed-off-by: Thomas Fleury <tfleury@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1552347
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Add abstraction of IO aperture accessors. Add new functions
gk20a_io_exists() and gk20a_io_valid_reg() to remove dependencies to
aperture fields from common code.
Implement Linux version of the abstraction by moving gk20a_readl()
and gk20a_writel() to new Linux specific io.c. Move the fields
defining IO aperture to nvgpu_os_linux.
Add t19x specific IO aperture initialization functions and add t19x
specific section to nvgpu_os_linux.
JIRA NVGPU-259
Change-Id: I09e79cda60d11a20d1099a9aaa6d2375236e94ce
Signed-off-by: Terje Bergstrom <tbergstrom@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1569698
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
For sync-point read map:
1. Added nvgpu_mem memory allocator in gk20a struct and
allocated memory for this in gk20a_finalize_poweron()
and freed this memory in gk20a_remove().
2. Added "u64 syncpt_ro_map_gpu_va" in vm_gk20a struct
for read map in vm.
Added nvgpu_quiesce() in nvgpu_remove() before freeing
syncpoint read map to ensure that nvgpu is idle.
JIRA GPUT19X-2
Change-Id: I7cbfec57f0992682dd833a1b0d92d694bcaf1eb3
Signed-off-by: seshendra Gadagottu <sgadagottu@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1514338
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>
The following changes are part of the porting of the bind/unbind
functionality.
These changes reuse the shutdown codepaths in iGPU and dGPU and fix a locking
issue with in gk20a_busy() where the usage count can lead to a deadlock during
the driver shutdown. It fixes a racing condition with the gr/mm code by
invalidating the sw ready flag while holding the busy lock
JIRA: EVLR-1739
Change-Id: I62ce47378436b21f447f4cd93388759ed3f9bad1
Signed-off-by: David Nieto <dmartineznie@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1554959
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: svccoveritychecker <svccoveritychecker@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
