Added new memory API _nvgpu_mem_create_from_phys for
creating nvgpu_mem from physical memory aperture.
With this new API, avoided usage of linux specific
"struct page" in general code and moved this code to
common linux code. This API internally uses
__nvgpu_mem_create_from_pages for creating nvgpu_mem
from physical pages.
JIRA GPUT19X-2
Change-Id: Iaf0193a7c33e71422e4ddabde01edf46f5a81794
Signed-off-by: seshendra Gadagottu <sgadagottu@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1571073
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>
While unbindin a channel from TSG, we first disable all the channels,
then examine the status of channel being removed in
gk20a_fifo_tsg_unbind_channel_verify_status(), and if this API fails we
re-enable all the channel and kill whole TSG
And in gk20a_fifo_tsg_unbind_channel_verify_status() we first check ctx_reload
and fault status and then check NEXT status
If channel has NEXT set we bail out
But since we have already changed the TSG ctx_reload status re-enabling all
channels in TSG might cause issues
Hence fix this by correcting sequence so that we first ensure that NEXT is
not set on channel and then only alter the status
Bug 200327095
Change-Id: I4f0786bc507fad5462d4cdd8d0ca91ea611ee3b5
Signed-off-by: Deepak Nibade <dnibade@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1575905
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Some GPUs require all channels to be on TSG and also have larger than 4GB
vidmem sizes which were not supported on the previous CE2 code.
This change creates a new property to track if the copy engine needs to
encapsulate its kernel context on tsg and also modifies the copy engine code
to support much larger copies without dramatically increasing the PB size.
JIRA: EVLR-1990
Change-Id: Ieb4acba0c787eb96cb9c7cd97f884d2119d445aa
Signed-off-by: David Nieto <dmartineznie@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1573216
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
Reviewed-by: Alex Waterman <alexw@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Nirav Patel <nipatel@nvidia.com>
Change the wait for work pending condition to interruptible in the
channel worker thread, as there's no reason to be noninterruptible.
A noninterruptible wait, even one with a timeout, causes the worker to
be printed in the Linux blocked tasks list which is confusing.
Change the cond signal to interruptible to match this.
Change-Id: I71d848b7f449a5d53fecae90c6a450c98c675c7f
Signed-off-by: Konsta Holtta <kholtta@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1570166
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
Reviewed-by: Alex Waterman <alexw@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Vijayakumar Subbu <vsubbu@nvidia.com>
Re-generate hardware headers so that all unsigned constants are
qualified with postfix U. This removes the need for compiler to do
implicit signed->unsigned conversions.
Change-Id: If3c1e25dcb07ce6857a4798f2c5308e2948fe5e0
Signed-off-by: Terje Bergstrom <tbergstrom@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1571163
Reviewed-by: Seshendra Gadagottu <sgadagottu@nvidia.com>
GVS: Gerrit_Virtual_Submit
Remove an SGL reference from the mm_gk20a.c code. This code is
common code and as such all linuxisms need to be fixed. It just
so happens that this particular function is only used by the
CDE code which is only present in Linux. So simply move this
function over to the CDE code.
JIRA NVGPU-30
JIRA NVGPU-225
JIRA NVGPU-226
Change-Id: Ifb0cb427c742c6d9cada382ace4a52f52474c379
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1576436
Reviewed-by: svccoveritychecker <svccoveritychecker@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Deepak Nibade <dnibade@nvidia.com>
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
When freeing kmalloc and vmalloc buffers memset them to zero
before freeing them with the kernel APIs. This is only done if
CONFIG_NVGPU_TRACK_MEM_USAGE is set since this adds obvious
overhead to the driver. However, it is an incredibly useful
debug tool, so it's nice to have.
This could be done by enabling Linux kernel configs as well,
but not all OSes may have such a feature so building it into
nvgpu may prove quite useful.
Change-Id: I7a6a9a6ab4f3606a73a90b354c5a4a7b9cd4d947
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1575565
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
Reviewed-by: svccoveritychecker <svccoveritychecker@nvidia.com>
Reviewed-by: Konsta Holtta <kholtta@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
Fix bitrot that was incurred when large amounts of the debugfs
stuff was moved the Linux struct. Since the debugfs debugging
is largely hidden under a config for the kmem code the necessary
changes for the kmem debugging were missed until now.
Change-Id: I105549ce39343e503212e302f39ede36c6ea5194
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1575564
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
Reviewed-by: svccoveritychecker <svccoveritychecker@nvidia.com>
Reviewed-by: Konsta Holtta <kholtta@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
As with GR's semaphore acquires that timeout, report
NVGPU_CHANNEL_GR_SEMAPHORE_TIMEOUT to userspace in the error notifier
also when a semaphore acquire timeout interrupt is received from PBDMA.
This timeout is used when the kernel watchdog timer is enabled.
Bug 1782480
Change-Id: I1ceb8632548c5e89febb2b80a5850116a2d4b670
Signed-off-by: Konsta Holtta <kholtta@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1574293
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
Reviewed-by: svccoveritychecker <svccoveritychecker@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
Reviewed-by: Seshendra Gadagottu <sgadagottu@nvidia.com>
Re-generate hardware headers so that all unsigned constants are
qualified with postfix U. This removes the need for compiler to do
implicit signed->unsigned conversions.
Change-Id: Ic93ef7f7a6beae57be7759c7eb3df9148afed824
Signed-off-by: Terje Bergstrom <tbergstrom@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1571162
Reviewed-by: Automatic_Commit_Validation_User
GVS: Gerrit_Virtual_Submit
Reviewed-by: Seshendra Gadagottu <sgadagottu@nvidia.com>
Below APIs are already declared in ioctl_channel.h, and hence remove duplicate
declaration from channel_gk20a.h
gk20a_channel_open()
gk20a_channel_ioctl()
gk20a_channel_release()
And move declaration of gk20a_channel_open_ioctl() from channel_gk20a.h to
ioctl_channel.h
Jira NVGPU-259
Change-Id: I46702ca481e41a19f92f4fe0169f95e31360abe0
Signed-off-by: Deepak Nibade <dnibade@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1573106
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Pre-os firmware takes care, among others, of the control of FAN till
the driver takes over its control. On some GPUs not enabling this FW can lead
tp physical board damage, hence it is needed to run this firmware.
JIRA: NVGPUGV100-9
Change-Id: I18d54cfd5eb64ecec79c5dae67ac8d5bb1facf36
Signed-off-by: David Nieto <dmartineznie@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1549035
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
Reviewed-by: svccoveritychecker <svccoveritychecker@nvidia.com>
Reviewed-by: Vijayakumar Subbu <vsubbu@nvidia.com>
Reviewed-by: Konsta Holtta <kholtta@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
These changes allow GV100 to init the basic HALs to pass
nvgpu_submit_twod
(1) Allocate fault buffer from vidmem instead of sysmem to prevent coherency
issues
(2) Properly enable FB
(3) Fan control requires the execution of the pre-os FW, without it the SKU201
is extremely noisy
JIRA: NVGPUGV100-9
Change-Id: I9b2072737e45432f957e7faae6d33bc0ab43b817
Signed-off-by: David Nieto <dmartineznie@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1539926
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
Reviewed-by: svccoveritychecker <svccoveritychecker@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
Split VIDMEM support into its own code files organized as such:
common/mm/vidmem.c - Base vidmem support
common/linux/vidmem.c - Linux specific user-space interaction
include/nvgpu/vidmem.h - Vidmem API definitions
Also use the config to enable/disable VIDMEM support in the makefile
and remove as many CONFIG_GK20A_VIDMEM preprocessor checks as possible
from the source code.
And lastly update a while-loop that iterated over an SGT to use the
new for_each construct for iterating over SGTs.
Currently this organization is not perfectly adhered to. More patches
will fix that.
JIRA NVGPU-30
JIRA NVGPU-138
Change-Id: Ic0f4d2cf38b65849c7dc350a69b175421477069c
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1540705
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Add an <nvgpu/errno.h> header file to explicitly include the -E* error
messages. Useful for header files with static inlines that return error
messages. In actual C code normally enough Linux/QNX headers bleed in
to get the error messages but header files with sparse includes do not
have this luxury.
Change-Id: I4833c7a6003578b9792bbf6b14cce3bd0adfec22
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1573307
Reviewed-by: svccoveritychecker <svccoveritychecker@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Konsta Holtta <kholtta@nvidia.com>
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
__nvgpu_gmmu_do_update_page_table() uses nvgpu_sgt_for_each_sgl to loop
through the entries of a buffer to be mapped, so when continue is used,
the sgl entry must not be reassigned again like it was before with a
pure while-and-update loop. Delete a reassignment to fix a case where
sgl = sgl->next could happen twice.
Bug 2002279
Bug 2001466
Change-Id: I47c8b853d4b35304740cd4e8a840df02fcd23054
Signed-off-by: Konsta Holtta <kholtta@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1575279
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Alex Waterman <alexw@nvidia.com>
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
Tested-by: Alex Waterman <alexw@nvidia.com>
Reviewed-by: Timo Alho <talho@nvidia.com>
We right now call gk20a_fifo_tsg_unbind_channel_verify_status() to verify
channel status while unbinding a channel from TSG while closing
Add support to do this verification per-platform and keep this disabled
for vgpu platforms
Bug 200327095
Change-Id: I6e2a6a09c784d24ac49477d5450b7d4b671878e3
Signed-off-by: Deepak Nibade <dnibade@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1572369
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
- Falcon HW header re-generate for gv11b.
- Re-generate hardware headers so that all unsigned
constants are qualified with postfix U. This removes
the need for compiler to do implicit signed->unsigned
conversions
Change-Id: I313945edac1112a32c965d9565b30dc95a002752
Signed-off-by: Mahantesh Kumbar <mkumbar@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1571352
Reviewed-by: Automatic_Commit_Validation_User
GVS: Gerrit_Virtual_Submit
Reviewed-by: Vijayakumar Subbu <vsubbu@nvidia.com>
- Constants are qualified with postfix U.
This removes the need for compiler to do
implicit signed->unsigned conversions
Change-Id: I039e269b18ea8aea48b30d3af84b347ae5509413
Signed-off-by: Mahantesh Kumbar <mkumbar@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1570998
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Vijayakumar Subbu <vsubbu@nvidia.com>
-Falcon HW header re-generate for gk20a, gm20b, gp10b & gp106.
-Re-generate hardware headers so that all unsigned constants are
qualified with postfix U. This removes the need for compiler to do
implicit signed->unsigned conversions
Change-Id: Ifdaac2c697ee7ba8be627e059bf18024a67bbd27
Signed-off-by: Mahantesh Kumbar <mkumbar@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1570775
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Vijayakumar Subbu <vsubbu@nvidia.com>
Currently if channel has NEXT bit set while closing the channel we just print
an error and continue channel unbind sequence from TSG
But since channel with NEXT set is active killing it can potentially corrupt
the TSG context and cause unpredictable errors on remaining channels/TSG
Hence fix this by killing whole TSG context if channel being closed has
NEXT bit set
if gk20a_fifo_tsg_unbind_channel() API returns error, kill the TSG
otherwise continue with channel unbind sequence
Bug 200327095
Change-Id: I2abf1a3db8ba6f105b6ca86e78006c7b2a7726cc
Signed-off-by: Deepak Nibade <dnibade@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1568566
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
Reviewed-by: svccoveritychecker <svccoveritychecker@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Konsta Holtta <kholtta@nvidia.com>
Reviewed-by: Vijayakumar Subbu <vsubbu@nvidia.com>
We right now call gk20a_fifo_tsg_unbind_channel_verify_status() to verify
channel status while unbinding a channel from TSG while closing
Add support to do this verification per-platform and keep this disabled
for vgpu platforms
Bug 200327095
Change-Id: I19fab41c74d10d528d22bd9b3982a4ed73c3b4ca
Signed-off-by: Deepak Nibade <dnibade@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1572368
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Combine the VIDMEM and SYSMEM GMMU mapping paths into one
single function. With the proper nvgpu_sgt abstraction in
place the high level mapping code can treat these two types
of maps identically. The benefits of this are immediate: a
change to the mapping code will have much more testing
coverage - the same code runs on both vidmem and sysmem.
This also makes it easier to make changes to mapping code
since the changes will be testable no matter what type of
mem is being mapped.
JIRA NVGPU-68
Change-Id: I308eda03d426966ba8e1d4892c99cf97b45c5993
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1566706
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
Reviewed-by: svccoveritychecker <svccoveritychecker@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Konsta Holtta <kholtta@nvidia.com>
Reviewed-by: Vijayakumar Subbu <vsubbu@nvidia.com>
For pre-silicon platforms, clock gating
should be skipped as it is not supported.
Added new flags "can_"x"lcg" to check platform
capability before programming SLCG,BLCG and ELCG.
Bug 200314250
Change-Id: Iec7564b00b988cdd50a02f3130662727839c5047
Signed-off-by: Deepak Goyal <dgoyal@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1566251
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
For Simualtion/Emulation platforms,clock gating
should be skipped as it is not supported.
Added new flags "can_"X"lcg" to check platform
capability before doing SLCG,BLCG and ELCG.
Bug 200314250
Change-Id: I4124d444a77a4c06df8c1d82c6038bfd457f3db0
Signed-off-by: Deepak Goyal <dgoyal@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1566049
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Rename get_physical_addr_bits and related functions to something that
more clearly conveys what they are doing. The basic idea of these
functions is to translate from a physical GPU address to a IOMMU GPU
address. To do that a particular bit (that varies from chip to chip)
is added to the physical address.
JIRA NVGPU-68
Change-Id: I536cc595c4397aad69a24f740bc74db03f52bc0a
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1542966
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Add a function to do address translation for IOMMU capable GPUs.
When an iGPU is behind and IOMMU it can pick whether to use that
IOMMU for translation by adding a bit to physical addresses. This
function takes care of that.
However, this required an abstracted nvgpu_iommuable() API to
check whether a GPU is behind an IOMMU. This patch adds that API
for Linux.
JIRA NVGPU-68
Change-Id: I489d14475167c019c294407372395df78c8b5feb
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1542965
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
Reviewed-by: svccoveritychecker <svccoveritychecker@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Sourab Gupta <sourabg@nvidia.com>
Reviewed-by: Konsta Holtta <kholtta@nvidia.com>
Reviewed-by: Vijayakumar Subbu <vsubbu@nvidia.com>
Remove the last sg_phys() call from the GMMU code and replace it
with a generic nvgpu_mem API. This new API, nvgpu_mem_get_phys_addr(),
returns the physical address of an nvgpu_mem struct.
Also, implement this new API in the Linux specific nvgpu_mem code
since it requires access to the underlying SGT/SGL.
JIRA NVGPU-68
Change-Id: Idf88701a2a8515464c658c26e0de493c82ff850d
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1542964
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
When the L2 flush IOCTL gets no l2 flush and no fb flush we now
return -EINVAL. This can sometimes happen if the user tries to just
invalidate. Currently we do not support L2 invalidates only.
Bug 1661242
Change-Id: I87f3259bfbd736b5f4222cfe7b3cfa4a6475389e
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1227125
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
- GR falcons bootstrap can be done using physical or
virtual address by setting flag usevamask in PMU interface
PMU_ACR_CMD_ID_BOOTSTRAP_MULTIPLE_FALCONS command
- With this change always setting to physical address support & removed
virtual address support along with code removal.
- Removed Linux specific code used to get info regarding WPR VA.
JIRA NVGPU-128
Change-Id: Id58f3ddc4418d61126f2a4eacb50713d278c10a0
Signed-off-by: Mahantesh Kumbar <mkumbar@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1572468
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
Reviewed-by: svccoveritychecker <svccoveritychecker@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Vijayakumar Subbu <vsubbu@nvidia.com>
Reviewed-by: Alex Waterman <alexw@nvidia.com>
