For some units that depend on libraries, tmake generates
what seems to be stub libraries, with all functions
implemented as a simple 'ret' instruction.
For instance, libfalcon_utf.so can be seen in multiple places:
du -b $(sudo find . -name libfalcon_utf.so)
7808 ./nvidia/kernel/nvgpu/userspace/units/pmu-l4t_64/libfalcon_utf.so
7808 ./nvidia/kernel/nvgpu/userspace/units/acr-l4t_64/libfalcon_utf.so
7808 ./nvidia/kernel/nvgpu/userspace/units/falcon/falcon_tests-l4t_64/libfalcon_utf.so
76968 ./nvidia/kernel/nvgpu/userspace/units/falcon-l4t_64/libfalcon_utf.so
76968 ./systemimage/home/nvgpu_unit/units/libfalcon_utf.so
The "real" library is in falcon-l4t_64/, and it contains actual code.
But install-unit.sh copies all *.so files to nvgpu_unit/units on target.
In some cases, we end up with a "stub" library being used instead of
the expected one.
To avoid this, make sure to copy the largest libraries (in bytes) last.
Also, remove the '-u' option of rsync to force the copy.
Jira NVGPU-4089
Change-Id: I3832373db9c5aa72ad9ce59eb09ec031609a69e4
Signed-off-by: Thomas Fleury <tfleury@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/2215319
GVS: Gerrit_Virtual_Submit
Reviewed-by: Philip Elcan <pelcan@nvidia.com>
Reviewed-by: Alex Waterman <alexw@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Add support gr_prepare for sw and hw.
Add needed registers using nvgpu_posix_io_add_reg_space calls.
Add unit tests covering following functions
nvgpu_gr_prepare_sw
nvgpu_gr_enable_hw
install-unit.sh modified to copy the firmware binaries under
nvgpu-unit/firmware directory
Currently the gr_init_prepare test is called as part of the
gr_config test, later this will be moved as separate unit test.
Jira NVGPU-3582
Bug 2693908
Change-Id: If8f5ac4988deba0db477b2177981f7912bdb8eec
Signed-off-by: vinodg <vinodg@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/2191254
GVS: Gerrit_Virtual_Submit
Reviewed-by: Alex Waterman <alexw@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>
This reverts commit 9bfdb2ba03f90f0cf828f08b99101a3a3e6c4532.
Bug 2693908
Change-Id: I3ef56773e46aad3626f16b84ea5e51c2fdcc3f1c
Signed-off-by: Bo Yan <byan@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/2189200
Add support gr_prepare for sw and hw.
Add needed registers using nvgpu_posix_io_add_reg_space calls.
Add unit tests covering following functions
nvgpu_gr_prepare_sw
nvgpu_gr_enable_hw
Copy the falcon ucode binaries under userspace/firmware
directory
install-unit.sh modified to copy the firmware binaries
under nvgpu-unnit/firmware directory
Jira NVGPU-3582
Change-Id: If2131d2c48e828251208da86688b0594e62de82e
Signed-off-by: vinodg <vinodg@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/2184293
Reviewed-by: Deepak Nibade <dnibade@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
When running unit tests, a log of all executed tests is
generated. A script along with the "required_tests.json"
will check that all required tests were effectively run, thus
ensuring that no tests are silently missing. The "testlist.py"
support script is also used to generate the list of required
tests, and can also generate an HTML-formatted output with links
to JAMA requirements wherever possible.
JIRA NVGPU-1717
Change-Id: Ic361b1d0a5f003f8b24952d63e20fb496dd1e787
Signed-off-by: Nicolas Benech <nbenech@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1983631
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>
Add a script to install unit tests on a target jetson board.
The installation consists of copying all the binaries generated
by tmake over to that target board and building a source tree
that matches the nvgpu code.
The reason the systemimage out directory is not used is this
is created by the `image createfs' command during the build
process. This script will work after just a simple `tmm[p]'
Change-Id: I1f2650e666a42c12762ab444159b69ba8fc582f8
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1850545
Reviewed-by: Philip Elcan <pelcan@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
