| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Fix an issue that results in:
Traceback (most recent call last):
File "Var <DEPENDS>", line 1, in <module>
bb.data_smart.ExpansionError: Failure expanding variable DEPENDS, expression was pkgconfig-native glib-2.0-native zlib-native ninja-native meson-native pixman-native qemu-xilinx-native bison-native ninja-native meson-native qemu-xilinx-multiarch-helper-native${@' pmu-rom-native' if 'xilinx' in d.getVar('LICENSE_FLAGS_ACCEPTED').split() else ''} alsa-lib-native dtc-native libgcrypt-native libslirp-native which triggered exception AttributeError: 'NoneType' object has no attribute 'split'
The variable dependency chain for the failure is: DEPENDS
This only occurs if no LICENSE_FLAGS_ACCEPTED is defined. This should not be
a blocking error. Instead the user is warned of reduced functionality and we
continue the build.
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
|
|
|
|
| |
Various AMD/Xilinx platforms will fail to start, unless the correct
amounts of memory are specified. The default OE ptests will try to
limit memory used, but this will cause boot failures with
qemu-xilinx.
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
| |
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
| |
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
|
|
| |
Since our qemu is not yet updated, we need the matching file versions
from YP Qemu 7.1.0 (langdale).
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
| |
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
|
|
| |
Previous commit split ai-engine code into a dynamic bbappend, but put the
code into the wrong directory. Move to the correct directory.
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
|
|
|
|
| |
Memory manger issue fixed for xbtest (#7492) (#7494)
Fixed CU context related issue. (#7487) (#7493)
2023.1 : Use correct BO handle to export and use the exported bo handle for AIE APIs for GMIO AIE Trace offload (#7491)
Signed-off-by: rbramand <rbramand@xilinx.com>
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
| |
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
|
|
|
| |
While systemd is not recommended for use on microblaze, it does work.
However, additional configuration items are requried in this case. So
enable the minimum options, as defined from the systemd README file.
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
|
|
|
| |
Adding meta-xilinx-core was contaminating the PACKAGECONFIG hash when
building mesa-native. Workaround this by only adjusting the value
when class-target is defined.
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
| |
Signed-off-by: rbramand <rbramand@xilinx.com>
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
|
|
|
|
| |
XRT requiring ZOCL will just ensure that the component is present,
however we've observed interface changes that make this not work
in all cases. Instead we now require the ZOCL version to be the same
as the XRT version.
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
|
|
|
|
|
|
| |
Added 0027-Updated-clock-name-and-structure-to-match-LIMA-drive.patch
Updated clock name and structure to match LIMA driver
Signed-off-by: Parth Gajjar <parth.gajjar@amd.com>
Fixup patch path to match recipe S value.
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
|
|
|
|
| |
Based on github Pull Request #42. While that pull request is no longer
valid with other changes, the actual bug still exists.
Originally reported by AnatoliiShablov.
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
|
|
|
| |
In addition update device tree bbappend to mimic how QEMU DTs use overlays
and only update to EXTRA_OVERLAYS if the proper conditions are met
Signed-off-by: Ben Levinsky <ben.levinsky@amd.com>
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
In case of Zynq-7000, using both "-process_bitstream bin" and "-o"
in bootgen flag to convert bit file to bin format, "-o" option will
not be effective and generated output file name is
${S}+${BIT_PATH}/<bit_file_name>.bit.bin file which causes build errors.
Hence we need to rename this file from <bit_file_name>.bit.bin to
${PN}.bit.bin which matches the firmware name in dtbo and move
${PN}.bit.bin to ${B} directory.
Signed-off-by: Sandeep Gundlupet Raju <sandeep.gundlupet-raju@amd.com>
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Qemu boots by loading ATF and jumping to it directory. A recent change
adds a verification that the data structure passed by the FSBL is correct
in a 'mode 5' (SD card) boot. If it's not valid ATF will fail to continue
to boot. (JTAG, 'mode 0' boot will proceed but other system behavior
changes.)
In order to emulate the FSBL handoff to ATF, we need to populate a data
structure in fffc0000, and tell ATF to look at that address. See the
comments for more information. This also ends up hard-coding the u-boot
address. If it changes booting will fail.
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
fpgamanager_custom bbclass allowed only accel.json file as user input
from firmware recipes, if user want to use flat or DFx static design
using shell.json, then shell.json is not deployed in
rootfs/${nonarch_base_libdir}/firmware/xilinx/${PN} directory.
Fix the logic to deploy shell.json (flat or DFx Static) or
accel.json (DFx RP) for flat, DFx Static and DFx RP use cases.
Signed-off-by: Sandeep Gundlupet Raju <sandeep.gundlupet-raju@amd.com>
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
|
|
| |
Update meta-xilinx-core README with langdale release info.
Signed-off-by: Sandeep Gundlupet Raju <sandeep.gundlupet-raju@amd.com>
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
|
|
|
|
|
|
|
| |
TMPDIR buildpaths warning are observed in libvcu-xlnx.
WARNING: libvcu-xlnx-1.0.0-xilinx-v+gitAUTOINC+83aabb84c2-r0 do_package_qa: QA Issue: File /usr/bin/ctrlsw_encoder in package libvcu-xlnx contains reference to TMPDIR
File /usr/bin/ctrlsw_decoder in package libvcu-xlnx contains reference to TMPDIR [buildpaths]
Hence disable buildpaths QA check for libvcu-xlnx.
Signed-off-by: Sandeep Gundlupet Raju <sandeep.gundlupet-raju@amd.com>
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
| |
Signed-off-by: Ben Levinsky <ben.levinsky@amd.com>
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
system.dtb is symlink to actual dtb file which is not present in linux
sysroot path. If specified dtb name for FIT_CONF_DEFAULT_DTB not present
fitimage.bbclass will not set default configuration variable.
To fix this, setting FIT_CONF_DEFAULT_DTB to empty if CONFIG_DTFILE not
set(xsct flow) so that which dtb comes first that will be the default
configuration in fitimage creation.
Signed-off-by: Raju Kumar Pothuraju <rajukumar.pothuraju@amd.com>
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
|
|
|
|
|
|
|
| |
TMPDIR buildpaths warning are observed in microblaze linux-xlnx-scr
WARNING: linux-xlnx-5.15.36-xilinx-v2022.2+gitAUTOINC+943565167d-r0 do_package_qa: QA Issue: File /usr/src/debug/linux-xlnx/5.15.36-xilinx-v2022.2+gitAUTOINC+943565167d-r0/lib/oid_registry_data.c in package linux-xlnx-src contains reference to TMPDIR [buildpaths]
Hence disable buildpaths QA check for microblaze linux-xlnx-scr
package.
Signed-off-by: Sandeep Gundlupet Raju <sandeep.gundlupet-raju@amd.com>
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
|
|
|
|
|
|
| |
TMPDIR buildpaths warning are observed in microblaze u-boot.elf
WARNING: u-boot-xlnx-1_v2021.01-xilinx-v2022.2+gitAUTOINC+f32d07b9b5-r0 do_package_qa: QA Issue: File /boot/u-boot-kcu105-microblazeel-v2021.01-xilinx-v2022.2+gitAUTOINC+f32d07b9b5-r0.elf in package u-boot-xlnx-elf contains reference to TMPDIR [buildpaths]
Hence disable buildpaths QA check for microblaze u-boot-elf package.
Signed-off-by: Sandeep Gundlupet Raju <sandeep.gundlupet-raju@amd.com>
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
|
|
|
|
|
|
| |
TMPDIR buildpaths warning are observed in zynqmp pmufw.
WARNING: pmufw-1.0-r0 do_package_qa: QA Issue: File /boot/pmufw.elf in package pmufw contains reference to TMPDIR [buildpaths]
Hence disable buildpaths QA check for zynqmp pmufw.
Signed-off-by: Sandeep Gundlupet Raju <sandeep.gundlupet-raju@amd.com>
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
|
|
|
|
|
|
| |
TMPDIR buildpaths warning are observed in fsbl for zynq and zynqmp.
WARNING: fsbl-1.0-r0 do_package_qa: QA Issue: File /boot/fsbl.elf in package fsbl contains reference to TMPDIR [buildpaths]
Hence disable buildpaths QA check for zynq and zynqmp fsbl.
Signed-off-by: Sandeep Gundlupet Raju <sandeep.gundlupet-raju@amd.com>
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
|
|
|
|
|
|
| |
TMPDIR buildpaths warning are observed in psmfw.
WARNING: psmfw-1.0-r0 do_package_qa: QA Issue: File /boot/psmfw.elf in package psmfw contains reference to TMPDIR [buildpaths]
Hence disable buildpaths QA check for versal psmfw.
Signed-off-by: Sandeep Gundlupet Raju <sandeep.gundlupet-raju@amd.com>
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
|
|
|
|
|
|
| |
TMPDIR buildpaths warning are observed in plmfw.
WARNING: plmfw-1.0-r0 do_package_qa: QA Issue: File /boot/plmfw.elf in package plmfw contains reference to TMPDIR [buildpaths]
Hence disable buildpaths QA check for versal plmfw.
Signed-off-by: Sandeep Gundlupet Raju <sandeep.gundlupet-raju@amd.com>
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
|
|
|
|
|
|
| |
TMPDIR buildpaths warning are observed.
WARNING: arm-trusted-firmware-2.6-xilinx-v2022.2+gitAUTOINC+a03759e829-r0 do_package_qa: QA Issue: File /boot/arm-trusted-firmware--2.6-xilinx-v2022.2+gitAUTOINC+a03759e829-r0-20230302182530.elf in package arm-trusted-firmware contains reference to TMPDIR [buildpaths]
Hence disable buildpaths QA check for versal.
Signed-off-by: Sandeep Gundlupet Raju <sandeep.gundlupet-raju@amd.com>
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
|
|
|
|
| |
Change CFLAGS to exclude thread-jumps and tree-pre optimizations.
Building driver with these optimizations is causing some driver calls to
hang.
Signed-off-by: Gregory Williams <gregory.williams@amd.com>
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Changelog:
fixed bug with finding MEM tile sets (#7392)
CR-1150169 mgmt-ioctl: fix lower bound check on xclbin size (#7389)
CR-1151862 xbutil validate fails testing m2m on OS RH 9.0 (#7391)
CR-1145201 Update GitHub documentation for xbmgmt dump (#7390)
Fixed CR-1152499 (#7380)
xbmgmt examine -r vmr outputs more helpful info from status (#7383)
CRs 1153404, 1154045: report incorrect settings and better handle compile-time trace (#7377)
Embedded metadata section is in XML format. So use read_xml (#7379)
Fix for showing Kernels to Global Memory data in Profile Summary (#7376)
CR-1135808 Some improvement suggestions for throttling feature (#7346)
fix xgq timestamp issue (#7375)
CR-1153785 Fix library load issues in hw emulation (#7374)
VITIS-5914 Add User space changes to fetch and parse aie column status dump info (#7328)
VITIS-6253 Part 1/3 Update tools help menu (#7293)
Fixed compile-time metrics on VCK5000 boards. (#7371)
CR-1142742 No interface counters mentioned in summary.csv when packet counters are used (#7367)
VITIS-5543 Support PL Deadlock Diagnosis (#7363)
Fix race in bo and device construction (#7361)
CRs: 1142742, 1150996, 1152826 (#7362)
CR-1152824: Fixing trace buffer allocation to not use full memory resources when user over allocates (#7356)
Adding checks to make sure bandwidth is not reported as infinity or over 100% (#7358)
Remove obsolete methods (#7357)
Update adf_profiling_api.cpp (#7351)
Refactor qos_type to cfg_param_type (#7350)
Palamida 175214 (#7355)
CR-1151706 XRT service bundle for collecting analytics data (#7340)
removed debug messages (#7348)
CR-1152339: Add flow to run summary file (#7352)
Fixed CU subdevice issue for legacy case (#7345)
CR-1137357 Remove libxrt_core from xbutil/xbmgmt (#7333)
Trace processor to create events from raw PL trace data and write to output csv (#7341)
Add FPT warning message for versal devices (#7315)
remove duplicate code (#7337)
CR-1152356 Remove xmc from V70 subdev resource list (#7347)
More AIE2 changes (#7344)
Changed trace and pprofiling plugins to build with xrt_coreutil instead of xrt_core (#7343)
Fixed coverity bugs in AIE Trace (#7342)
Added support for V70 AIE Trace PS kernel (#7307)
CR-1149236 Add support to dump APU's dimes and syslog to host (#7318)
Amend #7334 with missing file (#7336)
set default value of fast_nocddr_access as true. (#7319)
Added support for V70 AIE Profile PS kernel (#7335)
Palamida 175203,175210 (#7334)
fix a livelock in xrt_cu.c (#7332)
Updated xocl_kds_xgq_set_timestamp to use timespec64 and ktime_get_real_ts64 for newer Linux kernels (#7331)
Adding the trace wrapper to XRT AIE graph calls (#7330)
Updating comment (#7329)
AIE2 support follow-up changes (#7322)
Adjusting a check for large PL buffer size on Edge (#7323)
Use numbered build (#7327)
Update petalinux to 2023.1 version (#7326)
Align qdma driver with latest (#7324)
CR-1150349 move gmio reuse buffer warning (#7317)
Warning message when HSDP is used and monitors are only initialized and not read by XRT (#7316)
CR-1144989: Fixing issue where profiling generated incorrect CSV format (#7312)
Updated AIE Profiling PS kernel on VCK5000 (#7311)
CR-1151004 ASTeR TC 19.02 - v70 - server crash occurs shortly after a (#7306)
CR-1149626 sc warning (#7304)
Multislot final code support (#7269)
Initial APU time sync commit to synchronize APU date/time with host (#7273)
Signed-off-by: rbramand <rbramand@xilinx.com>
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
|
|
|
| |
This reverts commit 0e471db25311caf15b412e68eed4087eb7c664d3.
AIE has been moved to this layer. Restore previous configuration.
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
|
| |
Moves from meta-xilinx-tools as of commit 91c18d664dedc5c8e66f0744ec85a4d3c8bc300a.
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
|
|
|
|
|
|
|
| |
When we tried to create fitimage it will pick the dtb from linux recipe-sysroot path.
There are multiple dtb files in linux sysroot in sdt case and we did not specified any default dtb file to package in fitimage and it was picking baremetal dtb file which is leading to kernel panic.
There is a variable in yocto to specify the default dtb file to pick for fitimage FIT_CONF_DEFAULT_DTB.
If we specify the above then the fitimage is generated properly and able to boot successfully.
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
Signed-off-by: Varalaxmi Bingi <varalaxmi.bingi@amd.com>
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
| |
This reverts commit 6ad9836940fa9c3541b5339c84d6d86e36734e24.
|
|
|
|
|
|
|
| |
AIE components are only provided by meta-xilinx-tools layer, so the
configuration and dependencies need to be in the same layer.
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When using a system device tree build there will be more then one device-tree
that is constructed. If a fit image is constructed, then an DTB needs to be
selected. We use the CONFIG_DTFILE as the default, which should inherit
from the SYSTEM_DTFILE. The default for SYSTEM_DTFILE is empty as a fallback
which should restore the default behavior of first DTB found is the one used.
This also has ramifications for other recipes that use DTB files and unifies
everything.
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
| |
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
|
|
|
|
| |
We want to be able to install qemu-xilinx-dev package, if the user requests
qemu-dev. We can't just hard code a list, as the virtualization layer can
modify the packages produced based on the existence of Xen or not. So
dynamically compute the variable append.
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
|
|
|
|
|
|
| |
Fix git branch URL warnings and indentation.
WARNING: linux-firmware-ti-bt-1.0-r0 do_package_write_rpm_setscene:
URL: git://git.ti.com/ti-bt/service-packs.git;protocol=https does not set any branch parameter.
The future default branch used by tools and repositories is uncertain and we will therefore soon require this is set in all git urls.
Signed-off-by: Sandeep Gundlupet Raju <sandeep.gundlupet-raju@amd.com>
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
|
|
| |
If qemu ends up depending on libmali, we want to set the PACKAGE_ARCH to
machine value as this version is now machine specific.
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
|
|
|
| |
As we now support enabling for libmali and lima, we want to ensure that the
components are rebuilt if/when we switch between the configurations. This
is especially true on a non-graphics optimized part.
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
| |
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
|
|
|
|
| |
1. Enable versalnet -arch command option for versal-net devices.
2. Add COMPATIBLE_MACHINE for SOC and don't allow building for
Microblaze.
Signed-off-by: Sandeep Gundlupet Raju <sandeep.gundlupet-raju@amd.com>
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
|
|
| |
virtfs does not currently work on mingw32 due to a dependency on libcap-ng
with does not work.
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
|
|
|
| |
This patch resolves the macro redefination and
precision differ error in glmark2-es2 application.
Signed-off-by: Yash Ladani <yash.ladani@amd.com>
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
| |
Signed-off-by: Sandeep Gundlupet Raju <sandeep.gundlupet-raju@amd.com>
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
| |
Signed-off-by: rbramand <rbramand@xilinx.com>
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|
|
|
|
|
|
|
| |
Yocto Project upstream patches resolve integration issues, and resolve
various CVEs. Include these into the qemu-xilinx recipe.
Signed-off-by: Mark Hatle <mark.hatle@amd.com>
|