Age | Commit message (Collapse) | Author |
|
Signed-off-by: Sudheesh Mavila <sudheesh.mavila@amd.com>
|
|
Fix tearing issues for AMDGPU
|
|
Cleaned up indentation issues as the file was pretty
messed up. The OE guidelines suggest using 4 spaces
and no tabs.
Signed-off-by: Awais Belal <awais_belal@mentor.com>
|
|
This enables the Xorg TearFree option for the
AMDGPU to cope with various tearing issues.
Signed-off-by: Awais Belal <awais_belal@mentor.com>
|
|
Various Xorg configs are needed to be set in different scenarios
and can become painful when trying to configure one-by-one.
This now provides a way to simply populate an array and
use that to populate as many values as required.
Signed-off-by: Awais Belal <awais_belal@mentor.com>
|
|
common/xrandr: fixed scale issue
|
|
Whenever scale value is increased when it is between 0.0 - 1.0
and is kept within 0.0 - 1.0 range, the display does not expand
to fullscreen. It rather becomes smaller and leaves blank spaces on
the right and bottom sides. And the following error message shows up:
X Error of failed request: BadMatch (invalid parameter attributes)
Major opcode of failed request: 139 (RANDR)
Minor opcode of failed request: 7 (RRSetScreenSize)
Serial number of failed request: 36
Current serial number in output stream: 38
Before xrandr attempts to apply the transformation matrix requested
by "scale", "transform" or "scale-from" options, it disables the
CRTCs for which the transformed image's dimensions are larger than the
target display's resolution/mode. But, when the image is smaller, it
does not disable the CRTCs, and proceeds with applying the
transformation (involves framebuffer dimension changes). This works well
for all the scale values and transitions except for the above mentioned
transitions.
This patch fixes such issues!
Issue: INTAMDDET-2189
Signed-off-by: Arsalan-Awan <Arsalan_Awan@mentor.com>
|
|
Fixed build with EXTRA_IMAGE_FEATURES "dbg-pkgs"
|
|
The packaging and pkg splitting was not working as it should be.
"llvm6.0-dbg" pkg was eating up all available debug files/info and it
was leaving nothing for "libllvm6.0-llvm-6.0-dbg" pkg to get built.
libllvm6.0-llvm-6.0-dbg is a pkg that was not only getting set as a
dependency for llvm6.0-dbg by our recipe, but also actually is a
dependency for some mesa pkgs.
(as mentioned in this commit openembedded/meta-openembedded @ b95f9d5).
This commit syncs the recipe with upstream and that fixes the following
issue:
With dbg-pkgs added to EXTRA_IMAGE_FEATURES, build fails at do_rootfs with
the following error in log.do_rootfs:
Collected errors:
* calculate_dependencies_for: Cannot satisfy the following dependencies
for llvm6.0-dbg:
* libllvm6.0-llvm-6.0-dbg *
* opkg_solver_install: Cannot install package llvm6.0-dbg.
Issue: INTAMDDET-2285
Signed-off-by: Arsalan-Awan <Arsalan_Awan@mentor.com>
|
|
With dbg-pkgs added to EXTRA_IMAGE_FEATURES, build fails at do_bootimg with
an error that ROOTFS_SIZE exceeds 4GB.
Post install scripts and the package management system use disk space inside
an overhead area (which defaults to "1.3"). We will reduce it to 1.1 to
allow build with "dbg-pkgs". Otherwise ROOTFS_SIZE exceeds 4GB.
(Not possible for FAT fs used for HDDIMG)
where:
ROOTFS_SIZE = (actual-rootfs-dir-size * overhead-factor) + rootfs-extra-space
The following can be used to add extra available space in rootfs as required:
IMAGE_ROOTFS_EXTRA_SPACE_append = " + <size-in-kilobytes>"
Also, when using the above variable to add extra space, it may cause an error
similar to this:
The initramfs size 294088(K) overrides INITRAMFS_MAXSIZE: 131072(K)
You can set INITRAMFS_MAXSIZE a larger value. Usually, it should
be less than 1/2 of ram size, or you may fail to boot it.
You may need to adjust INITRAMFS_MAXSIZE variable to fix that, or as a
workaround, you could set it to an empty string, to avoid such error, as:
INITRAMFS_MAXSIZE = ""
Issue: INTAMDDET-2285
Signed-off-by: Arsalan-Awan <Arsalan_Awan@mentor.com>
|
|
Fix various build issues
|
|
v1000/linux-yocto: disable efi vars for pstore by default
|
|
This backports a fix for hosts using recent glibc,
the support is backported from
http://cgit.openembedded.org/openembedded-core/commit/?id=b1b7a46f0febe7ac17ec46ecdf88b4120a7d75c4
Upstream morty already has this fixed so we do a
conditional backport for MEL releases.
Signed-off-by: Awais Belal <awais_belal@mentor.com>
|
|
This is essentially a backport of
http://cgit.openembedded.org/openembedded-core/commit/?id=7687234ff0f36ae943ca17624da9cdd8da0da83b
for MEL releases so the patch is included
conditionally.
Signed-off-by: Awais Belal <awais_belal@mentor.com>
|
|
There are adverse effects of using the pstore backend for
this sort of usage and should only be enabled when there's
a real need so disable feature in default environment.
Signed-off-by: Awais Belal <awais_belal@mentor.com>
|
|
v1000: drop 4.4 kernel append
|
|
This got missed from the 4.9 upgrade.
Signed-off-by: Awais Belal <awais_belal@mentor.com>
|
|
vulkan-tools: fix build error with Qt5
|
|
The vulkan-tools configuration process checks for
Qt5 _and_ Qt5Svg availability before enabling the
traceviewer but our enablement of the feature only
checks for Qt5 layer which creates a discrepancy.
This ends up in an installation failure when we
try to manually install traceviewer binary which
is skipped during the build process due to the
above reason.
We fix this by depending upon the qtsvg package
as well which is checked for during the configuration
process otherwise a race condition would cause
undesireable/unknown condition.
Signed-off-by: Awais Belal <awais_belal@mentor.com>
|
|
V1000: Fixed audio/video desync in mpv with vdpau deint denoise
|
|
Updated mpv to v0.24.0 and ffmpeg to v3.2.2 for v1000 to fix
audio/video desynchronization message that is displayed when
trying to deinterlace/denoise an H264 Video through VDPAU.
Signed-off-by: Arsalan-Awan <Arsalan_Awan@mentor.com>
|
|
libva: upgrade to 1.7.3
|
|
Fix for Vulkan device info is set as "Unknown AMD GPU"
|
|
This upgrades libva to 1.7.3 to align with requirements.
Signed-off-by: Awais Belal <awais_belal@mentor.com>
|
|
Signed-off-by: Pavan Kumar Ramayanam <pavan.ramayanam@amd.com>
|
|
boost: fix build failure due to pch
|
|
This backports
http://cgit.openembedded.org/openembedded-core/commit/meta/recipes-support/boost?id=9f30160f8623aebe8459e5b155d01397ff0f13b3
The failure occurs due to a race condition so it is
hard to reproduce reliably but it is seen every now
and then.
Signed-off-by: Awais Belal <awais_belal@mentor.com>
|
|
v1000 kernel configs consolidation
|
|
Handle CONFIG_BT_* similar to the common layer.
Signed-off-by: Awais Belal <awais_belal@mentor.com>
|
|
Consolidating kgdb configs through fragments as it
is done in the common layer as well.
Signed-off-by: Awais Belal <awais_belal@mentor.com>
|
|
Consolidating wifi configurations just like we did
for HID.
Signed-off-by: Awais Belal <awais_belal@mentor.com>
|
|
Consolidating HID configuration to a fragment helps
maintaining the overall configuration on modular
level. This is the same approach used in the common
layer so follow suit.
Signed-off-by: Awais Belal <awais_belal@mentor.com>
|
|
snowyowl-user-config.cfg: set CONFIG_MMC_BLOCK=y
|
|
Snowyowl
|
|
The Snowy Owl board features an eMMC device that can
be used as a boot media. If the block device isn't
made part of the kernel the installation mechanism
which runs through initrd without kernel modules
fails to find the device which can be selected as
an installation candidate.
Signed-off-by: Awais Belal <awais_belal@mentor.com>
|
|
spirv-tools: fix build for LunarG SDK 1.0.65
|
|
Signed-off-by: Sudheesh Mavila <sudheesh.mavila@amd.com>
|
|
Signed-off-by: Sudheesh Mavila <sudheesh.mavila@amd.com>
|
|
Signed-off-by: Sudheesh Mavila <sudheesh.mavila@amd.com>
|
|
Signed-off-by: Sudheesh Mavila <sudheesh.mavila@amd.com>
|
|
Signed-off-by: Sudheesh Mavila <sudheesh.mavila@amd.com>
|
|
The CMake changes were made upstream as well so the patch
fails with the spirv revs that we're now using for LunarG
SDK 1.0.65. This was missed in the SDK upgrade process and
do_patch fails so drop the unneeded patch.
Signed-off-by: Awais Belal <awais_belal@mentor.com>
|
|
v1000: move to grub-efi
|
|
The v1000 platform does not support grub in legacy mode
with a dual boot setup which is required in case of the
MEL Lite release cycle. Drop grub and use grub-efi for
the BSP in order to cope up with the change.
Signed-off-by: Awais Belal <awais_belal@mentor.com>
|
|
The grub-install mechanism for grub-efi requires efibootmgr
at runtime so add a dependency accordingly.
Signed-off-by: Awais Belal <awais_belal@mentor.com>
|
|
update for Post-GA update 1 Release
|
|
Signed-off-by: Pavan Kumar Ramayanam <pavan.ramayanam@amd.com>
|
|
Signed-off-by: Pavan Kumar Ramayanam <pavan.ramayanam@amd.com>
|
|
v1000: amdgpu-firmware: add GPU firmware
|
|
Signed-off-by: Pavan Kumar Ramayanam <pavan.ramayanam@amd.com>
|