ti-sgx-ddk-um: Remove RDEPENDS on libdrm-omap
PATCH v2 1/6 ti-sgx-ddk-um: Remove RDEPENDS on libdrm-omap
We no longer use libdrm-omap so remove it.
--- .../recipes-graphics/libgles/ti-sgx-ddk-um_1.17.4948957.bb | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
meta-ti-bsp/recipes-graphics/libgles/ti-sgx-ddk-um_1.17.4948957.bb +1 -1
@@ -34,7 +34,7 @@ def use_initscript(d): inherit ${@oe.utils.ifelse(use_initscript(d), 'update-rc.d', '')} -RDEPENDS:${PN} += "libdrm libdrm-omap" +RDEPENDS:${PN} += "libdrm" RRECOMMENDS:${PN} += "ti-sgx-ddk-km"
Signature · 1 line
2.39.2
PATCH v2 2/6 ti-sgx-ddk-um: Remove no longer needed CLEANBROKEN and INSANE_SKIPs
The "clean" target has been fixed along with some other issues that caused us to need these set. Remove them now.
--- .../recipes-graphics/libgles/ti-sgx-ddk-um_1.17.4948957.bb | 6 ++---- 1 file changed, 2 insertions(+), 4 deletions(-)
meta-ti-bsp/recipes-graphics/libgles/ti-sgx-ddk-um_1.17.4948957.bb +2 -4
@@ -67,7 +67,5 @@ FILES:${PN} += "${sysconfdir}/*" FILES:${PN} += "${datadir}/*" FILES:${PN} += "${nonarch_base_libdir}/udev/rules.d" -INSANE_SKIP:${PN} += "dev-so ldflags useless-rpaths" -INSANE_SKIP:${PN} += "already-stripped dev-deps" - -CLEANBROKEN = "1" +INSANE_SKIP:${PN} += "ldflags" +INSANE_SKIP:${PN} += "already-stripped"
Signature · 1 line
2.39.2
PATCH v2 3/6 ti-sgx-ddk-km: Be specific on supported K3 machines
In reply to Andrew DavisOf the K3 family only the AM65x device is currently supported by this driver. COMPATIBLE_MACHINE should match.
--- .../recipes-bsp/powervr-drivers/ti-sgx-ddk-km_1.17.4948957.bb | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
meta-ti-bsp/recipes-bsp/powervr-drivers/ti-sgx-ddk-km_1.17.4948957.bb +2 -2
@@ -7,7 +7,7 @@ inherit module PROVIDES = "virtual/gpudriver" -COMPATIBLE_MACHINE = "ti33x|ti43x|omap-a15|k3" +COMPATIBLE_MACHINE = "ti33x|ti43x|omap-a15|am65xx" MACHINE_KERNEL_PR:append = "x" PR = "${MACHINE_KERNEL_PR}"
@@ -27,7 +27,7 @@ SRCREV = "eda7780bfd5277e16913c9bc0b0e6892b4e79063" TARGET_PRODUCT:omap-a15 = "jacinto6evm" TARGET_PRODUCT:ti33x = "ti335x" TARGET_PRODUCT:ti43x = "ti437x" -TARGET_PRODUCT:k3 = "ti654x" +TARGET_PRODUCT:am65xx = "ti654x" EXTRA_OEMAKE += 'KERNELDIR="${STAGING_KERNEL_DIR}" TARGET_PRODUCT=${TARGET_PRODUCT} WINDOW_SYSTEM=nulldrmws'
Signature · 1 line
2.39.2
PATCH v2 4/6 ti-sgx-ddk-km: Use PVR_BUILD and PVR_WS to match Rogue
In reply to Andrew DavisUse these to select build type and window system support to match how the Rouge driver does the same. No functional change.
--- .../powervr-drivers/ti-sgx-ddk-km_1.17.4948957.bb | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-)
meta-ti-bsp/recipes-bsp/powervr-drivers/ti-sgx-ddk-km_1.17.4948957.bb +6 -4
@@ -28,19 +28,21 @@ TARGET_PRODUCT:omap-a15 = "jacinto6evm" TARGET_PRODUCT:ti33x = "ti335x" TARGET_PRODUCT:ti43x = "ti437x" TARGET_PRODUCT:am65xx = "ti654x" +PVR_BUILD = "release" +PVR_WS = "nulldrmws" -EXTRA_OEMAKE += 'KERNELDIR="${STAGING_KERNEL_DIR}" TARGET_PRODUCT=${TARGET_PRODUCT} WINDOW_SYSTEM=nulldrmws' +EXTRA_OEMAKE += 'KERNELDIR="${STAGING_KERNEL_DIR}" BUILD=${PVR_BUILD} TARGET_PRODUCT=${TARGET_PRODUCT} WINDOW_SYSTEM=${PVR_WS}' do_compile:prepend() { cd ${S}/eurasia_km/eurasiacon/build/linux2/omap_linux } do_install() { - make -C ${STAGING_KERNEL_DIR} M=${B}/eurasia_km/eurasiacon/binary_omap_linux_nulldrmws_release/target_armhf/kbuild INSTALL_MOD_PATH=${D}${root_prefix} PREFIX=${STAGING_DIR_HOST} modules_install + make -C ${STAGING_KERNEL_DIR} M=${B}/eurasia_km/eurasiacon/binary_omap_linux_${PVR_WS}_${PVR_BUILD}/target_armhf/kbuild INSTALL_MOD_PATH=${D}${root_prefix} PREFIX=${STAGING_DIR_HOST} modules_install } -do_install:k3() { - make -C ${STAGING_KERNEL_DIR} M=${B}/eurasia_km/eurasiacon/binary_omap_linux_nulldrmws_release/target_aarch64/kbuild INSTALL_MOD_PATH=${D}${root_prefix} PREFIX=${STAGING_DIR_HOST} modules_install +do_install:am65xx() { + make -C ${STAGING_KERNEL_DIR} M=${B}/eurasia_km/eurasiacon/binary_omap_linux_${PVR_WS}_${PVR_BUILD}/target_aarch64/kbuild INSTALL_MOD_PATH=${D}${root_prefix} PREFIX=${STAGING_DIR_HOST} modules_install } RRECOMMENDS:${PN} += "ti-sgx-ddk-um"
Signature · 1 line
2.39.2
PATCH v2 5/6 meta-ti-bsp: Remove SGX driver from SIGGEN_EXCLUDERECIPES_ABISAFE
In reply to Andrew DavisNeither of recipes nor their ABI is all that stable. OpenGL might be slightly more stable, but that is not what these provide anymore. Remove these.
--- meta-ti-bsp/conf/layer.conf | 5 ----- 1 file changed, 5 deletions(-)
meta-ti-bsp/conf/layer.conf +0 -5
@@ -18,9 +18,4 @@ LAYERDEPENDS_meta-ti-bsp = " \ meta-arm \ " -SIGGEN_EXCLUDERECIPES_ABISAFE += " \ - ti-sgx-ddk-km \ - ti-sgx-ddk-um \ -" - HOSTTOOLS_NONFATAL += "truncate xxd comm"
Signature · 1 line
2.39.2
PATCH v2 6/6 meta-ti-bsp: graphics: Rename SGX driver recipes to match Rogue
In reply to Andrew DavisThe names we use for the GPU components are not consistent nor do the intuitively describe their function. Let's fix that.
We do 3 things, first powervr-drivers/ contains only the kernel-mode half of the driver, for GPUs most of the driver lives in userspace, and the "PowerVR" name is no longer correct for the latest GPUs. Rename this to ti-img-gpu-km/ which matches better with our naming for the user-mode half and the TI/IMG codecs.
Next, ti-img-rogue-driver is renamed ti-img-rogue-km, same reasoning as above and matches the SGX version of the same.
The usermode driver lives in a directory named "libgles", but the driver does not just provide libgles, it also can provide OpenCL, DRI, and Vulkan. Rename to match the kernel-mode half.
Lastly, the acronym "DDK" is short for "Driver Development Kit". This kit is provided by Imagination to TI and is what we use to "Develop" the "Driver". What we ship is not the DDK, it is the driver itself. The name DDK seems to have just been passed through. Remove use of "ddk" where we can.
To summarize:
* powervr-drivers -> ti-img-gpu-km
* ti-img-rogue-driver -> ti-img-rogue-km
* ti-sgx-ddk-km -> ti-img-sgx-km * libgles -> ti-img-gpu-umlibs * ti-sgx-ddk-um -> ti-img-sgx-umlibs
---
...compiler-support-OpenEmbedded-nodistro-internal-aarc.patch | 0
.../ti-img-rogue-km_1.18.6276027.bb} | 0
.../ti-img-sgx-km_1.17.4948957.bb} | 2 +-
meta-ti-bsp/recipes-graphics/mesa/mesa-gl_22.%.bbappend | 2 +-
meta-ti-bsp/recipes-graphics/mesa/mesa_22.%.bbappend | 2 +-
meta-ti-bsp/recipes-graphics/mesa/pvr-mesa.inc | 4 ++--
.../ti-img-rogue-umlibs_1.18.6276027.bb | 3 ++-
.../ti-img-sgx-umlibs}/pvrsrvkm.rules | 0
.../ti-img-sgx-umlibs_1.17.4948957.bb} | 2 +-
9 files changed, 8 insertions(+), 7 deletions(-)
rename meta-ti-bsp/recipes-bsp/{powervr-drivers/ti-img-rogue-driver => ti-img-gpu-km/ti-img-rogue-km}/0001-compiler-support-OpenEmbedded-nodistro-internal-aarc.patch (100%)
rename meta-ti-bsp/recipes-bsp/{powervr-drivers/ti-img-rogue-driver_1.18.6276027.bb => ti-img-gpu-km/ti-img-rogue-km_1.18.6276027.bb} (100%)
rename meta-ti-bsp/recipes-bsp/{powervr-drivers/ti-sgx-ddk-km_1.17.4948957.bb => ti-img-gpu-km/ti-img-sgx-km_1.17.4948957.bb} (97%)
rename meta-ti-bsp/recipes-graphics/{libgles => ti-img-gpu-umlibs}/ti-img-rogue-umlibs_1.18.6276027.bb (94%)
rename meta-ti-bsp/recipes-graphics/{libgles/ti-sgx-ddk-um => ti-img-gpu-umlibs/ti-img-sgx-umlibs}/pvrsrvkm.rules (100%)
rename meta-ti-bsp/recipes-graphics/{libgles/ti-sgx-ddk-um_1.17.4948957.bb => ti-img-gpu-umlibs/ti-img-sgx-umlibs_1.17.4948957.bb} (98%)meta-ti-bsp/recipes-bsp/ti-img-gpu-km/ti-img-rogue-km/0001-compiler-support-OpenEmbedded-nodistro-internal-aarc.patch renamed from meta-ti-bsp/recipes-bsp/powervr-drivers/ti-img-rogue-driver/0001-compiler-support-OpenEmbedded-nodistro-internal-aarc.patch +0 -0
meta-ti-bsp/recipes-bsp/ti-img-gpu-km/ti-img-rogue-km_1.18.6276027.bb renamed from meta-ti-bsp/recipes-bsp/powervr-drivers/ti-img-rogue-driver_1.18.6276027.bb +0 -0
meta-ti-bsp/recipes-bsp/ti-img-gpu-km/ti-img-sgx-km_1.17.4948957.bb renamed from meta-ti-bsp/recipes-bsp/powervr-drivers/ti-sgx-ddk-km_1.17.4948957.bb +1 -1
@@ -45,4 +45,4 @@ do_install:am65xx() { make -C ${STAGING_KERNEL_DIR} M=${B}/eurasia_km/eurasiacon/binary_omap_linux_${PVR_WS}_${PVR_BUILD}/target_aarch64/kbuild INSTALL_MOD_PATH=${D}${root_prefix} PREFIX=${STAGING_DIR_HOST} modules_install } -RRECOMMENDS:${PN} += "ti-sgx-ddk-um" +RRECOMMENDS:${PN} += "ti-img-sgx-umlibs"
meta-ti-bsp/recipes-graphics/mesa/mesa-gl_22.%.bbappend +1 -1
@@ -1 +1 @@ -require ${@bb.utils.contains_any('PREFERRED_PROVIDER_virtual/gpudriver', 'ti-img-rogue-driver ti-sgx-ddk-km', 'pvr-mesa.inc', '', d)} +require ${@bb.utils.contains_any('PREFERRED_PROVIDER_virtual/gpudriver', 'ti-img-rogue-km ti-img-sgx-km', 'pvr-mesa.inc', '', d)}
meta-ti-bsp/recipes-graphics/mesa/mesa_22.%.bbappend +1 -1
@@ -1 +1 @@ -require ${@bb.utils.contains_any('PREFERRED_PROVIDER_virtual/gpudriver', 'ti-img-rogue-driver ti-sgx-ddk-km', 'pvr-mesa.inc', '', d)} +require ${@bb.utils.contains_any('PREFERRED_PROVIDER_virtual/gpudriver', 'ti-img-rogue-km ti-img-sgx-km', 'pvr-mesa.inc', '', d)}
meta-ti-bsp/recipes-graphics/mesa/pvr-mesa.inc +2 -2
@@ -23,8 +23,8 @@ S = "${WORKDIR}/git" SRCREV = "44b82a7b6fa8a86243911b1b0d6a07b0fd8ecfba" PACKAGECONFIG:append = " \ - ${@bb.utils.contains('PREFERRED_PROVIDER_virtual/gpudriver', 'ti-img-rogue-driver', 'pvr', '', d)} \ - ${@bb.utils.contains('PREFERRED_PROVIDER_virtual/gpudriver', 'ti-sgx-ddk-km', 'sgx', '', d)} \ + ${@bb.utils.contains('PREFERRED_PROVIDER_virtual/gpudriver', 'ti-img-rogue-km', 'pvr', '', d)} \ + ${@bb.utils.contains('PREFERRED_PROVIDER_virtual/gpudriver', 'ti-img-sgx-km', 'sgx', '', d)} \ " PVR_DISPLAY_CONTROLLER_ALIAS ??= "tidss"
meta-ti-bsp/recipes-graphics/ti-img-gpu-umlibs/ti-img-rogue-umlibs_1.18.6276027.bb renamed from meta-ti-bsp/recipes-graphics/libgles/ti-img-rogue-umlibs_1.18.6276027.bb +2 -1
@@ -22,7 +22,8 @@ TARGET_PRODUCT:am62xx = "am62_linux" PVR_BUILD = "release" PVR_WS = "lws-generic" -RDEPENDS:${PN} += "mesa-megadriver libdrm ti-img-rogue-driver" +RDEPENDS:${PN} += "mesa-megadriver libdrm" +RDEPENDS:${PN} += "ti-img-rogue-km" do_install:append() { if ${@bb.utils.contains('DISTRO_FEATURES', 'usrmerge', 'true', 'false', d)}; then
meta-ti-bsp/recipes-graphics/ti-img-gpu-umlibs/ti-img-sgx-umlibs/pvrsrvkm.rules renamed from meta-ti-bsp/recipes-graphics/libgles/ti-sgx-ddk-um/pvrsrvkm.rules +0 -0
meta-ti-bsp/recipes-graphics/ti-img-gpu-umlibs/ti-img-sgx-umlibs_1.17.4948957.bb renamed from meta-ti-bsp/recipes-graphics/libgles/ti-sgx-ddk-um_1.17.4948957.bb +1 -1
@@ -36,7 +36,7 @@ inherit ${@oe.utils.ifelse(use_initscript(d), 'update-rc.d', '')} RDEPENDS:${PN} += "libdrm" -RRECOMMENDS:${PN} += "ti-sgx-ddk-km" +RRECOMMENDS:${PN} += "ti-img-sgx-km" S = "${WORKDIR}/git"
Signature · 1 line
2.39.2
Re: [meta-ti][master/kirkstone][PATCH v2 6/6] meta-ti-bsp: graphics: Rename SGX driver recipes to match Rogue
Quoted 22 lines · Andrew Davis via lists.yoctoproject.org
The names we use for the GPU components are not consistent nor do the intuitively describe their function. Let's fix that.
We do 3 things, first powervr-drivers/ contains only the kernel-mode half of the driver, for GPUs most of the driver lives in userspace, and the "PowerVR" name is no longer correct for the latest GPUs. Rename this to ti-img-gpu-km/ which matches better with our naming for the user-mode half and the TI/IMG codecs.
Next, ti-img-rogue-driver is renamed ti-img-rogue-km, same reasoning as above and matches the SGX version of the same.
The usermode driver lives in a directory named "libgles", but the driver does not just provide libgles, it also can provide OpenCL, DRI, and Vulkan. Rename to match the kernel-mode half.
Lastly, the acronym "DDK" is short for "Driver Development Kit". This kit is provided by Imagination to TI and is what we use to "Develop" the "Driver". What we ship is not the DDK, it is the driver itself. The name DDK seems to have just been passed through. Remove use of "ddk" where we can.
Quoted 7 lines
To summarize:
* powervr-drivers -> ti-img-gpu-km
* ti-img-rogue-driver -> ti-img-rogue-km* ti-sgx-ddk-km -> ti-img-sgx-km * libgles -> ti-img-gpu-umlibs * ti-sgx-ddk-um -> ti-img-sgx-umlibs
Renaming recipes will have a huge impact downstream! Last time we renamed omap5-sgx-ddk-um-linux to ti-sgx-ddk-um, there's been a lot of breakage and tons of complaints from customers and the community at large (in public and even more so in private)...
First of all, you are only advised to do such thing in master (and not right before the next release, which is now), definitely not in a stable branch, especially not in a LTS branch (e.g. kirkstone).
Second, do it during the version bump - keep the old version under the old name, add a new version with the new name.
Third, you have to provide backward compatibility and setup corresponding preemption/replacement policy (at least RPROVIDES, RREPLACES, RCONFLICTS). But that alone won't help with downstream layers adding their bbappends.
Just checking a few public OEM layers that are downstream to meta-ti (and that's just the tip of the iceberg, as majority of customer layers are not even public!):
Toradex (seems to be a bit outdated, I'm guessing they pin to specific SHA):
https://git.toradex.com/cgit/meta-toradex-ti.git/tree/recipes-bsp/powervr-drivers/ti-img-rogue-driver_1.15.6133109.bbappend?h=kirkstone-6.x.y
https://git.toradex.com/cgit/meta-toradex-ti.git/tree/recipes-graphics/libgles/ti-img-rogue-umlibs_1.15.6133109.bbappend?h=kirkstone-6.x.y
Variscite:
Their meta-variscite-bsp-ti layer doesn't appear to touch graphics and is still on dunfell...
SanCloud:
Also doesn't appear to modify graphics recipes directly.
AGL:
https://gerrit.automotivelinux.org/gerrit/gitweb?p=AGL/meta-agl.git;a=blob;f=meta-agl-bsp/meta-ti-bsp/recipes-bsp/ti-sgx-ddk-km/ti-sgx-ddk-km_%25.bbappend;hb=HEAD
https://gerrit.automotivelinux.org/gerrit/gitweb?p=AGL/meta-agl.git;a=blob;f=meta-agl-bsp/meta-ti-bsp/recipes-graphics/libgles/ti-sgx-ddk-um_%25.bbappend;hb=HEAD
These are just bbappends, but I'm sure there are other references to these current names in configs and/or their recipes...
AFAIK, AGL, for example, hasn't yet fully recovered from the latest graphics changes separating Mesa - there's work in progress going on right now.
I fully understand the desire to simplify and unify things, like directory and recipe names. But it comes with a lot of complications and require extra thought put into how not to break things for everyone around...
Yocto Project guidelines for public BSPs are much stricter than those for Distros or customer/product layers. It may be convenient to think of meta-ti as a "leaf" of the project, but it is actually quite a large "branch" with other "branches" and "leaves" downstream depending on it...
Signature · 1 line
Denys
Quoted 108 lines
Signed-off-by Andrew Davis <[email protected]>--- ...compiler-support-OpenEmbedded-nodistro-internal-aarc.patch | 0 .../ti-img-rogue-km_1.18.6276027.bb} | 0 .../ti-img-sgx-km_1.17.4948957.bb} | 2 +- meta-ti-bsp/recipes-graphics/mesa/mesa-gl_22.%.bbappend | 2 +- meta-ti-bsp/recipes-graphics/mesa/mesa_22.%.bbappend | 2 +- meta-ti-bsp/recipes-graphics/mesa/pvr-mesa.inc | 4 ++-- .../ti-img-rogue-umlibs_1.18.6276027.bb | 3 ++- .../ti-img-sgx-umlibs}/pvrsrvkm.rules | 0 .../ti-img-sgx-umlibs_1.17.4948957.bb} | 2 +- 9 files changed, 8 insertions(+), 7 deletions(-) rename meta-ti-bsp/recipes-bsp/{powervr-drivers/ti-img-rogue-driver => ti-img-gpu-km/ti-img-rogue-km}/0001-compiler-support-OpenEmbedded-nodistro-internal-aarc.patch (100%) rename meta-ti-bsp/recipes-bsp/{powervr-drivers/ti-img-rogue-driver_1.18.6276027.bb => ti-img-gpu-km/ti-img-rogue-km_1.18.6276027.bb} (100%) rename meta-ti-bsp/recipes-bsp/{powervr-drivers/ti-sgx-ddk-km_1.17.4948957.bb => ti-img-gpu-km/ti-img-sgx-km_1.17.4948957.bb} (97%) rename meta-ti-bsp/recipes-graphics/{libgles => ti-img-gpu-umlibs}/ti-img-rogue-umlibs_1.18.6276027.bb (94%) rename meta-ti-bsp/recipes-graphics/{libgles/ti-sgx-ddk-um => ti-img-gpu-umlibs/ti-img-sgx-umlibs}/pvrsrvkm.rules (100%) rename meta-ti-bsp/recipes-graphics/{libgles/ti-sgx-ddk-um_1.17.4948957.bb => ti-img-gpu-umlibs/ti-img-sgx-umlibs_1.17.4948957.bb} (98%)meta-ti-bsp/recipes-bsp/ti-img-gpu-km/ti-img-rogue-km/0001-compiler-support-OpenEmbedded-nodistro-internal-aarc.patch renamed from meta-ti-bsp/recipes-bsp/powervr-drivers/ti-img-rogue-driver/0001-compiler-support-OpenEmbedded-nodistro-internal-aarc.patch +0
-0meta-ti-bsp/recipes-bsp/ti-img-gpu-km/ti-img-rogue-km_1.18.6276027.bb renamed from meta-ti-bsp/recipes-bsp/powervr-drivers/ti-img-rogue-driver_1.18.6276027.bb +0
-0meta-ti-bsp/recipes-bsp/ti-img-gpu-km/ti-img-sgx-km_1.17.4948957.bb renamed from meta-ti-bsp/recipes-bsp/powervr-drivers/ti-sgx-ddk-km_1.17.4948957.bb +1
-1@@ -45,4 +45,4 @@ do_install:am65xx() { make -C ${STAGING_KERNEL_DIR} M=${B}/eurasia_km/eurasiacon/binary_omap_linux_${PVR_WS}_${PVR_BUILD}/target_aarch64/kbuild INSTALL_MOD_PATH=${D}${root_prefix} PREFIX=${STAGING_DIR_HOST} modules_install } -RRECOMMENDS:${PN} += "ti-sgx-ddk-um" +RRECOMMENDS:${PN} += "ti-img-sgx-umlibs"meta-ti-bsp/recipes-graphics/mesa/mesa-gl_22.%.bbappend +1
-1@@ -1 +1 @@ -require ${@bb.utils.contains_any('PREFERRED_PROVIDER_virtual/gpudriver', 'ti-img-rogue-driver ti-sgx-ddk-km', 'pvr-mesa.inc', '', d)} +require ${@bb.utils.contains_any('PREFERRED_PROVIDER_virtual/gpudriver', 'ti-img-rogue-km ti-img-sgx-km', 'pvr-mesa.inc', '', d)}meta-ti-bsp/recipes-graphics/mesa/mesa_22.%.bbappend +1
-1@@ -1 +1 @@ -require ${@bb.utils.contains_any('PREFERRED_PROVIDER_virtual/gpudriver', 'ti-img-rogue-driver ti-sgx-ddk-km', 'pvr-mesa.inc', '', d)} +require ${@bb.utils.contains_any('PREFERRED_PROVIDER_virtual/gpudriver', 'ti-img-rogue-km ti-img-sgx-km', 'pvr-mesa.inc', '', d)}meta-ti-bsp/recipes-graphics/mesa/pvr-mesa.inc +2
-2@@ -23,8 +23,8 @@ S = "${WORKDIR}/git" SRCREV = "44b82a7b6fa8a86243911b1b0d6a07b0fd8ecfba" PACKAGECONFIG:append = " \ - ${@bb.utils.contains('PREFERRED_PROVIDER_virtual/gpudriver', 'ti-img-rogue-driver', 'pvr', '', d)} \ - ${@bb.utils.contains('PREFERRED_PROVIDER_virtual/gpudriver', 'ti-sgx-ddk-km', 'sgx', '', d)} \ + ${@bb.utils.contains('PREFERRED_PROVIDER_virtual/gpudriver', 'ti-img-rogue-km', 'pvr', '', d)} \ + ${@bb.utils.contains('PREFERRED_PROVIDER_virtual/gpudriver', 'ti-img-sgx-km', 'sgx', '', d)} \ " PVR_DISPLAY_CONTROLLER_ALIAS ??= "tidss"meta-ti-bsp/recipes-graphics/ti-img-gpu-umlibs/ti-img-rogue-umlibs_1.18.6276027.bb renamed from meta-ti-bsp/recipes-graphics/libgles/ti-img-rogue-umlibs_1.18.6276027.bb +2
-1@@ -22,7 +22,8 @@ TARGET_PRODUCT:am62xx = "am62_linux" PVR_BUILD = "release" PVR_WS = "lws-generic" -RDEPENDS:${PN} += "mesa-megadriver libdrm ti-img-rogue-driver" +RDEPENDS:${PN} += "mesa-megadriver libdrm" +RDEPENDS:${PN} += "ti-img-rogue-km" do_install:append() { if ${@bb.utils.contains('DISTRO_FEATURES', 'usrmerge', 'true', 'false', d)}; thenmeta-ti-bsp/recipes-graphics/ti-img-gpu-umlibs/ti-img-sgx-umlibs/pvrsrvkm.rules renamed from meta-ti-bsp/recipes-graphics/libgles/ti-sgx-ddk-um/pvrsrvkm.rules +0
-0meta-ti-bsp/recipes-graphics/ti-img-gpu-umlibs/ti-img-sgx-umlibs_1.17.4948957.bb renamed from meta-ti-bsp/recipes-graphics/libgles/ti-sgx-ddk-um_1.17.4948957.bb +1
-1@@ -36,7 +36,7 @@ inherit ${@oe.utils.ifelse(use_initscript(d), 'update-rc.d', '')} RDEPENDS:${PN} += "libdrm" -RRECOMMENDS:${PN} += "ti-sgx-ddk-km" +RRECOMMENDS:${PN} += "ti-img-sgx-km" S = "${WORKDIR}/git"Signature · 1 line
2.39.2
Re: [meta-ti][master/kirkstone][PATCH v2 5/6] meta-ti-bsp: Remove SGX driver from SIGGEN_EXCLUDERECIPES_ABISAFE
In reply to Andrew DavisQuoted 3 lines · Andrew Davis via lists.yoctoproject.org
Neither of recipes nor their ABI is all that stable. OpenGL might be slightly more stable, but that is not what these provide anymore. Remove these.
This is not what you think it is... :) Actually, Rogue recipes should be added to the list.
This is specific to signatures/shared state. The ABI in this case is libgles and libegl (which are stable) and tells bitbake to avoid rebuilding generic apps depending on these packages between platfoms in the same architecture.
Quoted 21 lines
Signed-off-by Andrew Davis <[email protected]>--- meta-ti-bsp/conf/layer.conf | 5 ----- 1 file changed, 5 deletions(-)meta-ti-bsp/conf/layer.conf +0
-5@@ -18,9 +18,4 @@ LAYERDEPENDS_meta-ti-bsp = " \ meta-arm \ " -SIGGEN_EXCLUDERECIPES_ABISAFE += " \ - ti-sgx-ddk-km \ - ti-sgx-ddk-um \ -" - HOSTTOOLS_NONFATAL += "truncate xxd comm"Signature · 1 line
2.39.2
I'm fine with the first 4 patches in this series, if Ryan wants to merge them.
Quoted 22 lines · Andrew Davis via lists.yoctoproject.org
We no longer use libdrm-omap so remove it.
Signed-off-by Andrew Davis <[email protected]>--- .../recipes-graphics/libgles/ti-sgx-ddk-um_1.17.4948957.bb | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)meta-ti-bsp/recipes-graphics/libgles/ti-sgx-ddk-um_1.17.4948957.bb +1
-1@@ -34,7 +34,7 @@ def use_initscript(d): inherit ${@oe.utils.ifelse(use_initscript(d), 'update-rc.d', '')} -RDEPENDS:${PN} += "libdrm libdrm-omap" +RDEPENDS:${PN} += "libdrm" RRECOMMENDS:${PN} += "ti-sgx-ddk-km"Signature · 1 line
2.39.2
Re: [meta-ti][master/kirkstone][PATCH v2 5/6] meta-ti-bsp: Remove SGX driver from SIGGEN_EXCLUDERECIPES_ABISAFE
In reply to Denys DmytriyenkoQuoted 5 lines · Denys Dmytriyenko
Quoted 3 lines · Andrew Davis via lists.yoctoproject.org
Neither of recipes nor their ABI is all that stable. OpenGL might be slightly more stable, but that is not what these provide anymore. Remove these.
Quoted 7 lines
This is not what you think it is... :) Actually, Rogue recipes should be added to the list.
This is specific to signatures/shared state. The ABI in this case is libgles and libegl (which are stable) and tells bitbake to avoid rebuilding generic apps depending on these packages between platfoms in the same architecture.
These do not provide those API anymore, Mesa does. And nothing should depend on the KM package other than the UM libs.
Andrew
Quoted 22 lines
Quoted 22 lines
Signed-off-by Andrew Davis <[email protected]>--- meta-ti-bsp/conf/layer.conf | 5 ----- 1 file changed, 5 deletions(-)meta-ti-bsp/conf/layer.conf +0
-5@@ -18,9 +18,4 @@ LAYERDEPENDS_meta-ti-bsp = " \ meta-arm \ " -SIGGEN_EXCLUDERECIPES_ABISAFE += " \ - ti-sgx-ddk-km \ - ti-sgx-ddk-um \ -" - HOSTTOOLS_NONFATAL += "truncate xxd comm"Signature · 1 line
2.39.2
Re: [meta-ti][master/kirkstone][PATCH v2 5/6] meta-ti-bsp: Remove SGX driver from SIGGEN_EXCLUDERECIPES_ABISAFE
Quoted 9 lines · Andrew Davis
Quoted 8 lines · Denys Dmytriyenko
Quoted 3 lines · Andrew Davis via lists.yoctoproject.org
Neither of recipes nor their ABI is all that stable. OpenGL might be slightly more stable, but that is not what these provide anymore. Remove these.
This is not what you think it is... :) Actually, Rogue recipes should be added to the list.
Quoted 7 lines
Quoted 4 lines
This is specific to signatures/shared state. The ABI in this case is libgles and libegl (which are stable) and tells bitbake to avoid rebuilding generic apps depending on these packages between platfoms in the same architecture.
These do not provide those API anymore, Mesa does. And nothing should depend on the KM package other than the UM libs.
Still doesn't matter - if you have Mesa depend on these, you still need them listed in here to avoid rebuilding Mesa from one platform to another. And everything else downstream.
Quoted 20 lines
Quoted 20 lines
Quoted 20 lines
Signed-off-by Andrew Davis <[email protected]>--- meta-ti-bsp/conf/layer.conf | 5 ----- 1 file changed, 5 deletions(-)meta-ti-bsp/conf/layer.conf +0
-5@@ -18,9 +18,4 @@ LAYERDEPENDS_meta-ti-bsp = " \ meta-arm \ " -SIGGEN_EXCLUDERECIPES_ABISAFE += " \ - ti-sgx-ddk-km \ - ti-sgx-ddk-um \ -" - HOSTTOOLS_NONFATAL += "truncate xxd comm"Signature · 1 line
2.39.2
Re: [meta-ti][master/kirkstone][PATCH v2 5/6] meta-ti-bsp: Remove SGX driver from SIGGEN_EXCLUDERECIPES_ABISAFE
Quoted 15 lines · Denys Dmytriyenko
Quoted 14 lines · Andrew Davis
Quoted 12 lines · Denys Dmytriyenko
Quoted 3 lines · Andrew Davis via lists.yoctoproject.org
Neither of recipes nor their ABI is all that stable. OpenGL might be slightly more stable, but that is not what these provide anymore. Remove these.
This is not what you think it is... :) Actually, Rogue recipes should be added to the list.
This is specific to signatures/shared state. The ABI in this case is libgles and libegl (which are stable) and tells bitbake to avoid rebuilding generic apps depending on these packages between platfoms in the same architecture.
Quoted 7 lines
Quoted 2 lines
These do not provide those API anymore, Mesa does. And nothing should depend on the KM package other than the UM libs.
Still doesn't matter - if you have Mesa depend on these, you still need them listed in here to avoid rebuilding Mesa from one platform to another. And everything else downstream.
We *want* Mesa to be rebuilt if these change, these do not provide a stable ABI anymore, it can and does change, sometimes between platforms (SGX KM -> UM is based on plat).
Andrew
Quoted 21 lines
Quoted 21 lines
Quoted 21 lines
Quoted 21 lines
Signed-off-by Andrew Davis <[email protected]>--- meta-ti-bsp/conf/layer.conf | 5 ----- 1 file changed, 5 deletions(-)meta-ti-bsp/conf/layer.conf +0
-5@@ -18,9 +18,4 @@ LAYERDEPENDS_meta-ti-bsp = " \ meta-arm \ " -SIGGEN_EXCLUDERECIPES_ABISAFE += " \ - ti-sgx-ddk-km \ - ti-sgx-ddk-um \ -" - HOSTTOOLS_NONFATAL += "truncate xxd comm"Signature · 1 line
2.39.2
Re: [meta-ti][master/kirkstone][PATCH v2 6/6] meta-ti-bsp: graphics: Rename SGX driver recipes to match Rogue
In reply to Denys DmytriyenkoQuoted 31 lines · Denys Dmytriyenko
Quoted 22 lines · Andrew Davis via lists.yoctoproject.org
The names we use for the GPU components are not consistent nor do the intuitively describe their function. Let's fix that.
We do 3 things, first powervr-drivers/ contains only the kernel-mode half of the driver, for GPUs most of the driver lives in userspace, and the "PowerVR" name is no longer correct for the latest GPUs. Rename this to ti-img-gpu-km/ which matches better with our naming for the user-mode half and the TI/IMG codecs.
Next, ti-img-rogue-driver is renamed ti-img-rogue-km, same reasoning as above and matches the SGX version of the same.
The usermode driver lives in a directory named "libgles", but the driver does not just provide libgles, it also can provide OpenCL, DRI, and Vulkan. Rename to match the kernel-mode half.
Lastly, the acronym "DDK" is short for "Driver Development Kit". This kit is provided by Imagination to TI and is what we use to "Develop" the "Driver". What we ship is not the DDK, it is the driver itself. The name DDK seems to have just been passed through. Remove use of "ddk" where we can.
Quoted 7 lines
To summarize:
* powervr-drivers -> ti-img-gpu-km
* ti-img-rogue-driver -> ti-img-rogue-km* ti-sgx-ddk-km -> ti-img-sgx-km * libgles -> ti-img-gpu-umlibs * ti-sgx-ddk-um -> ti-img-sgx-umlibs
Quoted 9 lines
Renaming recipes will have a huge impact downstream! Last time we renamed omap5-sgx-ddk-um-linux to ti-sgx-ddk-um, there's been a lot of breakage and tons of complaints from customers and the community at large (in public and even more so in private)...
First of all, you are only advised to do such thing in master (and not right before the next release, which is now), definitely not in a stable branch, especially not in a LTS branch (e.g. kirkstone).
That's fine, this last patch doesn't need to go to kirkstone, I'll post it separately for master branch only.
Quoted 22 lines
Second, do it during the version bump - keep the old version under the old name, add a new version with the new name.
Third, you have to provide backward compatibility and setup corresponding preemption/replacement policy (at least RPROVIDES, RREPLACES, RCONFLICTS). But that alone won't help with downstream layers adding their bbappends.
Just checking a few public OEM layers that are downstream to meta-ti (and that's just the tip of the iceberg, as majority of customer layers are not even public!):
Toradex (seems to be a bit outdated, I'm guessing they pin to specific SHA):
https://git.toradex.com/cgit/meta-toradex-ti.git/tree/recipes-bsp/powervr-drivers/ti-img-rogue-driver_1.15.6133109.bbappend?h=kirkstone-6.x.y
https://git.toradex.com/cgit/meta-toradex-ti.git/tree/recipes-graphics/libgles/ti-img-rogue-umlibs_1.15.6133109.bbappend?h=kirkstone-6.x.yVariscite:
Their meta-variscite-bsp-ti layer doesn't appear to touch graphics and is still on dunfell...
Quoted 7 lines
SanCloud:
Also doesn't appear to modify graphics recipes directly.AGL:
https://gerrit.automotivelinux.org/gerrit/gitweb?p=AGL/meta-agl.git;a=blob;f=meta-agl-bsp/meta-ti-bsp/recipes-bsp/ti-sgx-ddk-km/ti-sgx-ddk-km_%25.bbappend;hb=HEAD
https://gerrit.automotivelinux.org/gerrit/gitweb?p=AGL/meta-agl.git;a=blob;f=meta-agl-bsp/meta-ti-bsp/recipes-graphics/libgles/ti-sgx-ddk-um_%25.bbappend;hb=HEAD
All three changes to the KM above are for adding compiler name support which is only needed due to our odd way of detection arch. That can be fixed then these should go away. The other two for UM are easy enough to fix and we should fix them in meta-ti for them.
Meta-ti is the upstream, fixes should go upstream. I get not doing this to the LTS branch, but for master branch we do need to make forward progress and clean things up like this patch does.
Quoted 6 lines
These are just bbappends, but I'm sure there are other references to these current names in configs and/or their recipes...
AFAIK, AGL, for example, hasn't yet fully recovered from the latest graphics changes separating Mesa - there's work in progress going on right now.
Well SGX isn't working at all right now now in kirkstone, so not sure what they are trying to fix at the moment..
Quoted 9 lines
I fully understand the desire to simplify and unify things, like directory and recipe names. But it comes with a lot of complications and require extra thought put into how not to break things for everyone around...
Yocto Project guidelines for public BSPs are much stricter than those for Distros or customer/product layers. It may be convenient to think of meta-ti as a "leaf" of the project, but it is actually quite a large "branch" with other "branches" and "leaves" downstream depending on it...
Re: [meta-ti][master/kirkstone][PATCH v2 5/6] meta-ti-bsp: Remove SGX driver from SIGGEN_EXCLUDERECIPES_ABISAFE
In reply to Andrew DavisQuoted 19 lines · Andrew Davis
Quoted 18 lines · Denys Dmytriyenko
Quoted 16 lines · Andrew Davis
Quoted 5 lines · Denys Dmytriyenko
Quoted 3 lines · Andrew Davis via lists.yoctoproject.org
Neither of recipes nor their ABI is all that stable. OpenGL might be slightly more stable, but that is not what these provide anymore. Remove these.
Quoted 7 lines
This is not what you think it is... :) Actually, Rogue recipes should be added to the list.
This is specific to signatures/shared state. The ABI in this case is libgles and libegl (which are stable) and tells bitbake to avoid rebuilding generic apps depending on these packages between platfoms in the same architecture.
These do not provide those API anymore, Mesa does. And nothing should depend on the KM package other than the UM libs.
Quoted 8 lines
Quoted 4 lines
Still doesn't matter - if you have Mesa depend on these, you still need them listed in here to avoid rebuilding Mesa from one platform to another. And everything else downstream.
We *want* Mesa to be rebuilt if these change, these do not provide a stable ABI anymore, it can and does change, sometimes between platforms (SGX KM -> UM is based on plat).
The name of the variable may be confusing. It's not about rebuilding Mesa once you make changes to the KM/UM pieces - that will happen automatically if the dependencies are tracked properly.
This variable is about re-use of the dependant generic packages. The issue it is trying to solve is when you have machine-specific packages early in the dependecy tree, everything down the dependency tree that depends on these packages will be treated as machine-specific, even if they are generic. Since bitbake will try to play safe here, you have to tell it otherwise.
Again, this is not about rebuilds of a dependent component between changes. This is about rebuilds of a dependent component between different machines (platforms) w/o making any changes.
For example, Wayland, Weston and all Qt5 modules are generic and should be re-used from shared state across all our Aarch64 platforms, but that's not the case and they are being rebuilt for each and every platform again and again.
I've done some experiments locally - back when SGX/Rogue UM libs provided libgles, libegl, libgbm (which API/ABI are supposed to be stable!), you certainly wanted to have them listed in this SIGGEN_EXCLUDERECIPES_ABISAFE list. Now, those ABIs have shifted to Mesa and we do mark that package as machine-specific (I had to add PACKAGE_ARCH = "${MACHINE_ARCH}" when redoing Reese's patches - see v8 changes section at [1]). With that in mind, this variable now has to list Mesa instead of individual UM packages - the patch is coming.
[1] https://patchwork.yoctoproject.org/project/ti/patch/[email protected]/
Signature · 1 line
Denys
Re: [meta-ti][master/kirkstone][PATCH v2 5/6] meta-ti-bsp: Remove SGX driver from SIGGEN_EXCLUDERECIPES_ABISAFE
Quoted 47 lines · Denys Dmytriyenko
Quoted 19 lines · Andrew Davis
Quoted 18 lines · Denys Dmytriyenko
Quoted 16 lines · Andrew Davis
Quoted 5 lines · Denys Dmytriyenko
Quoted 3 lines · Andrew Davis via lists.yoctoproject.org
Neither of recipes nor their ABI is all that stable. OpenGL might be slightly more stable, but that is not what these provide anymore. Remove these.
Quoted 7 lines
This is not what you think it is... :) Actually, Rogue recipes should be added to the list.
This is specific to signatures/shared state. The ABI in this case is libgles and libegl (which are stable) and tells bitbake to avoid rebuilding generic apps depending on these packages between platfoms in the same architecture.
These do not provide those API anymore, Mesa does. And nothing should depend on the KM package other than the UM libs.
Quoted 8 lines
Quoted 4 lines
Still doesn't matter - if you have Mesa depend on these, you still need them listed in here to avoid rebuilding Mesa from one platform to another. And everything else downstream.
We *want* Mesa to be rebuilt if these change, these do not provide a stable ABI anymore, it can and does change, sometimes between platforms (SGX KM -> UM is based on plat).
The name of the variable may be confusing. It's not about rebuilding Mesa once you make changes to the KM/UM pieces - that will happen automatically if the dependencies are tracked properly.
This variable is about re-use of the dependant generic packages. The issue it is trying to solve is when you have machine-specific packages early in the dependecy tree, everything down the dependency tree that depends on these packages will be treated as machine-specific, even if they are generic. Since bitbake will try to play safe here, you have to tell it otherwise.
Again, this is not about rebuilds of a dependent component between changes. This is about rebuilds of a dependent component between different machines (platforms) w/o making any changes.
For example, Wayland, Weston and all Qt5 modules are generic and should be re-used from shared state across all our Aarch64 platforms, but that's not the case and they are being rebuilt for each and every platform again and again.
Quoted 9 lines
I've done some experiments locally - back when SGX/Rogue UM libs provided libgles, libegl, libgbm (which API/ABI are supposed to be stable!), you certainly wanted to have them listed in this
SIGGEN_EXCLUDERECIPES_ABISAFElist. Now, those ABIs have shifted to Mesa and we do mark that package as machine-specific (I had to addPACKAGE_ARCH= "${MACHINE_ARCH}" when redoing Reese's patches - see v8 changes section at [1]). With that in mind, this variable now has to list Mesa instead of individual UM packages - the patch is coming.
So then we are saying the same thing, the UM libs should not be listed here. This patch should go in then, and another one that adds Mesa to the list can go in after.
If that is fine, then the first 5 patches in this series should be good to go.
Andrew
Re: [meta-ti][master/kirkstone][PATCH v2 5/6] meta-ti-bsp: Remove SGX driver from SIGGEN_EXCLUDERECIPES_ABISAFE
Quoted 59 lines · Andrew Davis
Quoted 48 lines · Denys Dmytriyenko
Quoted 24 lines · Andrew Davis
Quoted 15 lines · Denys Dmytriyenko
Quoted 14 lines · Andrew Davis
Quoted 12 lines · Denys Dmytriyenko
Quoted 3 lines · Andrew Davis via lists.yoctoproject.org
Neither of recipes nor their ABI is all that stable. OpenGL might be slightly more stable, but that is not what these provide anymore. Remove these.
This is not what you think it is... :) Actually, Rogue recipes should be added to the list.
This is specific to signatures/shared state. The ABI in this case is libgles and libegl (which are stable) and tells bitbake to avoid rebuilding generic apps depending on these packages between platfoms in the same architecture.
Quoted 7 lines
Quoted 2 lines
These do not provide those API anymore, Mesa does. And nothing should depend on the KM package other than the UM libs.
Still doesn't matter - if you have Mesa depend on these, you still need them listed in here to avoid rebuilding Mesa from one platform to another. And everything else downstream.
Quoted 4 lines
We *want* Mesa to be rebuilt if these change, these do not provide a stable ABI anymore, it can and does change, sometimes between platforms (SGX KM -> UM is
based on plat).The name of the variable may be confusing. It's not about rebuilding Mesa once you make changes to the KM/UM pieces - that will happen automatically if the dependencies are tracked properly.
This variable is about re-use of the dependant generic packages. The issue it is trying to solve is when you have machine-specific packages early in the dependecy tree, everything down the dependency tree that depends on these packages will be treated as machine-specific, even if they are generic. Since bitbake will try to play safe here, you have to tell it otherwise.
Again, this is not about rebuilds of a dependent component between changes. This is about rebuilds of a dependent component between different machines (platforms) w/o making any changes.
For example, Wayland, Weston and all Qt5 modules are generic and should be re-used from shared state across all our Aarch64 platforms, but that's not the case and they are being rebuilt for each and every platform again and again.
Quoted 9 lines
I've done some experiments locally - back when SGX/Rogue UM libs provided libgles, libegl, libgbm (which API/ABI are supposed to be stable!), you certainly wanted to have them listed in this
SIGGEN_EXCLUDERECIPES_ABISAFElist. Now, those ABIs have shifted to Mesa and we do mark that package as machine-specific (I had to addPACKAGE_ARCH= "${MACHINE_ARCH}" when redoing Reese's patches - see v8 changes section at [1]). With that in mind, this variable now has to list Mesa instead of individual UM packages - the patch is coming.
Quoted 5 lines
So then we are saying the same thing, the UM libs should not be listed here. This patch should go in then, and another one that adds Mesa to the list can go in after.
If that is fine, then the first 5 patches in this series should be good to go.
Denys, is there a follow on patch to this one that will correct your comments? Should I hold off on applying this patch until your patch is ready, or go ahead and apply it so that you have a basis for your patch?
Quoted 1 line
Andrew
Signature · 3 lines
Ryan Eatmon [email protected] ----------------------------------------- Texas Instruments, Inc. - LCPD - MGTS
Re: [meta-ti][master/kirkstone][PATCH v2 5/6] meta-ti-bsp: Remove SGX driver from SIGGEN_EXCLUDERECIPES_ABISAFE
Quoted 89 lines · Ryan Eatmon
Quoted 86 lines · Andrew Davis
Quoted 67 lines · Denys Dmytriyenko
Quoted 32 lines · Andrew Davis
Quoted 24 lines · Denys Dmytriyenko
Quoted 19 lines · Andrew Davis
Quoted 10 lines · Denys Dmytriyenko
Quoted 3 lines · Andrew Davis via lists.yoctoproject.org
Neither of recipes nor their ABI is all that stable. OpenGL might be slightly more stable, but that is not what these provide anymore. Remove these.
This is not what you think it is... :) Actually, Rogue recipes should be added
to the list.Quoted 7 lines
This is specific to signatures/shared state. The ABI in this case is libgles
and libegl (which are stable) and tells bitbake to avoid rebuilding generic
apps depending on these packages between platfoms in the same architecture.Quoted 3 lines
These do not provide those API anymore, Mesa does. And nothing should depend
on the KM package other than the UM libs.Quoted 6 lines
Still doesn't matter - if you have Mesa depend on these, you still need them
listed in here to avoid rebuilding Mesa from one platform to another. And
everything else downstream.Quoted 5 lines
We *want* Mesa to be rebuilt if these change, these do not provide a stable ABI
anymore, it can and does change, sometimes between platforms (SGX KM -> UM is
based on plat).The name of the variable may be confusing. It's not about rebuilding Mesa once
you make changes to the KM/UM pieces - that will happen automatically if the
dependencies are tracked properly.This variable is about re-use of the dependant generic packages. The issue it
is trying to solve is when you have machine-specific packages early in the
dependecy tree, everything down the dependency tree that depends on these packages will be treated as machine-specific, even if they are generic. Since
bitbake will try to play safe here, you have to tell it otherwise.Again, this is not about rebuilds of a dependent component between changes.
This is about rebuilds of a dependent component between different machines
(platforms) w/o making any changes.For example, Wayland, Weston and all Qt5 modules are generic and should be
re-used from shared state across all our Aarch64 platforms, but that's not the
case and they are being rebuilt for each and every platform again and again.Quoted 12 lines
I've done some experiments locally - back when SGX/Rogue UM libs provided libgles, libegl, libgbm (which API/ABI are supposed to be stable!), you certainly wanted to have them listed in this
SIGGEN_EXCLUDERECIPES_ABISAFE
list. Now, those ABIs have shifted to Mesa and we do mark that package as machine-specific (I had to addPACKAGE_ARCH= "${MACHINE_ARCH}" when redoing
Reese's patches - see v8 changes section at [1]). With that in mind, this variable now has to list Mesa instead of individual UM packages
- the patch
is coming.So then we are saying the same thing, the UM libs should not be listed here.
This patch should go in then, and another one that adds Mesa to the list can go in after.
Quoted 7 lines
Quoted 2 lines
If that is fine, then the first 5 patches in this series should be good to go.
Denys, is there a follow on patch to this one that will correct your comments? Should I hold off on applying this patch until your patch is ready, or go ahead and apply it so that you have a basis for your patch?
I can re-spin Andrew's patch with slight update in the commit message along with a new one to add Mesa.
Signature · 1 line
Denys
Re: [meta-ti][master/kirkstone][PATCH v2 5/6] meta-ti-bsp: Remove SGX driver from SIGGEN_EXCLUDERECIPES_ABISAFE
In reply to a message not in the archiveQuoted 93 lines · Denys Dmytriyenko
Quoted 92 lines · Ryan Eatmon
Quoted 88 lines · Andrew Davis
Quoted 67 lines · Denys Dmytriyenko
Quoted 32 lines · Andrew Davis
Quoted 24 lines · Denys Dmytriyenko
Quoted 19 lines · Andrew Davis
Quoted 10 lines · Denys Dmytriyenko
Quoted 3 lines · Andrew Davis via lists.yoctoproject.org
Neither of recipes nor their ABI is all that stable. OpenGL might be slightly more stable, but that is not what these provide anymore. Remove these.
This is not what you think it is... :) Actually, Rogue recipes should be added
to the list.Quoted 7 lines
This is specific to signatures/shared state. The ABI in this case is libgles
and libegl (which are stable) and tells bitbake to avoid rebuilding generic
apps depending on these packages between platfoms in the same architecture.Quoted 3 lines
These do not provide those API anymore, Mesa does. And nothing should depend
on the KM package other than the UM libs.Quoted 6 lines
Still doesn't matter - if you have Mesa depend on these, you still need them
listed in here to avoid rebuilding Mesa from one platform to another. And
everything else downstream.Quoted 5 lines
We *want* Mesa to be rebuilt if these change, these do not provide a stable ABI
anymore, it can and does change, sometimes between platforms (SGX KM -> UM is
based on plat).The name of the variable may be confusing. It's not about rebuilding Mesa once
you make changes to the KM/UM pieces - that will happen automatically if the
dependencies are tracked properly.This variable is about re-use of the dependant generic packages. The issue it
is trying to solve is when you have machine-specific packages early in the
dependecy tree, everything down the dependency tree that depends on these packages will be treated as machine-specific, even if they are generic. Since
bitbake will try to play safe here, you have to tell it otherwise.Again, this is not about rebuilds of a dependent component between changes.
This is about rebuilds of a dependent component between different machines
(platforms) w/o making any changes.For example, Wayland, Weston and all Qt5 modules are generic and should be
re-used from shared state across all our Aarch64 platforms, but that's not the
case and they are being rebuilt for each and every platform again and again.Quoted 12 lines
I've done some experiments locally - back when SGX/Rogue UM libs provided libgles, libegl, libgbm (which API/ABI are supposed to be stable!), you certainly wanted to have them listed in this
SIGGEN_EXCLUDERECIPES_ABISAFE
list. Now, those ABIs have shifted to Mesa and we do mark that package as machine-specific (I had to addPACKAGE_ARCH= "${MACHINE_ARCH}" when redoing
Reese's patches - see v8 changes section at [1]). With that in mind, this variable now has to list Mesa instead of individual UM packages
- the patch
is coming.So then we are saying the same thing, the UM libs should not be listed here.
This patch should go in then, and another one that adds Mesa to the list can go in after.If that is fine, then the first 5 patches in this series should be good to go.
Quoted 7 lines
Quoted 4 lines
Denys, is there a follow on patch to this one that will correct your comments? Should I hold off on applying this patch until your patch is ready, or go ahead and apply it so that you have a basis for your patch?
I can re-spin Andrew's patch with slight update in the commit message along with a new one to add Mesa.
Ok, v3 of this patch and a follow up patch have been submitted.
Signature · 1 line
Denys