| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
Refactor the kernel configuration fragment handling to shift all AGL
applied configuration fragments into a new AGL_KCONFIG_FRAGMENTS
variable that is used to generate SRC_URI and KERNEL_CONFIG_FRAGMENTS
additions for the various BSPs. The intent is to make it simple to
disable AGL provided configuration in downstream builds as the IC EG
has expressed as a requirement. Additionally, the rework has allowed
for some clean up of accumulated cruft.
In practice, clearing AGL_KCONFIG_FRAGMENTS drops all non-BSP provided
kernel configuration with the exception of some qemu BSP related
additions required for AGL CI and some explicitly configurable things
like netboot support.
Notable changes:
- Instead of always using AGL's own fragment merging logic on top of
the BSP kernel recipe, an effort is now made to leverage the BSP
recipes' own merging schemes, so there are now separate include
files for kernel-yocto.bbclass and plain kernel.bbclass based kernel
recipes, as well as a common include file that defines the
AGL_KCONFIG_FRAGMENTS variable and its derivations. That file can
be included directly in bbappends for BSP kernel recipes that use
the KERNEL_CONFIG_FRAGMENTS scheme (e.g. meta-ti, meta-qcom).
- The SMACK enabling configuration in meta-app-framework has been
updated to supply different fragments for enabling SMACK by default
for 4.x and 5.x kernels. This removes a warning from always
supplying the old configuration, and allows providing a CONFIG_LSM
definition to ensure over-riding any BSP modifications.
This allows removing the previous hack to handle CONFIG_LSM being
set in the defconfigs in linux-raspberrypi.
- By request, the linux-yocto support from meta-agl-bsp/meta-core has
been rationalized into meta-agl-core to improve the experience when
using meta-agl-core standalone for testing.
- All demo supporting kernel configuration has been removed, a
subsequent change to meta-agl-demo will add it there by leveraging
AGL_KCONFIG_FRAGMENTS.
- The hardware device support has been split out of the can-bus.cfg
fragment, in favor of shifting it to meta-agl-demo. A few other
stray non-CAN configuration options have also been removed from
can-bus.cfg, as they do not seem to be required.
Bug-AGL: SPEC-3983
Signed-off-by: Scott Murray <scott.murray@konsulko.com>
Change-Id: If6662fd36e26cec767b1d53b1188a74d01ef9dcf
Reviewed-on: https://gerrit.automotivelinux.org/gerrit/c/AGL/meta-agl/+/26460
Reviewed-by: Hiroyuki Ishii <ishii.hiroyuki002@jp.panasonic.com>
Reviewed-by: Jan-Simon Moeller <jsmoeller@linuxfoundation.org>
Tested-by: Jenkins Job builder account
ci-image-build: Jenkins Job builder account
ci-image-boot-test: Jenkins Job builder account
|
|
The meta-rcar layer activates by default the configuration MTD_RENESAS_RPC_HYPERFLASH in kernel.
MTD_RENESAS_RPC_HYPERFLASH is not upstream, it's a meta-rcar specific feature.
MTD_RENESAS_RPC_HYPERFLASH feature in the kernel need an firmware update on the board and
can be considered (IMHO) as a security issues in production.
We need to set DISABLE_RPC_ACCESS to deactivate it.
v2 (jsmoeller): Use ?= for DISABLE_RPC_ACCESS .
Bug-AGL: SPEC-3674
Signed-off-by: ronan <ronan.lemartret@iot.bzh>
Change-Id: Iefaf9ca115d1fb805b0d3766fc0c800b74daecee
Reviewed-on: https://gerrit.automotivelinux.org/gerrit/c/AGL/meta-agl/+/25815
Tested-by: Jenkins Job builder account <agl-jobbuilder@automotivelinux.org>
ci-image-build: Jenkins Job builder account <agl-jobbuilder@automotivelinux.org>
ci-image-boot-test: Jenkins Job builder account <agl-jobbuilder@automotivelinux.org>
Reviewed-by: Jan-Simon Moeller <jsmoeller@linuxfoundation.org>
|
|
* In RCAR BSP 4.14 the perf fix is upstream, we don't need to
keep this fix in AGL.
Bug-AGL: SPEC-3516
Signed-off-by: Ronan Le Martret <ronan.lemartret@iot.bzh>
Change-Id: Ic014347319318122f99eabf48170a5b393ab46c5
Reviewed-on: https://gerrit.automotivelinux.org/gerrit/c/AGL/meta-agl/+/25069
Reviewed-by: Jan-Simon Moeller <jsmoeller@linuxfoundation.org>
Tested-by: Jan-Simon Moeller <jsmoeller@linuxfoundation.org>
|
|
The local patch to fix the KingFisher HDMI connector ordering and a
configuration fragment to disable its onboard MOST support were
accidentally dropped in the previous change for the perf BFD patch
issue. Add them back to restore previous behavior.
Bug-AGL: SPEC-3443
Signed-off-by: Scott Murray <scott.murray@konsulko.com>
Change-Id: I527d073901bad4860475df81da36bbf60d8c5be5
Reviewed-on: https://gerrit.automotivelinux.org/gerrit/c/AGL/meta-agl/+/25011
Tested-by: Jenkins Job builder account <agl-jobbuilder@automotivelinux.org>
ci-image-build: Jenkins Job builder account <agl-jobbuilder@automotivelinux.org>
Reviewed-by: Jan-Simon Moeller <jsmoeller@linuxfoundation.org>
Reviewed-by: Stéphane Desneux <stephane.desneux@iot.bzh>
ci-image-boot-test: Jenkins Job builder account <agl-jobbuilder@automotivelinux.org>
|
|
Currently the upstream rcar bsp does need the fix for building perf.
The meta-rcar-gen3-adas layer contains a fix in the big patchqueue.
Thus conditionally apply the perf build fix.
Bug-AGL: SPEC-3443
Signed-off-by: Jan-Simon Möller <jsmoeller@linuxfoundation.org>
Change-Id: Icbf53add71ed731c35dec149841691c82ae4372c
Reviewed-on: https://gerrit.automotivelinux.org/gerrit/c/AGL/meta-agl/+/24914
Reviewed-by: Stéphane Desneux <stephane.desneux@iot.bzh>
|
|
Thanks to the work done by Renesas and members of our community, we can
now directly use meta-renesas from upstream github.
A shout-out to everyone involved (Kurokawa-san, Ronan, Stephane).
It turned-out it is simpler to create a symlink-based slim version
of meta-rcar-gen3-adas layer instead of BBMASK'ing what should
not be in there. See the comments in
meta-agl-bsp/meta-rcar-gen3-adas/conf/NOTES
as to why.
Remaining work is to include the packages outside of the layer.conf
in a follow-up.
Bug-AGL: SPEC-3380
Change-Id: I2be1b44049d74146546380e2eb866f0558a26822
Signed-off-by: Jan-Simon Moeller <jsmoeller@linuxfoundation.org>
Reviewed-on: https://gerrit.automotivelinux.org/gerrit/c/AGL/meta-agl/+/24796
Reviewed-by: Stéphane Desneux <stephane.desneux@iot.bzh>
|
|
- If we want to build AGL for ulcb board without
the meta-rcar (kingficher board support), we need to
split the patch.
Change-Id: I145887d4619a7132e46f98f30e1fc0dd3de6b29a
Signed-off-by: Ronan Le Martret <ronan.lemartret@iot.bzh>
|
|
Bug-AGL: SPEC-2486
Change-Id: Ic0cf9128d1630ede1af8024e3bd5258b7529b683
Signed-off-by: Harunobu Kurokawa <harunobu.kurokawa.dn@renesas.com>
|
|
Bug-AGL: SPEC-1945
Latest meta-rcar update SDIO frequency for TI's WLAN driver.
However, in M3 case, the this patch raise ti/wlcore driver WANING.
[ 107.009706] ------------[ cut here ]------------
[ 107.014342] WARNING: CPU: 0 PID: 4224 at /kernel-source//drivers/net/wireless/ti/wlcore/sdio.c:145 wl12xx_sdio_raw_write+0xb0/0x160 [wlcore_sdio]
[ 107.027375] Modules linked in: rfcomm 8021q garp stp mrp llc can_bcm can bnep nfc arc4 wl18xx wlcore mac80211 cfg80211 crc32_ce crct10dif_ce btwilink bluetooth ecdh_generic wlcore_sdio rcar_can st_drv rfkill can_dev vspm_if(O) vsp2(O) vspm(O) uvcs_drv(O) sllin(O) pvrsrvkm(O) mmngrbuf(O) mmngr(O) hdm_usb(O) hdm_i2c(O) hdm_dim2(O) aim_v4l2(O) aim_sound(O) aim_network(O) aim_cdev(O) mostcore(O) ipv6
[ 107.062537] CPU: 0 PID: 4224 Comm: wpa_supplicant Tainted: G O 4.14.35-yocto-standard #1
[ 107.071749] Hardware name: Renesas M3ULCB Kingfisher board based on r8a7796 (DT)
[ 107.079138] task: ffff8005fc610e00 task.stack: ffff000020bc8000
[ 107.085054] PC is at wl12xx_sdio_raw_write+0xb0/0x160 [wlcore_sdio]
[ 107.091316] LR is at wl12xx_sdio_raw_write+0x7c/0x160 [wlcore_sdio]
<snip>
[ 107.467808] ---[ end trace 0140d637a5afd50b ]---
[ 107.472613] wl1271_sdio mmc2:0001:2: sdio write failed (-84)
[ 107.478340] ------------[ cut here ]------------
[ 107.483008] WARNING: CPU: 0 PID: 4224 at /kernel-source//drivers/net/wireless/ti/wlcore/main.c:796 wl12xx_queue_recovery_work.part.26+0x58/0x60 [wlcore]
[ 107.496648] Modules linked in: rfcomm 8021q garp stp mrp llc can_bcm can bnep nfc arc4 wl18xx wlcore mac80211 cfg80211 crc32_ce crct10dif_ce btwilink bluetooth ecdh_generic wlcore_sdio rcar_can st_drv rfkill can_dev vspm_if(O) vsp2(O) vspm(O) uvcs_drv(O) sllin(O) pvrsrvkm(O) mmngrbuf(O) mmngr(O) hdm_usb(O) hdm_i2c(O) hdm_dim2(O) aim_v4l2(O) aim_sound(O) aim_network(O) aim_cdev(O) mostcore(O) ipv6
[ 107.531806] CPU: 0 PID: 4224 Comm: wpa_supplicant Tainted: G W O 4.14.35-yocto-standard #1
[ 107.541018] Hardware name: Renesas M3ULCB Kingfisher board based on r8a7796 (DT)
[ 107.548406] task: ffff8005fc610e00 task.stack: ffff000020bc8000
[ 107.554363] PC is at wl12xx_queue_recovery_work.part.26+0x58/0x60 [wlcore]
[ 107.561265] LR is at wl12xx_queue_recovery_work+0x1c/0x28 [wlcore]
Change-Id: I6f15251e6997f52892efd118e8941af5930c5ca9
Signed-off-by: Harunobu Kurokawa <harunobu.kurokawa.dn@renesas.com>
|
|
Kingfisher has MedlaLB Interface for most, kernel tree has
most driver as build-in driver. So these configs is enabled.
Howemver AGL use most driver in gerrit repository. So AGL
should be disable it.
Change-Id: Ia46500d58fc16f348a098c58f5e807cf37913e5e
Signed-off-by: Harunobu Kurokawa <harunobu.kurokawa.dn@renesas.com>
|
|
Change-Id: Id1b6b88e606d0a102eefbdb9c8deae2dfe0cf73b
Signed-off-by: Ronan Le Martret <ronan.lemartret@iot.bzh>
Signed-off-by: Harunobu Kurokawa <harunobu.kurokawa.dn@renesas.com>
|
Scott Murray
ronan
Jan-Simon Möller
Harunobu Kurokawa