The last 2 things to do are to fix the build system so that when you build concurrently the modules get into the system.tar.bz2.
Things should get installed to:
out/target/product/pandaboard/system/modules
...and they do when you build without -j
The file:
device/linaro/common/tasks/kernel.mk
Contains the recipe to build the modules and copy them over. The problem is that the kernel may get built before or after the system.tar.bz2 gets produced. The system.tar.bz2 gets built via a recipe:
core/Makefile:786:system_tar := $(PRODUCT_OUT)/system.tar core/Makefile:787:INSTALLED_SYSTEMTARBALL_TARGET := $(system_tar).$(SYSTEM_TARBALL_FORMAT) core/Makefile:788:$(INSTALLED_SYSTEMTARBALL_TARGET): PRIVATE_SYSTEM_TAR := $(system_tar) core/Makefile-789-$(INSTALLED_SYSTEMTARBALL_TARGET): $(FS_GET_STATS) $(INTERNAL_SYSTEMIMAGE_FILES) core/Makefile-790- $(build-systemtarball-target) core/Makefile-791- core/Makefile-792-.PHONY: systemtarball-nodeps core/Makefile-793-systemtarball-nodeps: $(FS_GET_STATS) \ core/Makefile-794- $(filter-out systemtarball-nodeps stnod,$(MAKECMDGOALS)) core/Makefile-795- $(build-systemtarball-target)
I had tried a few things, but I think if:
core/Makefile-793-systemtarball-nodeps: $(FS_GET_STATS) \ core/Makefile-794- $(filter-out systemtarball-nodeps stnod,$(MAKECMDGOALS))
...is updated to depend on a kernel module rule, possibly broken out from device/linaro/common/tasks/kernel.mk things will probably work okay.
Once this is done the insmod commands that are already a part of:
out/target/product/pandaboard/root/init.omap4430.rc
namely:
insmod /system/modules/st_drv.ko insmod /system/modules/bluetooth.ko insmod /system/modules/bnep.ko insmod /system/modules/hci_uart.ko insmod /system/modules/rfcomm.ko insmod /system/modules/btwilink.ko
insmod /system/modules/wl12xx.ko insmod /system/modules/wl12xx_sdio.ko
will work against the already being installed binaries at:
out/target/product/pandaboard/system/lib/firmware/ti-connectivity/wl128x-fw-ap.bin out/target/product/pandaboard/system/lib/firmware/ti-connectivity/wl1271-fw.bin out/target/product/pandaboard/system/lib/firmware/ti-connectivity/wl128x-fw.bin out/target/product/pandaboard/system/lib/firmware/ti-connectivity/wl1271-nvs.bin out/target/product/pandaboard/system/lib/firmware/ti-connectivity/wl1271-fw-2.bin out/target/product/pandaboard/system/lib/firmware/ti-connectivity/wl1271-nvs-example-wl128x.bin out/target/product/pandaboard/system/lib/firmware/ti-connectivity/wl1271-fw-ap.bin
and
ls out/target/product/pandaboard/system/lib/firmware ti-connectivity TIInit_7.2.31.bts
which does get installed correctly.
The second part is to get the BT daemon pulled into Android. Dave this may be a good job for you to get up to speed on Android.
The BPs are here:
https://blueprints.launchpad.net/linaro-android/+spec/linaro-android-panda-w... https://blueprints.launchpad.net/linaro-android/+spec/linaro-android-panda-b...
Vishal, please update these.
FYI BPs are all filed on a monthly basis. Each BP should track the production of tangible result every month. If a BP isn't finished and there's nothing tangible to show for it, it gets moved (like these) to the next month. If there is something tangible, then it gets split.
Each BP has 3 required sections:
Headline: Something punchy about what this BP is going to produce like, "Panda-LEB is now shipping with a 3.0 kernel"
Acceptance: What's considered done.
Work items: A list of items with TODO, INPROGRESS and DONE.
You can also put a notes section together
Notes:
Feel free to look through:
https://launchpad.net/linaro-android/+milestone/11.08 https://launchpad.net/linaro-android/+milestone/11.07 https://launchpad.net/linaro-android/+milestone/11.06
for examples.
Welcome to the Panda-LEB PoC role!
-Zach
Andy Doan had the observation:
I hit a problem similar to this last year and discovered that something is a bit broken in make 3.81 related to this. We reverted to make 3.80 and things started working as we expected.
so that may be something. I certainly ran into something when I tried to get the dependencies worked out.
On 29 August 2011 13:57, Zach Pfeffer zach.pfeffer@linaro.org wrote:
The last 2 things to do are to fix the build system so that when you build concurrently the modules get into the system.tar.bz2.
Things should get installed to:
out/target/product/pandaboard/system/modules
...and they do when you build without -j
The file:
device/linaro/common/tasks/kernel.mk
Contains the recipe to build the modules and copy them over. The problem is that the kernel may get built before or after the system.tar.bz2 gets produced. The system.tar.bz2 gets built via a recipe:
core/Makefile:786:system_tar := $(PRODUCT_OUT)/system.tar core/Makefile:787:INSTALLED_SYSTEMTARBALL_TARGET := $(system_tar).$(SYSTEM_TARBALL_FORMAT) core/Makefile:788:$(INSTALLED_SYSTEMTARBALL_TARGET): PRIVATE_SYSTEM_TAR := $(system_tar) core/Makefile-789-$(INSTALLED_SYSTEMTARBALL_TARGET): $(FS_GET_STATS) $(INTERNAL_SYSTEMIMAGE_FILES) core/Makefile-790- $(build-systemtarball-target) core/Makefile-791- core/Makefile-792-.PHONY: systemtarball-nodeps core/Makefile-793-systemtarball-nodeps: $(FS_GET_STATS) \ core/Makefile-794- $(filter-out systemtarball-nodeps stnod,$(MAKECMDGOALS)) core/Makefile-795- $(build-systemtarball-target)
I had tried a few things, but I think if:
core/Makefile-793-systemtarball-nodeps: $(FS_GET_STATS) \ core/Makefile-794- $(filter-out systemtarball-nodeps stnod,$(MAKECMDGOALS))
...is updated to depend on a kernel module rule, possibly broken out from device/linaro/common/tasks/kernel.mk things will probably work okay.
Once this is done the insmod commands that are already a part of:
out/target/product/pandaboard/root/init.omap4430.rc
namely:
insmod /system/modules/st_drv.ko insmod /system/modules/bluetooth.ko insmod /system/modules/bnep.ko insmod /system/modules/hci_uart.ko insmod /system/modules/rfcomm.ko insmod /system/modules/btwilink.ko
insmod /system/modules/wl12xx.ko insmod /system/modules/wl12xx_sdio.ko
will work against the already being installed binaries at:
out/target/product/pandaboard/system/lib/firmware/ti-connectivity/wl128x-fw-ap.bin out/target/product/pandaboard/system/lib/firmware/ti-connectivity/wl1271-fw.bin out/target/product/pandaboard/system/lib/firmware/ti-connectivity/wl128x-fw.bin out/target/product/pandaboard/system/lib/firmware/ti-connectivity/wl1271-nvs.bin out/target/product/pandaboard/system/lib/firmware/ti-connectivity/wl1271-fw-2.bin out/target/product/pandaboard/system/lib/firmware/ti-connectivity/wl1271-nvs-example-wl128x.bin out/target/product/pandaboard/system/lib/firmware/ti-connectivity/wl1271-fw-ap.bin
and
ls out/target/product/pandaboard/system/lib/firmware ti-connectivity TIInit_7.2.31.bts
which does get installed correctly.
The second part is to get the BT daemon pulled into Android. Dave this may be a good job for you to get up to speed on Android.
The BPs are here:
https://blueprints.launchpad.net/linaro-android/+spec/linaro-android-panda-w... https://blueprints.launchpad.net/linaro-android/+spec/linaro-android-panda-b...
Vishal, please update these.
FYI BPs are all filed on a monthly basis. Each BP should track the production of tangible result every month. If a BP isn't finished and there's nothing tangible to show for it, it gets moved (like these) to the next month. If there is something tangible, then it gets split.
Each BP has 3 required sections:
Headline: Something punchy about what this BP is going to produce like, "Panda-LEB is now shipping with a 3.0 kernel"
Acceptance: What's considered done.
Work items: A list of items with TODO, INPROGRESS and DONE.
You can also put a notes section together
Notes:
Feel free to look through:
https://launchpad.net/linaro-android/+milestone/11.08 https://launchpad.net/linaro-android/+milestone/11.07 https://launchpad.net/linaro-android/+milestone/11.06
for examples.
Welcome to the Panda-LEB PoC role!
-Zach
On 08/30/2011 03:24 AM, Somebody in the thread at some point said:
Andy Doan had the observation:
I hit a problem similar to this last year and discovered that something is a bit broken in make 3.81 related to this. We reverted to make 3.80 and things started working as we expected.
I found a similar issue in kernel with Fedora Rawhide make upgrading to 3.81, they tightened their syntax checking in the makefile to error out now on illegal stuff they used to wink at. The errors reflected actual errors in makefile syntax.
The second part is to get the BT daemon pulled into Android. Dave this may be a good job for you to get up to speed on Android.
Actually Dave has other important fish to fry at the moment, thanks.
-Andy
On 29 August 2011 21:14, Andy Green andy.green@linaro.org wrote:
On 08/30/2011 03:24 AM, Somebody in the thread at some point said:
Andy Doan had the observation:
I hit a problem similar to this last year and discovered that something is a bit broken in make 3.81 related to this. We reverted to make 3.80 and things started working as we expected.
I found a similar issue in kernel with Fedora Rawhide make upgrading to 3.81, they tightened their syntax checking in the makefile to error out now on illegal stuff they used to wink at. The errors reflected actual errors in makefile syntax.
The second part is to get the BT daemon pulled into Android. Dave this may be a good job for you to get up to speed on Android.
Actually Dave has other important fish to fry at the moment, thanks.
What's his availability looking like? Will he be able to lend a hand soon?
-Andy
-- Andy Green | TI Landing Team Leader Linaro.org │ Open source software for ARM SoCs | Follow Linaro http://facebook.com/pages/Linaro/155974581091106 - http://twitter.com/#%21/linaroorg - http://linaro.org/linaro-blog
On 08/31/2011 12:58 AM, Somebody in the thread at some point said:
On 29 August 2011 21:14, Andy Greenandy.green@linaro.org wrote:
On 08/30/2011 03:24 AM, Somebody in the thread at some point said:
Andy Doan had the observation:
I hit a problem similar to this last year and discovered that something is a bit broken in make 3.81 related to this. We reverted to make 3.80 and things started working as we expected.
I found a similar issue in kernel with Fedora Rawhide make upgrading to 3.81, they tightened their syntax checking in the makefile to error out now on illegal stuff they used to wink at. The errors reflected actual errors in makefile syntax.
The second part is to get the BT daemon pulled into Android. Dave this may be a good job for you to get up to speed on Android.
Actually Dave has other important fish to fry at the moment, thanks.
What's his availability looking like? Will he be able to lend a hand soon?
I don't think so... sorry if you got the impression that he's some assignee to Android team because I asked him to get a gerrit account, it really is not the case.
I'm still, after some months now, looking for the experts in the Android team to implement glue to get stuff working in our kernel functional on Android.
-Andy
On 30 August 2011 13:56, Andy Green andy.green@linaro.org wrote:
On 08/31/2011 12:58 AM, Somebody in the thread at some point said:
On 29 August 2011 21:14, Andy Greenandy.green@linaro.org wrote:
On 08/30/2011 03:24 AM, Somebody in the thread at some point said:
Andy Doan had the observation:
I hit a problem similar to this last year and discovered that something is a bit broken in make 3.81 related to this. We reverted to make 3.80 and things started working as we expected.
I found a similar issue in kernel with Fedora Rawhide make upgrading to 3.81, they tightened their syntax checking in the makefile to error out now on illegal stuff they used to wink at. The errors reflected actual errors in makefile syntax.
The second part is to get the BT daemon pulled into Android. Dave this may be a good job for you to get up to speed on Android.
Actually Dave has other important fish to fry at the moment, thanks.
What's his availability looking like? Will he be able to lend a hand soon?
I don't think so... sorry if you got the impression that he's some assignee to Android team because I asked him to get a gerrit account, it really is not the case.
I'm still, after some months now, looking for the experts in the Android team to implement glue to get stuff working in our kernel functional on Android.
It should be better now. With Vishal Bhoj coming online he'll be able to focus on TI issues. I'm assuming Dave will be is PoC? Its extremely important that TI has an Android PoC that is actively working with Android builds and can work with Gerrit so that we can land features quickly.
-Andy
-- Andy Green | TI Landing Team Leader Linaro.org │ Open source software for ARM SoCs | Follow Linaro http://facebook.com/pages/Linaro/155974581091106 - http://twitter.com/#%21/linaroorg - http://linaro.org/linaro-blog
On 08/31/2011 05:22 AM, Somebody in the thread at some point said:
What's his availability looking like? Will he be able to lend a hand soon?
I don't think so... sorry if you got the impression that he's some assignee to Android team because I asked him to get a gerrit account, it really is not the case.
I'm still, after some months now, looking for the experts in the Android team to implement glue to get stuff working in our kernel functional on Android.
It should be better now. With Vishal Bhoj coming online he'll be able to focus on TI issues. I'm assuming Dave will be is PoC? Its extremely important that TI has an Android PoC that is actively working with Android builds and can work with Gerrit so that we can land features quickly.
Well, I spent several man-days on these Android hacking sessions personally - at first got the 3D driver going which was great - but latterly didn't bear any other fruit.
IMO the issue is not a lack of POC since you have had me for a four hour block per week for however long that's been going, but your side actually doing the integration work. Unless something's busted on kernel side or needs adaptation, in which case someone will help since we also would like to twiddle Android UI menus and see WLAN coming etc, you probably don't even need us around to do it since I already explained the firmware, daemon and module requirements for these assets.
It's nice Bohj is looking after it now but do we actually know how to do, eg, WLAN / BT / HDMI Audio / Onboard audio integration to Android, between us? I haven't seen it done for any board yet and it hasn't shown signs of advancing until now. If we are unsure how to do these somewhat specialized actions, do we have a plan to find out how?
-Andy
-
Andy Green -
Zach Pfeffer