Kernel config fragments are not applied
Hello
I am trying to add a custom kernel config to my layer. I followed the following guide:
http://www.yoctoproject.org/docs/1.6/dev-manual/dev-manual.html#creating-config-fragments
I have the following files in my layer:
recipes-kernel/linux/linux-cubox-i_3.0.35.bbappend:
FILESEXTRAPATHS_prepend := "${THISDIR}/${PN}-${PV}:"
inherit kernel
COMPATIBLE_MACHINE_cubox-i = "(cubox-i)"
SRC_URI += "file://videoin.cfg"
recipes-kernel/linux/linux-cubox-i-3.0.35/videoin.cfg:
CONFIG_VIDEOBUF_VMALLOC=yCONFIG_VIDEO_TVEEPROM=yCONFIG_VIDEO_TUNER=yCONFIG_VIDEO_ADV7180=yCONFIG_VIDEO_MXC_CAMERA=yCONFIG_MXC_TVIN_ADV7180=yCONFIG_USB_VIDEO_CLASS=yCONFIG_USB_GSPCA=yCONFIG_USB_M5602=yCONFIG_USB_STV06XX=y
...
The layer is included in conf/bblayers.conf. The file videoin.cfg is found and "installed" in the correct tmp/work directory. However the config is never added to the kernel config file.
What am I doing wrong?
I use Yocto "master-next", cubox-i4pro.
Thanks
Michael
Quoted 9 lines · Neuer User
Hello
I am trying to add a custom kernel config to my layer. I followed the following guide:
http://www.yoctoproject.org/docs/1.6/dev-manual/dev-manual.html#creating-config-fragmentsI have the following files in my layer:
recipes-kernel/linux/linux-cubox-i_3.0.35.bbappend:
The answer to your question depends on the base linux-cubox-i recipe.
Config fragment support works for kernels that inherit from linux-yocto, versus ones that inherit directly from kernel.bbclass. So if you pop open that base recipe, and look for a reference to linux-yocto.inc, you'll have the answer.
It is possible to bbappend the linux-yocto support into the kernel recipe via your layer, and might be worth a try.
The separation between the two classes exists to not break compatibility with existing kernel recipes, or change long standing oe workflow with the existing classes.
Cheers,
Bruce
Quoted 35 lines
FILESEXTRAPATHS_prepend:= "${THISDIR}/${PN}-${PV}:"inherit kernel
COMPATIBLE_MACHINE_cubox-i = "(cubox-i)"
SRC_URI+= "file://videoin.cfg"recipes-kernel/linux/linux-cubox-i-3.0.35/videoin.cfg:
CONFIG_VIDEOBUF_VMALLOC=yCONFIG_VIDEO_TVEEPROM=yCONFIG_VIDEO_TUNER=yCONFIG_VIDEO_ADV7180=yCONFIG_VIDEO_MXC_CAMERA=yCONFIG_MXC_TVIN_ADV7180=yCONFIG_USB_VIDEO_CLASS=yCONFIG_USB_GSPCA=yCONFIG_USB_M5602=yCONFIG_USB_STV06XX=y
...The layer is included in
conf/bblayers.conf. The file videoin.cfg is found and "installed" in the correct tmp/work directory. However the config is never added to the kernel config file.What am I doing wrong?
I use Yocto "master-next", cubox-i4pro.
Thanks
Michael
Am 28.04.2014 15:03, schrieb Bruce Ashfield:
Quoted 12 lines
The answer to your question depends on the base linux-cubox-i recipe.
Config fragment support works for kernels that inherit from linux-yocto, versus ones that inherit directly from kernel.bbclass. So if you pop open that base recipe, and look for a reference to linux-yocto.inc, you'll have the answer.
It is possible to bbappend the linux-yocto support into the kernel recipe via your layer, and might be worth a try.
Quoted 8 lines
The separation between the two classes exists to not break compatibility with existing kernel recipes, or change long standing oe workflow with the existing classes.
Cheers,
Bruce
Ahh, I see. That is probably the reason then. It includes from linux-imx:
include recipes-kernel/linux/linux-imx.inc
Now, should I just forget the fragment idea and use my own defconfig file? Or can I just include linux-yocto.inc in my bbappend?
Quoted 19 lines · Neuer User
Am 28.04.2014 15:03, schrieb Bruce Ashfield:
Quoted 18 lines
The answer to your question depends on the base linux-cubox-i recipe.
Config fragment support works for kernels that inherit from linux-yocto, versus ones that inherit directly from kernel.bbclass. So if you pop open that base recipe, and look for a reference to linux-yocto.inc, you'll have the answer.
It is possible to bbappend the linux-yocto support into the kernel recipe via your layer, and might be worth a try.
The separation between the two classes exists to not break compatibility with existing kernel recipes, or change long standing oe workflow with the existing classes.
Cheers,
Quoted 8 lines
Quoted 2 lines
Bruce
Ahh, I see. That is probably the reason then. It includes from linux-imx:
include recipes-kernel/linux/linux-imx.inc
Now, should I just forget the fragment idea and use my own defconfig file? Or can I just include linux-yocto.inc in my bbappend?
It all depends on how adventurous you are today :) Using a defconfig will work, and you can generate one without much trouble. Using fragments will allow you to keep your changes separate from the baseline configuration, but you'll need to tweak the recipe .. which always carries a risk of some unexpected side effects.
I'm around to help regardless of which way you chose.
Cheers,
Bruce
Quoted 1 line
Am 28.04.2014 15:16, schrieb Bruce Ashfield:
Quoted 6 lines
It all depends on how adventurous you are today :) Using a defconfig will work, and you can generate one without much trouble. Using fragments will allow you to keep your changes separate from the baseline configuration, but you'll need to tweak the recipe .. which always carries a risk of some unexpected side effects.
Quoted 6 lines
I'm around to help regardless of which way you chose.
Cheers,
Bruce
Well, I looked at the linux-yocto file and tried a couple of combinations in my bbappend. All of them geenrated lots of different errors.
e.g. I used:
FILESEXTRAPATHS_prepend := "${THISDIR}/${PN}-${PV}:"
COMPATIBLE_MACHINE_cubox-i = "(cubox-i)"
SRC_URI += "file://videoin.cfg"
addtask kernel_configme before do_configure after do_patch
# Pick up shared functions
inherit kernel
inherit kernel-yocto
#require linux-dtb.inc
# extra tasks addtask kernel_configcheck after do_configure before do_compile
The error messages are too long to include here :-/
Guess, it is probably best to go the defconfig way.
Quoted 9 lines · Neuer User
Am 28.04.2014 15:16, schrieb Bruce Ashfield:
Quoted 8 lines
It all depends on how adventurous you are today :) Using a defconfig will work, and you can generate one without much trouble. Using fragments will allow you to keep your changes separate from the baseline configuration, but you'll need to tweak the recipe .. which always carries a risk of some unexpected side effects.
I'm around to help regardless of which way you chose.
Quoted 7 lines
Quoted 4 lines
Cheers,
Bruce
Well, I looked at the linux-yocto file and tried a couple of combinations in my bbappend. All of them geenrated lots of different errors.
If you send me pointers to your bblayers and local.conf settings, I'll run a build to see what happens here as well.
I'm always interested in tracking these down regardless.
Bruce
Quoted 24 lines
e.g. I used:
FILESEXTRAPATHS_prepend:= "${THISDIR}/${PN}-${PV}:"COMPATIBLE_MACHINE_cubox-i = "(cubox-i)"
SRC_URI+= "file://videoin.cfg"addtask
kernel_configmebeforedo_configureafterdo_patch# Pick up shared functions
inherit kernel
inherit kernel-yocto
#require linux-dtb.inc# extra tasks addtask kernel_configcheck after do_configure before do_compileThe error messages are too long to include here :-/
Guess, it is probably best to go the defconfig way.
Am 28.04.2014 15:37, schrieb Bruce Ashfield:
Quoted 8 lines
If you send me pointers to your bblayers and local.conf settings, I'll run a build to see what happens here as well.
I'm always interested in tracking these down regardless.
Bruce
Testing this should be pretty straightforward, if see that correctly.
Just init the repo with yocto master-next. Then export MACHINE=cubox-i.
My layer has:
recipes-kernel/linux/linux-cubox-i_3.0.35.bbappend:
FILESEXTRAPATHS_prepend := "${THISDIR}/${PN}-${PV}:"
inherit kernel
COMPATIBLE_MACHINE_cubox-i = "(cubox-i)"
SRC_URI += "file://videoin.cfg"
And in recipes-kernel/linux/linux-cubox-i-3.0.35/videoin.cfg:
CONFIG_VIDEO_ADV7180=y
(for a test this should suffice)
Then after a "bitbake linux-cubix-i -c configure" you can grep ADV7180 in tmp/work/cubox_i-poky-linux-gnueabi/linux-cubox-i/3.0.35-r0/git/.config
Normally it is "=n", now it should be "=y".
Thanks
Michael
Quoted 10 lines · Neuer User
Am 28.04.2014 15:37, schrieb Bruce Ashfield:
Quoted 8 lines
If you send me pointers to your bblayers and local.conf settings, I'll run a build to see what happens here as well.
I'm always interested in tracking these down regardless.
Bruce
Quoted 7 lines
Testing this should be pretty straightforward, if see that correctly.
Just init the repo with yocto master-next. Then export MACHINE=cubox-i.
My layer has:
recipes-kernel/linux/linux-cubox-i_3.0.35.bbappend:
Can you send me your bblayers out of band ? I'm using meta-fsl-arm and meta-fsl-arm-extra to get a build going, and running into expansion errors in the kernel recipes. Rather than hunt around more than necessary, I'll check against your layers.
Bruce
Quoted 24 lines
FILESEXTRAPATHS_prepend:= "${THISDIR}/${PN}-${PV}:"inherit kernel
COMPATIBLE_MACHINE_cubox-i = "(cubox-i)"
SRC_URI+= "file://videoin.cfg"And in recipes-kernel/linux/linux-cubox-i-3.0.35/videoin.cfg:
CONFIG_VIDEO_ADV7180=y(for a test this should suffice)
Then after a "bitbake linux-cubix-i -c configure" you can grep ADV7180 in tmp/work/cubox_i-poky-linux-gnueabi/linux-cubox-i/3.0.35-r0/git/.config
Normally it is "=n", now it should be "=y".
Thanks
Michael
Am 30.04.2014 20:44, schrieb Bruce Ashfield:
Quoted 7 lines
Can you send me your bblayers out of band ? I'm using meta-fsl-arm and meta-fsl-arm-extra to get a build going, and running into expansion errors in the kernel recipes. Rather than hunt around more than necessary, I'll check against your layers.
Bruce
Sure:
LCONF_VERSION = "6"
BBPATH = "${TOPDIR}"
BSPDIR := "${@os.path.abspath(os.path.dirname(d.getVar('FILE', True)) + '/../..')}"
BBFILES ?= ""
BBLAYERS = " \
${BSPDIR}/sources/poky/meta \
${BSPDIR}/sources/poky/meta-yocto \
\
${BSPDIR}/sources/meta-openembedded/meta-oe \
${BSPDIR}/sources/meta-openembedded/meta-multimedia \
\
${BSPDIR}/sources/meta-fsl-arm \
${BSPDIR}/sources/meta-fsl-arm-extra \
${BSPDIR}/sources/meta-fsl-demos \
${BSPDIR}/sources/meta-openembedded/meta-ruby \
${BSPDIR}/sources/meta-qt5 \
${BSPDIR}/sources/meta-mylayer \
"
Not sure that you need all. meta-ruby and qt5 is needed for my application.
Quoted 30 lines · Neuer User
Am 30.04.2014 20:44, schrieb Bruce Ashfield:
Quoted 7 lines
Can you send me your bblayers out of band ? I'm using meta-fsl-arm and meta-fsl-arm-extra to get a build going, and running into expansion errors in the kernel recipes. Rather than hunt around more than necessary, I'll check against your layers.
Bruce
Sure:
LCONF_VERSION= "6"BBPATH = "${TOPDIR}"
BSPDIR := "${@os.path.abspath(os.path.dirname(d.getVar('FILE', True)) + '/../..')}"BBFILES ?= ""
BBLAYERS = " \ ${BSPDIR}/sources/poky/meta \ ${BSPDIR}/sources/poky/meta-yocto \ \ ${BSPDIR}/sources/meta-openembedded/meta-oe \ ${BSPDIR}/sources/meta-openembedded/meta-multimedia \ \ ${BSPDIR}/sources/meta-fsl-arm \ ${BSPDIR}/sources/meta-fsl-arm-extra \ ${BSPDIR}/sources/meta-fsl-demos \ ${BSPDIR}/sources/meta-openembedded/meta-ruby \ ${BSPDIR}/sources/meta-qt5 \ ${BSPDIR}/sources/meta-mylayer \
Quoted 1 line
"
Hrm. I have all the key layers right, which leaves me with the expansion issue. The README indicates the danny branch for the meta-fsl-* layers, is that what you are using ?
Bruce
Quoted 4 lines
Not sure that you need all. meta-ruby and qt5 is needed for my application.
Am 30.04.2014 21:21, schrieb Bruce Ashfield:
Quoted 7 lines
Hrm. I have all the key layers right, which leaves me with the expansion issue. The README indicates the danny branch for the meta-fsl-* layers, is that what you are using ?
Bruce
No, I'm using master-next. Only in master-next you can find the cubox-i machine definition.
Quoted 11 lines · Neuer User
Am 28.04.2014 15:37, schrieb Bruce Ashfield:
Quoted 8 lines
If you send me pointers to your bblayers and local.conf settings, I'll run a build to see what happens here as well.
I'm always interested in tracking these down regardless.
Bruce
Testing this should be pretty straightforward, if see that correctly.
I'm heading out, but wanted to send an update. I can confirm that there are some issues on master as it currently stands .. but the good news, is that in my queue of changes for Yocto 1.7, you can seamlessly add fragement support to that recipe and it works.
I'll be looking at this again later, and will send a branch reference for testing. I need to bisect for the particular change first, since the entire queue can't go into the tree yet (it breaks some use cases). But I will submit the ones that fix this case.
Cheers,
Bruce
Quoted 30 lines
Just init the repo with yocto master-next. Then export MACHINE=cubox-i.
My layer has:
recipes-kernel/linux/linux-cubox-i_3.0.35.bbappend:
FILESEXTRAPATHS_prepend:= "${THISDIR}/${PN}-${PV}:"inherit kernel
COMPATIBLE_MACHINE_cubox-i = "(cubox-i)"
SRC_URI+= "file://videoin.cfg"And in recipes-kernel/linux/linux-cubox-i-3.0.35/videoin.cfg:
CONFIG_VIDEO_ADV7180=y(for a test this should suffice)
Then after a "bitbake linux-cubix-i -c configure" you can grep ADV7180 in tmp/work/cubox_i-poky-linux-gnueabi/linux-cubox-i/3.0.35-r0/git/.config
Normally it is "=n", now it should be "=y".
Thanks
Michael
Quoted 18 lines · Bruce Ashfield
Quoted 11 lines · Neuer User
Am 28.04.2014 15:37, schrieb Bruce Ashfield:
Quoted 8 lines
If you send me pointers to your bblayers and local.conf settings, I'll run a build to see what happens here as well.
I'm always interested in tracking these down regardless.
Bruce
Testing this should be pretty straightforward, if see that correctly.
I'm heading out, but wanted to send an update. I can confirm that there are some issues on master as it currently stands .. but the good news, is that in my queue of changes for Yocto 1.7, you can seamlessly add fragement support to that recipe and it works.
Quoted 4 lines
I'll be looking at this again later, and will send a branch reference for testing. I need to bisect for the particular change first, since the entire queue can't go into the tree yet (it breaks some use cases). But I will submit the ones that fix this case.
I'm glad that I looked again, I though there was an error without my clean up series .. but what I was seeing was a legitimate double application of the patch.
What error where you seeing when you tried a bbappend like so:
------------FILESEXTRAPATHS_prepend := "${THISDIR}/${PN}-${PV}:"
inherit kernel
require recipes-kernel/linux/linux-yocto.inc
COMPATIBLE_MACHINE_cubox-i = "(cubox-i)"
SRC_URI += "file://videoin.cfg"
--------------
Your error message will set me straight .. and tell me if it is just the late night hacking fooling my eyes .. or not!
Bruce
Quoted 37 lines
Cheers,
Bruce
Quoted 31 lines
Just init the repo with yocto master-next. Then export MACHINE=cubox-i.
My layer has:
recipes-kernel/linux/linux-cubox-i_3.0.35.bbappend:
FILESEXTRAPATHS_prepend:= "${THISDIR}/${PN}-${PV}:"inherit kernel
COMPATIBLE_MACHINE_cubox-i = "(cubox-i)"
SRC_URI+= "file://videoin.cfg"And in recipes-kernel/linux/linux-cubox-i-3.0.35/videoin.cfg:
CONFIG_VIDEO_ADV7180=y(for a test this should suffice)
Then after a "bitbake linux-cubix-i -c configure" you can grep ADV7180 in tmp/work/cubox_i-poky-linux-gnueabi/linux-cubox-i/3.0.35-r0/git/.config
Normally it is "=n", now it should be "=y".
Thanks
Michael
Hi Bruce
Here is the error messega, I get after a "bitbake linux-cubox-i -c cleansstate; bitbake linux-cubox-i" using the recipe you posted:
ERROR: Function failed: do_patch (log file is located at /home/ubuntu/yocto/build/tmp/work/cubox_i-poky-linux-gnueabi/linux-cubox-i/3.0.35-r0/temp/log.do_patch.5660)
ERROR: Logfile of failure stored in:/home/ubuntu/yocto/build/tmp/work/cubox_i-poky-linux-gnueabi/linux-cubox-i/3.0.35-r0/temp/log.do_patch.5660 Log data follows:
| DEBUG: Executing shell function do_patch | WARNING: no meta data branch found ... | ls: cannot access .meta*: No such file or directory | Already on 'imx_3.0.35_4.1.0' | [INFO] validating against known patches (cubox-i-standard-meta) error: patch failed: arch/arm/lib/memset.S:19#############] (/)(110 %) | error: arch/arm/lib/memset.S: patch does not apply | ERROR. could not update git tree | ERROR. Could not apply patches for cubox-i. | Patch failures can be resolved in the devshell (bitbake -c
devshell linux-cubox-i)
| WARNING: exit code 1 from a shell command.
| ERROR: Function failed: do_patch (log file is located at /home/ubuntu/yocto/build/tmp/work/cubox_i-poky-linux-gnueabi/linux-cubox-i/3.0.35-r0/temp/log.do_patch.5660)
ERROR: Task 3 (/home/ubuntu/yocto/sources/meta-fsl-arm-extra/recipes-kernel/linux/linux-cubox-i_3.0.35.bb, do_patch) failed with exit code '1'
Hope, it helps!
Michael
Am 01.05.2014 06:29, schrieb Bruce Ashfield:
Quoted 22 lines
I'm glad that I looked again, I though there was an error without my clean up series .. but what I was seeing was a legitimate double application of the patch.
What error where you seeing when you tried a bbappend like so:
------------
FILESEXTRAPATHS_prepend:= "${THISDIR}/${PN}-${PV}:"inherit kernel
require recipes-kernel/linux/linux-yocto.incCOMPATIBLE_MACHINE_cubox-i = "(cubox-i)"
SRC_URI+= "file://videoin.cfg"
--------------Your error message will set me straight .. and tell me if it is just the late night hacking fooling my eyes .. or not!
Bruce
Quoted 4 lines · Neuer User
Hi Bruce
Here is the error messega, I get after a "bitbake linux-cubox-i -c cleansstate; bitbake linux-cubox-i" using the recipe you posted:
Aha. This is what I was seeing as well, and there's some issues and assumptions that are built into those kernel recipes that are causing the yocto tools to attempt a patch (which is already on the branch) and hence the failure (the recipes should really be tweaked .. but I digress).
I'm prepping a small patch that will have this particular recipe and use case working out of the box.
Bruce
Quoted 24 lines
ERROR: Function failed:
do_patch(log file is located at/home/ubuntu/yocto/build/tmp/work/cubox_i-poky-linux-gnueabi/linux-cubox-i/3.0.35-r0/temp/log.do_patch.5660)
ERROR: Logfile of failure stored in:/home/ubuntu/yocto/build/tmp/work/cubox_i-poky-linux-gnueabi/linux-cubox-i/3.0.35-r0/temp/log.do_patch.5660Log data follows:| DEBUG: Executing shell function do_patch | WARNING: no meta data branch found ... | ls: cannot access .meta*: No such file or directory | Already on 'imx_3.0.35_4.1.0' | [INFO] validating against known patches (cubox-i-standard-meta) error: patch failed: arch/arm/lib/memset.S:19#############] (/)(110 %) | error: arch/arm/lib/memset.S: patch does not apply | ERROR. could not update git tree | ERROR. Could not apply patches for cubox-i. | Patch failures can be resolved in the devshell (bitbake -cdevshell linux-cubox-i)
| WARNING: exit code 1 from a shell command.
| ERROR: Function failed:do_patch(log file is located at/home/ubuntu/yocto/build/tmp/work/cubox_i-poky-linux-gnueabi/linux-cubox-i/3.0.35-r0/temp/log.do_patch.5660)
ERROR: Task 3 (/home/ubuntu/yocto/sources/meta-fsl-arm-extra/recipes-kernel/linux/linux-cubox-i_3.0.35.bb,do_patch) failed with exit code '1'
Quoted 11 lines
Hope, it helps!
Michael
Am 01.05.2014 06:29, schrieb Bruce Ashfield:
Quoted 6 lines
I'm glad that I looked again, I though there was an error without my clean up series .. but what I was seeing was a legitimate double application of the patch.
What error where you seeing when you tried a bbappend like so:
ll ticking.. ;-)
If anyone has more suggestions or info, please let me know! I’ll keep you posted.
Update: Just reached 4 days of uptime!
Update 2: Today it’s 4 month’s after I wrote this blog. The machine is still up & running and has reached 106 days of uptime!
Update 3: After 7 months (222 days of uptime) I finally retired the machine since we migrated to our CloudStack cloud. The fix described ab
Quoted 18 lines
Quoted 16 lines
------------
FILESEXTRAPATHS_prepend:= "${THISDIR}/${PN}-${PV}:"inherit kernel
require recipes-kernel/linux/linux-yocto.incCOMPATIBLE_MACHINE_cubox-i = "(cubox-i)"
SRC_URI+= "file://videoin.cfg"
--------------Your error message will set me straight .. and tell me if it is just the late night hacking fooling my eyes .. or not!
Bruce
Quoted 6 lines · Bruce Ashfield
Quoted 4 lines · Neuer User
Hi Bruce
Here is the error messega, I get after a "bitbake linux-cubox-i -c cleansstate; bitbake linux-cubox-i" using the recipe you posted:
Quoted 8 lines
Aha. This is what I was seeing as well, and there's some issues and assumptions
that are built into those kernel recipes that are causing the yocto tools to attempt a patch (which is already on the branch) and hence the failure (the recipes should really be tweaked .. but I digress).I'm prepping a small patch that will have this particular recipe and use case working out of the box.
... and I managed to get this working without a patch. You just need the following as your bbappend:
----------------FILESEXTRAPATHS_prepend := "${THISDIR}/${PN}-${PV}:"
inherit kernel
require recipes-kernel/linux/linux-yocto.inc
COMPATIBLE_MACHINE_cubox-i = "(cubox-i)"
SRC_URI += "file://videoin.cfg"
SRCREV_machine = "${SRCREV}"
----------------
The key is the addition of SRCREV_machine. It deals with the issue that I mentioned with the kernel recipes. They have a series of patches that are already on the branch being built. But yet the SRCREV backs the branch off and allows the patches to apply (something I wouldn't do .. but that's not the point).
The linux-yocto validation can do a similar back off, but it currently triggers on SRCREV_machine, which wasn't set .. this is already fixed in my current series that will go out for Yocto 1.7.
After this:
]> pwd /home/bruce/poky/build/tmp/work/cubox_i-poky-linux-gnueabi/linux-cubox-i/3.0.35-r0 yow-bashfiel-d3 [/home/bruc.../3.0.35-r0]> grep CONFIG_VIDEO_ADV7180 linux-cubox_i-standard-build/.config CONFIG_VIDEO_ADV7180=y
Note, the fragment you were working with needs this:
Quoted 1 line
cat videoin.cfg
CONFIG_VIDEO_CAPTURE_DRIVERS=yCONFIG_VIDEO_ADV7180=y
Or you'll be missing a dependency and have the config dropped from the final .config.
Bruce
Quoted 72 lines
Bruce
Quoted 26 lines
ERROR: Function failed:
do_patch(log file is located at/home/ubuntu/yocto/build/tmp/work/cubox_i-poky-linux-gnueabi/linux-cubox-i/3.0.35-r0/temp/log.do_patch.5660)ERROR: Logfile of failure stored in:
/home/ubuntu/yocto/build/tmp/work/cubox_i-poky-linux-gnueabi/linux-cubox-i/3.0.35-r0/temp/log.do_patch.5660Log data follows:
| DEBUG: Executing shell function do_patch | WARNING: no meta data branch found ... | ls: cannot access .meta*: No such file or directory | Already on 'imx_3.0.35_4.1.0' | [INFO] validating against known patches (cubox-i-standard-meta) error: patch failed: arch/arm/lib/memset.S:19#############] (/)(110 %) | error: arch/arm/lib/memset.S: patch does not apply | ERROR. could not update git tree | ERROR. Could not apply patches for cubox-i. | Patch failures can be resolved in the devshell (bitbake -cdevshell linux-cubox-i)
| WARNING: exit code 1 from a shell command.
| ERROR: Function failed:do_patch(log file is located at/home/ubuntu/yocto/build/tmp/work/cubox_i-poky-linux-gnueabi/linux-cubox-i/3.0.35-r0/temp/log.do_patch.5660)ERROR: Task 3 (
/home/ubuntu/yocto/sources/meta-fsl-arm-extra/recipes-kernel/linux/linux-cubox-i_3.0.35.bb,Quoted 13 lines
do_patch) failed with exit code '1'Hope, it helps!
Michael
Am 01.05.2014 06:29, schrieb Bruce Ashfield:
Quoted 6 lines
I'm glad that I looked again, I though there was an error without my clean up series .. but what I was seeing was a legitimate double application of the patch.
What error where you seeing when you tried a bbappend like so:
ll ticking.. ;-)If anyone has more suggestions or info, please let me know! I’ll keep you posted.
Update: Just reached 4 days of uptime!
Update 2: Today it’s 4 month’s after I wrote this blog. The machine is still up & running and has reached 106 days of uptime!
Update 3: After 7 months (222 days of uptime) I finally retired the machine since we migrated to our CloudStack cloud. The fix described ab
Quoted 18 lines
Quoted 16 lines
------------
FILESEXTRAPATHS_prepend:= "${THISDIR}/${PN}-${PV}:"inherit kernel
require recipes-kernel/linux/linux-yocto.incCOMPATIBLE_MACHINE_cubox-i = "(cubox-i)"
SRC_URI+= "file://videoin.cfg"
--------------Your error message will set me straight .. and tell me if it is just the late night hacking fooling my eyes .. or not!
Bruce
That's great! And just one line to add. Perfect solution!
Thanks a lot!
Michael
Am 02.05.2014 19:24, schrieb Bruce Ashfield:
Quoted 62 lines · Bruce Ashfield
Quoted 6 lines · Neuer User
Quoted 4 lines
Hi Bruce
Here is the error messega, I get after a "bitbake linux-cubox-i -c cleansstate; bitbake linux-cubox-i" using the recipe you posted:
Quoted 8 lines
Aha. This is what I was seeing as well, and there's some issues and assumptions
that are built into those kernel recipes that are causing the yocto tools to attempt a patch (which is already on the branch) and hence the failure (the recipes should really be tweaked .. but I digress).I'm prepping a small patch that will have this particular recipe and use case working out of the box.
... and I managed to get this working without a patch. You just need the following as your bbappend:
----------------
FILESEXTRAPATHS_prepend:= "${THISDIR}/${PN}-${PV}:"inherit kernel
require recipes-kernel/linux/linux-yocto.incCOMPATIBLE_MACHINE_cubox-i = "(cubox-i)"
SRC_URI+= "file://videoin.cfg"
SRCREV_machine= "${SRCREV}"
----------------The key is the addition of
SRCREV_machine. It deals with the issue that I mentioned with the kernel recipes. They have a series of patches that are already on the branch being built. But yet the SRCREV backs the branch off and allows the patches to apply (something I wouldn't do .. but that's not the point).The linux-yocto validation can do a similar back off, but it currently triggers on
SRCREV_machine, which wasn't set .. this is already fixed in my current series that will go out for Yocto 1.7.After this:
]> pwd
/home/bruce/poky/build/tmp/work/cubox_i-poky-linux-gnueabi/linux-cubox-i/3.0.35-r0yow-bashfiel-d3 [
/home/bruc.../3.0.35-r0]> grepCONFIG_VIDEO_ADV7180linux-cubox_i-standard-build/.configCONFIG_VIDEO_ADV7180=yNote, the fragment you were working with needs this:
Quoted 1 line
cat videoin.cfg
CONFIG_VIDEO_CAPTURE_DRIVERS=yCONFIG_VIDEO_ADV7180=yOr you'll be missing a dependency and have the config dropped from the final .config.
Bruce
Hi Bruce
Have you got any further?
I currently use a "defconfig" replacement, but honestly, it feels very dumb. Fragments seem to be much smarter to combine and test different configurations.
Cheers
Michael
Am 28.04.2014 15:37, schrieb Bruce Ashfield:
Quoted 51 lines · Neuer User
Quoted 9 lines
Am 28.04.2014 15:16, schrieb Bruce Ashfield:
Quoted 8 lines
It all depends on how adventurous you are today :) Using a defconfig will work, and you can generate one without much trouble. Using fragments will allow you to keep your changes separate from the baseline configuration, but you'll need to tweak the recipe .. which always carries a risk of some unexpected side effects.
I'm around to help regardless of which way you chose.
Quoted 8 lines
Quoted 4 lines
Cheers,
Bruce
Well, I looked at the linux-yocto file and tried a couple of combinations in my bbappend. All of them geenrated lots of different errors.
If you send me pointers to your bblayers and local.conf settings, I'll run a build to see what happens here as well.
I'm always interested in tracking these down regardless.
Bruce
Quoted 24 lines
e.g. I used:
FILESEXTRAPATHS_prepend:= "${THISDIR}/${PN}-${PV}:"COMPATIBLE_MACHINE_cubox-i = "(cubox-i)"
SRC_URI+= "file://videoin.cfg"addtask
kernel_configmebeforedo_configureafterdo_patch# Pick up shared functions
inherit kernel
inherit kernel-yocto
#require linux-dtb.inc# extra tasks addtask kernel_configcheck after do_configure before do_compileThe error messages are too long to include here :-/
Guess, it is probably best to go the defconfig way.
Quoted 3 lines · Neuer User
Hi Bruce
Have you got any further?
Your timing is good. I was preeempted by another issue yesterday, but I have the fsl-imx layers cloned and am about to start a build. I'll send a follow up when I know more.
Bruce
Quoted 63 lines
I currently use a "defconfig" replacement, but honestly, it feels very dumb. Fragments seem to be much smarter to combine and test different configurations.
Cheers
Michael
Am 28.04.2014 15:37, schrieb Bruce Ashfield:
Quoted 51 lines · Neuer User
Quoted 9 lines
Am 28.04.2014 15:16, schrieb Bruce Ashfield:
Quoted 8 lines
It all depends on how adventurous you are today :) Using a defconfig will work, and you can generate one without much trouble. Using fragments will allow you to keep your changes separate from the baseline configuration, but you'll need to tweak the recipe .. which always carries a risk of some unexpected side effects.
I'm around to help regardless of which way you chose.
Quoted 8 lines
Quoted 4 lines
Cheers,
Bruce
Well, I looked at the linux-yocto file and tried a couple of combinations in my bbappend. All of them geenrated lots of different errors.
If you send me pointers to your bblayers and local.conf settings, I'll run a build to see what happens here as well.
I'm always interested in tracking these down regardless.
Bruce
Quoted 24 lines
e.g. I used:
FILESEXTRAPATHS_prepend:= "${THISDIR}/${PN}-${PV}:"COMPATIBLE_MACHINE_cubox-i = "(cubox-i)"
SRC_URI+= "file://videoin.cfg"addtask
kernel_configmebeforedo_configureafterdo_patch# Pick up shared functions
inherit kernel
inherit kernel-yocto
#require linux-dtb.inc# extra tasks addtask kernel_configcheck after do_configure before do_compileThe error messages are too long to include here :-/
Guess, it is probably best to go the defconfig way.