[yocto] where to put IMAGE_INSTALL??
There seems to be some sensitivity to where I put the statement IMAGE_INSTALL += web-webkit. At first I just modifed the meta/receipes-sato/image/core-image-sato.bb file to include it. That caused the webkit to be built. Since that is not an ideal location, I removed it and put it into the local.conf file. That also caused the webkit to be built, but I wanted it to be part of my BSP permanently. I thought about creating a core-image-sato.bbappend file with the IMAGE_INSTALL += web-webkit in it and put that into the BSP, but that didn't cause the webkit to be built. Under the meta-intel/meta-n450 directory, I created a receipes-jfa directory and put the core-image-sato.bbappend file there. I must be missing a key rule about directory parsing and .bbappend files. Jim A ___ yocto mailing list yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/yocto
Re: [yocto] where to put IMAGE_INSTALL??
On 2012-01-12 05:53, Jim Abernathy wrote: There seems to be some sensitivity to where I put the statement IMAGE_INSTALL += web-webkit. At first I just modifed the meta/receipes-sato/image/core-image-sato.bb file to include it. That caused the webkit to be built. Since that is not an ideal location, I removed it and put it into the local.conf file. That also caused the webkit to be built, but I wanted it to be part of my BSP permanently. I thought about creating a core-image-sato.bbappend file with the IMAGE_INSTALL += web-webkit in it and put that into the BSP, but that didn't cause the webkit to be built. Under the meta-intel/meta-n450 directory, I created a receipes-jfa directory and put the core-image-sato.bbappend file there. I must be missing a key rule about directory parsing and .bbappend files. Look at the how the layers are parsed - this info is in layer/conf/layer.conf Try putting your .bbappend file in .../meta-intel/meta-n450/recipes-jfa/images/core-image-sato.bbappend -- Gary Thomas | Consulting for the MLB Associates |Embedded world ___ yocto mailing list yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/yocto
Re: [yocto] Request to enable SMP and virtio for qemux86/x86-64
Bruce, Thanks for your efforts! Edwin On 2012/1/11 20:43, Bruce Ashfield wrote: On 12-01-11 02:01 AM, Zhai, Edwin wrote: Bruce, Can we enable SMP and virtio by default for qemux86/x86-64? This can achieve huge perf boost for workload inside qemu. E.g. we enabled self-hosted image, where we build yocto inside qemu. Attached patch showes the kernel config option. Is it reasonable? It should be. I just need to look at the fallback modes, and how this impacts different host systems. That being said, I've run with similar configs in the past (depending on the workload) with no issues. qemux8-64 is already SMP, so it would just need the addition of virtio devices, which just means we'd look at this as any BSP/target config update. But with graceful degradation (i.e maxcpus with SMP set) and agreement that we'd like to simulate SMP by default, then we can make the change and I'll merge it into the base config for the target and pull it into the kernel tree. Cheers, Bruce Thanks, Edwin ___ yocto mailing list yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/yocto
Re: [yocto] where to put IMAGE_INSTALL??
On 01/12/2012 08:06 AM, Gary Thomas wrote: On 2012-01-12 05:53, Jim Abernathy wrote: There seems to be some sensitivity to where I put the statement IMAGE_INSTALL += web-webkit. At first I just modifed the meta/receipes-sato/image/core-image-sato.bb file to include it. That caused the webkit to be built. Since that is not an ideal location, I removed it and put it into the local.conf file. That also caused the webkit to be built, but I wanted it to be part of my BSP permanently. I thought about creating a core-image-sato.bbappend file with the IMAGE_INSTALL += web-webkit in it and put that into the BSP, but that didn't cause the webkit to be built. Under the meta-intel/meta-n450 directory, I created a receipes-jfa directory and put the core-image-sato.bbappend file there. I must be missing a key rule about directory parsing and .bbappend files. Look at the how the layers are parsed - this info is in layer/conf/layer.conf Try putting your .bbappend file in .../meta-intel/meta-n450/recipes-jfa/images/core-image-sato.bbappend I think you are correct now that I look more closely at the meta-n450/conf/layer.conf file. It has: BBFILES := ${BBFILES} ${LAYERDIR}/recipes-*/*/*.bb \ ${LAYERDIR}/recipes-*/*/*.bbappend which tells me that it will not find recipes-jfa/core-image-sato.bbappend, but will find meta-n450/recipes-jfa/images/core-image-sato.bbappend Thanks, JIm A ___ yocto mailing list yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/yocto
Re: [yocto] Does Edison work with Beagleboard linux-yocto-3.0 kernel?
Yet another follow up. I finally found my C3 Beagleboard and the default kernel built off master yesterday works on that platform too. I was able to do another build with the tips you guys gave and it looks like it is picking up the kernel from my local git repo now. I did the calibrate example and while I couldn't see the printk's due to the silent boot being turned on somehow, uname -a was different than the default image mine is now: root@beagleboard:~# uname -a Linux beagleboard 3.0.14-yocto-standard+ #1 PREEMPT Thu Jan 12 08:45:10 EST 2012 armv7l GNU/Linux And just for sanity, I made another copy of my bare clone and verified that I pushed the calibrate changes correctly. So it looks like I'm good now thanks to your help! Do I have to do a cleanall every time I'm finished pushing changes back to my local kernel repo? Is there a document that gives clues as to how to setup a local u-boot repo for making changes to it? Is is simply changing the u-boot recipe SRC_URI to use my local u-boot git repo in poky/meda/recipes-bsp/u-boot or is it more involved than that? Is there a u-boot dev layer like the poky-extras/meta-kernel-dev? Regards, Brian ___ yocto mailing list yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/yocto
Re: [yocto] Does Edison work with Beagleboard linux-yocto-3.0 kernel?
On 12/01/12 15:21, Brian Hutchinson wrote: Yet another follow up. I finally found my C3 Beagleboard and the default kernel built off master yesterday works on that platform too. I was able to do another build with the tips you guys gave and it looks like it is picking up the kernel from my local git repo now. I did the calibrate example and while I couldn't see the printk's due to the silent boot being turned on somehow, uname -a was different than the default image mine is now: root@beagleboard:~# uname -a Linux beagleboard 3.0.14-yocto-standard+ #1 PREEMPT Thu Jan 12 08:45:10 EST 2012 armv7l GNU/Linux And just for sanity, I made another copy of my bare clone and verified that I pushed the calibrate changes correctly. So it looks like I'm good now thanks to your help! Do I have to do a cleanall every time I'm finished pushing changes back to my local kernel repo? Is there a document that gives clues as to how to setup a local u-boot repo for making changes to it? Is is simply changing the u-boot recipe SRC_URI to use my local u-boot git repo in poky/meda/recipes-bsp/u-boot or is it more involved than that? Is there a u-boot dev layer like the poky-extras/meta-kernel-dev? Regards, Brian ___ yocto mailing list yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/yocto Brian, What image are you building, as I cannot get core-image-minimal to boot at all on my xM. I am currently trying to fix a hosed sd card (who knows what happened to it!) and then I will see if it has made any difference. Cheers, Jack. ___ yocto mailing list yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/yocto
Re: [yocto] Does Edison work with Beagleboard linux-yocto-3.0 kernel?
On 12-01-12 10:21 AM, Brian Hutchinson wrote: Yet another follow up. I finally found my C3 Beagleboard and the default kernel built off master yesterday works on that platform too. I was able to do another build with the tips you guys gave and it looks like it is picking up the kernel from my local git repo now. I did the calibrate example and while I couldn't see the printk's due to the silent boot being turned on somehow, uname -a was different than the default image mine is now: root@beagleboard:~# uname -a Linux beagleboard 3.0.14-yocto-standard+ #1 PREEMPT Thu Jan 12 08:45:10 EST 2012 armv7l GNU/Linux And just for sanity, I made another copy of my bare clone and verified that I pushed the calibrate changes correctly. So it looks like I'm good now thanks to your help! Do I have to do a cleanall every time I'm finished pushing changes back to my local kernel repo? The bitbake AUTOREV code should take care of updating the clone of your local repo in downloads/git2. I take it that this isn't happening ? Is there a document that gives clues as to how to setup a local u-boot repo for making changes to it? Is is simply changing the u-boot recipe SRC_URI to use my local u-boot git repo in poky/meda/recipes-bsp/u-boot or is it more involved than that? Is there a u-boot dev layer like the poky-extras/meta-kernel-dev? It should be (largely) as simple as that. We could create something simple and throw it in with the meta-kernel-dev layer if there's any interest in adding it. Cheers, Bruce Regards, Brian ___ yocto mailing list yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/yocto
[yocto] qemu EFI build failure
I tried to build a qemux86 EFI image, by setting: - in my local.conf: IMAGE_FSTYPES += live - in poly/meta/conf/machine/qemux86.conf: MACHINE_FEATURES += efi The build gave me the following error: bash-3.00$ bitbake core-image-minimal Loading cache: 100% |###| ETA: 00:00:00 Loaded 1083 entries from dependency cache. OE Build Configuration: BB_VERSION= 1.15.0 TARGET_ARCH = i586 TARGET_OS = linux MACHINE = qemux86 DISTRO= poky DISTRO_VERSION= 1.1+snapshot-20120111 TUNE_FEATURES = m32 i586 TARGET_FPU= meta meta-yocto= (nobranch):468998cddbe1a803096c9b357e1b5daa3b7e8c2e NOTE: Resolving any missing task queue dependencies NOTE: Preparing runqueue NOTE: Executing SetScene Tasks NOTE: Executing RunQueue Tasks NOTE: Running task 1082 of 1701 (ID: 139, /tmp/localdisk/josef/yocto-stable/poky/meta/recipes-bsp/grub/grub-efi-native_1.99.bb, do_configure) NOTE: Running task 1682 of 1701 (ID: 97, /tmp/localdisk/josef/yocto-stable/poky/meta/recipes-core/images/core-image-minimal-initramfs.bb, do_rootfs) NOTE: package grub-efi-i586-native-1.99-r2: task do_configure: Started NOTE: package core-image-minimal-initramfs-1.0-r0: task do_rootfs: Started NOTE: package core-image-minimal-initramfs-1.0-r0: task do_rootfs: Succeeded ERROR: Function 'do_configure' failed (see /tmp/localdisk/josef/yocto-stable/mytestbuild/tmp/work/x86_64-linux/grub-efi-i586-native-1.99-r2/temp/log.do_configure.8536 for further information) ERROR: Logfile of failure stored in: /tmp/localdisk/josef/yocto-stable/mytestbuild/tmp/work/x86_64-linux/grub-efi-i586-native-1.99-r2/temp/log.do_configure.8536 Log data follows: | automake (GNU automake) 1.11.2 | Copyright (C) 2011 Free Software Foundation, Inc. | License GPLv2+: GNU GPL version 2 or later http://gnu.org/licenses/gpl-2.0.html | This is free software: you are free to change and redistribute it. | There is NO WARRANTY, to the extent permitted by law. | | Written by Tom Tromey tro...@redhat.com |and Alexandre Duret-Lutz a...@gnu.org. | AUTOV is 1.11 | NOTE: Executing autoreconf --verbose --install --force --exclude=autopoint -I /tmp/localdisk/josef/yocto-stable/mytestbuild/tmp/work/x86_64-linux/grub-efi-i586-native-1.99-r2/grub-1.99/aclocal-copy/ -I /tmp/localdisk/josef/yocto-stable/mytestbuild/tmp/work/x86_64-linux/grub-efi-i586-native-1.99-r2/grub-1.99/m4/ -I/tmp/localdisk/josef/yocto-stable/mytestbuild/tmp/sysroots/x86_64-linux/usr/share/aclocal-1.11 -I /tmp/localdisk/josef/yocto-stable/mytestbuild/tmp/work/x86_64-linux/grub-efi-i586-native-1.99-r2/grub-1.99/aclocal-copy/ | autoreconf: Entering directory `.' | autoreconf: configure.ac: not using Gettext | autoreconf: running: aclocal -I /tmp/localdisk/josef/yocto-stable/mytestbuild/tmp/work/x86_64-linux/grub-efi-i586-native-1.99-r2/grub-1.99/aclocal-copy/ -I /tmp/localdisk/josef/yocto-stable/mytestbuild/tmp/work/x86_64-linux/grub-efi-i586-native-1.99-r2/grub-1.99/m4/ -I /tmp/localdisk/josef/yocto-stable/mytestbuild/tmp/sysroots/x86_64-linux/usr/share/aclocal-1.11 -I /tmp/localdisk/josef/yocto-stable/mytestbuild/tmp/work/x86_64-linux/grub-efi-i586-native-1.99-r2/grub-1.99/aclocal-copy/ -I /tmp/localdisk/josef/yocto-stable/mytestbuild/tmp/work/x86_64-linux/grub-efi-i586-native-1.99-r2/grub-1.99/aclocal-copy/ -I /tmp/localdisk/josef/yocto-stable/mytestbuild/tmp/work/x86_64-linux/grub-efi-i586-native-1.99-r2/grub-1.99/m4/ -I /tmp/localdisk/josef/yocto-stable/mytestbuild/tmp/sysroots/x86_64-linux/usr/share/aclocal-1.11 -I /tmp/localdisk/josef/yocto-stable/mytestbuild/tmp/work/x86_64-linux/grub-efi-i586-native-1.99-r2/grub-1.99/aclocal-copy/ --force --warnings=cross -I m4 | aclocal: unknown warning category `cross' | configure.ac:515: warning: AC_LANG_CONFTEST: no AC_LANG_SOURCE call detected in body | ../../lib/autoconf/lang.m4:194: AC_LANG_CONFTEST is expanded from... | acinclude.m4:317: grub_CHECK_ENABLE_EXECUTE_STACK is expanded from... | configure.ac:515: the top level | configure.ac:518: warning: AC_LANG_CONFTEST: no AC_LANG_SOURCE call detected in body | ../../lib/autoconf/lang.m4:194: AC_LANG_CONFTEST is expanded from... | acinclude.m4:397: grub_CHECK_PIE is expanded from... | configure.ac:518: the top level | configure.ac:526: warning: AC_LANG_CONFTEST: no AC_LANG_SOURCE call detected in body | ../../lib/autoconf/lang.m4:194: AC_LANG_CONFTEST is expanded from... | acinclude.m4:344: grub_CHECK_STACK_PROTECTOR is expanded from... | configure.ac:526: the top level | configure.ac:532: warning: AC_LANG_CONFTEST: no AC_LANG_SOURCE call detected in body | ../../lib/autoconf/lang.m4:194: AC_LANG_CONFTEST is expanded from... | acinclude.m4:363: grub_CHECK_STACK_ARG_PROBE is expanded from... | configure.ac:532: the top level | autoreconf: configure.ac: tracing | configure.ac:515: warning: AC_LANG_CONFTEST: no AC_LANG_SOURCE call detected
Re: [yocto] Does Edison work with Beagleboard linux-yocto-3.0 kernel?
On Thu, Jan 12, 2012 at 11:00 AM, Bruce Ashfield bruce.ashfi...@windriver.com wrote: The bitbake AUTOREV code should take care of updating the clone of your local repo in downloads/git2. I take it that this isn't happening ? ... haven't tried ... was just sticking to the example in the documentation which says to push changes and do a cleanall. The spin time on my dev machine wasn't too bad but long enough to get old quick if I had to do it often. It should be (largely) as simple as that. We could create something simple and throw it in with the meta-kernel-dev layer if there's any interest in adding it. Just wanting to know what the best practice is as I have lots of platforms to support. Regards, Brian ___ yocto mailing list yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/yocto
Re: [yocto] Does Edison work with Beagleboard linux-yocto-3.0 kernel?
On 12-01-12 11:20 AM, Brian Hutchinson wrote: On Thu, Jan 12, 2012 at 11:00 AM, Bruce Ashfield bruce.ashfi...@windriver.com wrote: The bitbake AUTOREV code should take care of updating the clone of your local repo in downloads/git2. I take it that this isn't happening ? ... haven't tried ... was just sticking to the example in the documentation which says to push changes and do a cleanall. The spin Good point. I hadn't noticed that it was in there. Worth trying without it, and raising a bug if it isn't required. It's bitbake internals rather than kernel at play here, so I'm far from authoritative about what should or shouldn't work. time on my dev machine wasn't too bad but long enough to get old quick if I had to do it often. Agreed! It should be (largely) as simple as that. We could create something simple and throw it in with the meta-kernel-dev layer if there's any interest in adding it. Just wanting to know what the best practice is as I have lots of platforms to support. I prefer to work this way, since managing patches in a source git repository is much easier for me. If you only have a few patches, then they can just be pushed on top and added to the SRC_URI, but you'd be doing more development under tmp/work/ in that case .. and I'm paranoid about losing things :) Cheers, Bruce Regards, Brian ___ yocto mailing list yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/yocto
Re: [yocto] Does Edison work with Beagleboard linux-yocto-3.0 kernel?
On Thu, Jan 12, 2012 at 10:56 AM, Jack Mitchell m...@communistcode.co.uk wrote: Brian, What image are you building, as I cannot get core-image-minimal to boot at all on my xM. I am currently trying to fix a hosed sd card (who knows what happened to it!) and then I will see if it has made any difference. Probably the same images as you. I switched my branch to master and the default kernel built was 3.0.12. You might have to smoke your SD card (re-partition it for boot root) and copy MLO first, then u-boot.bin and uImage. Copying MLO first is key. I know this can be a pain. Google for beagleboard SD Card script to find one of the scripts that automates setting up the SD Card. They just create the partitions etc., you still have to copy the contents to the card in the right order. MLO is the first you copy on the boot partition, the order the rest are copied don't matter AFAIK. I also have a boot.scr with: setenv bootcmd 'mmc init; fatload mmc 0:1 0x8030 uImage; bootm 0x8030' setenv setenv bootargs 'console=ttyO2,115200n8 root=/dev/mmcblk0p2 rootwait rootfstype=ext3 rw' boot Hope that helps! Regards, Brian ___ yocto mailing list yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/yocto
Re: [yocto] Does Edison work with Beagleboard linux-yocto-3.0 kernel?
On Thu, Jan 12, 2012 at 11:24 AM, Bruce Ashfield bruce.ashfi...@windriver.com wrote: I prefer to work this way, since managing patches in a source git repository is much easier for me. If you only have a few patches, then they can just be pushed on top and added to the SRC_URI, but you'd be doing more development under tmp/work/ in that case .. and I'm paranoid about losing things :) I know it probably isn't the best way (did it mostly because I was in a hurry and never took the time to think about the right way to do it) but I've made changes to the kernel u-boot before in the tmp directories (with Angstrom and Arago) and build images manually (make uImage etc.) and that was OK for quick/dirty things but I've gotten burned before by forgetting my changes and doing a bitbake build and wiping out my changes. I like the method of pointing yocto to local git repos and think that is better for long term development as it will contain history and would be easy to determine what changed etc. I'd like to do something similar with u-boot and thought you guys had worked out a defacto standard way but in looking through all the docs it doesn't look that way. Regards, Brian ___ yocto mailing list yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/yocto
Re: [yocto] qemu EFI build failure
On 01/12/2012 08:19 AM, Josef Ahmad wrote: I tried to build a qemux86 EFI image, by setting: - in my local.conf: IMAGE_FSTYPES += live - in poly/meta/conf/machine/qemux86.conf: MACHINE_FEATURES += efi I haven't tried live images with QEMU. For one thing, they aren't really necessary as you can specify all the boot parameters on the qemu command line. Is there are reason you want to use the live image specifically? Also, in order to properly test EFI in QEMU, you will need to use an EFI BIOS - I believe you're aware of this already - but this isn't currently supported by the runqemu scripts that ship with yocto. The build gave me the following error: I'll do some test builds - it isn't clear to me what is going on here. snip Has anyone encountered the same error? I'm not sure I set up the correct configuration. Also, is there another way to append efi to MACHINE_FEATURES rather than by modifying qemux86.conf? You should be able to do something like: MACHINE_FEATURES_append_qemux86 = efi Note that you will also need to enable the efi support in the kernel, which is done with the KERNEL_FEATURES variable, something like: KERNEL_FEATURES_append_qemux86 = conf/efi-ext.scc Either of these can be set in your local.conf. -- Darren Hart Intel Open Source Technology Center Yocto Project - Linux Kernel ___ yocto mailing list yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/yocto
Re: [yocto] Does Edison work with Beagleboard linux-yocto-3.0 kernel?
On 12-01-12 11:48 AM, Brian Hutchinson wrote: On Thu, Jan 12, 2012 at 11:24 AM, Bruce Ashfield bruce.ashfi...@windriver.com wrote: I prefer to work this way, since managing patches in a source git repository is much easier for me. If you only have a few patches, then they can just be pushed on top and added to the SRC_URI, but you'd be doing more development under tmp/work/ in that case .. and I'm paranoid about losing things :) I know it probably isn't the best way (did it mostly because I was in a hurry and never took the time to think about the right way to do it) but I've made changes to the kernel u-boot before in the tmp directories (with Angstrom and Arago) and build images manually (make uImage etc.) and that was OK for quick/dirty things but I've gotten burned before by forgetting my changes and doing a bitbake build and wiping out my changes. I like the method of pointing yocto to local git repos and think that is better for long term development as it will contain history and would be easy to determine what changed etc. I'd like to do something similar with u-boot and thought you guys had worked out a defacto standard way but in looking through all the docs it doesn't look that way. Ah yes. Nothing in particular. I was a non-oe import to yocto, so a lot of what I've done is from being bitten in the past and making sure that it didn't happen to me in a new environment. Of course, I'm all for general solutions and getting feedback as well! Cheers, Bruce Regards, Brian ___ yocto mailing list yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/yocto
[yocto] [PATCH] meta-intel/common: fix ptr-int and Werror=address compile errors
A couple of things that had previously been warnings are now errors, so they need to be fixed up. The first problem is a comparison between the address of a static struct and NULL, which can never be valid. A different fix for this is upstream, which includes an API usage change; we don't need that to fix this problem. The second problem is a cast from pointer to integer in fbdevhw.c. This also is fixed upstream by removing the whole section of code which is bogus anyway, which is also done here. This also adds a missing PR to the xserver-xorg recipe. Signed-off-by: Tom Zanussi tom.zanu...@intel.com --- .../xorg-xserver/xserver-xorg-1.9.3.inc|5 + .../xserver-xorg-1.9.3/ptr-to-int-cast-fix.patch | 92 .../xserver-xorg-1.9.3/werror-address-fix.patch| 49 +++ 3 files changed, 146 insertions(+), 0 deletions(-) create mode 100644 common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3/ptr-to-int-cast-fix.patch create mode 100644 common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3/werror-address-fix.patch diff --git a/common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3.inc b/common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3.inc index 888445d..8c7009f 100644 --- a/common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3.inc +++ b/common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3.inc @@ -4,6 +4,10 @@ SRC_URI += file://nodolt.patch \ # Misc build failure for master HEAD SRC_URI += file://fix_open_max_preprocessor_error.patch +# What once were warnings now are errors, fix those up +SRC_URI += file://werror-address-fix.patch \ +file://ptr-to-int-cast-fix.patch + PROTO_DEPS += xf86driproto dri2proto DEPENDS += font-util EXTRA_OECONF += --enable-dri --enable-dri2 --enable-dga @@ -13,3 +17,4 @@ LIC_FILES_CHKSUM = file://COPYING;md5=3dd2bbe3563837f80ed8926b06c1c353 SRC_URI[md5sum] = 5bef6839a76d029204ab31aa2fcb5201 SRC_URI[sha256sum] = 864831f51e841ff37f2445d1c85b86b559c8860a435fb496aead4f256a2b141d +PR = r1 diff --git a/common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3/ptr-to-int-cast-fix.patch b/common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3/ptr-to-int-cast-fix.patch new file mode 100644 index 000..705cffc --- /dev/null +++ b/common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3/ptr-to-int-cast-fix.patch @@ -0,0 +1,92 @@ +Upstream-Status: Inappropriate [already upstream] + +It's broken for devices with BARs above 4G, and the sysfs method should +work everywhere anyway. As a pleasant side effect, this fixes some +warnings: + +fbdevhw.c: In function 'fbdev_open_pci': +fbdevhw.c:333:4: warning: cast from pointer to integer of different size +fbdevhw.c:334:4: warning: cast from pointer to integer of different size +fbdevhw.c:336:4: warning: cast from pointer to integer of different size +fbdevhw.c:337:4: warning: cast from pointer to integer of different size + +Signed-off-by: Adam Jackson ajax (a] redhat.com +Integrated-by: Tom Zanussi tom.zanussi (a] intel.com + +Index: xorg-server-1.9.3/hw/xfree86/fbdevhw/fbdevhw.c +=== +--- xorg-server-1.9.3.orig/hw/xfree86/fbdevhw/fbdevhw.c2012-01-12 10:32:07.097729262 -0600 xorg-server-1.9.3/hw/xfree86/fbdevhw/fbdevhw.c 2012-01-12 10:32:55.076732780 -0600 +@@ -291,14 +291,7 @@ + { + structfb_fix_screeninfo fix; + char filename[256]; +-int fd,i,j; +- +- +-/* There are two ways to that we can determine which fb device is +- * associated with this PCI device. The more modern way is to look in +- * the sysfs directory for the PCI device for a file named +- * graphics/fb* +- */ ++int fd, i; + + for (i = 0; i 8; i++) { + sprintf(filename, +@@ -331,55 +324,10 @@ + } + } + +- +-/* The other way is to examine the resources associated with each fb +- * device and see if there is a match with the PCI device. This technique +- * has some problems on certain mixed 64-bit / 32-bit architectures. +- * There is a flaw in the fb_fix_screeninfo structure in that it only +- * returns the low 32-bits of the address of the resources associated with +- * a device. However, on a mixed architecture the base addresses of PCI +- * devices, even for 32-bit applications, may be higher than 0x0f000. +- */ +- +-for (i = 0; i 8; i++) { +- sprintf(filename,/dev/fb%d,i); +- if (-1 == (fd = open(filename,O_RDWR,0))) { +- xf86DrvMsg(-1, X_WARNING, +- open %s: %s\n, filename,
Re: [yocto] where to put IMAGE_INSTALL??
On 01/12/2012 08:06 AM, Gary Thomas wrote: On 2012-01-12 05:53, Jim Abernathy wrote: There seems to be some sensitivity to where I put the statement IMAGE_INSTALL += web-webkit. At first I just modifed the meta/receipes-sato/image/core-image-sato.bb file to include it. That caused the webkit to be built. Since that is not an ideal location, I removed it and put it into the local.conf file. That also caused the webkit to be built, but I wanted it to be part of my BSP permanently. I thought about creating a core-image-sato.bbappend file with the IMAGE_INSTALL += web-webkit in it and put that into the BSP, but that didn't cause the webkit to be built. Under the meta-intel/meta-n450 directory, I created a receipes-jfa directory and put the core-image-sato.bbappend file there. I must be missing a key rule about directory parsing and .bbappend files. Look at the how the layers are parsed - this info is in layer/conf/layer.conf Try putting your .bbappend file in .../meta-intel/meta-n450/recipes-jfa/images/core-image-sato.bbappend I just tested your method of putting the .bbappend file in the directory above and it work just fine. I also found out that you should not put the IMAGE_INSTALL in the local.conf. It builds okay, but the resulting image will not boot. Tom pointing me the a statement in the Poky reference manual about not doing this as well. JIm A ___ yocto mailing list yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/yocto
[yocto] QtCreator and Yocto SDK
Hi, I'm currently looking for instructions on how to integrate a Yocto SDK with QtCreator. I've successfully built my application by sourcing the environment file and simply calling qmake/make, but QtCreator refuses to build the project. Any ideas/tips? Marc ___ yocto mailing list yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/yocto
Re: [yocto] kernel-tools failure for linux-yoctort_3.0.bb for poky/edison branch.
On Fri, 2011-12-16 at 19:03 -0800, Bruce Ashfield wrote: On 11-12-16 8:34 PM, Darren Hart wrote: On 12/16/2011 05:22 PM, Bodke, Kishore K wrote: With the attached Darren's kern-tools-native_git.bb file, build was success. Thanks Kishore. See the patch I just sent:kern-tools: Include do_install() in bbappend for what I believe to be the solution. I suppose the proper solution would be to version the kern-tools-native recipe when major things like that change, but given the nature of this layer, I think the above patch is adequate. That's not the solution to either problem (well, yes, it's a solution and there's nothing inherently wrong with it, it's just covering up the underlying issue) which is a layer dependency issue. In this case, meta-kernel-dev tracks the master branch of poky, not any other branch or version. And that is implicit, not explicit, hence why it got into this tangle. We could simply branch meta-kernel-dev and have an edison branch for it, not modify the appends to contain something that will just break again in the future. But meta-kernel-dev isn't officially used by much at the moment, so I'm not convinced that branching it is worth the trouble (but I can be convinced otherwise). I'd rather either just remove the kern-tools bbappend (I'm probably the only one on the planet that really needs it), or just trust people to know what they are up to with the meta-kernel-dev layer. Revisiting this - meta-kernel-dev is used in the documentation, and Scott has now run up against this when testing and updating it. Since the documentation does use edison, and the walk-throughs have the user check out edison branches of both poky and the BSP layers, it would be consistent to have the user also do the same for meta-kernel-dev. Probably just an edison branch with the kern-tools bbappend removed would work fine... Tom Alternatively, we wait for layer dependency tooling and use that :) Cheers, Bruce -- Darren -Original Message- From: Bodke, Kishore K Sent: Friday, December 16, 2011 5:07 PM To: Hart, Darren Cc: Bruce Ashfield; yocto@yoctoproject.org Subject: RE: kernel-tools failure for linux-yoctort_3.0.bb for poky/edison branch. Hi, I still get the same error below after changing the kern-tools-native_git.bb file. SRCREV ?= 364437739c45a5e771d1f7b3ac73c35f1328fd97 PR = r11 ERROR: Function 'do_kernel_configme' failed (see /usr/local/src/yocto_1_1/poky/build/tmp/work/crystalforest-poky-linux/linux-yocto-rt-3.0.4+git1+d05450e4aef02c1b7137398ab3a9f8f96da74f52_1+0936e13cc65d816f1759e2322c5e3fc82a5037f3-r1/temp/log.do_kernel_configme.27942 for further information) ERROR: Logfile of failure stored in: /usr/local/src/yocto_1_1/poky/build/tmp/work/crystalforest-poky-linux/linux-yocto-rt-3.0.4+git1+d05450e4aef02c1b7137398ab3a9f8f96da74f52_1+0936e13cc65d816f1759e2322c5e3fc82a5037f3-r1/temp/log.do_kernel_configme.27942 Log data follows: | [INFO] doing kernel configme | [INFO] Branch yocto/standard/preempt-rt/base used by common-pc-preempt-rt.scc | [INFO] collecting configs in ./meta/meta-series | mv: cannot stat `/usr/local/src/yocto_1_1/poky/build/tmp/work/crystalforest-poky-linux/linux-yocto-rt-3.0.4+git1+d05450e4aef02c1b7137398ab3a9f8f96da74f52_1+0936e13cc65d816f1759e2322c5e3fc82a5037f3-r1/linux-crystalforest-preempt-rt-build/.tmp.config*': No such file or directory | creation of pre-processed config data failed | config of yocto/standard/preempt-rt/base (common-pc-preempt-rt.scc) failed | ERROR: Function 'do_kernel_configme' failed (see /usr/local/src/yocto_1_1/poky/build/tmp/work/crystalforest-poky-linux/linux-yocto-rt-3.0.4+git1+d05450e4aef02c1b7137398ab3a9f8f96da74f52_1+0936e13cc65d816f1759e2322c5e3fc82a5037f3-r1/temp/log.do_kernel_configme.27942 for further information) NOTE: package linux-yocto-rt-3.0.4+git1+d05450e4aef02c1b7137398ab3a9f8f96da74f52_1+0936e13cc65d816f1759e2322c5e3fc82a5037f3-r1: task do_kernel_configme: Failed NOTE: package gcc-cross-4.6.1+svnr175454-r10: task do_configure: Started ERROR: Task 716 (/usr/local/src/yocto_1_1/poky/meta/recipes-kernel/linux/linux-yocto-rt_3.0.bb, do_kernel_configme) failed with exit code '1' Thanks Kishore. -Original Message- From: Hart, Darren Sent: Friday, December 16, 2011 3:08 PM To: Bodke, Kishore K Cc: Bruce Ashfield; yocto@yoctoproject.org Subject: Re: kernel-tools failure for linux-yoctort_3.0.bb for poky/edison branch. On 12/16/2011 02:12 PM, Bodke, Kishore K wrote: I am building core-image-sato. I have not changed any boot parameters. Everything was default settings. By image format I mean are you using a live image? core-image-sato-*.hddimg or are you formatting a disk and copying over contents of the rootfs? The kernel parameters are specified in meta-intel and meta-romley. My guess is something has gone wrong in
Re: [yocto] kernel-tools failure for linux-yoctort_3.0.bb for poky/edison branch.
On 01/12/2012 12:39 PM, Tom Zanussi wrote: On Fri, 2011-12-16 at 19:03 -0800, Bruce Ashfield wrote: On 11-12-16 8:34 PM, Darren Hart wrote: On 12/16/2011 05:22 PM, Bodke, Kishore K wrote: With the attached Darren's kern-tools-native_git.bb file, build was success. Thanks Kishore. See the patch I just sent:kern-tools: Include do_install() in bbappend for what I believe to be the solution. I suppose the proper solution would be to version the kern-tools-native recipe when major things like that change, but given the nature of this layer, I think the above patch is adequate. That's not the solution to either problem (well, yes, it's a solution and there's nothing inherently wrong with it, it's just covering up the underlying issue) which is a layer dependency issue. In this case, meta-kernel-dev tracks the master branch of poky, not any other branch or version. And that is implicit, not explicit, hence why it got into this tangle. We could simply branch meta-kernel-dev and have an edison branch for it, not modify the appends to contain something that will just break again in the future. But meta-kernel-dev isn't officially used by much at the moment, so I'm not convinced that branching it is worth the trouble (but I can be convinced otherwise). I'd rather either just remove the kern-tools bbappend (I'm probably the only one on the planet that really needs it), or just trust people to know what they are up to with the meta-kernel-dev layer. Revisiting this - meta-kernel-dev is used in the documentation, and Scott has now run up against this when testing and updating it. Since the documentation does use edison, and the walk-throughs have the user check out edison branches of both poky and the BSP layers, it would be consistent to have the user also do the same for meta-kernel-dev. Probably just an edison branch with the kern-tools bbappend removed would work fine... Tom Works for me. -- Darren Hart Intel Open Source Technology Center Yocto Project - Linux Kernel ___ yocto mailing list yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/yocto
Re: [yocto] qemu EFI build failure
On 01/12/2012 10:27 AM, Darren Hart wrote: On 01/12/2012 08:19 AM, Josef Ahmad wrote: I tried to build a qemux86 EFI image, by setting: - in my local.conf: IMAGE_FSTYPES += live - in poly/meta/conf/machine/qemux86.conf: MACHINE_FEATURES += efi I haven't tried live images with QEMU. For one thing, they aren't really necessary as you can specify all the boot parameters on the qemu command line. Is there are reason you want to use the live image specifically? Also, in order to properly test EFI in QEMU, you will need to use an EFI BIOS - I believe you're aware of this already - but this isn't currently supported by the runqemu scripts that ship with yocto. The build gave me the following error: I'll do some test builds - it isn't clear to me what is going on here. snip Has anyone encountered the same error? I'm not sure I set up the correct configuration. Also, is there another way to append efi to MACHINE_FEATURES rather than by modifying qemux86.conf? You should be able to do something like: MACHINE_FEATURES_append_qemux86 = efi Note that you will also need to enable the efi support in the kernel, which is done with the KERNEL_FEATURES variable, something like: KERNEL_FEATURES_append_qemux86 = conf/efi-ext.scc Either of these can be set in your local.conf. More accurately: MACHINE_FEATURES_append_qemux86= efi pcbios KERNEL_FEATURES_append_qemux86= cfg/efi-ext.scc IMAGE_FSTYPES += live With these added to local.conf and building for qemux86 I see the same configure failures reported by Kishore and Josef: | ../../lib/autoconf/lang.m4:194: AC_LANG_CONFTEST is expanded from... | acinclude.m4:363: grub_CHECK_STACK_ARG_PROBE is expanded from... I've discussed in IRC with Nitin and he thinks this may be related to the automake update and is looking into it. -- Darren Hart Intel Open Source Technology Center Yocto Project - Linux Kernel ___ yocto mailing list yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/yocto
Re: [yocto] [PATCH] meta-intel/common: fix ptr-int and Werror=address compile errors
On 01/12/2012 10:33 AM, Tom Zanussi wrote: A couple of things that had previously been warnings are now errors, so they need to be fixed up. The first problem is a comparison between the address of a static struct and NULL, which can never be valid. A different fix for this is upstream, which includes an API usage change; we don't need that to fix this problem. The second problem is a cast from pointer to integer in fbdevhw.c. This also is fixed upstream by removing the whole section of code which is bogus anyway, which is also done here. This also adds a missing PR to the xserver-xorg recipe. Signed-off-by: Tom Zanussi tom.zanu...@intel.com I'm fine with the patch given your upstream API change comment above. I am curious about just dropping the struct == NULL test as opposed to testing for some value in the structure that would indicate the intended condition. If you're happy with it as is, then: Acked-by: Darren Hart dvh...@linux.intel.com --- .../xorg-xserver/xserver-xorg-1.9.3.inc|5 + .../xserver-xorg-1.9.3/ptr-to-int-cast-fix.patch | 92 .../xserver-xorg-1.9.3/werror-address-fix.patch| 49 +++ 3 files changed, 146 insertions(+), 0 deletions(-) create mode 100644 common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3/ptr-to-int-cast-fix.patch create mode 100644 common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3/werror-address-fix.patch diff --git a/common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3.inc b/common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3.inc index 888445d..8c7009f 100644 --- a/common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3.inc +++ b/common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3.inc @@ -4,6 +4,10 @@ SRC_URI += file://nodolt.patch \ # Misc build failure for master HEAD SRC_URI += file://fix_open_max_preprocessor_error.patch +# What once were warnings now are errors, fix those up +SRC_URI += file://werror-address-fix.patch \ +file://ptr-to-int-cast-fix.patch + PROTO_DEPS += xf86driproto dri2proto DEPENDS += font-util EXTRA_OECONF += --enable-dri --enable-dri2 --enable-dga @@ -13,3 +17,4 @@ LIC_FILES_CHKSUM = file://COPYING;md5=3dd2bbe3563837f80ed8926b06c1c353 SRC_URI[md5sum] = 5bef6839a76d029204ab31aa2fcb5201 SRC_URI[sha256sum] = 864831f51e841ff37f2445d1c85b86b559c8860a435fb496aead4f256a2b141d +PR = r1 diff --git a/common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3/ptr-to-int-cast-fix.patch b/common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3/ptr-to-int-cast-fix.patch new file mode 100644 index 000..705cffc --- /dev/null +++ b/common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3/ptr-to-int-cast-fix.patch @@ -0,0 +1,92 @@ +Upstream-Status: Inappropriate [already upstream] + +It's broken for devices with BARs above 4G, and the sysfs method should +work everywhere anyway. As a pleasant side effect, this fixes some +warnings: + +fbdevhw.c: In function 'fbdev_open_pci': +fbdevhw.c:333:4: warning: cast from pointer to integer of different size +fbdevhw.c:334:4: warning: cast from pointer to integer of different size +fbdevhw.c:336:4: warning: cast from pointer to integer of different size +fbdevhw.c:337:4: warning: cast from pointer to integer of different size + +Signed-off-by: Adam Jackson ajax (a] redhat.com +Integrated-by: Tom Zanussi tom.zanussi (a] intel.com + +Index: xorg-server-1.9.3/hw/xfree86/fbdevhw/fbdevhw.c +=== +--- xorg-server-1.9.3.orig/hw/xfree86/fbdevhw/fbdevhw.c 2012-01-12 10:32:07.097729262 -0600 xorg-server-1.9.3/hw/xfree86/fbdevhw/fbdevhw.c 2012-01-12 10:32:55.076732780 -0600 +@@ -291,14 +291,7 @@ + { + struct fb_fix_screeninfo fix; + charfilename[256]; +-int fd,i,j; +- +- +-/* There are two ways to that we can determine which fb device is +- * associated with this PCI device. The more modern way is to look in +- * the sysfs directory for the PCI device for a file named +- * graphics/fb* +- */ ++int fd, i; + + for (i = 0; i 8; i++) { + sprintf(filename, +@@ -331,55 +324,10 @@ + } + } + +- +-/* The other way is to examine the resources associated with each fb +- * device and see if there is a match with the PCI device. This technique +- * has some problems on certain mixed 64-bit / 32-bit architectures. +- * There is a flaw in the fb_fix_screeninfo structure in that it only +-
Re: [yocto] kernel-tools failure for linux-yoctort_3.0.bb for poky/edison branch.
On 12-01-12 03:52 PM, Darren Hart wrote: On 01/12/2012 12:39 PM, Tom Zanussi wrote: On Fri, 2011-12-16 at 19:03 -0800, Bruce Ashfield wrote: On 11-12-16 8:34 PM, Darren Hart wrote: On 12/16/2011 05:22 PM, Bodke, Kishore K wrote: With the attached Darren's kern-tools-native_git.bb file, build was success. Thanks Kishore. See the patch I just sent:kern-tools: Include do_install() in bbappend for what I believe to be the solution. I suppose the proper solution would be to version the kern-tools-native recipe when major things like that change, but given the nature of this layer, I think the above patch is adequate. That's not the solution to either problem (well, yes, it's a solution and there's nothing inherently wrong with it, it's just covering up the underlying issue) which is a layer dependency issue. In this case, meta-kernel-dev tracks the master branch of poky, not any other branch or version. And that is implicit, not explicit, hence why it got into this tangle. We could simply branch meta-kernel-dev and have an edison branch for it, not modify the appends to contain something that will just break again in the future. But meta-kernel-dev isn't officially used by much at the moment, so I'm not convinced that branching it is worth the trouble (but I can be convinced otherwise). I'd rather either just remove the kern-tools bbappend (I'm probably the only one on the planet that really needs it), or just trust people to know what they are up to with the meta-kernel-dev layer. Revisiting this - meta-kernel-dev is used in the documentation, and Scott has now run up against this when testing and updating it. Since the documentation does use edison, and the walk-throughs have the user check out edison branches of both poky and the BSP layers, it would be consistent to have the user also do the same for meta-kernel-dev. Probably just an edison branch with the kern-tools bbappend removed would work fine... Tom Works for me. Seems ok. I can create a branch shortly, once I'm slightly out from under my 3.2 tree creation rush. Bruce ___ yocto mailing list yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/yocto
Re: [yocto] qemu EFI build failure
On 01/12/2012 01:28 PM, Bodke, Kishore K wrote: I have only MACHINE_FEATURES += efi KERNEL_FEATURES_append_cedartrail += cfg/efi-ext.scc. I am not having IMAGE_FSTYPES += live. The live change is just to trigger the live image type which builds grub-efi-native for efi systems. qemux86 needs this added explicitly. -- Darren Thanks Kishore. -Original Message- From: Hart, Darren Sent: Thursday, January 12, 2012 1:20 PM To: Ahmad, Josef Cc: yocto@yoctoproject.org; Bodke, Kishore K; Kamble, Nitin A Subject: Re: [yocto] qemu EFI build failure On 01/12/2012 10:27 AM, Darren Hart wrote: On 01/12/2012 08:19 AM, Josef Ahmad wrote: I tried to build a qemux86 EFI image, by setting: - in my local.conf: IMAGE_FSTYPES += live - in poly/meta/conf/machine/qemux86.conf: MACHINE_FEATURES += efi I haven't tried live images with QEMU. For one thing, they aren't really necessary as you can specify all the boot parameters on the qemu command line. Is there are reason you want to use the live image specifically? Also, in order to properly test EFI in QEMU, you will need to use an EFI BIOS - I believe you're aware of this already - but this isn't currently supported by the runqemu scripts that ship with yocto. The build gave me the following error: I'll do some test builds - it isn't clear to me what is going on here. snip Has anyone encountered the same error? I'm not sure I set up the correct configuration. Also, is there another way to append efi to MACHINE_FEATURES rather than by modifying qemux86.conf? You should be able to do something like: MACHINE_FEATURES_append_qemux86 = efi Note that you will also need to enable the efi support in the kernel, which is done with the KERNEL_FEATURES variable, something like: KERNEL_FEATURES_append_qemux86 = conf/efi-ext.scc Either of these can be set in your local.conf. More accurately: MACHINE_FEATURES_append_qemux86= efi pcbios KERNEL_FEATURES_append_qemux86= cfg/efi-ext.scc IMAGE_FSTYPES += live With these added to local.conf and building for qemux86 I see the same configure failures reported by Kishore and Josef: | ../../lib/autoconf/lang.m4:194: AC_LANG_CONFTEST is expanded from... | acinclude.m4:363: grub_CHECK_STACK_ARG_PROBE is expanded from... I've discussed in IRC with Nitin and he thinks this may be related to the automake update and is looking into it. -- Darren Hart Intel Open Source Technology Center Yocto Project - Linux Kernel ___ yocto mailing list yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/yocto
Re: [yocto] [PATCH] meta-intel/common: fix ptr-int and Werror=address compile errors
On Thu, 2012-01-12 at 13:24 -0800, Darren Hart wrote: On 01/12/2012 10:33 AM, Tom Zanussi wrote: A couple of things that had previously been warnings are now errors, so they need to be fixed up. The first problem is a comparison between the address of a static struct and NULL, which can never be valid. A different fix for this is upstream, which includes an API usage change; we don't need that to fix this problem. The second problem is a cast from pointer to integer in fbdevhw.c. This also is fixed upstream by removing the whole section of code which is bogus anyway, which is also done here. This also adds a missing PR to the xserver-xorg recipe. Signed-off-by: Tom Zanussi tom.zanu...@intel.com I'm fine with the patch given your upstream API change comment above. I am curious about just dropping the struct == NULL test as opposed to testing for some value in the structure that would indicate the intended condition. If you're happy with it as is, then: Well, the struct == NULL test can obviously never affect anything at run-time, and by definition never has, so can be safely removed. Changing the code to actually test something inside the struct that may or may not be meaningful would mean auditing all the code to find out whether the contents are always in a sane state for that test. Given the choice between a fix that makes no run-time changes and has no possibility of causing a regression vs spending a lot of time making an old version of obviously buggy code I know nothing about perfect, I chose the former. While I'm at it, I should also probably fix the 7000 other warnings - the only reason I'm in here in the first place is that a couple of them suddenly turned into errors... Tom Acked-by: Darren Hart dvh...@linux.intel.com --- .../xorg-xserver/xserver-xorg-1.9.3.inc|5 + .../xserver-xorg-1.9.3/ptr-to-int-cast-fix.patch | 92 .../xserver-xorg-1.9.3/werror-address-fix.patch| 49 +++ 3 files changed, 146 insertions(+), 0 deletions(-) create mode 100644 common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3/ptr-to-int-cast-fix.patch create mode 100644 common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3/werror-address-fix.patch diff --git a/common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3.inc b/common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3.inc index 888445d..8c7009f 100644 --- a/common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3.inc +++ b/common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3.inc @@ -4,6 +4,10 @@ SRC_URI += file://nodolt.patch \ # Misc build failure for master HEAD SRC_URI += file://fix_open_max_preprocessor_error.patch +# What once were warnings now are errors, fix those up +SRC_URI += file://werror-address-fix.patch \ +file://ptr-to-int-cast-fix.patch + PROTO_DEPS += xf86driproto dri2proto DEPENDS += font-util EXTRA_OECONF += --enable-dri --enable-dri2 --enable-dga @@ -13,3 +17,4 @@ LIC_FILES_CHKSUM = file://COPYING;md5=3dd2bbe3563837f80ed8926b06c1c353 SRC_URI[md5sum] = 5bef6839a76d029204ab31aa2fcb5201 SRC_URI[sha256sum] = 864831f51e841ff37f2445d1c85b86b559c8860a435fb496aead4f256a2b141d +PR = r1 diff --git a/common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3/ptr-to-int-cast-fix.patch b/common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3/ptr-to-int-cast-fix.patch new file mode 100644 index 000..705cffc --- /dev/null +++ b/common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3/ptr-to-int-cast-fix.patch @@ -0,0 +1,92 @@ +Upstream-Status: Inappropriate [already upstream] + +It's broken for devices with BARs above 4G, and the sysfs method should +work everywhere anyway. As a pleasant side effect, this fixes some +warnings: + +fbdevhw.c: In function 'fbdev_open_pci': +fbdevhw.c:333:4: warning: cast from pointer to integer of different size +fbdevhw.c:334:4: warning: cast from pointer to integer of different size +fbdevhw.c:336:4: warning: cast from pointer to integer of different size +fbdevhw.c:337:4: warning: cast from pointer to integer of different size + +Signed-off-by: Adam Jackson ajax (a] redhat.com +Integrated-by: Tom Zanussi tom.zanussi (a] intel.com + +Index: xorg-server-1.9.3/hw/xfree86/fbdevhw/fbdevhw.c +=== +--- xorg-server-1.9.3.orig/hw/xfree86/fbdevhw/fbdevhw.c2012-01-12 10:32:07.097729262 -0600
Re: [yocto] [PATCH] meta-intel/common: fix ptr-int and Werror=address compile errors
On 01/12/2012 01:47 PM, Tom Zanussi wrote: On Thu, 2012-01-12 at 13:24 -0800, Darren Hart wrote: On 01/12/2012 10:33 AM, Tom Zanussi wrote: A couple of things that had previously been warnings are now errors, so they need to be fixed up. The first problem is a comparison between the address of a static struct and NULL, which can never be valid. A different fix for this is upstream, which includes an API usage change; we don't need that to fix this problem. The second problem is a cast from pointer to integer in fbdevhw.c. This also is fixed upstream by removing the whole section of code which is bogus anyway, which is also done here. This also adds a missing PR to the xserver-xorg recipe. Signed-off-by: Tom Zanussi tom.zanu...@intel.com I'm fine with the patch given your upstream API change comment above. I am curious about just dropping the struct == NULL test as opposed to testing for some value in the structure that would indicate the intended condition. If you're happy with it as is, then: Well, the struct == NULL test can obviously never affect anything at run-time, and by definition never has, so can be safely removed. Changing the code to actually test something inside the struct that may or may not be meaningful would mean auditing all the code to find out whether the contents are always in a sane state for that test. Ah, yes, good point. Looks good to me as is. -- Darren Given the choice between a fix that makes no run-time changes and has no possibility of causing a regression vs spending a lot of time making an old version of obviously buggy code I know nothing about perfect, I chose the former. While I'm at it, I should also probably fix the 7000 other warnings - the only reason I'm in here in the first place is that a couple of them suddenly turned into errors... Tom Acked-by: Darren Hart dvh...@linux.intel.com --- .../xorg-xserver/xserver-xorg-1.9.3.inc|5 + .../xserver-xorg-1.9.3/ptr-to-int-cast-fix.patch | 92 .../xserver-xorg-1.9.3/werror-address-fix.patch| 49 +++ 3 files changed, 146 insertions(+), 0 deletions(-) create mode 100644 common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3/ptr-to-int-cast-fix.patch create mode 100644 common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3/werror-address-fix.patch diff --git a/common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3.inc b/common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3.inc index 888445d..8c7009f 100644 --- a/common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3.inc +++ b/common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3.inc @@ -4,6 +4,10 @@ SRC_URI += file://nodolt.patch \ # Misc build failure for master HEAD SRC_URI += file://fix_open_max_preprocessor_error.patch +# What once were warnings now are errors, fix those up +SRC_URI += file://werror-address-fix.patch \ +file://ptr-to-int-cast-fix.patch + PROTO_DEPS += xf86driproto dri2proto DEPENDS += font-util EXTRA_OECONF += --enable-dri --enable-dri2 --enable-dga @@ -13,3 +17,4 @@ LIC_FILES_CHKSUM = file://COPYING;md5=3dd2bbe3563837f80ed8926b06c1c353 SRC_URI[md5sum] = 5bef6839a76d029204ab31aa2fcb5201 SRC_URI[sha256sum] = 864831f51e841ff37f2445d1c85b86b559c8860a435fb496aead4f256a2b141d +PR = r1 diff --git a/common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3/ptr-to-int-cast-fix.patch b/common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3/ptr-to-int-cast-fix.patch new file mode 100644 index 000..705cffc --- /dev/null +++ b/common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3/ptr-to-int-cast-fix.patch @@ -0,0 +1,92 @@ +Upstream-Status: Inappropriate [already upstream] + +It's broken for devices with BARs above 4G, and the sysfs method should +work everywhere anyway. As a pleasant side effect, this fixes some +warnings: + +fbdevhw.c: In function 'fbdev_open_pci': +fbdevhw.c:333:4: warning: cast from pointer to integer of different size +fbdevhw.c:334:4: warning: cast from pointer to integer of different size +fbdevhw.c:336:4: warning: cast from pointer to integer of different size +fbdevhw.c:337:4: warning: cast from pointer to integer of different size + +Signed-off-by: Adam Jackson ajax (a] redhat.com +Integrated-by: Tom Zanussi tom.zanussi (a] intel.com + +Index: xorg-server-1.9.3/hw/xfree86/fbdevhw/fbdevhw.c +=== +--- xorg-server-1.9.3.orig/hw/xfree86/fbdevhw/fbdevhw.c2012-01-12 10:32:07.097729262
Re: [yocto] [PATCH] meta-intel/common: fix ptr-int and Werror=address compile errors
On Thu, 2012-01-12 at 13:49 -0800, Darren Hart wrote: On 01/12/2012 01:47 PM, Tom Zanussi wrote: On Thu, 2012-01-12 at 13:24 -0800, Darren Hart wrote: On 01/12/2012 10:33 AM, Tom Zanussi wrote: A couple of things that had previously been warnings are now errors, so they need to be fixed up. The first problem is a comparison between the address of a static struct and NULL, which can never be valid. A different fix for this is upstream, which includes an API usage change; we don't need that to fix this problem. The second problem is a cast from pointer to integer in fbdevhw.c. This also is fixed upstream by removing the whole section of code which is bogus anyway, which is also done here. This also adds a missing PR to the xserver-xorg recipe. Signed-off-by: Tom Zanussi tom.zanu...@intel.com I'm fine with the patch given your upstream API change comment above. I am curious about just dropping the struct == NULL test as opposed to testing for some value in the structure that would indicate the intended condition. If you're happy with it as is, then: Well, the struct == NULL test can obviously never affect anything at run-time, and by definition never has, so can be safely removed. Changing the code to actually test something inside the struct that may or may not be meaningful would mean auditing all the code to find out whether the contents are always in a sane state for that test. Ah, yes, good point. Looks good to me as is. OK, thanks for the ack.. Tom -- Darren Given the choice between a fix that makes no run-time changes and has no possibility of causing a regression vs spending a lot of time making an old version of obviously buggy code I know nothing about perfect, I chose the former. While I'm at it, I should also probably fix the 7000 other warnings - the only reason I'm in here in the first place is that a couple of them suddenly turned into errors... Tom Acked-by: Darren Hart dvh...@linux.intel.com --- .../xorg-xserver/xserver-xorg-1.9.3.inc|5 + .../xserver-xorg-1.9.3/ptr-to-int-cast-fix.patch | 92 .../xserver-xorg-1.9.3/werror-address-fix.patch| 49 +++ 3 files changed, 146 insertions(+), 0 deletions(-) create mode 100644 common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3/ptr-to-int-cast-fix.patch create mode 100644 common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3/werror-address-fix.patch diff --git a/common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3.inc b/common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3.inc index 888445d..8c7009f 100644 --- a/common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3.inc +++ b/common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3.inc @@ -4,6 +4,10 @@ SRC_URI += file://nodolt.patch \ # Misc build failure for master HEAD SRC_URI += file://fix_open_max_preprocessor_error.patch +# What once were warnings now are errors, fix those up +SRC_URI += file://werror-address-fix.patch \ +file://ptr-to-int-cast-fix.patch + PROTO_DEPS += xf86driproto dri2proto DEPENDS += font-util EXTRA_OECONF += --enable-dri --enable-dri2 --enable-dga @@ -13,3 +17,4 @@ LIC_FILES_CHKSUM = file://COPYING;md5=3dd2bbe3563837f80ed8926b06c1c353 SRC_URI[md5sum] = 5bef6839a76d029204ab31aa2fcb5201 SRC_URI[sha256sum] = 864831f51e841ff37f2445d1c85b86b559c8860a435fb496aead4f256a2b141d +PR = r1 diff --git a/common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3/ptr-to-int-cast-fix.patch b/common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3/ptr-to-int-cast-fix.patch new file mode 100644 index 000..705cffc --- /dev/null +++ b/common/recipes-graphics/xorg-xserver/xserver-xorg-1.9.3/ptr-to-int-cast-fix.patch @@ -0,0 +1,92 @@ +Upstream-Status: Inappropriate [already upstream] + +It's broken for devices with BARs above 4G, and the sysfs method should +work everywhere anyway. As a pleasant side effect, this fixes some +warnings: + +fbdevhw.c: In function 'fbdev_open_pci': +fbdevhw.c:333:4: warning: cast from pointer to integer of different size +fbdevhw.c:334:4: warning: cast from pointer to integer of different size +fbdevhw.c:336:4: warning: cast from pointer to integer of different size +fbdevhw.c:337:4: warning: cast from pointer to integer of different size + +Signed-off-by: Adam Jackson ajax (a] redhat.com +Integrated-by: Tom Zanussi tom.zanussi (a] intel.com + +Index:
Re: [yocto] qemu EFI build failure
I have only MACHINE_FEATURES += efi KERNEL_FEATURES_append_cedartrail += cfg/efi-ext.scc. I am not having IMAGE_FSTYPES += live. Thanks Kishore. -Original Message- From: Hart, Darren Sent: Thursday, January 12, 2012 1:20 PM To: Ahmad, Josef Cc: yocto@yoctoproject.org; Bodke, Kishore K; Kamble, Nitin A Subject: Re: [yocto] qemu EFI build failure On 01/12/2012 10:27 AM, Darren Hart wrote: On 01/12/2012 08:19 AM, Josef Ahmad wrote: I tried to build a qemux86 EFI image, by setting: - in my local.conf: IMAGE_FSTYPES += live - in poly/meta/conf/machine/qemux86.conf: MACHINE_FEATURES += efi I haven't tried live images with QEMU. For one thing, they aren't really necessary as you can specify all the boot parameters on the qemu command line. Is there are reason you want to use the live image specifically? Also, in order to properly test EFI in QEMU, you will need to use an EFI BIOS - I believe you're aware of this already - but this isn't currently supported by the runqemu scripts that ship with yocto. The build gave me the following error: I'll do some test builds - it isn't clear to me what is going on here. snip Has anyone encountered the same error? I'm not sure I set up the correct configuration. Also, is there another way to append efi to MACHINE_FEATURES rather than by modifying qemux86.conf? You should be able to do something like: MACHINE_FEATURES_append_qemux86 = efi Note that you will also need to enable the efi support in the kernel, which is done with the KERNEL_FEATURES variable, something like: KERNEL_FEATURES_append_qemux86 = conf/efi-ext.scc Either of these can be set in your local.conf. More accurately: MACHINE_FEATURES_append_qemux86= efi pcbios KERNEL_FEATURES_append_qemux86= cfg/efi-ext.scc IMAGE_FSTYPES += live With these added to local.conf and building for qemux86 I see the same configure failures reported by Kishore and Josef: | ../../lib/autoconf/lang.m4:194: AC_LANG_CONFTEST is expanded from... | acinclude.m4:363: grub_CHECK_STACK_ARG_PROBE is expanded from... I've discussed in IRC with Nitin and he thinks this may be related to the automake update and is looking into it. -- Darren Hart Intel Open Source Technology Center Yocto Project - Linux Kernel ___ yocto mailing list yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/yocto
Re: [yocto] qemu EFI build failure
I just sent a patch to oecore mailing list to fix the grub issue with automake 1.11.2. Thanks, Nitin -Original Message- From: Hart, Darren Sent: Thursday, January 12, 2012 1:39 PM To: Bodke, Kishore K Cc: Ahmad, Josef; yocto@yoctoproject.org; Kamble, Nitin A Subject: Re: [yocto] qemu EFI build failure On 01/12/2012 01:28 PM, Bodke, Kishore K wrote: I have only MACHINE_FEATURES += efi KERNEL_FEATURES_append_cedartrail += cfg/efi-ext.scc. I am not having IMAGE_FSTYPES += live. The live change is just to trigger the live image type which builds grub-efi-native for efi systems. qemux86 needs this added explicitly. -- Darren Thanks Kishore. -Original Message- From: Hart, Darren Sent: Thursday, January 12, 2012 1:20 PM To: Ahmad, Josef Cc: yocto@yoctoproject.org; Bodke, Kishore K; Kamble, Nitin A Subject: Re: [yocto] qemu EFI build failure On 01/12/2012 10:27 AM, Darren Hart wrote: On 01/12/2012 08:19 AM, Josef Ahmad wrote: I tried to build a qemux86 EFI image, by setting: - in my local.conf: IMAGE_FSTYPES += live - in poly/meta/conf/machine/qemux86.conf: MACHINE_FEATURES += efi I haven't tried live images with QEMU. For one thing, they aren't really necessary as you can specify all the boot parameters on the qemu command line. Is there are reason you want to use the live image specifically? Also, in order to properly test EFI in QEMU, you will need to use an EFI BIOS - I believe you're aware of this already - but this isn't currently supported by the runqemu scripts that ship with yocto. The build gave me the following error: I'll do some test builds - it isn't clear to me what is going on here. snip Has anyone encountered the same error? I'm not sure I set up the correct configuration. Also, is there another way to append efi to MACHINE_FEATURES rather than by modifying qemux86.conf? You should be able to do something like: MACHINE_FEATURES_append_qemux86 = efi Note that you will also need to enable the efi support in the kernel, which is done with the KERNEL_FEATURES variable, something like: KERNEL_FEATURES_append_qemux86 = conf/efi-ext.scc Either of these can be set in your local.conf. More accurately: MACHINE_FEATURES_append_qemux86= efi pcbios KERNEL_FEATURES_append_qemux86= cfg/efi-ext.scc IMAGE_FSTYPES += live With these added to local.conf and building for qemux86 I see the same configure failures reported by Kishore and Josef: | ../../lib/autoconf/lang.m4:194: AC_LANG_CONFTEST is expanded from... | acinclude.m4:363: grub_CHECK_STACK_ARG_PROBE is expanded from... I've discussed in IRC with Nitin and he thinks this may be related to the automake update and is looking into it. -- Darren Hart Intel Open Source Technology Center Yocto Project - Linux Kernel ___ yocto mailing list yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/yocto
Re: [yocto] qemu EFI build failure
Here is the commit: http://git.yoctoproject.org/cgit.cgi/poky-contrib/log/?h=nitin/misc Nitin -Original Message- From: yocto-boun...@yoctoproject.org [mailto:yocto- boun...@yoctoproject.org] On Behalf Of Kamble, Nitin A Sent: Thursday, January 12, 2012 2:32 PM To: Hart, Darren; Bodke, Kishore K Cc: Ahmad, Josef; yocto@yoctoproject.org Subject: Re: [yocto] qemu EFI build failure I just sent a patch to oecore mailing list to fix the grub issue with automake 1.11.2. Thanks, Nitin -Original Message- From: Hart, Darren Sent: Thursday, January 12, 2012 1:39 PM To: Bodke, Kishore K Cc: Ahmad, Josef; yocto@yoctoproject.org; Kamble, Nitin A Subject: Re: [yocto] qemu EFI build failure On 01/12/2012 01:28 PM, Bodke, Kishore K wrote: I have only MACHINE_FEATURES += efi KERNEL_FEATURES_append_cedartrail += cfg/efi-ext.scc. I am not having IMAGE_FSTYPES += live. The live change is just to trigger the live image type which builds grub-efi-native for efi systems. qemux86 needs this added explicitly. -- Darren Thanks Kishore. -Original Message- From: Hart, Darren Sent: Thursday, January 12, 2012 1:20 PM To: Ahmad, Josef Cc: yocto@yoctoproject.org; Bodke, Kishore K; Kamble, Nitin A Subject: Re: [yocto] qemu EFI build failure On 01/12/2012 10:27 AM, Darren Hart wrote: On 01/12/2012 08:19 AM, Josef Ahmad wrote: I tried to build a qemux86 EFI image, by setting: - in my local.conf: IMAGE_FSTYPES += live - in poly/meta/conf/machine/qemux86.conf: MACHINE_FEATURES += efi I haven't tried live images with QEMU. For one thing, they aren't really necessary as you can specify all the boot parameters on the qemu command line. Is there are reason you want to use the live image specifically? Also, in order to properly test EFI in QEMU, you will need to use an EFI BIOS - I believe you're aware of this already - but this isn't currently supported by the runqemu scripts that ship with yocto. The build gave me the following error: I'll do some test builds - it isn't clear to me what is going on here. snip Has anyone encountered the same error? I'm not sure I set up the correct configuration. Also, is there another way to append efi to MACHINE_FEATURES rather than by modifying qemux86.conf? You should be able to do something like: MACHINE_FEATURES_append_qemux86 = efi Note that you will also need to enable the efi support in the kernel, which is done with the KERNEL_FEATURES variable, something like: KERNEL_FEATURES_append_qemux86 = conf/efi-ext.scc Either of these can be set in your local.conf. More accurately: MACHINE_FEATURES_append_qemux86= efi pcbios KERNEL_FEATURES_append_qemux86= cfg/efi-ext.scc IMAGE_FSTYPES += live With these added to local.conf and building for qemux86 I see the same configure failures reported by Kishore and Josef: | ../../lib/autoconf/lang.m4:194: AC_LANG_CONFTEST is expanded from... | acinclude.m4:363: grub_CHECK_STACK_ARG_PROBE is expanded from... I've discussed in IRC with Nitin and he thinks this may be related to the automake update and is looking into it. -- Darren Hart Intel Open Source Technology Center Yocto Project - Linux Kernel ___ yocto mailing list yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/yocto ___ yocto mailing list yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/yocto
[yocto] Developer Manual Appendix B example
I'm trying to do the Developer Manual Appendix B example exactly as written only changing the path to my home directory. When I get to B.1.8.3, bitbake after patching the kernel calibrate.c and modifiing the path to the local git repository for linux-yocto, I get an error on the build. The log file of the error is below: [INFO] doing kernel configme [INFO] Branch yocto/standard/common-pc/base used by common-pc-standard.scc [INFO] collecting configs in ./meta/meta-series mv: cannot stat `/home/jim/poky/build/tmp/work/qemux86-poky-linux/linux-yocto-3.0.4+git1+d05450e4aef02c1b7137398ab3a9f8f96da74f52_1+72671808fdbe69a9fe03fd8f094e7c59da04a28c-r2/linux-qemux86-standard-build/.tmp.config*': No such file or directory creation of pre-processed config data failed config of yocto/standard/common-pc/base (common-pc-standard.scc) failed ERROR: Function 'do_kernel_configme' failed (see /home/jim/poky/build/tmp/work/qemux86-poky-linux/linux-yocto-3.0.4+git1+d05450e4aef02c1b7137398ab3a9f8f96da74f52_1+72671808fdbe69a9fe03fd8f094e7c59da04a28c-r2/temp/log.do_kernel_configme.15214 for further information) Any help would be appreciated. Jim A ___ yocto mailing list yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/yocto
Re: [yocto] Developer Manual Appendix B example
On 12-01-12 09:50 PM, James Abernathy wrote: I'm trying to do the Developer Manual Appendix B example exactly as written only changing the path to my home directory. When I get to B.1.8.3, bitbake after patching the kernel calibrate.c and modifiing the path to the local git repository for linux-yocto, I get an error on the build. The log file of the error is below: Funny how things tend to arrive in bunches. We just discussed this today on the list. The meta-kernel-dev layer is tied to master, and causes this problem when used with edison. And that's what you are seeing here. I'm going to create a edison branch for meta-kernel-dev to match up the tool SRCREVs. In the meantime, if you remove the kern-tools-native_git.bbappend from the meta-kernel-dev layer, the AUTOREV setting will be removed, and you won't pull in tools that are too new for the matching kernel. Cheers, Bruce [INFO] doing kernel configme [INFO] Branch yocto/standard/common-pc/base used by common-pc-standard.scc [INFO] collecting configs in ./meta/meta-series mv: cannot stat `/home/jim/poky/build/tmp/work/qemux86-poky-linux/linux-yocto-3.0.4+git1+d05450e4aef02c1b7137398ab3a9f8f96da74f52_1+72671808fdbe69a9fe03fd8f094e7c59da04a28c-r2/linux-qemux86-standard-build/.tmp.config*': No such file or directory creation of pre-processed config data failed config of yocto/standard/common-pc/base (common-pc-standard.scc) failed ERROR: Function 'do_kernel_configme' failed (see /home/jim/poky/build/tmp/work/qemux86-poky-linux/linux-yocto-3.0.4+git1+d05450e4aef02c1b7137398ab3a9f8f96da74f52_1+72671808fdbe69a9fe03fd8f094e7c59da04a28c-r2/temp/log.do_kernel_configme.15214 for further information) Any help would be appreciated. Jim A ___ yocto mailing list yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/yocto ___ yocto mailing list yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/yocto
Re: [yocto] Developer Manual Appendix B example
On Thu, Jan 12, 2012 at 9:54 PM, Bruce Ashfield bruce.ashfi...@windriver.com wrote: ** On 12-01-12 09:50 PM, James Abernathy wrote: I'm trying to do the Developer Manual Appendix B example exactly as written only changing the path to my home directory. When I get to B.1.8.3, bitbake after patching the kernel calibrate.c and modifiing the path to the local git repository for linux-yocto, I get an error on the build. The log file of the error is below: Funny how things tend to arrive in bunches. We just discussed this today on the list. The meta-kernel-dev layer is tied to master, and causes this problem when used with edison. And that's what you are seeing here. I'm going to create a edison branch for meta-kernel-dev to match up the tool SRCREVs. In the meantime, if you remove the kern-tools-native_git.bbappend from the meta-kernel-dev layer, the AUTOREV setting will be removed, and you won't pull in tools that are too new for the matching kernel. That worked as you said. I thought it was odd how we did a git checkout for Linux and Poky, but not the poky-extras repositories. Jim A Cheers, Bruce [INFO] doing kernel configme [INFO] Branch yocto/standard/common-pc/base used by common-pc-standard.scc [INFO] collecting configs in ./meta/meta-series mv: cannot stat `/home/jim/poky/build/tmp/work/qemux86-poky-linux/linux-yocto-3.0.4+git1+d05450e4aef02c1b7137398ab3a9f8f96da74f52_1+72671808fdbe69a9fe03fd8f094e7c59da04a28c-r2/linux-qemux86-standard-build/.tmp.config*': No such file or directory creation of pre-processed config data failed config of yocto/standard/common-pc/base (common-pc-standard.scc) failed ERROR: Function 'do_kernel_configme' failed (see /home/jim/poky/build/tmp/work/qemux86-poky-linux/linux-yocto-3.0.4+git1+d05450e4aef02c1b7137398ab3a9f8f96da74f52_1+72671808fdbe69a9fe03fd8f094e7c59da04a28c-r2/temp/log.do_kernel_configme.15214 for further information) Any help would be appreciated. Jim A ___ yocto mailing listyocto@yoctoproject.orghttps://lists.yoctoproject.org/listinfo/yocto ___ yocto mailing list yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/yocto
Re: [yocto] Developer Manual Appendix B example
On Thu, Jan 12, 2012 at 9:54 PM, Bruce Ashfield bruce.ashfi...@windriver.com wrote: Funny how things tend to arrive in bunches. We just discussed this today Ha! Thanks Jim for making me feel better at least I wasn't the only one that had trouble ;-) Brian ___ yocto mailing list yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/yocto
Re: [yocto] Unable to set system clock
Hello guys, On 01/11/2012 01:36 PM, Gary Thomas wrote: There is a bug #1767 that can make the system always reset the clock to the saved timestamp on boot. I am not pretty sure for this because I am able to set system clock using freescale provided rootfs. This problem I am getting only with poky rootfs. I guess I need to change some recipe. It would be a great help if you can help me in that. The problem is that even when you set the date, your Poky system does not use it when the system is rebooted because of this bug. The simplest work-around (until the bug fix is finalized) is to add the 'coreutils' package to your image. I found the same bug and I fixed it in December. I didn't know there was open bug related to that. Anyway patch is available in master branch of poky: http://git.yoctoproject.org/cgit/cgit.cgi/poky/commit/?id=91367de75575e79a94b2f90b6b7aa8740a039fc5 Regards, Lauri Hintsala ___ yocto mailing list yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/yocto