| Submitter | Nitin A Kamble |
|---|---|
| Date | July 12, 2011, 7:02 p.m. |
| Message ID | <cover.1310497097.git.nitin.a.kamble@intel.com> |
| Download | mbox |
| Permalink | /patch/7453/ |
| State | New, archived |
| Headers | show |
Pull-request
git://git.pokylinux.org/poky-contrib nitin/x32Comments
On Tue, 2011-07-12 at 12:02 -0700, nitin.a.kamble@intel.com wrote: > From: Nitin A Kamble <nitin.a.kamble@intel.com> > > These are commits for yocto tree. I came up with as part of the x32 work. > I think these commits does not belong in the x32 layer and should go > in the yocto/oecore tree. > > The siteinfo.bbclass commit is a hacky way to get the the right siteinfo for > x32 now. In the future a better solution will be implemented for that purpose. > > The following changes since commit 7354fc9213f27aa1b643dbe88070437f1ee4c063: > > insane.bbclass: skip rdepends QA checks for kernel / modules (2011-07-12 15:22:09 +0100) > > are available in the git repository at: > git://git.pokylinux.org/poky-contrib nitin/x32 > http://git.pokylinux.org/cgit.cgi/poky-contrib/log/?h=nitin/x32 > > Nitin A Kamble (5): > siteinfo.bbclass: hack for x32 > udev-172: add a newer version for newer kernel > runqemu: support qemux32 machine These first three need a little more thought. They're fine to as proof of concept work but we need to find better ways to integrate them. Specifically: The siteinfo change is a hack and can't be merged as is. We need to find a way to change the target_arch or target_os in such a way we can signal to siteinfo to do something different. I also wonder if using i686-linux is actually correct for x32 in all cases since x32 does support 64 bit data types. The udev update needs various other work as people have commented. We need to resolve the udev differences between meta-oe and oe-core. runqemu looks like it can be simplified as it looks like just the kernel name is different? Is x32 using a different machine config at present? Can we use the qemux86-64 machine directly and just add some tweaks to it? That would avoid the need to change the qemu scripting? Cheers, Richard
On 07/13/2011 04:20 AM, Richard Purdie wrote: > On Tue, 2011-07-12 at 12:02 -0700, nitin.a.kamble@intel.com wrote: >> From: Nitin A Kamble <nitin.a.kamble@intel.com> >> >> These are commits for yocto tree. I came up with as part of the x32 work. >> I think these commits does not belong in the x32 layer and should go >> in the yocto/oecore tree. >> >> The siteinfo.bbclass commit is a hacky way to get the the right siteinfo for >> x32 now. In the future a better solution will be implemented for that purpose. >> >> The following changes since commit 7354fc9213f27aa1b643dbe88070437f1ee4c063: >> >> insane.bbclass: skip rdepends QA checks for kernel / modules (2011-07-12 15:22:09 +0100) >> >> are available in the git repository at: >> git://git.pokylinux.org/poky-contrib nitin/x32 >> http://git.pokylinux.org/cgit.cgi/poky-contrib/log/?h=nitin/x32 >> >> Nitin A Kamble (5): >> siteinfo.bbclass: hack for x32 >> udev-172: add a newer version for newer kernel >> runqemu: support qemux32 machine > > These first three need a little more thought. They're fine to as proof > of concept work but we need to find better ways to integrate them. > > Specifically: > > The siteinfo change is a hack and can't be merged as is. We need to find > a way to change the target_arch or target_os in such a way we can signal > to siteinfo to do something different. I also wonder if using i686-linux > is actually correct for x32 in all cases since x32 does support 64 bit > data types. Hopefully this will be easier with the siteinfo dictionary bits we've got in oe.dev that I'm working on moving to oe-core (world finished overnight so time to diff the config.log's and kick off some of the automated tests).
On 07/13/2011 04:20 AM, Richard Purdie wrote: > On Tue, 2011-07-12 at 12:02 -0700, nitin.a.kamble@intel.com wrote: >> From: Nitin A Kamble<nitin.a.kamble@intel.com> >> >> These are commits for yocto tree. I came up with as part of the x32 work. >> I think these commits does not belong in the x32 layer and should go >> in the yocto/oecore tree. >> >> The siteinfo.bbclass commit is a hacky way to get the the right siteinfo for >> x32 now. In the future a better solution will be implemented for that purpose. >> >> The following changes since commit 7354fc9213f27aa1b643dbe88070437f1ee4c063: >> >> insane.bbclass: skip rdepends QA checks for kernel / modules (2011-07-12 15:22:09 +0100) >> >> are available in the git repository at: >> git://git.pokylinux.org/poky-contrib nitin/x32 >> http://git.pokylinux.org/cgit.cgi/poky-contrib/log/?h=nitin/x32 >> >> Nitin A Kamble (5): >> siteinfo.bbclass: hack for x32 >> udev-172: add a newer version for newer kernel >> runqemu: support qemux32 machine > > These first three need a little more thought. They're fine to as proof > of concept work but we need to find better ways to integrate them. > > Specifically: > > The siteinfo change is a hack and can't be merged as is. We need to find > a way to change the target_arch or target_os in such a way we can signal > to siteinfo to do something different. I also wonder if using i686-linux > is actually correct for x32 in all cases since x32 does support 64 bit > data types. I think it should have its own bits in separate file. > > The udev update needs various other work as people have commented. We > need to resolve the udev differences between meta-oe and oe-core. > > runqemu looks like it can be simplified as it looks like just the kernel > name is different? Is x32 using a different machine config at present? > Can we use the qemux86-64 machine directly and just add some tweaks to > it? That would avoid the need to change the qemu scripting? > > Cheers, > > Richard > > > > > > _______________________________________________ > Openembedded-core mailing list > Openembedded-core@lists.openembedded.org > http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-core
From: Nitin A Kamble <nitin.a.kamble@intel.com> These are commits for yocto tree. I came up with as part of the x32 work. I think these commits does not belong in the x32 layer and should go in the yocto/oecore tree. The siteinfo.bbclass commit is a hacky way to get the the right siteinfo for x32 now. In the future a better solution will be implemented for that purpose. The following changes since commit 7354fc9213f27aa1b643dbe88070437f1ee4c063: insane.bbclass: skip rdepends QA checks for kernel / modules (2011-07-12 15:22:09 +0100) are available in the git repository at: git://git.pokylinux.org/poky-contrib nitin/x32 http://git.pokylinux.org/cgit.cgi/poky-contrib/log/?h=nitin/x32 Nitin A Kamble (5): siteinfo.bbclass: hack for x32 udev-172: add a newer version for newer kernel runqemu: support qemux32 machine kernel.bbclass: fix the broken lines glibc: fix false failure meta/classes/kernel.bbclass | 4 +- meta/classes/siteinfo.bbclass | 3 + meta/recipes-core/glibc/glibc.inc | 2 +- meta/recipes-core/udev/udev-172/init | 59 +++++++++ meta/recipes-core/udev/udev-172/local.rules | 35 ++++++ meta/recipes-core/udev/udev-172/permissions.rules | 131 +++++++++++++++++++++ meta/recipes-core/udev/udev-172/run.rules | 14 +++ meta/recipes-core/udev/udev-172/udev.rules | 116 ++++++++++++++++++ meta/recipes-core/udev/udev_172.bb | 6 + scripts/qemuimage-testlib | 2 +- scripts/runqemu | 12 ++- scripts/runqemu-internal | 25 ++++ 12 files changed, 401 insertions(+), 8 deletions(-) create mode 100644 meta/recipes-core/udev/udev-172/init create mode 100644 meta/recipes-core/udev/udev-172/local.rules create mode 100644 meta/recipes-core/udev/udev-172/permissions.rules create mode 100644 meta/recipes-core/udev/udev-172/run.rules create mode 100644 meta/recipes-core/udev/udev-172/udev.rules create mode 100644 meta/recipes-core/udev/udev_172.bb