On 14-03-28 09:32 AM, Richard Purdie wrote:
On Fri, 2014-03-28 at 08:52 -0400, Bruce Ashfield wrote:
On 14-03-28 05:45 AM, Richard Purdie wrote:
On Thu, 2014-03-27 at 14:25 -0400, Bruce Ashfield wrote:
The release kernel for Yocto 1.5 is the 3.14 kernel, so we introduce
the versioned recipes here.
Signed-off-by: Bruce Ashfield <[email protected]>
---
meta/recipes-kernel/linux/linux-yocto-tiny_3.14.bb | 21 ++++++++++++
meta/recipes-kernel/linux/linux-yocto_3.14.bb | 37
++++++++++++++++++++++
2 files changed, 58 insertions(+)
create mode 100644 meta/recipes-kernel/linux/linux-yocto-tiny_3.14.bb
create mode 100644 meta/recipes-kernel/linux/linux-yocto_3.14.bb
diff --git a/meta/recipes-kernel/linux/linux-yocto-tiny_3.14.bb
b/meta/recipes-kernel/linux/linux-yocto-tiny_3.14.bb
new file mode 100644
index 000000000000..8faa9584acf0
--- /dev/null
+++ b/meta/recipes-kernel/linux/linux-yocto-tiny_3.14.bb
@@ -0,0 +1,21 @@
+require recipes-kernel/linux/linux-yocto.inc
+
+KBRANCH = "standard/tiny/base"
+LINUX_KERNEL_TYPE = "tiny"
+KCONFIG_MODE = "--allnoconfig"
+
+LINUX_VERSION ?= "3.14-rc8"
+
+KMETA = "meta"
+
+SRCREV_machine ?= "fecc3fd7d31bd93766ff4f0431fecdbbfa4c3a7c"
+SRCREV_meta ?= "3689f99f4d2a051e8d3ff72345a67d4d04a88020"
+
+PV = "${LINUX_VERSION}+git${SRCPV}"
I'm afraid we can't do this in such a simple way. The big issue is that
according to our version comparison functions (in bitbake and in the
package managers), "3.14-rc8" > "3.14". This will then break updates
from package feeds.
The traditional way we've worked around this in times gone by is the
rather ugly:
PV = "3.13+3.14-rc8+git${SRCPV}"
since in this case "3.14" > "3.13+3.14-rc8"
The better way to handle this would be "3.14~rc8" however not all of our
package backends have support for "~" at this point so we can't do that.
There is a long standing bug open about this.
The same issue also applies to the libc-headers recipe.
eeee. Since I don't do package feeds, it's never been a problem for me.
Wouldn't PE bump when I go to 3.14 work as well ? Or does that go away
with the PR sever as well ?
A PE bump would work however we try not to use those, they're a last
resort to correct things we can't otherwise fix. The PR server doesn't
help here although in future I do hope we can teach it about PE too.
I can muck with PV, or we can hold until Linus releases over the
weekend, and I go with the released PVs.
What's your preference ?
Might as well wait over the weekend. If Linus does go for a rc9, we'll
need to merge a version with the modified PV though.
In the meantime I might pull this series into master-next for testing
but it will not get added to master in its current form.
FWIW: I just fixed the PVs of libc-headers and linux-yocto_3.14 to be the
3.13+ variety.
All other comments are also incorporated and the branches repushed.
Outside of needing to override the TC libc headers default of 3.14 (that
I left
in, expecting the full update) in local.conf, all builds continue to
work.
Bruce
Cheers,
Richard
--
_______________________________________________
Openembedded-core mailing list
[email protected]
http://lists.openembedded.org/mailman/listinfo/openembedded-core
