linux/poky.git/meta/classes/package.bbclass, branch 1.4_M3.rc1

package.bbclass: drop unnecessary and broken packages.remove()

2013-01-21T10:22:08+00:00

(From OE-Core rev: 5480371d0473482437ce4350e6593f006101ef40) Signed-off-by: Christopher Larson Signed-off-by: Richard Purdie

package.bbclass: Fix shlibs cross package-arch contamination

2013-01-20T12:58:58+00:00

I found that if I had two packages architectures built (say core2 and i586), that the shlibs data from i586 was being used in the core2 build. This lead to odd dependency issues if the i586 build was out of date and core2 was being targetted. When loading shlibs files, the last loaded wins so we need to iterate the lists from least to strongest weighting, not the other way around. With this patch applied, the contamination issue is resolved. (From OE-Core rev: acd7b1c2ee9f3ff84c45a723dba5e0bc5f6a100c) Signed-off-by: Richard Purdie

package.bbclass: replace AUTOINC directly in PKGV

2013-01-20T12:58:57+00:00

b27b438221e16ac3df6ac66d761b77e3bd43db67 introduced a regression when PKGV is modified by the recipe. E.g. when having something like | PV = "1.7.33+gitr${SRCPV}" | PKGV = "1.7.33+gitr${GITPKGV}" | RDEPENDS_${PN}-subpkg += "${PN} (= ${EXTENDPKGV})" | inherit gitpkgv The resulting core package has | Version: 1.7.33+git234+cd01e8c-r11 and because the subpackage has | Depends: vdr (= 1.7.33+git0+cd01e8c18d494150730214f033f84e5282ce7639-r11) image build fails with | * satisfy_dependencies_for: Cannot satisfy the following dependencies ... | * vdr (= 1.7.33+git0+cd01e8c18d494150730214f033f84e5282ce7639-r11) Patch avoids the 'PKGV = "cleanup(${PV})"' operation and replaces AUTOINC directly within PKGV. This might introduce a regression too when 'AUTOINC' is added to PKGV intentionally but impact is small and can be fixed trivially by using another string. (From OE-Core rev: 6e46664085fa49bc204fc944ca23ed6383e2ec51) Signed-off-by: Enrico Scholz Cc: Constantin Musca Signed-off-by: Richard Purdie

package.bbclass: Skip testing "packages" a second time.

2013-01-07T11:09:44+00:00

Since the "packages" variable is already tested a few lines above in the same routine: if len(packages) < 1: bb.debug(1, "No packages to build, skipping do_package") return there is no point testing it again, and the error message doesn't even refer to "packages", anyway. (From OE-Core rev: 371fbacff47828cfff99f1152f9e269e27c474da) Signed-off-by: Robert P. J. Day Signed-off-by: Saul Wold Signed-off-by: Richard Purdie

package.bbclass: don't prepend MLPREFIX to LOCALEBASEPN

2013-01-07T11:09:37+00:00

- all the recipes that overwrite LOCALEBASEPN must consider also the MLPREFIX - if the LOCALEBASEPN variable is not overwritten then it will have the correct prefix (LOCALEBASEPN ??= "${PN}") (From OE-Core rev: 855d64a3091c5358e2e1a7f4247929515b068708) Signed-off-by: Constantin Musca Signed-off-by: Saul Wold Signed-off-by: Richard Purdie

multilib: fix allarch/kernel/module-base multilib issues

2012-12-31T09:43:25+00:00

- skip the non-packagegroup allarch recipes in multilib_virtclass_handler - extend PROVIDES/RPROVIDES for allarch recipes which are not packagegroups - use variants from MULTILIB_GLOBAL_VARIANTS (lib32 lib64 libx32) to create additional pkgdata files for multilib allarch: ${pkgdatadir}/${variant}-${PN} and ${pkgdatadir}/runtime/${variant}-${pkg} - use variants from MULTILIB_VARIANTS to create additional pkgdata files for multilib kernel/module-base recipes - add a sanity check to determine if the current multilib is in MULTILIB_GLOBAL_VARIANTS [YOCTO #2918] [YOCTO #3440] [YOCTO #3565] [YOCTO #3568] (From OE-Core rev: bc4da2573dfb59ea2fc4af359701818df20f7663) Signed-off-by: Constantin Musca Signed-off-by: Saul Wold Signed-off-by: Richard Purdie

package.bbclass: prepend MLPREFIX to LOCALEBASEPN

2012-12-11T15:58:35+00:00

We need to prepend MLPREFIX to LOCALEBASEPN in order to fully enable multilib [YOCTO #3440] (From OE-Core rev: 91fe09a9a8b317d27145efff4201f1138d6a45d4) Signed-off-by: Constantin Musca Signed-off-by: Richard Purdie

package.bbclass: make do_split_packages handle non-existent root directories

2012-12-11T15:54:36+00:00

This function has different behaviour if the split directory doesn't exist depending on the recursive argument: non-recursive uses os.listdirs which throws an exception, recursive uses os.walk which doesn't. do_split_packages should silently handle non-existent directories because it's mainly used for plugin directories, which may end up being empty though changing the distro configuration (for example, connman without wifi distro feature). So, add an early exit if the split root doesn't exist. (From OE-Core rev: 937101e3fdd6afd00f6f8a8be411a67110c4ae78) Signed-off-by: Ross Burton Signed-off-by: Saul Wold Signed-off-by: Richard Purdie

prs: use the PRServer to replace the BB_URI_LOCALCOUNT functionality

2012-12-06T12:31:16+00:00

- adapt the PRServer to generate an incremental number for every revision set associated with a (PN, PV) - use one incrementing value rather than several - PV becomes 0.1+gitAUTOINC+deadbeefdecafbad_decafbaddeadbeef - if PV contains 'AUTOINC' set PKGV to the same thing with AUTOINC replaced with the "auto PR" value - if the PRServer isn't running replace AUTOINC with 0 (no package feeds, but usual image creation would work) - use the PR table with "AUTOINC-${PV-without-SRCREVs}" as key - modify prexport/primport to address the generated srcrev entries - this patch addresses the following proposal: http://lists.linuxtogo.org/pipermail/bitbake-devel/2012-November/003878.html (From OE-Core rev: b27b438221e16ac3df6ac66d761b77e3bd43db67) Signed-off-by: Constantin Musca Signed-off-by: Saul Wold Signed-off-by: Richard Purdie

package.bbclass: Adapt debugsrc code to workdir layout changes

2012-11-21T09:10:02+00:00

With the change to WORKDIR layout, splutting PN and PV into two directories, the debugsrc splutting code layout became suboptimal. This changes things to include the information as it was before. Ideally this code would be written to more generically support other layouts buts it not clear that the tools would even support that right now so this is the best immediate fix. (From OE-Core rev: 432cfbb403f0e864d1fad383c2bbb6f9bdb80770) Signed-off-by: Richard Purdie