This reverts commit aa66ef6598c84231577d139ec7be413e73fac2b1 since
bdwgc-native fails to build after it. Anything which runs with a
sub-configure will fail after this change. It therefore needs
rethinking.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

Fix autotools.bbclass to use relative paths for acpaths instead of
absolute ones. Since absolute paths may cause potential autoreconf
error like:

    Can't exec "/bin/sh": Argument list too long ...

This error occurs while building coreutils with long TMPDIR, because
it has bunch of m4 files need to be expanded.

[YOCTO #2766]

Signed-off-by: Wenzong Fan <wenzong.fan@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

For reasons that are now shrouded in obscurity, autotools.bbclass
has long contained a special heuristic to avoid attempting to run
autoreconf when building autoconf or automake themselves.  However,
the wildcard test against PN which is used there is problematic when
trying to build another package whose name happens to start with
"autoconf", and in any case it is silly to do this test at runtime
for every package.  The individual recipes for autoconf and automake
can just as easily suppress the behaviour that they don't want by
providing a custom do_configure() method which just runs configure.

Signed-off-by: Phil Blundell <pb@pbcl.net>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

Unfortunately whilst rerunning configure and make against a project will mostly
work there are situations where it does not correctly do the right thing.

In particular, eglibc and gcc will fail out with errors where settings
do not match a previously built configuration. It could be argued they are
broken but the situation is what it is. There is the possibility of more subtle
errors too.

This patch adds removal of the build directory (${B}) when configure is
rerunning, the sstate checksum for do_configure has changed and ${S} != ${B}.
We could simply use a stamp but saving out the previous configuration checksum
adds some data at no real overhead.

If we find there are things where we want to disable this behaviour with
CONFIGURESTAMPFILE = "" in the recipe, or users could disable it globally.

[YOCTO #2774]
[YOCTO #2848]

This is particularly helpful for eglibc and gcc which use split builds by default and
are a particular source of reconfigure type problems.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Signed-off-by: Saul Wold <sgw@linux.intel.com>

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

When configure fails, it usually says "see config.log" yet nobody ever shares
the config.log file meaning the person trying to help invariably has to ask
for more information.

This patch dumps all the config.log files into the main bitbake log files when
configure fails, meaning all the information is present to help someone debug
such failures. It does make the log rather larger but this is preferable to
not having enough information in most cases.

[YOCTO #2463]

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

Currently we copy the aclocal directory to the build so that autotools
doesn't see .m4 files disappear when its processing them. This can happen
if for example, package X is being rebuilt at the same time as Y and it
gets uninstalled from sstate (assuming there are no dependencies between
X and Y). This code making the copy was added to avoid races but introduces
a race of its own, namely that the files can disappear during the copy.

This patch adds a cp-noerror script which silently ignores such errors
and gives the behaviour we need in this case. It hence fixes issues which
crop up for users and the autobuilder occasionally.

[YOCTO #2485]

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

-nativesdk recipes should not have dependency to lib-cross, which is
never used. This unnecessary dependency would result different task hash
values in sstate for different MACHINE settings.

Signed-off-by: Lianhao Lu <lianhao.lu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

Using "1" with getVar is bad coding style and "True" is preferred.
This patch is a sed over the meta directory of the form:

sed \
 -e 's:\(\.getVar([^,()]*, \)1 *):\1True):g' \
 -e 's:\(\.getVarFlag([^,()]*, [^,()]*, \)1 *):\1True):g' \
 -i `grep -ril getVar *`

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

We need to ensure any aclocal-copy directory is removed before we s
earch for .m4 files, else the locations .m4 files are found from can
vary depending on whether its the first or second time we run configure.

Clearing any existing aclocal-copy directory before we start resolves
this issue and makes builds deterministic again.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>