<feed xmlns='http://www.w3.org/2005/Atom'>
<title>buildroot/toolchain, branch 2015.11</title>
<subtitle>OpenPOWER buildroot sources</subtitle>
<id>https://git.raptorcs.com/git/buildroot/atom?h=2015.11</id>
<link rel='self' href='https://git.raptorcs.com/git/buildroot/atom?h=2015.11'/>
<link rel='alternate' type='text/html' href='https://git.raptorcs.com/git/buildroot/'/>
<updated>2015-11-05T22:21:38+00:00</updated>
<entry>
<title>toolchain/external: fix gdbserver install with Linaro 2015.08</title>
<updated>2015-11-05T22:21:38+00:00</updated>
<author>
<name>Trent Piepho</name>
<email>tpiepho@kymetacorp.com</email>
</author>
<published>2015-11-05T20:12:32+00:00</published>
<link rel='alternate' type='text/html' href='https://git.raptorcs.com/git/buildroot/commit/?id=d7a92aa2fbe1cfd13c8d0bf7c7942b2ead747ffe'/>
<id>urn:sha1:d7a92aa2fbe1cfd13c8d0bf7c7942b2ead747ffe</id>
<content type='text'>
In the latest Linaro toolchain, the gdbserver has moved (surprise!)
and is now located side-by-side with the toolchain executables.

This commit adds this path as a new location where to search for a
gdbserver, and while at it wraps the line that has become too long in
the process.

[Thomas: rework commit log according to Yann's suggestion.]

Signed-off-by: Trent Piepho &lt;tpiepho@kymetacorp.com&gt;
Acked-by: "Yann E. MORIN" &lt;yann.morin.1998@free.fr&gt;
Signed-off-by: Thomas Petazzoni &lt;thomas.petazzoni@free-electrons.com&gt;
</content>
</entry>
<entry>
<title>toolchain-external: make extraction idempotent</title>
<updated>2015-11-04T07:31:09+00:00</updated>
<author>
<name>Arnout Vandecappelle</name>
<email>arnout@mind.be</email>
</author>
<published>2015-11-03T23:09:35+00:00</published>
<link rel='alternate' type='text/html' href='https://git.raptorcs.com/git/buildroot/commit/?id=b731dc7bfb9c8ce7be502711f0b44ccab5515f1d'/>
<id>urn:sha1:b731dc7bfb9c8ce7be502711f0b44ccab5515f1d</id>
<content type='text'>
Commit 23ffa7ec first extracts to the toolchain-external build
directory and then moves everything to $(HOST_DIR)/opt/ext-toolchain.
However, this is not idempotent, because moving directories over
existing ones doesn't always work, particularly if the target is on
another device.

Simply remove the destination contents before moving.

Signed-off-by: Arnout Vandecappelle (Essensium/Mind) &lt;arnout@mind.be&gt;
Signed-off-by: Thomas Petazzoni &lt;thomas.petazzoni@free-electrons.com&gt;
</content>
</entry>
<entry>
<title>toolchain/external: use generic extract commands (blackfin case)</title>
<updated>2015-11-03T21:32:03+00:00</updated>
<author>
<name>Yann E. MORIN</name>
<email>yann.morin.1998@free.fr</email>
</author>
<published>2015-10-24T12:48:57+00:00</published>
<link rel='alternate' type='text/html' href='https://git.raptorcs.com/git/buildroot/commit/?id=d0185582d0e24281640579cb3ea1d67e519febae'/>
<id>urn:sha1:d0185582d0e24281640579cb3ea1d67e519febae</id>
<content type='text'>
The backfin toolchains come in two archives.

We extract the first (main) archive using the generic extract commands,
while the second is extracted as a post-extract hook.

Signed-off-by: "Yann E. MORIN" &lt;yann.morin.1998@free.fr&gt;
Cc: Thomas Petazzoni &lt;thomas.petazzoni@free-electrons.com&gt;
Cc: Vicente Olivert Riera &lt;Vincent.Riera@imgtec.com&gt;
Reviewed-by: Romain Naour &lt;romain.naour@openwide.fr&gt;
Signed-off-by: Thomas Petazzoni &lt;thomas.petazzoni@free-electrons.com&gt;
</content>
</entry>
<entry>
<title>toolchain/external: use generic extract commands (!blackfin case)</title>
<updated>2015-11-03T21:30:08+00:00</updated>
<author>
<name>Yann E. MORIN</name>
<email>yann.morin.1998@free.fr</email>
</author>
<published>2015-10-24T12:48:56+00:00</published>
<link rel='alternate' type='text/html' href='https://git.raptorcs.com/git/buildroot/commit/?id=23ffa7ecf763206f570689a0644784d555948e67'/>
<id>urn:sha1:23ffa7ecf763206f570689a0644784d555948e67</id>
<content type='text'>
Now that packages can provide a list of files to be excluded when
extracting their archive, downloaded external toolchains are no longer
special in this respect.

Still, those toolchains are currently extracted directly into their
final location, $(HOST_DIR)/opt/ext-toolchain/ which means we still
need a custom extract command.

Except, we don't really need it: we can just move the toolchain, after
it's been extracted by the generic extract command, with a post-extract
hook.

This means that:

  - we now extract the toolchain with the generic extract command,

  - the toolchain is thus extracted into $(@D) ,

  - fixup commands are run against $(@D), as a post-extract hook,
    instead of against $(HOST_DIR)/opt/ext-toolchain ,

  - once this is done, we move $(@D)/* into the final location with a
    new post-extract hook.

Note: the blackfin case is special, and will be handled in a follow-up
patch.

[Thomas: register the TOOLCHAIN_EXTERNAL_FIXUP_CMDS only for the Arago
case, add some additional comments in the code about why we're moving
the toolchain around.]

Signed-off-by: "Yann E. MORIN" &lt;yann.morin.1998@free.fr&gt;
Cc: Thomas Petazzoni &lt;thomas.petazzoni@free-electrons.com&gt;
Cc: Vicente Olivert Riera &lt;Vincent.Riera@imgtec.com&gt;
Reviewed-by: Romain Naour &lt;romain.naour@openwide.fr&gt;
Reviewed-by: Vicente Olivert Riera &lt;Vincent.Riera@imgtec.com&gt;
Tested-by: Vicente Olivert Riera &lt;Vincent.Riera@imgtec.com&gt;
Signed-off-by: Thomas Petazzoni &lt;thomas.petazzoni@free-electrons.com&gt;
</content>
</entry>
<entry>
<title>toolchain-external/blackfin: drop --hard-dereference</title>
<updated>2015-11-03T21:21:16+00:00</updated>
<author>
<name>Yann E. MORIN</name>
<email>yann.morin.1998@free.fr</email>
</author>
<published>2015-10-24T12:48:54+00:00</published>
<link rel='alternate' type='text/html' href='https://git.raptorcs.com/git/buildroot/commit/?id=dbdc241d6a476855f6548ba282dfd18382245745'/>
<id>urn:sha1:dbdc241d6a476855f6548ba282dfd18382245745</id>
<content type='text'>
Currently, for the blackfin external toolchains, we tell tar to
extract files with the --hard-dereference. However, --hard-dereference
is only meaningful when creating an archive, not when extracting
it. Therefore, let's drop this option.

[Thomas: rework commit title and commit log, after some suggestions
from Arnout.]

Signed-off-by: "Yann E. MORIN" &lt;yann.morin.1998@free.fr&gt;
Cc: Thomas Petazzoni &lt;thomas.petazzoni@free-electrons.com&gt;
Cc: Vicente Olivert Riera &lt;Vincent.Riera@imgtec.com&gt;
Reviewed-by: Arnout Vandecappelle (Essensium/Mind) &lt;arnout@mind.be&gt;
Signed-off-by: Thomas Petazzoni &lt;thomas.petazzoni@free-electrons.com&gt;
</content>
</entry>
<entry>
<title>toolchain/external: bump Linaro AArch64 to 2015.08</title>
<updated>2015-11-03T21:02:58+00:00</updated>
<author>
<name>Yann E. MORIN</name>
<email>yann.morin.1998@free.fr</email>
</author>
<published>2015-10-26T23:49:41+00:00</published>
<link rel='alternate' type='text/html' href='https://git.raptorcs.com/git/buildroot/commit/?id=24bfce0ebc18e6ed7dda868250455a3566c0c116'/>
<id>urn:sha1:24bfce0ebc18e6ed7dda868250455a3566c0c116</id>
<content type='text'>
That toolchain is built for an x86_64 host, so we make it available only
for x86_64, and we keep the old 2014.09 toolchain for x86 hosts.

To avoid dealing with legacy symbols and introduce versioned options,
we reuse the same symbol for both toolchains. Thanks to the different
depednencies (on the host), we can give them different prompts and
different help texts.

Signed-off-by: "Yann E. MORIN" &lt;yann.morin.1998@free.fr&gt;
Cc: Thomas Petazzoni &lt;thomas.petazzoni@free-electrons.com&gt;
Cc: Arnout Vandecappelle &lt;arnout@mind.be&gt;
Signed-off-by: Thomas Petazzoni &lt;thomas.petazzoni@free-electrons.com&gt;
</content>
</entry>
<entry>
<title>toolchain/external: bump Linaro ARMEB to 2015.08</title>
<updated>2015-11-03T21:02:11+00:00</updated>
<author>
<name>Yann E. MORIN</name>
<email>yann.morin.1998@free.fr</email>
</author>
<published>2015-10-26T23:49:40+00:00</published>
<link rel='alternate' type='text/html' href='https://git.raptorcs.com/git/buildroot/commit/?id=997ef60d902a9a50a401c9f0e597339ebd6c45c7'/>
<id>urn:sha1:997ef60d902a9a50a401c9f0e597339ebd6c45c7</id>
<content type='text'>
That toolchain is built for an x86_64 host, so we make it available only
for x86_64, and we keep the old 2014.09 toolchain for x86 hosts.

To avoid dealing with legacy symbols and introduce versioned options,
we reuse the same symbol for both toolchains. Thanks to the different
depednencies (on the host), we can give them different prompts and
different help texts.

[Thomas: tweak Config.in help text to actually match this toolchain
instead of being a wrong copy/paste from the old Linaro toolchain for
ARMeb.]

Signed-off-by: "Yann E. MORIN" &lt;yann.morin.1998@free.fr&gt;
Cc: Thomas Petazzoni &lt;thomas.petazzoni@free-electrons.com&gt;
Cc: Arnout Vandecappelle &lt;arnout@mind.be&gt;
Signed-off-by: Thomas Petazzoni &lt;thomas.petazzoni@free-electrons.com&gt;
</content>
</entry>
<entry>
<title>toolchain/external: bump Linaro ARM to 2015.08</title>
<updated>2015-11-03T20:52:44+00:00</updated>
<author>
<name>Yann E. MORIN</name>
<email>yann.morin.1998@free.fr</email>
</author>
<published>2015-10-26T23:49:39+00:00</published>
<link rel='alternate' type='text/html' href='https://git.raptorcs.com/git/buildroot/commit/?id=9b3b98bf5a0f5953ff86b59d8d76e3386d3be4af'/>
<id>urn:sha1:9b3b98bf5a0f5953ff86b59d8d76e3386d3be4af</id>
<content type='text'>
That toolchain is built for an x86_64 host, so we make it available only
for x86_64, and we keep the old 2014.09 toolchain for x86 hosts.

To avoid dealing with legacy symbols and introduce versioned options,
we reuse the same symbol for both toolchains. Thanks to the different
depednencies (on the host), we can give them different prompts and
different help texts.

[Thomas: s/eglibc/glibc/ as noticed by Baruch.]

Signed-off-by: "Yann E. MORIN" &lt;yann.morin.1998@free.fr&gt;
Cc: Thomas Petazzoni &lt;thomas.petazzoni@free-electrons.com&gt;
Cc: Arnout Vandecappelle &lt;arnout@mind.be&gt;
Signed-off-by: Thomas Petazzoni &lt;thomas.petazzoni@free-electrons.com&gt;
</content>
</entry>
<entry>
<title>toolchain: add 4.3.x choice for headers</title>
<updated>2015-11-02T19:56:22+00:00</updated>
<author>
<name>Vicente Olivert Riera</name>
<email>Vincent.Riera@imgtec.com</email>
</author>
<published>2015-11-02T14:56:51+00:00</published>
<link rel='alternate' type='text/html' href='https://git.raptorcs.com/git/buildroot/commit/?id=aef2df82e1eace53fef0c998ec360299a5135b9b'/>
<id>urn:sha1:aef2df82e1eace53fef0c998ec360299a5135b9b</id>
<content type='text'>
Signed-off-by: Vicente Olivert Riera &lt;Vincent.Riera@imgtec.com&gt;
Reviewed-by: "James Knight" &lt;james.knight@rockwellcollins.com&gt;
Signed-off-by: Thomas Petazzoni &lt;thomas.petazzoni@free-electrons.com&gt;
</content>
</entry>
<entry>
<title>toolchain/wrapper: fix potential bug in foreach loop</title>
<updated>2015-10-25T22:01:48+00:00</updated>
<author>
<name>Yann E. MORIN</name>
<email>yann.morin.1998@free.fr</email>
</author>
<published>2015-10-21T21:21:07+00:00</published>
<link rel='alternate' type='text/html' href='https://git.raptorcs.com/git/buildroot/commit/?id=32e2636c51e409f108f24174257ea4867ec6d74d'/>
<id>urn:sha1:32e2636c51e409f108f24174257ea4867ec6d74d</id>
<content type='text'>
In Makefile, the comma ',' is used to separate the arguments passed to
functions, so we should not be allowed to use straight commas in strings
we want to expand.

For the toolchain wrapper, we need to transform a list:
    -mfoo -mbar -mbuz

into something acceptable for a C array assignment:
    "-mfoo", "-mbar", "-mbuz",

So, we use a $(foreach ...) loop for that. However, we do have a
straight comma in there.

It does not cause any issue in practice, since $(foreach) is a make
builtin function that accepts three and only three parameters.

However, this is not sane.

Change the straight comma to the usual $(comma) expansion, like we would
do for a call to any other function.

At the same time, make the code a bit easier to read, by first creating
the transformed list, and then creating the define.

Signed-off-by: "Yann E. MORIN" &lt;yann.morin.1998@free.fr&gt;
Cc: Arnout Vandecappelle &lt;arnout@mind.be&gt;
Acked-by: Arnout Vandecappelle (Essensium/Mind) &lt;arnout@mind.be&gt;
Tested-by: Arnout Vandecappelle (Essensium/Mind) &lt;arnout@mind.be&gt;
Signed-off-by: Peter Korsgaard &lt;peter@korsgaard.com&gt;
</content>
</entry>
</feed>
