Merge lp:~zhenqiang-chen/gcc-linaro/linaro-local-multilib into lp:gcc-linaro

Proposed by Zhenqiang Chen
Status: Merged
Merge reported by: Christophe Lyon
Merged at revision: not available
Proposed branch: lp:~zhenqiang-chen/gcc-linaro/linaro-local-multilib
Merge into: lp:gcc-linaro
To merge this branch: bzr merge lp:~zhenqiang-chen/gcc-linaro/linaro-local-multilib
Reviewer Review Type Date Requested Status
Christophe Lyon Approve
Review via email: mp+201877@code.launchpad.net

Description of the change

Linaro local patch form armv4t multilib support.

To post a comment you must log in.
Revision history for this message
Linaro Toolchain Builder (cbuild) wrote :

cbuild has taken a snapshot of this branch at r122152 and queued it for build.

The diff against the ancestor r122151 is available at:
 http://cbuild.validation.linaro.org/snapshots/gcc-linaro-4.8+bzr122152~zhenqiang-chen~linaro-local-multilib.diff

and will be built on the following builders:
 a9-builder a9hf-builder armv5-builder i686 x86_64 xaarch64 xaarch64_bare xcortexa15hf

You can track the build queue at:
 http://cbuild.validation.linaro.org/helpers/scheduler

cbuild-snapshot: gcc-linaro-4.8+bzr122152~zhenqiang-chen~linaro-local-multilib
cbuild-ancestor: lp:gcc-linaro+bzr122151
cbuild-state: check

Revision history for this message
Christophe Lyon (christophe-lyon) wrote :

All targets are clean:

What is the purpose of the comma and space variables in t-mlibs?
Could you make sure to avoid blank line changes?

Could you state why is this patch needed? (why isn't the modification of config.gcc and t-mlibs sufficient?)

review: Needs Information
Revision history for this message
Zhenqiang Chen (zhenqiang-chen) wrote :

On 17 January 2014 06:27, Christophe Lyon <email address hidden> wrote:
> Review: Needs Information
>
> All targets are clean:
>
> What is the purpose of the comma and space variables in t-mlibs?

comma and space variables are "useless" for current multilib options.

When there are lots of multilib options, we can use
--with-multilib-list to select what we want to have for a special
build.
e.g.
ifneq (,$(findstring armv5t,$(subst $(comma),$(space),$(with_multilib_list))))
...

> Could you make sure to avoid blank line changes?

The last blank line should be removed. But other blank lines are
expected to separate different MICROs. This is similar with thoese in
config/arm/t-aprofile

> Could you state why is this patch needed? (why isn't the modification of config.gcc and t-mlibs sufficient?)

The patch comes from a customer's requirement to build U-boot with
Linaro toolchain binaries. By default, Linaro toolchain binaries are
for Cortex-A9, THUMB and with HARD fp. These options can not fit in
u-boot build, which only supports ARMV4T. To meet the requirement in
one toolchain, we need multilib support. So we change config.gcc and
add t-mlibs.

To align with Linaro image, Linaro toolchain should support multiarch
sysroot layout. But multiarch is conflict with multilib on where to
find the header files and where to link to libraries. So we change
gcc.c and incpath.c to make sure: If there is multilib option
(armv4T), try to search multilib dir first, then multiarch dir, then
default dir. Otherwise search multiarch dir first, then default dir.

Revision history for this message
Christophe Lyon (christophe-lyon) wrote :

Maybe we should think about possible ways to upstream it.

review: Approve
Revision history for this message
Christophe Lyon (christophe-lyon) wrote :

Committed as r206704.

Subscribers

People subscribed via source and target branches

to all changes: