Subzero: Add a "make check-presubmit" target.

Subzero: Add a "make check-presubmit" target.

BUG= none
R=jpp@chromium.org

Committed: https://gerrit.chromium.org/gerrit/gitweb?p=native_client/pnacl-subzero.git;a=commit;h=6c4ad849c747eedda1e1e7568781f95cd1176c49

[Jim Stichnoth, 2015-11-16 05:43:32]

Description was changed from

==========
Subzero: Add a "make check-presubmit" target.

BUG= none
==========

to

==========
Subzero: Add a "make check-presubmit" target.

BUG= none
==========

[Jim Stichnoth, 2015-11-16 05:43:32]

stichnot@chromium.org changed reviewers:
+ jpp@chromium.org, kschimpf@google.com, sehr@chromium.org

[John, 2015-11-16 14:03:23]

lgtm

https://codereview.chromium.org/1452553002/diff/40001/Makefile.standalone
File Makefile.standalone (right):

https://codereview.chromium.org/1452553002/diff/40001/Makefile.standalone#new...
Makefile.standalone:421: SPECBUILDONLY := false
\o/

https://codereview.chromium.org/1452553002/diff/40001/Makefile.standalone#new...
Makefile.standalone:437: make -f Makefile.standalone -j$(NPROCS) \
Too aggressive? Maybe -j$(NPROCS - 5), or something?

https://codereview.chromium.org/1452553002/diff/40001/Makefile.standalone#new...
Makefile.standalone:451: echo Success
Maybe run some light-weight benchmarks as a smoke test? mcf, gcc, and bzip (for
arm) would require an extra minute.

[Jim Stichnoth, 2015-11-16 18:17:59]

https://codereview.chromium.org/1452553002/diff/40001/Makefile.standalone
File Makefile.standalone (right):

https://codereview.chromium.org/1452553002/diff/40001/Makefile.standalone#new...
Makefile.standalone:421: SPECBUILDONLY := false
On 2015/11/16 14:03:23, John wrote:
> \o/

Is there a complaint or suggestion here?

https://codereview.chromium.org/1452553002/diff/40001/Makefile.standalone#new...
Makefile.standalone:437: make -f Makefile.standalone -j$(NPROCS) \
On 2015/11/16 14:03:23, John wrote:
> Too aggressive? Maybe -j$(NPROCS - 5), or something?

FWIW, I always run -j32 on my z620 and -j4 on my laptop, with no problems
(insofar as building on one's laptop is "no problem").  Also, lit defaults to
using all processors.  Pretty sure that's common practice across the group.

That said, the right thing to do would be to "make -j32 check-presubmit" and get
the sub-make command to be passed the -j32.  But the interwebs seem to indicate
the /proc/cpuinfo trick as the "best" approach.  Any other ideas?

https://codereview.chromium.org/1452553002/diff/40001/Makefile.standalone#new...
Makefile.standalone:451: echo Success
On 2015/11/16 14:03:23, John wrote:
> Maybe run some light-weight benchmarks as a smoke test? mcf, gcc, and bzip
(for
> arm) would require an extra minute.

That's a good idea.

However, for now we have a problem since szbuild_spec2k.py creates binaries
named <component>.gcc.opt.arm32 but run_all.sh expects <component>.gcc.opt.arm .
 I'll tweak that in the script, but be aware of that when you go about building
and running on the ARM hardware.

[Jim Stichnoth, 2015-11-16 20:48:01]

Committed patchset #8 (id:140001) manually as
6c4ad849c747eedda1e1e7568781f95cd1176c49 (presubmit successful).