Can't build Bazel with Intel compiler (2)

[mj-harvey]

Following on from case #662
Commenting out the static_asserts() that break the build, the compile then breaks at:

'''
$ ./compile.sh
INFO: You can skip this first step by providing a path to the bazel binary as second argument:
INFO: ./compile.sh build /path/to/bazel
🍃 Building Bazel from scratch............
🍃 Building Bazel with Bazel.
.Extracting Bazel installation...
...........
WARNING: Sandboxed execution is not supported on your system and thus hermeticity of actions cannot be guaranteed. See http://bazel.io/docs/bazel-user-manual.html#sandboxing for more information. You can turn off this warning via --ignore_unsupported_sandboxing.
INFO: Found 1 target...
INFO: From Compiling src/main/cpp/blaze_util.cc:

Error: A license for CCompL could not be obtained (-1,359,2).

Is your license file in the right location and readable?
The location of your license file should be specified via
the $INTEL_LICENSE_FILE environment variable.

License file(s) used were (in this order):
1. Trusted Storage
** 2. /raid-1/apps/intel/2013/composer_xe_2013.0.079/Licenses
** 3. /opt/intel/licenses
** 4. /Users/Shared/Library/Application Support/Intel/Licenses
* 5. /raid-1/apps/intel/2013/composer_xe_2013.0.079/bin/intel64/.lic

Please visit http://software.intel.com/sites/support/ if you require technical assistance.

icpc: error #10052: could not checkout FLEXlm license
ERROR: /raid-1/apps/bazelbuild/bazel/src/main/cpp/BUILD:7:1: C++ compilation of rule '//src/main/cpp:blaze_util' failed: icpc failed: error executing command /apps/intel/2013/bin/icpc -U_FORTIFY_SOURCE '-D_FORTIFY_SOURCE=1' -fstack-protector -Wall -Wunused-but-set-parameter -Wno-free-nonheap-object -fno-omit-frame-pointer '-std=c++0x' -DBLAZE_OPENSOURCE ... (remaining 24 argument(s) skipped): com.google.devtools.build.lib.shell.BadExitStatusException: Process exited with status 1.
Target //src:bazel failed to build
Use --verbose_failures to see the command lines of failed build steps.
INFO: Elapsed time: 8.513s, Critical Path: 2.18s

Building output/bazel
'''

You are running the compiler in such a way that the value of the required environment variable INTEL_LICENSE_FILE is not being propagated. the compiler thus can't find its license server.

[ulfjack]

Well, we intentionally don't propagate environment variables from outside the bazel binary; generally speaking, they can cause builds to be non-deterministic and non-reproducible. However, we probably want to add a mechanism to poke specific holes into that policy to allow things such as license files (it's not just the Intel C++ compiler). As a workaround, you can change your crosstool to use a wrapper script that sets the env variable and then calls through to the actual binary.

If we make C++-specific changes here, we'll want to make sure that we can also handle the Apple C++ toolchain requirements for iOS and OSX builds.

[GearFear]

Is there a work-around to solve the issue?

[ulfjack]

You should try setting --action_env=INTEL_LICENSE_FILE on the command line (or in your bazelrc file). It may not work yet, since not all actions handle --action_env correctly (#3320) in which case I'll prioritize getting it to work for the relevant C++ actions.

[GearFear]

I tried - unfortunately, it did not help :( I really need to build TensorFlow with ICC, but I can't deal with this issue.

[mj-harvey]

Wrapper icc in a shell script that sets the envvar appropriately #!/bin/sh export INTEL_LICENSE_FILE=xxx /path/to/icc.bin "$*" or similar.

…

On 21 July 2017 at 09:55, GearFear ***@***.***> wrote: I tried - unfortunately, it did not help :( I really need to build TensorFlow with ICC, but I can't deal with this issue. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#663 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AHh-DU1T3Yotoe-zVBdsCP8Nrm-mjHZ3ks5sQGdogaJpZM4GrnZR> .

-- <https://www.acellera.com> <https://twitter.com/acellera?ref_src=twsrc%5Egoogle%7Ctwcamp%5Eserp%7Ctwgr%5Eauthor> <https://www.youtube.com/user/acelleralive> <https://www.linkedin.com/company/acellera> <https://www.acellera.com/md-simulation-blog-news/> [image: skype:acellera] <http://is.gd/1eXkbS>

[hlopko]

Is this issue still valid?

[euler89]

Can you confirm this is still an issue?

[mj-harvey]

It turned out to be trivially worked-around by shimming the compiler in a wrapper.