32 bit build...

[Crayder]

My laptop is 32 bit and all of the releases are 64 bit.

So basically I'm requesting a 32 bit .deb build.

[meteorcloudy]

Can you try to compile Bazel from source?

$ git clone https://github.com/bazelbuild/bazel.git
$ cd bazel
$ ./compile.sh

This will create a bazel binary in bazel-bin/src/bazel.
http://www.bazel.io/docs/install.html#h2

[Crayder]

I know how to compile... And I already tried. It just wouldn't.

[meteorcloudy]

Can you show me the error you got when running ./compile.sh?

[Crayder]

No, I won't have my laptop back for a few days (probably at least a week, maybe two). I don't recall what they were either.

[laszlocsomor]

Bazel only supports 64-bit platforms. Unfortunately we can't help with building it on 32-bit platforms.

[jirkadanek]

If there is a 32bit compatible Remote Executor (https://docs.bazel.build/versions/2.0.0/remote-execution.html), it might be possible to build and test on 32 bit with bazel that way.

[jiridanek]

As per this issue, Bazel project does not release binaries for i686 and as per #8652 the binaries are not expected to work on RHEL6x. And they indeed do not work:

[root@rhel6x ~]# bazel
/usr/local/bin/bazel: /usr/lib64/libstdc++.so.6: version `CXXABI_1.3.5' not found (required by /usr/local/bin/bazel)
/usr/local/bin/bazel: /usr/lib64/libstdc++.so.6: version `GLIBCXX_3.4.18' not found (required by /usr/local/bin/bazel)
/usr/local/bin/bazel: /usr/lib64/libstdc++.so.6: version `GLIBCXX_3.4.14' not found (required by /usr/local/bin/bazel)
/usr/local/bin/bazel: /usr/lib64/libstdc++.so.6: version `GLIBCXX_3.4.15' not found (required by /usr/local/bin/bazel)
/usr/local/bin/bazel: /usr/lib64/libstdc++.so.6: version `GLIBCXX_3.4.19' not found (required by /usr/local/bin/bazel)
/usr/local/bin/bazel: /lib64/libc.so.6: version `GLIBC_2.14' not found (required by /usr/local/bin/bazel)
/usr/local/bin/bazel: /lib64/libc.so.6: version `GLIBC_2.15' not found (required by /usr/local/bin/bazel)
/usr/local/bin/bazel: /lib64/libc.so.6: version `GLIBC_2.17' not found (required by /usr/local/bin/bazel)

Let me drop here notes about getting Bazel on RHEL6i. I tried first running the binaries under qemu-x86_64 and then I tried remote execution. Remote execution seems to work and solves my goal of performning RHEL6i builds with Bazel. I was not able to get the qemu-x86-64 solution to work.

Remote execution

https://docs.bazel.build/versions/master/remote-execution.html

This looks to be the way.

Scoot

There is Go toolchain for i686, awesome! When I compiled the project, I hit a few issues, the last of which looked hard to fix, and I got Buildfarm working next, so I did not investigate.

• explicitly report missing dependencies in the environment, e.g. nc twitter/scoot#481
• setup-cloud-scoot fails to run and error is not clearly given twitter/scoot#480
• Missing Bazel 'capabilities' endpoint twitter/scoot#456

Scoot does not work for me.

Buildfarm

That worked fine. I built the _dist.jars on another machine, shipped them onto rhel6i machine, together with sample configs, and was able to start remote build of Buildfarm itself. I am still watching for issues.

jdanek@nixos ~/repos/bazel-buildfarm (git)-[master] % bazel build //src/main/java/build/buildfarm:buildfarm-server_deploy.jar
jdanek@nixos ~/repos/bazel-buildfarm (git)-[master] % bazel build //src/main/java/build/buildfarm:buildfarm-operationqueue-worker_deploy.jar

Seems to be working fine so far.

Qemu

Compile qemu, as per https://wiki.qemu.org/Hosts/Linux

yum install -y git glib2-devel libfdt-devel pixman-devel zlib-devel
https://download.qemu.org/qemu-2.12.1.tar.xz
tar xf qemu-2.12.1.tar.xz
cd qemu-2.12.1
./configure --target-list=x86_64-linux-user
make install

Register qemu-x86_64 as the interpreter for 64bit binaries, from https://k3a.me/running-x86_64-amd64-executables-on-ppc64le-ibm-power-9-gentoo/

echo ':x86_64:M::\x7fELF\x02\x01\x01\x00\x00\x00\x00\x00\x00\x00\x00\x00\x02\x00\x3e\x00:\xff\xff\xff\xff\xff\xfe\xfe\xfc\xff\xff\xff\xff\xff\xff\xff\xff\xfe\xff\xff\xff:/usr/local/bin/qemu-x86_64:OC' > /proc/sys/fs/binfmt_misc/register

Copy 64bit libraries from Ubuntu Trusty docker image (docker --rm -it library/ubuntu:trusty) onto the rhel6i system.

Taking libraries from a rhel6x does not help (glibc and libstdc++ are too old there) and with libraries from rhel7x I get a segfault. So Trusty it is.

[rhel6i] mkdir -p /lib64 /lib/x86_64-linux-gnu /usr/lib/x86_64-linux-gnu/ 

[trusty] for lib in /usr/lib/x86_64-linux-gnu/libstdc++.so.6 /lib/x86_64-linux-gnu/libm.so.6 /lib/x86_64-linux-gnu/librt.so.1 /lib/x86_64-linux-gnu/libdl.so.2 /lib/x86_64-linux-gnu/libpthread.so.0 /lib/x86_64-linux-gnu/libgcc_s.so.1 /lib/x86_64-linux-gnu/libc.so.6 /lib64/ld-linux-x86-64.so.2
    do scp $lib root@rhel6i:/${lib}
done

I saw "cannot mmap brk" error from qemu, apparently caused by running out of RAM, since adding more memory to the rhel6i VM solved the problem.

qemu 0.14.0, 1.5.1

[root@rhel6i qemu-1.5.3]# bazel
qemu: Unsupported syscall: 202
qemu: Unsupported syscall: 202
qemu: Unsupported syscall: 202
qemu: Unsupported syscall: 202
qemu: Unsupported syscall: 202
qemu: Unsupported syscall: 202
file is invalid or corrupted (missing end of central directory record)
Opening zip "/proc/self/exe": Cannot find central directory
FATAL: Failed to open '/proc/self/exe' as a zip file: (error: 5): Input/output error

qemu 2.3.1

[root@rhel6i ~]# bazel
Opening zip "/proc/self/exe": lseek(): Bad file descriptor
FATAL: Failed to open '/proc/self/exe' as a zip file: (error: 9): Bad file descriptor

And I stopped version climbing at qemu 3.0.1 because I don't have packaged dependemcies in repo for it any more

[root@rhel6i qemu-3.0.1]# ./configure --target-list=x86_64-linux-user --python=/opt/miniconda3/bin/python

ERROR: glib-2.40 gthread-2.0 is required to compile QEMU

As a last resort, qemu 4.2.0, the latest version, also needs a custom kernel (due to qemu 4.2, linux/if_alg.h: No such file or directory; installed http://elrepo.org/tiki/kernel-lt) and updated glibc. I had two issues there

• Version 2.48 of glibc claims to need lipbcre >= 8.13, but it does not work with that version (It needs PCRE_INFO_MAXLOOKBEHIND, which is undeclared). I found reports that pcre 8.21 des not work either. I installed libpcre 8.38 from sources and glib compiled fine.

• I had to do yum install devtoolset-2-binutils, it did not get installed automatically, an I had mismatched as when compiling. Devtoolset is similar to what Red Hat calls software collection. Devtoolsets 3+ are part of SCL, but devtoolsets 1 and 2 are separate repositories. To activate the scl I had to do scl enable devtoolset-2 bash.

  CC      /root/qemu-4.2.0/slirp/src/mbuf.o
/tmp/ccodbApI.s: Assembler messages:
/tmp/ccodbApI.s:762: Error: expecting string instruction after `rep'

Never the less, after doing all this, I got a segfault this time

# LD_LIBRARY_PATH=/usr/local/lib strace -f /usr/local/bin/qemu-x86_64 /opt/baz/lib/bazel/bin/bazel-real
[pid 24103] write(6, "tty=/dev/pts/0\n", 15) = 15
[pid 24103] munmap(0x44f06000, 4096)    = 0
[pid 24103] statfs64("/root/.cache/bazel/_bazel_root/d41d8cd98f00b204e9800998ecf8427e", 84, {f_type="EXT2_SUPER_MAGIC", f_bsize=4096, f_blocks=10287747, f_bfree=8297009, f_bavail=7776393, f_files=2600960, f_ffree=2316214, f_fsid={1083499738, 1387305835}, f_namelen=255, f_frsize=4096, f_flags=4128}) = 0
[pid 24103] access("/usr/gnemul/qemu-x86_64/root/.cache/bazel/_bazel_root/install/a968348e0a72afc534c7baff88ca46f1", F_OK) = -1 ENOENT (No such file or directory)
[pid 24103] access("/root/.cache/bazel/_bazel_root/install/a968348e0a72afc534c7baff88ca46f1", F_OK) = -1 ENOENT (No such file or directory)
[pid 24103] clock_gettime(CLOCK_MONOTONIC, {157344, 550953409}) = 0
[pid 24103] access("/usr/gnemul/qemu-x86_64/root/.cache/bazel/_bazel_root/install", F_OK) = -1 ENOENT (No such file or directory)
[pid 24103] stat64("/root/.cache/bazel/_bazel_root/install", {st_mode=S_IFDIR|0755, st_size=4096, ...}) = 0
[pid 24103] lstat64("/root/.cache/bazel/_bazel_root/install", {st_mode=S_IFDIR|0755, st_size=4096, ...}) = 0
[pid 24103] geteuid32()                 = 0
[pid 24103] umask(022)                  = 022
[pid 24103] umask(022)                  = 022
[pid 24103] --- SIGSEGV {si_signo=SIGSEGV, si_code=SEGV_MAPERR, si_addr=0xffba4000} ---
[pid 24103] rt_sigprocmask(SIG_SETMASK, [], NULL, 8) = 0
[pid 24103] rt_sigprocmask(SIG_SETMASK, ~[RTMIN RT_1], NULL, 8) = 0
[pid 24103] getrlimit(RLIMIT_CORE, {rlim_cur=0, rlim_max=RLIM_INFINITY}) = 0
[pid 24103] getrlimit(RLIMIT_CORE, {rlim_cur=0, rlim_max=RLIM_INFINITY}) = 0
[pid 24103] setrlimit(RLIMIT_CORE, {rlim_cur=0, rlim_max=RLIM_INFINITY}) = 0
[pid 24103] futex(0x413900, FUTEX_WAKE_PRIVATE, 2147483647) = 0
[pid 24103] write(2, "qemu: uncaught target signal 11 "..., 67qemu: uncaught target signal 11 (Segmentation fault) - core dumped
) = 67
[pid 24103] rt_sigaction(SIGSEGV, {SIG_DFL, ~[RTMIN RT_1], 0}, NULL, 8) = 0
[pid 24103] kill(24103, SIGSEGV)        = 0
[pid 24103] rt_sigsuspend(~[SEGV RTMIN RT_1] <unfinished ...>
[pid 24104] <... futex resumed> )       = ? ERESTARTSYS (To be restarted if SA_RESTART is set)
[pid 24103] <... rt_sigsuspend resumed> ) = ? ERESTARTNOHAND (To be restarted if no handler)
[pid 24104] --- SIGSEGV {si_signo=SIGSEGV, si_code=SI_USER, si_pid=24103, si_uid=0} ---
[pid 24103] rt_sigsuspend(~[SEGV RTMIN RT_1] <unfinished ...>
[pid 24104] +++ killed by SIGSEGV (core dumped) +++
+++ killed by SIGSEGV (core dumped) +++
Segmentation fault (core dumped)

Qemu way did not work for me.