[Bug 242266] CLang 6.0.1 (Part of the System 12.0) can not Build CLang 8.0.1 from the Port Collection
bugzilla-noreply at freebsd.org
bugzilla-noreply at freebsd.org
Thu Nov 28 19:29:31 UTC 2019
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=242266
Dimitry Andric <dim at FreeBSD.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |dim at FreeBSD.org
--- Comment #3 from Dimitry Andric <dim at FreeBSD.org> ---
It looks like you are tickling some old bug in 6.0.1 which makes it use way too
much memory for this particular compile command. I tried on 12.0-RELEASE amd64
with a lot of RAM, and I got the following time(1) output:
84.83 real 31.36 user 53.31 sys
6560428 maximum resident set size
40472 average shared memory size
2587 average unshared data size
255 average unshared stack size
3201539 page reclaims
1370 page faults
0 swaps
1379 block input operations
5 block output operations
0 messages sent
0 messages received
0 signals received
1383 voluntary context switches
1152 involuntary context switches
E.g, it needs 6.3 GiB of memory and more than a minute to compile. This is
obviously never going to work on i386, which can use up to 4 GiB of memory for
user processes.
Compare this to clang 8.0.1 on recent stable/12:
23.17 real 18.12 user 3.36 sys
314944 maximum resident set size
42981 average shared memory size
358 average unshared data size
253 average unshared stack size
69152 page reclaims
1090 page faults
0 swaps
1100 block input operations
5 block output operations
3 messages sent
0 messages received
0 signals received
1105 voluntary context switches
346 involuntary context switches
That needs only ~308 MiB of memory, and a little more than 20 seconds.
I'm unsure what will help here. You could try leaving out the -march=native
stuff, or lowering the optimization level. The former should be easy enough,
but for the latter, I don't know how to achieve that with the port.
Another alternative is to do a binary upgrade to 12.1-R, and work from there.
--
You are receiving this mail because:
You are the assignee for the bug.
More information about the freebsd-toolchain
mailing list