| 1 |
Hello, |
| 2 |
|
| 3 |
Since LLVM builds have grown in size lately, I wanted to add some of |
| 4 |
check-reqs-r1 checks to it. However, I'm having real trouble guessing |
| 5 |
what the correct sizes should be. |
| 6 |
|
| 7 |
Most importantly, as bug #479356 points out, using '-g' greatly |
| 8 |
increases the build size. My small measures show that without that |
| 9 |
option, LLVM has peek build space consumption around: |
| 10 |
|
| 11 |
- 400-550M without clang (depending on targets), |
| 12 |
- 950-1200M with clang, |
| 13 |
- 16G with clang & USE=debug (assertions, checks). |
| 14 |
|
| 15 |
Those measures were done with '-O2', USE=-debug and single ABI build. |
| 16 |
multilib build consumes almost twice as much. |
| 17 |
|
| 18 |
But if we change the flags, for complete llvm+clang: |
| 19 |
|
| 20 |
- 1.2G for -O2 (as shown above), |
| 21 |
- 12G for -O0 -g. |
| 22 |
|
| 23 |
With such a difference, I don't really see using one of the two values. |
| 24 |
If I go for the lower one, '-g' users may still hit unexpected |
| 25 |
out-of-space issues (like the bug shows). If I go for the higher one, |
| 26 |
many users (including me) will unnecessarily hit the space constraints. |
| 27 |
|
| 28 |
What can we do to improve this? I'm not really happy to have LLVM |
| 29 |
ebuild analyze CFLAGS to set proper space constraints. Maybe we should |
| 30 |
make check-reqs-r1 automatically bump the constraints by some |
| 31 |
statistical multiplier when it detects -g? |
| 32 |
|
| 33 |
[1]:https://bugs.gentoo.org/show_bug.cgi?id=479356 |
| 34 |
|
| 35 |
-- |
| 36 |
Best regards, |
| 37 |
Michał Górny |
Archives