Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

GCC 8 Link Time and Interprocedural Optimization

hubicka.blogspot.com

Post reply on HN
Print view View original

1–10 of 14 posts

Re: GCC 8 Link Time and Interprocedural Optimization

#4

Post by ebikelaw »

ezoe wrote: post #2

>so you can rebuild binary with debug info after your program core dumped and use it to debug the core dump I've never thought it's possible. Rebuild after the core dump ? Rather than build with debug info and strip it away.

If your build system is reproducible there is no reason this would not work.

Re: GCC 8 Link Time and Interprocedural Optimization

#5

Post by jlarocco »

ebikelaw wrote: post #4
ezoe wrote: post #2

>so you can rebuild binary with debug info after your program core dumped and use it to debug the core dump I've never thought it's possible. Rebuild after the core dump ? Rather than build with debug info and strip it away.

If your build system is reproducible there is no reason this would not work.

Maybe that's the case now, with GCC 8, but historically that hasn't been true.

In the past the effects of using both -g and -O* flags could interact each other because they can both affect how registers are allocated, how call stacks are setup, how functions are inlined, etc. In other words, adding -g by itself made the build non-reproducible.

If that's no longer true in GCC 8 it's great news.

Re: GCC 8 Link Time and Interprocedural Optimization

#6

Post by glandium »

jlarocco wrote: post #5
ebikelaw wrote: post #4

Earlier quoted context omitted.

If your build system is reproducible there is no reason this would not work.

Maybe that's the case now, with GCC 8, but historically that hasn't been true. In the past the effects of using both -g and -O* flags could interact each other because they can both affect how registers are allocated, how call stacks are setup, how functions are inlined, etc. In other words, adding -g by itself made the build non-reproducible. If that's no longer true in GCC 8 it's great news.

> In other words, adding -g by itself made the build non-reproducible.

In my experience, this is largely not true. However, what is plausible is that before GCC 8, LTO, which, for speed reasons, can split work across multiple threads, was not deterministic, and GCC 8 might have made that better.

Re: GCC 8 Link Time and Interprocedural Optimization

#7

Post by readittwice »

jlarocco wrote: post #5
ebikelaw wrote: post #4

Earlier quoted context omitted.

If your build system is reproducible there is no reason this would not work.

Maybe that's the case now, with GCC 8, but historically that hasn't been true. In the past the effects of using both -g and -O* flags could interact each other because they can both affect how registers are allocated, how call stacks are setup, how functions are inlined, etc. In other words, adding -g by itself made the build non-reproducible. If that's no longer true in GCC 8 it's great news.

AFAIR GCC bootstrap builds by default with "-g -O2". There are 3 stages: 1) use system compiler to compile GCC's source code, 2) use resulting binary to build same source, 3) use binary from 2) to build source one more time. After all stages the binaries from stage 2 and 3 are compared byte for byte if they are identical.

Bootstrap failures are taken very seriously by the GCC developers. So I really doubt that -g and -O* lead to non-deterministic builds.

Post reply on HN