Created attachment 564034 [details] emerge --info Had to unmask it to build firefox-65.0. Using libressl 2.6.5 (but everything is instelled from main repo except nodejs and qtnetwork are installed from libressl overlay). Attaching build.log and emerge-info.log.
Created attachment 564036 [details] /var/tmp/portage/dev-lang/rust-1.32.0/temp/build.log
No detailed compilation logs, so cannot tell for usre, but this may be my hardware issue (segfault; I've got early buggy Ryzen7): some attempts resulted in compilation failures on different modules (not rustc_driver), until (N+1)th attempt finally succeeded.
Hi, can you try reproducing again? but before the emerge/ebuild please export this: export RUSTFLAGS="-v" RUST_BACKTRACE=1 this should provide a bit more detail, current log is useless. it's such a pain to use those buggy Ryzens with gentoo, do you also see segfaults in gcc from time to time?
> can you try reproducing again? but before the emerge/ebuild please export > this: > > export RUSTFLAGS="-v" RUST_BACKTRACE=1 Build.log is still short, even if I put this into ebuild's src_compile(). > it's such a pain to use those buggy Ryzens with gentoo, do you also see > segfaults in gcc from time to time? Yep, but with ccache, sequence of build runs is always convergent, haha. :)))
Created attachment 564168 [details] build.log with 'export RUST....'
probably something in the build system unsets it... different crate again. you already know it, it's your cpu. the only thing I can suggest (aside from trying to RMA that Ryzen) is check out dev-util/sccache. I've recently added it to gentoo and it can work as ccache/distcc for rust. But it can do more harm that do good, so use with care. If a miscompiled crate gets into cache breakage will spread like cancer across all the rust packages using that crate. Closing, feel free to reopen if you find it's not the faulty cpu to blame.
