| 30 Apr 2024 |
 John Ericson | I made it and I don't like it | 15:50:53 |
| John Ericson | and no one has done me the favor of replacing it with something better alas | 15:51:07 |
| John Ericson | :) | 15:51:10 |
 rhelmot | alas indeed | 15:51:23 |
 dramforever | unrelated to the above, on mips i'm getting this when building openssh
fmt_scaled.c: In function 'scan_scaled':
fmt_scaled.c:214:1: sorry, unimplemented: argument 'used' is not supported for '-fzero-call-used-regs' on this target
214 | }
| ^
i think this is just ... not completely implemented in gcc, like it's just broken depending on what the code is doing and fmt_scaled.c just happens to deal with floats and stuff maybe. i figured out if you --without-hardening you can get the configure script to not use the broken -fzero-call-used-regs... what do i do now?
| 15:58:30 |
 artemist | In reply to @rhelmot:matrix.org
Did you build with flake.nix or ./configure && make? Configure. I couldn't actually get it to finish build without some more patches but the system in config.h was just x86_64-freebsd | 15:59:21 |
| John Ericson | rhelmot: oh I found the issue | 15:59:31 |
| John Ericson | the dot in the version | 15:59:35 |
| John Ericson | is confusing some stupid code | 15:59:40 |
| dramforever | In reply to @dramforever:matrix.org
unrelated to the above, on mips i'm getting this when building openssh
fmt_scaled.c: In function 'scan_scaled':
fmt_scaled.c:214:1: sorry, unimplemented: argument 'used' is not supported for '-fzero-call-used-regs' on this target
214 | }
| ^
i think this is just ... not completely implemented in gcc, like it's just broken depending on what the code is doing and fmt_scaled.c just happens to deal with floats and stuff maybe. i figured out if you --without-hardening you can get the configure script to not use the broken -fzero-call-used-regs... what do i do now?
i think ideally i get the openssh folks to notice this because they seem to still care https://github.com/openssh/openssh-portable/commit/f9193f03db0029fc9c31fbdb5c66a2737446bd8f | 16:00:19 |
| John Ericson | rhelmot artemist : OK I rebased (force pushed!) your branch, and fixed the splicing thing that was bugging me | 16:07:47 |
| John Ericson | you should now be able to remove lots of inherit (buildFreebsd) .... | 16:07:59 |
| rhelmot | on it!! | 16:08:09 |
| John Ericson | I think we can decide on https://github.com/NixOS/nixpkgs/pull/308032 too | 16:08:25 |
| John Ericson | and I think I can extract freebsdBranches from your main PR too, further shrinking things down | 16:08:42 |
| John Ericson | (in general I like splitting up the PR more, but I am happy to do the work pulling things out) | 16:08:52 |
| John Ericson | (rather than make you both slice and dice into tiny morsals) | 16:09:05 |
| John Ericson | Also I am curious what the deal with FreeBSD and the smallest version number | 16:09:21 |
| John Ericson | do they have that or just two version numbers? | 16:09:29 |
| John Ericson | noticed that the bash script to spit out the version skippped it | 16:09:40 |
| rhelmot | what does that mean exactly? are you talking about pkgs/os-specific/bsd/freebsd/update.py ? | 16:13:54 |
| artemist | I'm guessing sys/conf/newvers.sh in the FreeBSD source tree? The update.py script calls that to figure out the version | 16:15:11 |
| artemist | I think they put the patch number in BRANCH (e.g. RELEASE-p6) instead of revision for some reason | 16:18:14 |
| artemist | Not sure why but it's what they do | 16:18:23 |
| John Ericson | artemist: ah OK | 16:21:36 |
| John Ericson | thought the patch number was gone altogether | 16:21:43 |
| John Ericson | that helps | 16:21:46 |
| John Ericson | thanks both of you | 16:21:53 |
| rhelmot | In reply to @Ericson2314:matrix.org
you should now be able to remove lots of inherit (buildFreebsd) .... I pushed a change removing everything that did not cause a recursion error when removed | 16:23:43 |
| rhelmot | there's still quite a lot though. I don't know how to determine if it's fixable or just due to the netbsd-mentioned problem | 16:24:10 |
