gcc8: init at 8.1.0 #39866

Synthetica9 · 2018-05-02T16:36:28Z

Motivation for this change

GCC 8.1 was just released.

Things done

~~This is still compiling on my machine. I see no reason why it wouldn't finish, but I wanted to get the WIP PR out.~~ Okay; it compiled. Ship it?
This does not set GCC8 as the default version.
For good measure, this derivation is mostly copied from GCC7
Tested using sandboxing (nix.useSandbox on NixOS, or option build-use-sandbox in nix.conf on non-NixOS)
Built on platform(s)
- NixOS
- macOS
- other Linux distributions
Tested via one or more NixOS test(s) if existing and applicable for the change (look inside nixos/tests)
Tested compilation of all pkgs that depend on this change using nix-shell -p nox --run "nox-review wip"
Tested execution of all binary files (usually in ./result/bin/)
Fits CONTRIBUTING.md.

dtzWill · 2018-05-02T18:34:30Z

I vote getting this in the repository first, with separate PR for changing the default (and dealing with any breakage/fixes needed).

Synthetica9 · 2018-05-02T18:36:16Z

Sounds fair enough

dtzWill · 2018-05-02T18:46:46Z

pkgs/development/compilers/gcc/8/default.nix

+
+  src = fetchurl {
+    url = "mirror://gcc/releases/gcc-${version}/gcc-${version}.tar.xz";
+    /* sha256 = "0p71bij6bfhzyrs8676a8jmpjsfz392s2rg862sdnsk30jpacb43"; */


Remove the gcc7 hash, please :).

Whoops, ✔️

dtzWill · 2018-05-02T18:48:19Z

cc @shlevy re:riscV patches and testing (but no need to block on a response or this being resolved)
Let's see what borg says first :)
I'll try w/musl, but this isn't a blocker either :)

Log: ``` patching sources applying patch /nix/store/6m27y27zvzsjn1ir4y8mm9nc9xnh2sgx-riscv-no-relax.patch patching file gcc/config/riscv/riscv.c Reversed (or previously applied) patch detected! Assume -R? [n] Apply anyway? [n] Skipping patch. 1 out of 1 hunk ignored -- saving rejects to file gcc/config/riscv/riscv.c.rej patching file gcc/config/riscv/riscv.opt Reversed (or previously applied) patch detected! Assume -R? [n] Apply anyway? [n] Skipping patch. 1 out of 1 hunk ignored -- saving rejects to file gcc/config/riscv/riscv.opt.rej patching file gcc/doc/invoke.texi Reversed (or previously applied) patch detected! Assume -R? [n] Apply anyway? [n] Skipping patch. ```

Synthetica9 · 2018-05-02T18:55:48Z

@GrahamcOfBorg build gcc8

dtzWill · 2018-05-02T19:46:22Z

Consider adding these commits:

add cc-wrapper variants for gcc8: dtzWill@9ecd2e2
drop self-tests patch as it doesn't apply, we'll see how things work!
dtzWill@ec4a460

(for testing purposes I made a branch that contains this as well as setting gcc8 as default:
https://github.com/dtzWill/nixpkgs/commits/experimental/gcc8)

Synthetica9 · 2018-05-02T20:00:55Z

@dtzWill The drop-self-test patch was available in 7.3.0 (https://git.busybox.net/buildroot/plain/package/gcc/7.3.0), so I don't we should remove that yet (unless it was somehow confirmed by the GCC dev team that it won't be needed)

dtzWill · 2018-05-02T20:25:18Z

@dtzWill The drop-self-test patch was available in 7.3.0 (https://git.busybox.net/buildroot/plain/package/gcc/7.3.0), so I don't we should remove that yet (unless it was somehow confirmed by the GCC dev team that it won't be needed)

Well since the patch wasn't issued by the GCC dev team I'm not sure why they're the authority to ask :).

But leaving it commented-out works for me-- the real answer is to get some builders testing cross to see if it is needed or not. That can be handled separately, sorry for piling on :).

Synthetica9 · 2018-05-02T20:30:20Z

@dtzWill Nah, it's fine. I'll change the hash to a placeholder value and make the url dependent on the version, but leave it commented.

How's the gcc7->gcc8 switch looking with regards to broken packages?

dtzWill · 2018-05-02T20:40:03Z

How's the gcc7->gcc8 switch looking with regards to broken packages?

Not sure--and I won't be able to provide a good answer for this, someone would need to setup a new jobset on Hydra. But that can be done in the follow-up PR that changes the default?

I can confirm that with gcc8 as default we're at least able to build

x86_64-linux: through cc-test-wrapper
x86_64-musl: through test-wrapper
x86_64 -> aarch64-musl:stdenv.cc (no self-tests patch!)
x86_64 -> aarch64-glibc: errror building glibc itself D:

(I have tested very little beyond basics so far, so dunno re:package compatibility at all :))

dtzWill · 2018-05-02T21:10:05Z

LGTM, borg is probably having same problem I was having with mirrors (ended up manually using nix-prefetch-url). We can at least make it available for testing with this!

gebner · 2018-05-03T08:25:59Z

pkgs/development/compilers/gcc/8/default.nix

+, buildPackages
+}:
+
+assert langJava     -> zip != null && unzip != null


I thought upstream had removed java support in gcc 6?

Oh, yeah, right you are. Should the arguments stay in place and go unused (apart from maybe an assert not langJava), or should I remove them altogether (

nixpkgs/pkgs/development/compilers/gcc/7/default.nix

Line 30 in ba52ae5

, cloog # unused; just for compat with gcc4, as we override the parameter on some places

seems to suggest that the first option would be more conventional.)

I think we should remove them completely. AFAICT the only place where we override enableJava et al. is in the gcj6 attribute (which uses gcc6).

Synthetica9 · 2018-05-03T11:12:46Z

@gebner I removed all mentions of Java. I'll also create a PR to do the same for GCC7.

shlevy · 2018-05-03T13:52:27Z

Unable to test at the moment, but IIRC gcc 8 contains the RISC-V stuff already

Synthetica9 · 2018-05-03T14:07:00Z

@shlevy I ctrl-f'd the release notes for RISC, but it isn't mentioned AFAICS.

shlevy · 2018-05-03T14:12:03Z

Just checked, both patches are on the 8.1.0 tag on subversion.

gebner · 2018-05-03T14:18:56Z

pkgs/development/compilers/gcc/8/default.nix

+, libelf                      # optional, for link-time optimizations (LTO)
+, isl ? null # optional, for the Graphite optimization framework.
+, zlib ? null
+, pkgconfig ? null


With java support removed, all of these options (pkgconfig until x11Support) are now unused as well.

grabs hatchet

See also: NixOS@51ac1d8 NixOS#39866 (comment)

dtzWill · 2018-05-04T14:35:54Z

This is ready to be merged, just waiting to confirm doesn't trigger any mass-rebuilds.

dtzWill · 2018-05-04T14:52:17Z

LGTM, apologies if I missed something. Thanks @Synthetica9!

Synthetica9 · 2018-05-04T15:01:40Z

@dtzWill: I think the next step would be a PR to use GCC8 as the default GCC version? Would it be sufficient to just git cherry-pick dtzWill@2e7772a from master and throw that up for PR?

dtzWill · 2018-05-04T15:04:27Z

@Synthetica9 I think so! There definitely is some breakage but we can address that in that PR :).

(and if my commit isn't sufficient for whatever reason that can be discussed/reviewed there as well)

matthewbauer · 2018-05-05T21:39:00Z

Hi,

gcc8 is failing on x86_64-darwin:

Job: https://hydra.nixos.org/job/nixpkgs/trunk/gcc8.x86_64-darwin

Not sure what's going on. Clang gives this error:

clang-5.0: error: unsupported option '--gstabs'

Synthetica9 · 2018-05-06T10:55:32Z

@matthewbauer I think it would be best to open a separate issue for that; I don't have access to a macos box.

Synthetica9 added 2 commits May 2, 2018 18:34

gcc8: init at 8.1.0

659363f

gcc8: register to all-packages

736dab4

GrahamcOfBorg added 10.rebuild-darwin: 0 10.rebuild-linux: 0 labels May 2, 2018

gcc8: added myself as maintainer

728eb27

GrahamcOfBorg added 8.has: package (new) 10.rebuild-darwin: 1-10 10.rebuild-linux: 1-10 and removed 10.rebuild-darwin: 0 10.rebuild-linux: 0 labels May 2, 2018

dtzWill reviewed May 2, 2018

View reviewed changes

Synthetica9 added 2 commits May 2, 2018 20:51

gcc8: removed commented gcc7 hash

c7a6b36

add cc-wrapper test variants for gcc7, gcc8, add to release.nix

48e4449

gcc8: made the remove-selftests patch more robust, removed old hash

5794d94

dtzWill approved these changes May 2, 2018

View reviewed changes

Synthetica9 changed the title ~~[WIP] gcc8: init at 8.1.0~~ gcc8: init at 8.1.0 May 2, 2018

gebner reviewed May 3, 2018

View reviewed changes

Synthetica9 changed the title ~~gcc8: init at 8.1.0~~ [WIP] gcc8: init at 8.1.0 May 3, 2018

gcc8: Removed java as a target language

d21ff87

This was referenced May 3, 2018

gcc7: Removed java as a target language #39917

Closed

gcc7: Removed java as a target language #39924

Merged

gebner reviewed May 3, 2018

View reviewed changes

gcc8: Removed X11 components, since they were only used by langJava

51ac1d8

Synthetica9 added a commit to Synthetica9/nixpkgs that referenced this pull request May 3, 2018

gcc7: Removed X11 components, since they were only used by langJava

719810d

See also: NixOS@51ac1d8 NixOS#39866 (comment)

Synthetica9 changed the title ~~[WIP] gcc8: init at 8.1.0~~ gcc8: init at 8.1.0 May 3, 2018

dtzWill merged commit e02dfb5 into NixOS:master May 4, 2018

dtzWill mentioned this pull request Jul 27, 2018

gcc8: 8.1 -> 8.2 #44151

Merged

9 tasks

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

gcc8: init at 8.1.0 #39866

gcc8: init at 8.1.0 #39866

Synthetica9 commented May 2, 2018 •

edited by dtzWill

dtzWill commented May 2, 2018

Synthetica9 commented May 2, 2018

dtzWill May 2, 2018

Synthetica9 May 2, 2018

dtzWill commented May 2, 2018

Synthetica9 commented May 2, 2018

dtzWill commented May 2, 2018

Synthetica9 commented May 2, 2018

dtzWill commented May 2, 2018

Synthetica9 commented May 2, 2018

dtzWill commented May 2, 2018 •

edited

dtzWill commented May 2, 2018

gebner May 3, 2018

Synthetica9 May 3, 2018

gebner May 3, 2018

Synthetica9 commented May 3, 2018

shlevy commented May 3, 2018

Synthetica9 commented May 3, 2018

shlevy commented May 3, 2018

gebner May 3, 2018

Synthetica9 May 3, 2018

dtzWill commented May 4, 2018

dtzWill commented May 4, 2018

Synthetica9 commented May 4, 2018

dtzWill commented May 4, 2018

matthewbauer commented May 5, 2018

Synthetica9 commented May 6, 2018

gcc8: init at 8.1.0 #39866

gcc8: init at 8.1.0 #39866

Conversation

Synthetica9 commented May 2, 2018 • edited by dtzWill

Motivation for this change

Things done

dtzWill commented May 2, 2018

Synthetica9 commented May 2, 2018

dtzWill May 2, 2018

Choose a reason for hiding this comment

Synthetica9 May 2, 2018

Choose a reason for hiding this comment

dtzWill commented May 2, 2018

Synthetica9 commented May 2, 2018

dtzWill commented May 2, 2018

Synthetica9 commented May 2, 2018

dtzWill commented May 2, 2018

Synthetica9 commented May 2, 2018

dtzWill commented May 2, 2018 • edited

dtzWill commented May 2, 2018

gebner May 3, 2018

Choose a reason for hiding this comment

Synthetica9 May 3, 2018

Choose a reason for hiding this comment

gebner May 3, 2018

Choose a reason for hiding this comment

Synthetica9 commented May 3, 2018

shlevy commented May 3, 2018

Synthetica9 commented May 3, 2018

shlevy commented May 3, 2018

gebner May 3, 2018

Choose a reason for hiding this comment

Synthetica9 May 3, 2018

Choose a reason for hiding this comment

dtzWill commented May 4, 2018

dtzWill commented May 4, 2018

Synthetica9 commented May 4, 2018

dtzWill commented May 4, 2018

matthewbauer commented May 5, 2018

Synthetica9 commented May 6, 2018

Synthetica9 commented May 2, 2018 •

edited by dtzWill

dtzWill commented May 2, 2018 •

edited