Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

fflas-ffpack: 2.2.2 -> 2.3.2 #36437

Merged
merged 1 commit into from Apr 23, 2018
Merged

fflas-ffpack: 2.2.2 -> 2.3.2 #36437

merged 1 commit into from Apr 23, 2018

Conversation

timokau
Copy link
Member

@timokau timokau commented Mar 7, 2018 •
edited

Motivation for this change

fflas-ffpack was outdated.

This updates it, makes it more reproducible / less error-prone and enables checks.

Things done
  • 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.

@timokau timokau mentioned this pull request Mar 7, 2018
Merged
8 tasks
@timokau timokau force-pushed the origin/fflas-ffpack-update branch 3 times, most recently from 89e6559 to c3db41b Compare March 16, 2018 00:03
@7c6f434c
Copy link
Member

I am not sure Nixpkgs is usable on any x86 CPU without SSE. SSE4 is another story, of course. Not sure where to draw the line…

@timokau
Copy link
Member Author

timokau commented Apr 12, 2018

For what its worth, arch and debian both disable it in their packages. Gentoo has this

local simd_opt="--disable-avx"
if( (use cpu_flags_x86_avx) || (use cpu_flags_x86_avx2) || (use cpu_flags_x86_sse4_1) ); then
	simd_opt="--enable-simd"
fi

--enable-simd used to be the way to set that option, until somebody upstream complained that its not fine-grained enough and they replaced it with all those other options :/

@timokau
Copy link
Member Author

timokau commented Apr 12, 2018

I rebased on master and changed the commit message (since the update part was already merged in #37628).

@matthewbauer
Copy link
Member

@GrahamcOfBorg build fflas-ffpack

@GrahamcOfBorg
Copy link

Failure on aarch64-linux (full log)

Attempted: fflas-ffpack

Partial log (click to expand)

../configure: line 508: can't create lib/Makefile: nonexistent directory
../configure: line 508: can't create lib/Makefile: nonexistent directory
../configure: line 508: can't create lib/Makefile: nonexistent directory
../configure: line 509: can't create lib/Makefile: nonexistent directory
DONE configure
/nix/store/9vknvj1gnbd2mbci7sw07cy1lgyxxy02-stdenv-linux/setup: line 100: include/atlas_cacheedge.h: No such file or directory
builder for '/nix/store/ppbhwsd89zfhqv9l8j8zrz64m1bwnhvy-atlas-3.10.2.drv' failed with exit code 1
cannot build derivation '/nix/store/cfxdrhbk198r2dr2ivwgzrsw4cmww2ni-liblapack-3.4.1.drv': 1 dependencies couldn't be built
cannot build derivation '/nix/store/8fmb81p6llvqffmxgwapp8xfr908q6vl-fflas-ffpack-2.3.2.drv': 1 dependencies couldn't be built
�[31;1merror:�[0m build of '/nix/store/8fmb81p6llvqffmxgwapp8xfr908q6vl-fflas-ffpack-2.3.2.drv' failed

@GrahamcOfBorg
Copy link

No attempt on x86_64-darwin (full log)

The following builds were skipped because they don't evaluate on x86_64-darwin: fflas-ffpack

Partial log (click to expand)


a) For `nixos-rebuild` you can set
  { nixpkgs.config.allowUnsupportedSystem = true; }
in configuration.nix to override this.

b) For `nix-env`, `nix-build`, `nix-shell` or any other Nix command you can add
  { allowUnsupportedSystem = true; }
to ~/.config/nixpkgs/config.nix.

@GrahamcOfBorg
Copy link

Success on x86_64-linux (full log)

Attempted: fflas-ffpack

Partial log (click to expand)

 /nix/store/jy9knxp7nmw80jkf932axrs1b4p9k4hi-coreutils-8.29/bin/install -c -m 644 fflas-ffpack.pc '/nix/store/arl8bgjkm35fxpcy1wq7rdgqmsb247qc-fflas-ffpack-2.3.2/lib/pkgconfig'
make[2]: Leaving directory '/build/source'
make[1]: Leaving directory '/build/source'
post-installation fixup
shrinking RPATHs of ELF executables and libraries in /nix/store/arl8bgjkm35fxpcy1wq7rdgqmsb247qc-fflas-ffpack-2.3.2
strip is /nix/store/j75dgadrff2d1fyc4fczmcgqkid2imdx-binutils-2.30/bin/strip
stripping (with command strip and flags -S) in /nix/store/arl8bgjkm35fxpcy1wq7rdgqmsb247qc-fflas-ffpack-2.3.2/lib  /nix/store/arl8bgjkm35fxpcy1wq7rdgqmsb247qc-fflas-ffpack-2.3.2/bin
patching script interpreter paths in /nix/store/arl8bgjkm35fxpcy1wq7rdgqmsb247qc-fflas-ffpack-2.3.2
/nix/store/arl8bgjkm35fxpcy1wq7rdgqmsb247qc-fflas-ffpack-2.3.2/bin/fflas-ffpack-config: interpreter directive changed from "/usr/bin/env bash" to "/nix/store/xn5gv3lpfy91yvfy9b0i7klfcxh9xskz-bash-4.4-p19/bin/bash"
checking for references to /build in /nix/store/arl8bgjkm35fxpcy1wq7rdgqmsb247qc-fflas-ffpack-2.3.2...

@matthewbauer matthewbauer merged commit d6b1e81 into NixOS:master Apr 23, 2018
@timokau
Copy link
Member Author

timokau commented Apr 23, 2018

Apparently this fails on aarch64 because its dependency liblapack fails which again fails because its dependency atlas fails.

liblapack never ends up in the fflas-ffpack build closure and apparently openblas provides lapack. Is there any reason to keep that in buildInpugs @7c6f434c? And should we mark atlas as broken for aarch64?

@timokau timokau deleted the origin/fflas-ffpack-update branch April 23, 2018 18:34
@7c6f434c
Copy link
Member

ATLAS seems to have some support for ARM64, but maybe building it is complicated.

Frankly, I haven't yet needed to use something where I cared about specific choice of LAS. But sometimes library build systems have interesting preferences (and I forget to revisit the situation when they get better). So, dropping the dependency is fine with me if it builds.

@timokau
Copy link
Member Author

timokau commented Apr 23, 2018

In #10808 it sounds like it would be relatively unproblematic to just disable atlas for aarch64.

@timokau
Copy link
Member Author

timokau commented Apr 23, 2018

Dropped the dependency in #39384.

@Ericson2314
Copy link
Member

Ericson2314 commented Apr 24, 2018 •
edited

SSE4 is another story, of course. Not sure where to draw the line…

I've been meaning to add more x86_32 variants. Someday...

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@Ericson2314 Ericson2314 Awaiting requested review from Ericson2314

Assignees
No one assigned
Labels
10.rebuild-darwin: 0 10.rebuild-linux: 1-10
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.

None yet
5 participants
@timokau @7c6f434c @matthewbauer @GrahamcOfBorg @Ericson2314