Commit graph

175 commits

Author SHA1 Message Date
John Ericson e755a8a27d treewide: Use targetPrefix instead of prefix for platform name prefixes
Certain tools, e.g. compilers, are customarily prefixed with the name of
their target platform so that multiple builds can be used at once
without clobbering each other on the PATH. I was using identifiers named
`prefix` for this purpose, but that conflicts with the standard use of
`prefix` to mean the directory where something is installed. To avoid
conflict and confusion, I renamed those to `targetPrefix`.
2017-11-27 03:15:50 -05:00
John Ericson 0101856765
Merge pull request #30549 from obsidiansystems/bintools
treewide: Introduce stdenv.cc.bintools
2017-11-08 14:20:48 -05:00
Vladimír Čunát c8c6a1edb5
Merge #31209: cc-wrapper: Fix if dynamicLinker not found 2017-11-06 13:07:43 +01:00
Bojan Nikolic 3a63fc1258 Remove trailing line 2017-11-05 23:00:24 +00:00
Bojan Nikolic 2ed054885b Remove the unnecessary second conditional 2017-11-05 22:57:44 +00:00
John Ericson 70d91badf5 treewide: Depend on stdenv.cc.bintools instead of binutils directly
One should do this when needed executables at build time. It is more
honest and cross-friendly than refering to binutils directly.
2017-11-05 17:10:53 -05:00
Nikolay Amiantov 5f3b84e979 cc-wrapper: disable POSIX compatibility 2017-11-05 12:19:37 +02:00
Bojan Nikolic 3e9daece1d nixos/cc-wrapper: Fix bug if dynamicLinker not found
If a dynamic linker for target is not found the generated script fails
due to unbound variable error (due to "set -u"). Correct by specifying
default value with dynamicLinker:- and not generating ldflagsBefore if
no linker is found.

This problem was found when cross compiling to mingw32 targets
2017-11-03 21:36:43 +00:00
Dan Peebles b426c85ce2 Get rid of most @rpath nonsense on Darwin
This requires some small changes in the stdenv, then working around the
weird choice LLVM made to hardcode @rpath in its install name, and then
lets us remove a ton of annoying workaround hacks in many of our Go
packages. With any luck this will mean less hackery going forward.
2017-10-08 16:13:46 -04:00
John Ericson fdbda216b1 cc-wrapper: Clean up dynamic linking with x86 multilib
It's better layering to do everything in ld-wrapper. Also, use numeric
comparisons for `relocatable`.
2017-09-28 20:05:26 -04:00
John Ericson d349f9a340 cc-wrapper: Use stdenvNoCC to build
cc-wrapper may wrap a cc-compiler, but it doesn't need one to build
itself. (c.f. expand-response-params is a separate derivation.) This
helps avoid cycles on the cross stuff, in addition to removing a
useless dependency edge.

I could have been super careful with overrides in the stdenv to avoid
the mass rebuild, but I don't think it's worth it.
2017-09-26 14:08:21 -04:00
John Ericson 0d3d2a01d2 cc-wrapper: Add set -x tracing for NIX_DEBUG >= 7 2017-09-26 11:24:19 -04:00
John Ericson 127a5f3357 treewide: Use (( "${NIX_DEBUG:-0}" >= 1) )) consistently 2017-09-26 11:24:19 -04:00
John Ericson 6c74ee68b9 cc-wrapper, cc-wrapper-old: Simplify shell logic
1. `crossDrv` is now the default so we don't need to worry about that in
   build != host builds.

2. shell is the build time shell, so `wrapCCCross` doesn't need to
   worry, as build == host.

3. `shell.shellPath` will always be appended where useful.

4. Complicated `shell == ""` logic served no purpose.
2017-09-19 16:45:24 -04:00
John Ericson 13fc982e65 cc-wrapper: Use same dynamic loader on all Darwin, not just x86_64
In practice, this is correct because iOS is on ARM and puts the loader
there.
2017-09-18 19:25:58 -04:00
John Ericson a44bbc72e6 cc-wrapper: Remove obsolete assertion
This was just causing evaluation problems on cross.
2017-09-18 19:22:57 -04:00
Eelco Dolstra ec8d41f08c
Revert "Merge pull request #28557 from obsidiansystems/binutils-wrapper"
This reverts commit 0a944b345e, reversing
changes made to 61733ed6cc.

I dislike these massive stdenv changes with unclear motivation,
especially when they involve gratuitous mass renames like NIX_CC ->
NIX_BINUTILS. The previous such rename (NIX_GCC -> NIX_CC) caused
months of pain, so let's not do that again.
2017-09-07 12:51:21 +02:00
John Ericson dbf6d20d64 binutils-wrapper: Import separately from cc-wrapper 2017-09-01 11:44:56 -04:00
John Ericson 40e9b2a7e6 binutils-wrapper: Init
Factor a binutils wrapper out of cc-wrapper. While only LD is wrapped,
the setup hook defines environment variables on behalf of other
utilites.
2017-09-01 11:44:55 -04:00
John Ericson fbb7d335db cc-wrapper: Use separate mangler for "bool" variables
This avoids any `NIX_FOOBAR=1 1` not triggering conditions.
2017-09-01 11:44:54 -04:00
John Ericson 1f5807d760 cc-wrapper: Pull variable mangler into utils.sh
In preparation for splitting out binutils-wrapper
2017-09-01 11:44:54 -04:00
John Ericson 94c0267fc1 cc-wrapper: Clean up dynamic linking with x86 multilib
It's better layering to do everything in ld-wrapper.
2017-09-01 11:44:54 -04:00
John Ericson 3d3a6e0fac cc-wrapper: Remove support for NIX_LDFLAGS_HARDEN
It has long been deprecated
2017-08-31 13:54:32 -04:00
John Ericson 46fd4bcb14 cc-wrapper: Remove {START,EXEC}_HOOK
These are no longer used by anything
2017-08-31 13:54:07 -04:00
John Ericson 97a48835b7 mkDerivation, cc-wrapper: Check hardening flag validity in Nix
This becomes necessary if more wrappers besides cc-wrapper start
supporting hardening flags. Also good to make the warning into an
error.

Also ensure interface is being used right: Not as a string, not just in
bash.
2017-08-30 17:53:42 +02:00
John Ericson 822a8d0148 cc-wrapper: Remove redundant hardening
GCC just passes `-z ...` flags to ld unaltered, and they are already
passed to LD anyways. On the other hand, `-pie` affects gcc behavior
too.
2017-08-30 17:53:42 +02:00
Daiderd Jordan 92652b4d79
darwin-CF: use @rpath for library id and add an rpath entry for CF based on NIX_COREFOUNDATION_RPATH 2017-08-28 23:24:58 +02:00
John Ericson df7c305c4c cc-wrapper: Leverage the setup script instead of buildCommand 2017-08-25 15:10:05 -04:00
John Ericson dc47e763c1 cc-wrapper: Use set -u for better maintainability 2017-08-25 15:10:05 -04:00
John Ericson a470be5a16 expand-response-params: Build more normally 2017-08-25 15:10:02 -04:00
John Ericson 287fce6402 expand-response-params: Pull out of cc-wrapper
No hashes were changed
2017-08-25 11:21:09 -04:00
Daniel Peebles 4f7f48fb1a Merge pull request #28521 from obsidiansystems/cc-wrapper-dead-code
cc-wrapper: Remove dead code
2017-08-24 16:05:51 -04:00
Tuomas Tynkkynen 0c0fad6141 treewide: Consistently call ARM 'arm'
No need for silly differences.
2017-08-24 01:17:01 +03:00
John Ericson 08f3a60e7b cc-wrapper: Remove dead code
ccPath is only defined below, so this condition would never be true.

Worse, that's not quite true: what if somebody happend to have `/clang`
and no sandboxing. Boy, wouldn't that be annoying to debug!
2017-08-23 17:09:47 -04:00
John Ericson 0135e61b4c Merge remote-tracking branch 'upstream/master' into staging
That way the tarball job succeeds
2017-08-21 18:42:57 -04:00
John Ericson c035711072 cc-wrapper: Remove unused params
Ensured hashes unchanged and eval succeeds in tarball job
2017-08-21 18:40:41 -04:00
Bojan Nikolic ad8d5649a2 set +u workaround for empty arrays
libDirs can be empty, which in combination with "set -u" of
9f1e009975 will cause a variable unbound
error on old bash versions
2017-08-21 12:27:26 +01:00
Daiderd Jordan 5d3527c148
cc-wrapper: don't export CPP
CPP defaults to $(CC) -E and setting it explicitly seems to cause
issues in a number of packages like gcc and gdb.

https://www.gnu.org/software/make/manual/html_node/Implicit-Variables.html
2017-08-16 21:38:32 +02:00
John Ericson 3c9cf282bb cc-wrapper: Improve set -u compliance 2017-08-14 14:44:32 -04:00
Vladimír Čunát 6899c7fdb9
Merge branch 'master' into gcc-6 2017-08-12 10:09:41 +02:00
Daiderd Jordan bf8c125170
cc-wrapper: fix LD_DYLD_PATH on darwin
Having multiple compilers in the build environment would result in an
invalid LD_DYLD_PATH like /usr/lib/dyld/usr/lib/dyld.
Since the path is hardcoded in XNU it can't be anything but
/usr/lib/dyld anyway.
2017-08-10 00:22:58 +02:00
Orivej Desh 771f28bac8 cc-wrapper: Fix support for qtbase-setup-hook
Revert https://github.com/NixOS/nixpkgs/pull/27657#issuecomment-318161946
due to https://github.com/NixOS/nixpkgs/pull/28021#issuecomment-321045542
2017-08-09 18:17:46 +02:00
Orivej Desh a6e7dffd1f cc-wrapper: Fix standalone ld 2017-08-08 11:03:51 +02:00
Orivej Desh c8e9dcc8a4 cc-wrapper: Fix standalone gcc
This ensures that all salted variables are defined even if the wrapped program
is invoked outside nix-build environment.
2017-08-08 11:03:51 +02:00
Orivej Desh c8f7f18e69 cc-wrapper: Fix adding directories to rpath
This fixes a bug introduced in #27831: `for path in "$dir"/lib*.so` assumed that
all libs match `lib*.so`, but 07674788d6 started
adding libs that match `*.so` and `*.so.*`.
2017-08-08 11:03:51 +02:00
Robin Gloster 3e981b9e33
Revert "cc-wrapper: fix set -u errors"
This reverts commit 67a41eafe9.

see #28021 for discussion
2017-08-08 10:20:00 +02:00
Robin Gloster 67a41eafe9
cc-wrapper: fix set -u errors
cc @Ericson2314
2017-08-08 10:15:34 +02:00
John Ericson 810fb0c968 cc-wrapper: Don't slurp NIX_DONT_SET_RPATH twice
Otherwise we end with a definition like `1 1`, which doesn't trigger the right
conditional.
2017-08-07 11:09:25 -04:00
John Ericson 2493454e13 cc-wrapper: Use set -u throughout
Now is an opportune time to do this, as the infixSalt conversion in
`add-flags.sh` ensures that all the relevant `NIX_*` vars will be
defined even if empty.
2017-08-07 03:05:51 -04:00
John Ericson 42f35503b5 cc-wrapper: Make hygienic
See the added comments for what exactly has been done.
2017-08-07 03:05:50 -04:00