ce21a784a4
stage1: More fixes for BE targets
...
* Fix packed struct alignment
* Adjust some tests
2020-04-13 17:35:14 -04:00
20b2457388
only use -nostdinc++ on c++ files
...
This is a follow-up to d27ef1aaef
2020-04-12 01:03:27 -04:00
d27ef1aaef
Zig treats -lstdc++ identically to -lc++
...
Thanks to Michael Dusan for figuring out what was happening here.
closes #5012
2020-04-12 00:50:16 -04:00
22dbeab29d
Merge pull request #4857 from LemonBoy/fix-4777
...
Rewrite the bound checks in slice operator
2020-04-07 14:51:25 -04:00
cc0fca9d83
stage1: Respect the specified name for extern var
...
Extend the logic used for function definitions to variables.
Closes #4947
2020-04-07 14:48:37 -04:00
64d0960244
zig cc: recognize a few more linker options
...
* `--major-image-version`
* `--minor-image-version`
* `--stack`
2020-04-06 19:13:36 -04:00
f6cdc94a50
ir: Fix error checking for vector ops
...
The extra logic that's needed was lost during a refactoring, now it
should be fine.
2020-04-05 18:34:31 -04:00
fe77c38247
ir: Remove unused and commented out code
2020-04-05 18:34:31 -04:00
d2d97e55cc
ir: Support shift left/right on vectors
2020-04-05 18:34:31 -04:00
54ffcf95a8
ir: Support div/mod/rem on vector types
...
Closes #4050
2020-04-05 18:34:31 -04:00
c4a5f519f2
Do not parse native_libc.txt anymore when linking on native target, always run detection of libc.
...
Fixes #4772
2020-04-05 11:01:35 -04:00
11b50e3ad8
change the default ABI of riscv64-linux-musl
...
Before, this would cause a link failure when mixing Zig and C code for
RISC-V targets.
Now, the ABIs match and Zig and C code can be mixed successfully.
I will file a follow-up issue for the ability to deal more explicitly
with ABIs.
closes #4863
2020-04-03 13:13:09 -04:00
855edd2949
ir: Rewrite the bound checks in slice operator
...
Closes #4777
2020-04-03 10:58:39 +02:00
5314641e11
zig cc: support more linker args
2020-04-02 17:29:22 -04:00
e4edc6d118
zig cc: respect -MF -MV -MD options
...
Zig disables its caching and forwards these args when any are provided.
see #4784
2020-04-02 15:47:27 -04:00
783f73c7e3
zig cc properly handles -S flag and .ll, .bc extensions
2020-04-01 16:01:06 -04:00
212e2354b8
stage1: make C++ switch fallthrough an error
...
Make fallthrough an error when compiler supports it. This requires a new
macro that is defined with such compilers to be used as a statement, at
all fallthrough sites:
switch (...) {
case 0:
...
ZIG_FALLTHROUGH;
case 1:
...
break;
default:
...
break;
}
If we ever move to C++17 as minimal requirement, then the macro can be
replaced with `[[fallthrough]];` at statement sites.
2020-04-01 15:56:00 -04:00
0f1f56bb69
Merge pull request #4896 from FireFox317/fix-arm32-stuff
...
fix some nullptr dereferences on arm-linux-musleabhif
2020-04-01 15:55:31 -04:00
d33766e6c7
Make sure that ZigTypeVector and ZigTypeArray have the same memory layout
...
Throughout the stage1 code it is assumed that these have the same layout,
but that was not the case. This caused an issue on 32-bit hardware.
2020-04-01 20:50:13 +02:00
553f0e0546
fixups and revert a few things
2020-04-01 11:56:39 -04:00
7eb938c909
Use length field as passed in stage2 libc_installation instead of relying on zero termination
2020-04-01 10:36:38 -04:00
cfedd3aca2
revert detection of rtti and exceptions
...
This caused link errors in c++ code because it was not correct to pass
these flags to child codegens. And that was the only reason to detect
these flags. Otherwise we can safely rely on non-explicitly-detected
flag forwarding.
2020-03-30 17:42:30 -04:00
f407109070
zig c++: get it working with musl and mingw-w64
2020-03-27 12:38:52 -04:00
db17c0d88c
ability to compile c++ hello world with zig c++
...
closes #4786
2020-03-26 22:48:37 -04:00
fae6cf0961
improved handling of native system directories
...
* `-isystem` instead of `-I` for system include directories
fixes a problem with native system directories interfering with zig's
bundled libc.
* separate Stage2Target.is_native into Stage2Target.is_native_os and
Stage2Target.is_native_cpu.
2020-03-25 20:34:15 -04:00
94f7c56001
riscv: add -mrelax arg for C to work around upstream issue
...
See #4485
2020-03-22 21:47:19 -04:00
e5e5196d8e
Merge remote-tracking branch 'origin/master' into llvm10
2020-03-22 20:48:21 -04:00
23c263776c
Merge remote-tracking branch 'origin/master' into llvm10
2020-03-22 15:09:29 -04:00
2b65dc1032
zig cc: detect optimization and debug flags
2020-03-21 22:30:46 -04:00
4b0ddb817b
zig cc: better support for the preprocessor option (-E)
2020-03-21 20:32:48 -04:00
a4eaeee720
ability to use zig cc as a drop-in C compiler
...
The basics are working
2020-03-21 15:39:39 -04:00
7438d0fc31
glibc: include ld symbols and proper soname for ld
2020-03-20 14:39:05 -04:00
53b5aa812b
Merge remote-tracking branch 'origin/master' into llvm10
2020-03-19 22:19:24 -04:00
1d7861a36e
fix incorrect sentinel check
2020-03-19 13:18:14 -04:00
72a261b4d3
fix runtime slice of pointer not setting length
2020-03-19 09:53:55 -04:00
4435b05b6b
fix regression when slicing 0-bit pointers
2020-03-19 09:53:55 -04:00
0707be8de8
fixes in semantic analysis needed to support this feature
2020-03-19 09:53:54 -04:00
2182d28cb0
slicing with comptime start and end results in array
...
implements #863
2020-03-19 09:53:51 -04:00
e3c92d0532
ir: More changes to sentinel-terminated const arrays
...
* Don't add an extra slot for the sentinel. Most of the code keeps using
the constant value from the type descriptor, let's harmonize all the
code dealing with sentinels.
* Properly write out sentinel values when reinterpreting pointers at
comptime.
* Allow the reading of the 0th element in a `[0:S]T` type.
2020-03-19 09:53:30 -04:00
1479c28b49
ir: Correct ABI size calculation for arrays
...
Zero-length array with a sentinel may not have zero size.
Closes #4749
2020-03-18 11:10:45 -04:00
a77386eb98
for build-obj with only 1 C file, name .o file after root_out_name
2020-03-14 17:11:51 -04:00
66d7370fac
special case when doing build-obj with just one source file
...
When building an object file from only one source file, instead of
having a two-stage cache system, we special case it and use the cache
directory that the .o file is output to as the final cache directory for
all the build artifacts.
When there are more than 1 source file, the linker has to merge objects
into one, and so the two stage approach makes sens. But in the case of
only one source file, this prevents needlessly copying the object file.
This commit fixes an issue with the previous one, where zig with cache
enabled would print a directory that actually did not have any build
artifacts in it.
2020-03-13 23:59:36 -04:00
656ba530d8
Merge remote-tracking branch 'origin/master' into llvm10
2020-03-13 15:17:53 -04:00
f51bec321b
Merge pull request #4707 from Vexu/small-atomics
...
Support atomic operations with bools and non power of two integers
2020-03-12 18:55:16 -04:00
6dde769279
Simplify stores, use sext for signed ints
2020-03-12 22:02:58 +02:00
bfebc11d06
fix zig-cache to treat cpu-features as raw-bytes
...
- add Stage2Target.cache_hash_len
- add cache_mem(ch, ptr, len)
- update call sites to use { ptr, len }
2020-03-11 19:33:12 -04:00
9262f065f5
Move abi size checking to codegen
2020-03-11 16:48:18 +02:00
ec906a9771
fix codegen, update docs
2020-03-11 13:55:52 +02:00
21809c3300
support non power of two integers in atomic ops
2020-03-11 09:24:53 +02:00
8dc188ebe0
support atomic operations with bools
2020-03-10 22:33:32 +02:00
LemonBoy