Commit Graph

99 Commits

Author SHA1 Message Date
Bob Wilson 7184d9fc33 Change ARM vfp assembly functions to use unified syntax.
llvm-svn: 138332
2011-08-23 16:40:18 +00:00
Bob Wilson 66cdd22ff4 Refactor DEFINE_COMPILERRT_FUNCTION.
llvm-svn: 138277
2011-08-22 21:49:47 +00:00
Joerg Sonnenberger 8b068943a4 Don't redeclare sr.
llvm-svn: 136424
2011-07-28 23:15:41 +00:00
Joerg Sonnenberger cf2996dc22 Don't redeclare sr.
llvm-svn: 136377
2011-07-28 19:47:33 +00:00
Eric Christopher 1d1809437e Implement mulo<mode>4 for use in signed overflow checking.
Fixes rdar://9219742 and rdar://9218244

llvm-svn: 133284
2011-06-17 20:17:05 +00:00
Anton Korobeynikov 3c5199042a Add some macro magic: user __USER_LABEL_PREFIX__ instead of hardcoding it for each platform
llvm-svn: 129818
2011-04-19 21:22:14 +00:00
Anton Korobeynikov 4322a7d2c6 More fixes for missed includes
llvm-svn: 129807
2011-04-19 20:50:32 +00:00
Anton Korobeynikov 147ecfad82 Add missed include
llvm-svn: 129804
2011-04-19 20:45:29 +00:00
Anton Korobeynikov 0529d4da93 Add missed files from my last commit.
llvm-svn: 129777
2011-04-19 18:14:19 +00:00
Anton Korobeynikov e63da933eb Move abi bits to separate header. Force AAPCS for EABI in accordance with ARM RTABI.
llvm-svn: 129769
2011-04-19 17:52:09 +00:00
Anton Korobeynikov 75e3c1993c Add ARM EABI function aliases to routines
llvm-svn: 129768
2011-04-19 17:51:24 +00:00
Anton Korobeynikov bdadd87c4d Make use of LOCAL_LABEL macro
llvm-svn: 129767
2011-04-19 17:50:42 +00:00
Anton Korobeynikov 1653610449 Get rid of darwin'isms. Provide convenient macro for assembler local lables.
llvm-svn: 129766
2011-04-19 17:50:09 +00:00
Nick Kledzik e7e288c805 don't add version info for static library
llvm-svn: 129489
2011-04-14 00:40:20 +00:00
Nick Kledzik aa60d6ac01 fix typo
llvm-svn: 129389
2011-04-12 21:22:48 +00:00
Nick Kledzik 191c2804e7 <rdar://problem/9185449> update version info for arm
llvm-svn: 129386
2011-04-12 21:11:47 +00:00
Daniel Dunbar 4c70f3e6ac Revert "Split single & double comparison routines into separate implementation
files," for now, I missed some necesary updates.

llvm-svn: 128296
2011-03-25 18:45:39 +00:00
Daniel Dunbar 1102c9ae58 Split single & double comparison routines into separate implementation files,
for consistency.

llvm-svn: 128282
2011-03-25 15:52:51 +00:00
Daniel Dunbar f3798a397d compiler-rt: Split subdf3 and subsf3 out of add implementations, for
consistency.

llvm-svn: 128038
2011-03-21 23:30:19 +00:00
Stephen Canon 32acbef33f slight re-arrangement to maybe pick up one cycle on dual-issue ARM cores
llvm-svn: 128003
2011-03-21 17:35:26 +00:00
Stephen Canon 5abb5c14c4 Carefully written implementations of the 32-bit integer divide and modulus functions for ARM. These are still using a naive digit-by-digit algorithm, but the core loop has been carefully written.
llvm-svn: 127882
2011-03-18 16:35:02 +00:00
Nick Kledzik 47a823b281 <rdar://problem/8914924> implement udivmodsi4 and divmodsi4 for ARM
llvm-svn: 127778
2011-03-17 00:09:13 +00:00
Chris Lattner 15bc34c7c2 don't compile modsi3 into an infinite loop, patch by Matt Johnson!
llvm-svn: 127429
2011-03-10 22:11:46 +00:00
Nick Kledzik fc1586cf9f <rdar://problem/9091300> update Apple packaging of compiler-rt
llvm-svn: 127057
2011-03-05 02:15:16 +00:00
Nick Kledzik 9f1bd06ae9 Add missing <stdint.h> to build properly on Solaris. Patch by Joakim Johansson
llvm-svn: 123006
2011-01-07 19:09:06 +00:00
Howard Hinnant 5b791f6e64 license change
llvm-svn: 119397
2010-11-16 22:13:33 +00:00
Stephen Canon 5f0e6e7b92 Adds an extra explicit cast to fix Bug 7931 and removes codepaths that were never used
llvm-svn: 111269
2010-08-17 19:13:45 +00:00
Nick Kledzik cceb1f2e67 add assembly implementation of modsi3 so compiler does not have to special case a - (a / b) * b optimization
llvm-svn: 109492
2010-07-27 06:24:32 +00:00
Stephen Canon 8770d3d367 Fix error in rep_clz on non-LP64 targets. Patch by Christoph Gerum
llvm-svn: 109416
2010-07-26 18:17:00 +00:00
Stephen Canon df69264765 Hand-tuning of single-precision soft-float comparison routines for ARM
llvm-svn: 107891
2010-07-08 17:45:05 +00:00
Nick Kledzik 9445ef8f4a fix arvm5 builds
llvm-svn: 107588
2010-07-04 18:29:01 +00:00
Stephen Canon 644ab08dec Initial implementation of double-precision soft-float division, moved a couple utility functions from the multiplications into the utility header
llvm-svn: 107586
2010-07-04 16:53:39 +00:00
Stephen Canon 9ae6fd58c4 Single-precision soft-float division
llvm-svn: 107579
2010-07-04 06:15:44 +00:00
Stephen Canon 909f6c7bc4 Further cleanup of ARM bswap implementations, thanks to John Tytgat
llvm-svn: 107576
2010-07-03 21:47:50 +00:00
Stephen Canon b84e7dc80e Marginally smaller implementations of bswap for ARMv4, ARMv5 that avoid the need to establish a frame
llvm-svn: 107555
2010-07-03 01:00:49 +00:00
Stephen Canon 7973295a19 Renamed unsigned->float routines to match GCC convention
llvm-svn: 107554
2010-07-03 00:56:03 +00:00
Nick Kledzik 209ac43b76 ARM before armv6 did not supprt 'rev' instruction
llvm-svn: 107548
2010-07-03 00:12:47 +00:00
Stephen Canon 04b97967dd [unsigned] int -> float soft-float conversions for single- and double-precision
llvm-svn: 107535
2010-07-02 23:05:46 +00:00
Stephen Canon 4be0d7ead9 float- and double-to-integer soft-float conversions for compiler-rt
llvm-svn: 107524
2010-07-02 22:10:58 +00:00
Stephen Canon e9946d0d3f ... and one more fix to remove some warnings in the new double -> float conversion
llvm-svn: 107409
2010-07-01 18:02:15 +00:00
Stephen Canon 74eaf1f66c Correction to previous commit which mistakenly included older versions of some files; now includes the correct LLVM license header
llvm-svn: 107408
2010-07-01 17:58:24 +00:00
Stephen Canon b1fdde1d01 Adding soft-float comparisons, addition, subtraction, multiplication and negation
llvm-svn: 107400
2010-07-01 15:52:42 +00:00
Nick Kledzik c6cacf9858 sjlj based personality requires a different name
llvm-svn: 103330
2010-05-07 23:44:20 +00:00
Nick Kledzik 8e7ebea0b9 add explicit alignment directives to assure arm code is 4-byte aligned
llvm-svn: 102030
2010-04-21 22:36:23 +00:00
Nick Kledzik 5b0a0629a3 add include of int_lib.h to match change to use compilerrt_abort()
llvm-svn: 101396
2010-04-15 20:37:56 +00:00
Nick Kledzik b3d3ec7091 <rdar://problem/7813254> check enable_execute_stack implementation
llvm-svn: 100036
2010-03-31 19:52:01 +00:00
Daniel Dunbar 587b1d419d Add new build option KERNEL_USE, which compiles with -mkernel and gets propogated to CFLAGS. Use this to call panic() instead of abort() when enabled.
llvm-svn: 100015
2010-03-31 17:00:48 +00:00
Daniel Dunbar f287008791 Use a private compilerrt_abort() define instead of calling abort directly.
- Fiddling with abort directly is annoying given the way we use system includes, although it would be nice to fix this so we could make sure calling abort directly is verboten.

llvm-svn: 100014
2010-03-31 17:00:45 +00:00
Nick Kledzik debb901045 add __sync_synchronize. Needed by compiler when emitting thumb1 with -fno-builtin
llvm-svn: 95132
2010-02-02 21:34:04 +00:00
Nick Kledzik ac02b5a09c fix return type of _Unwind_Exception.exception_cleanup
llvm-svn: 93983
2010-01-20 06:13:20 +00:00