Rehash but don't grow when full of tombstones.

Patch by José Fonseca!

llvm-svn: 128566

Rehash but don't grow when full of tombstones.

Patch by José Fonseca!

llvm-svn: 128565

Avoid turning a floating point division with a constant power of two into a denormal multiplication.

Some platforms may treat denormals as zero, on other platforms multiplication
with a subnormal is slower than dividing by a normal.

llvm-svn: 128555

The idea is, that if an ieee 754 float is divided by a power of two, we can
turn the division into a cheaper multiplication. This function sees if we can
get an exact multiplicative inverse for a divisor and returns it if possible.

This is the hard part of PR9587.

I tested many inputs against llvm-gcc's frotend implementation of this
optimization and didn't find any difference. However, floating point is the
land of weird edge cases, so any review would be appreciated.

llvm-svn: 128545

Constant folding support for calls to umul.with.overflow(), basically identical to the smul.with.overflow() code.

llvm-svn: 128379

llvm-svn: 128218

llvm-svn: 128199

llvm-svn: 128098

Properly initialize all fields in CrashReporterCleanupContext.  This caused the buildbot failure earlier.

llvm-svn: 128071

Temporarily stop recovering resources in CrashRecoveryContext while I investigate further why this works on my machine and not on others.

llvm-svn: 128065

Provide a means for CrashRecovery clients to determine if code is currently running while crash recovery cleanups are being processed.

llvm-svn: 128008

- glibc falls back to fork+exec if a file actions object is present.
- On BSDs this saves a malloc.

llvm-svn: 127969

Tweak CrashRecoveryContextCleanup to provide an easy method for clients to select between 'delete' and 'destructor' cleanups, and allow the destructor of CrashRecoveryContextCleanupRegister to be pseudo re-entrant.

llvm-svn: 127929

Tweak CrashRecoveryContext::GetCurrent() to return quickly if 'gCrsahRecoveryEnabled' is false.  This avoids us needing to go to thread local storage for
the performance sensitive case where we are compiling code.

llvm-svn: 127928

llvm-svn: 127891

On MSVCRT and compatible, output of %e is incompatible to Posix by default. Number of exponent digits should be at least 2. "%+03d"

FIXME: Implement our formatter in future!
llvm-svn: 127872

makes valgrind stop complaining about uninitialized variables being read when it
accesses a bitfield (category) that shares its bits with these variables.

llvm-svn: 127871

llvm-svn: 127853

Augment CrashRecoveryContext to have registered "cleanup" objects that can be used to release resources during a crash.

llvm-svn: 127849

instead of copying.

llvm-svn: 127835

Add an argument to APInt's magic udiv calculation to specify the number of bits that are known zero in the divided number.

This will come in handy soon.

llvm-svn: 127828

report_fatal_error() invokes exit(). We know report_fatal_error() might not write messages to stderr when any errors were detected on FD == 2.

llvm-svn: 127726

FIXME: It is a temporal hack. We should detect as many "special file name" as possible.
llvm-svn: 127724

FIXME: We should use sys::fs::unique_file() in future.
llvm-svn: 127723

llvm-svn: 127705

llvm-svn: 127426

llvm-svn: 127417

llvm-svn: 127416

llvm-svn: 127413

support for creating buffers that cover only a part of a file.

llvm-svn: 127409

llvm-svn: 127241

llvm-svn: 127140

inefficient file system buffering if the writes are not a multiple of the desired
buffer size. Avoid this by limiting the large write to a multiple of the buffer
size and copying the remainder into the buffer.

Thanks to Dan for pointing this out.

llvm-svn: 127026

raw_ostream: If writing a string that is larger than the buffer, write it directly instead of doing many buffer-sized writes.

This caps the number of write(2) calls per string to a maximum of 2.

llvm-svn: 127010

llvm-svn: 126558

of a constant had a minor typo introduced when copying it from the book, which
caused it to favor negative approximations over positive approximations in many
cases. Positive approximations require fewer operations beyond the multiplication.

In the case of division by 3, we still generate code that is a single instruction
larger than GCC's code.

llvm-svn: 126097

llvm-svn: 125986

llvm-svn: 125943

llvm-svn: 125804

Add getAllOnesValue of FP numbers to Constants and APFloat.
Add more tests.

llvm-svn: 125776