a new ilist_node class, and remove them. Unlike alist_node,
ilist_node doesn't attempt to manage storage itself, so it avoids
the associated problems, including being opaque in gdb.
Adjust the Recycler class so that it doesn't depend on alist_node.
Also, change it to use explicit Size and Align parameters, allowing
it to work when the largest-sized node doesn't have the greatest
alignment requirement.
Change MachineInstr's MachineMemOperand list from a pool-backed
alist to a std::list for now.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@54146 91177308-0d34-0410-b5e6-96231b3b80d8
use a timer group for the timers in SelectionDAGISel. Also,
Split scheduling out from emitting, to give each their own
timer.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@53476 91177308-0d34-0410-b5e6-96231b3b80d8
for handling bookkeeping for deleted objects, as well as the alist class
template, for keeping lists of objects allocated from Recyclers, and some
related utilities.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@53210 91177308-0d34-0410-b5e6-96231b3b80d8
of after, so that any reallocation it does doesn't get counted for the pass
being timed. This probably doesn't account for a timing discrepancy I was
looking into, but I'm fixing it anyway.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@52693 91177308-0d34-0410-b5e6-96231b3b80d8
multiplicative inverse of a given number. Modify udivrem to allow input and
output pairs of arguments to overlap. Patch is based on the work by Chandler
Carruth.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@52638 91177308-0d34-0410-b5e6-96231b3b80d8
over-shift-right should return -1. So here it should be signed-extended,
when bitwidth larger than 64.
test case: llvm/test/ExecutionEngine/2008-06-05-APInt-OverAShr.ll
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@51999 91177308-0d34-0410-b5e6-96231b3b80d8
several things that were neither in an anonymous namespace nor static
but not intended to be global.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@51017 91177308-0d34-0410-b5e6-96231b3b80d8
be truncated to 32 bits. This fixes the recent Benchmarks/McCat/09-vor
regression on x86-64, among other things.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@50372 91177308-0d34-0410-b5e6-96231b3b80d8
ppc long double, which doesn't work.
This may break some stuff temporarily, but I want to
avoid the propagation of code that assumes this works.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@49983 91177308-0d34-0410-b5e6-96231b3b80d8
start of a filename, not a filename+length. All clients can produce a
null terminated name, and the system api's require null terminated
strings anyway.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@49041 91177308-0d34-0410-b5e6-96231b3b80d8
1) stop using MappedFile.
2) if profitable use the sys::path::MapInFilePages api to
read the file.
3) otherwise fallback to read.
When sys::path::MapInFilePages is implemented, this provides
several benefits:
#1: this avoids fragmenting memory for small files.
#2: this avoids extraneous stat calls when the file size is known.
#3: this only keeps the file descriptor open while reading the
file, not for the duration of the lifetime of the memory
buffer. This fixes a serious clang FD 'leak' problem.
I believe that this will work on a win32 machine, but I don't have
one to test on. I'd appreciate it if someone could check.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@49031 91177308-0d34-0410-b5e6-96231b3b80d8
not an mmapped file. This more closely matches its requirements and
provides an implicitly null terminated buffer, something this
routine had to emulate itself before.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@49024 91177308-0d34-0410-b5e6-96231b3b80d8
profile of the APSInt object. This caused unexpected Profile collisions where
none should have occurred.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@47338 91177308-0d34-0410-b5e6-96231b3b80d8
we had reached the "fake bucket" after the last bucket, allowing the iterator
in some cases to run off the end of the hashtable.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@47178 91177308-0d34-0410-b5e6-96231b3b80d8
Added member template "Add" to FoldingSetNodeID that allows "adding" arbitrary
objects to a profile via dispatch to FoldingSetTrait<T>::Profile().
Removed FoldingSetNodeID::AddAPFloat and FoldingSetNodeID::APInt, as their
functionality is now replaced using the above mentioned member template.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46957 91177308-0d34-0410-b5e6-96231b3b80d8
exposed a bug in APFloat's long double->double conversion of
NaNs. Broke several things in the ieee part of gcc testsuite.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46617 91177308-0d34-0410-b5e6-96231b3b80d8
'FoldingSetNodeImpl' (previously 'FoldingSetNodeID' was a typedef of
'FoldingSetNodeImpl::NodeID').
Why? Clients can now easily forward declare 'FoldingSetNodeID' without having
to include FoldingSet.h.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46187 91177308-0d34-0410-b5e6-96231b3b80d8
memory rather than in a copy of the APFloat. This avoids problems
when the destination is wider than our significand and is cleaner.
Also provide deterministic values in all cases where conversion
fails, namely zero for NaNs and the minimal or maximal value
respectively for underflow or overflow.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@43626 91177308-0d34-0410-b5e6-96231b3b80d8
integer conversion. In some such cases this makes us one or two orders
of magnitude faster than NetBSD's libc. Glibc seems to have a similar
fast path.
Also, tighten up some upper bounds to save a bit of memory.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@42984 91177308-0d34-0410-b5e6-96231b3b80d8
memory for the significand once up-front. Also ignore insignificant
trailing zeroes; this saves unnecessary multiplications later.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@42964 91177308-0d34-0410-b5e6-96231b3b80d8
from user input strings.
Such conversions are more intricate and subtle than they may appear;
it is unlikely I have got it completely right first time. I would
appreciate being informed of any bugs and incorrect roundings you
might discover.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@42912 91177308-0d34-0410-b5e6-96231b3b80d8
No compile-time support for constant operations yet,
just format transformations. Make readers and
writers work. Split constants into 2 doubles in
Legalize.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@42865 91177308-0d34-0410-b5e6-96231b3b80d8
arbitrary range of bits embedded in the middle of another bignum.
This kind of operation is desirable in many cases of software
floating point, e.g. converting bignum integers to floating point
numbers of fixed precision (you want to extract the precision most
significant bits).
Elsewhere, add an assertion, and exit the shift functions early if
the shift count is zero.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@42745 91177308-0d34-0410-b5e6-96231b3b80d8
convertFromZeroExtendedInteger can be simplified as it doesn't need
to make a copy of the source bignum.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@42734 91177308-0d34-0410-b5e6-96231b3b80d8
It used to modify its argument in-place.
This interface is saner and the implementation more efficient. It will
be needed for decimal->binary conversion.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@42733 91177308-0d34-0410-b5e6-96231b3b80d8
input. APInt unfortunately zero-extends signed integers, so Dale
modified the function to expect zero-extended input. Make this
assumption explicit in the function name.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@42732 91177308-0d34-0410-b5e6-96231b3b80d8
part widths. Also, return the number of parts actually required to
hold the result's value.
Remove an over-cautious condition from rounding of float->hex conversion.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@42669 91177308-0d34-0410-b5e6-96231b3b80d8
bit width instead of number of words allocated, which
makes it actually work for int->APF conversions.
Adjust callers. Add const to one of the APInt constructors
to prevent surprising match when called with const
argument.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@42210 91177308-0d34-0410-b5e6-96231b3b80d8
double from some of the many places in the optimizers
it appears, and do something reasonable with x86
long double.
Make APInt::dump() public, remove newline, use it to
dump ConstantSDNode's.
Allow APFloats in FoldingSet.
Expand X86 backend handling of long doubles (conversions
to/from int, mostly).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@41967 91177308-0d34-0410-b5e6-96231b3b80d8
access to bits). Use them in place of float and
double interfaces where appropriate.
First bits of x86 long double constants handling
(untested, probably does not work).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@41858 91177308-0d34-0410-b5e6-96231b3b80d8
Add APFloat interfaces to ConstantFP, SelectionDAG.
Fix integer bit in double->APFloat conversion.
Convert LegalizeDAG to use APFloat interface in
ConstantFPSDNode uses.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@41587 91177308-0d34-0410-b5e6-96231b3b80d8
two's complement bignum arithmetic. They could be used to
implement much of APInt, but the idea is they are enough to
implement APFloat as well, which the current APInt interface
is not suited for.
Patch by Neil Booth!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@41124 91177308-0d34-0410-b5e6-96231b3b80d8
contents of the set were small, deallocate and shrink the set. This
avoids having us to memset as much data, significantly speeding up
some pathological cases. For example, this speeds up the verifier
from 0.3899s to 0.0763 (5.1x) on the testcase from PR1432 in a
release build.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@40837 91177308-0d34-0410-b5e6-96231b3b80d8
would result in calling realloc() on a null pointer. Instead, if we encounter
this situation, make a normal call to malloc().
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@40014 91177308-0d34-0410-b5e6-96231b3b80d8
1. Make SmallPtrSet::erase faster in the small case by replacing a memmove
with a pointer copy.
2. Fix a bug where the null terminator at the end of the array in the small
case was not copied
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@37696 91177308-0d34-0410-b5e6-96231b3b80d8
can't use getZExtValue() to extract the low order bits for each digit.
Instead, we need to access the low order word directly.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@37242 91177308-0d34-0410-b5e6-96231b3b80d8
not being generated correctly because the shl operator does not mutate its
object but returns a new value. Also, make the distinction between radix
16 and the others more clear.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@37111 91177308-0d34-0410-b5e6-96231b3b80d8
types:
1. Functions to compute div/rem at the same time.
2. Further assurance that an APInt with 0 bitwidth cannot be constructed.
3. Left and right rotate operations.
4. An exactLogBase2 function which requires an exact power of two or it
returns -1.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@37025 91177308-0d34-0410-b5e6-96231b3b80d8
shift amount is 0. Previously this code would do a lshr by the bit width
which can lead to incorrect results.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@37010 91177308-0d34-0410-b5e6-96231b3b80d8
handlers (like the pass list). My previous fix only supported *new* command
line options, not additions to old ones.
This fixes test/Feature/load_module.ll
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@35935 91177308-0d34-0410-b5e6-96231b3b80d8
Change uses of sys::Path class to sys::PathWithStatus in those places where
the file status information is needed.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@35743 91177308-0d34-0410-b5e6-96231b3b80d8
This dramatically reduce the amount of memory allocated by the commandline stuff
at static init time, changing it to build local data structures when ParseCommandLineOptions
is called. In a dummy empty program that links some llvm libraries, this reduces
the number of malloc'd bytes from 4864 to 3360 on entry to main. Most of that
memory is now allocated by non-commandline related stuff.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@35701 91177308-0d34-0410-b5e6-96231b3b80d8
Make the sys::Path::getFileStatus function more efficient by having it
return a pointer to the FileStatus structure rather than copy it. Adjust
uses of the function accordingly. Also, fix some memory issues in sys::Path.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@35476 91177308-0d34-0410-b5e6-96231b3b80d8
this to a NOTE: because pari/gp results start to get rounded incorrectly
after 192 bits of precision. APInt and pari/gp never differ by more than
1, but APInt is more accurate because it does not lose precision after 192
bits as does pari/gp.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34834 91177308-0d34-0410-b5e6-96231b3b80d8
First round of ConstantRange changes. This makes all CR constructors use
only APInt and not use ConstantInt. Clients are adjusted accordingly.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34756 91177308-0d34-0410-b5e6-96231b3b80d8
Convert ConstantRange class to use APInt internally as its value type for
the constant range, instead of ConstantInt.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34745 91177308-0d34-0410-b5e6-96231b3b80d8
the bit width of negative numbers by computing the minimum bit width for a
negative value. E.g. 0x1800000000000000 could be just 0x8000000000000000
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34695 91177308-0d34-0410-b5e6-96231b3b80d8
1. Add unsigned and signed versions of methods so a "bool" argument doesn't
need to be passed in.
2. Make the various getMin/getMax functions all be inline since they are
so simple.
3. Simplify sdiv and srem code.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34680 91177308-0d34-0410-b5e6-96231b3b80d8
2. Change RoundDoubleToAPInt to take a bit width parameter. Use that
parameter to limit the bit width of the result.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34673 91177308-0d34-0410-b5e6-96231b3b80d8
2. Rewrite operator=(const APInt& RHS) to allow the RHS to be a different
bit width than the LHS. This makes it possible to use APInt as the key
of a DenseMap, as needed for the IntConstants map in Constants.cpp
3. Fix operator=(uint64_t) to clear unused bits in case the client assigns
a value that has more bits than the APInt allows.
4. Assert that bit widths are equal in operator==
5. Revise getHashValue() to put the bit width in the low order six bits.
This should help to make i1 0, i2 0, ... i64 0 all distinct in the
IntConstants DenseMap.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34646 91177308-0d34-0410-b5e6-96231b3b80d8
2. Fix countTrailingZeros to use a faster algorithm.
3. Simplify sext() slightly by using isNegative().
4. Implement ashr using word-at-a-time logic instead of bit-at-a-time
5. Rename locals named isNegative so they don't clash with method name.
6. Fix fromString to compute negated value correctly.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34629 91177308-0d34-0410-b5e6-96231b3b80d8
2. Implement the trunc, sext, and zext operations.
3. Improve fromString to accept negative values as input.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34616 91177308-0d34-0410-b5e6-96231b3b80d8
2. Move comments for methods to .h file, delete them in .cpp file.
3. All places that were doing manual clear of high order bits now call the
clearUnusedBits() method in order to not depend on undefined behavior
of the >> operator when the number of bits shifted equals the word size.
4. Reduced # of loc by using the new result of clearUnusedBits() method.
5. Simplified logic (decreased indentation) in a few places.
6. Added code comments to larger functions that needed them.
7. Added FIXME notes about weak implementations of things (e.g. bit-by-bit
shift right is sub-optimal).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34603 91177308-0d34-0410-b5e6-96231b3b80d8
2. Clean up comments, style, coding standards, etc.
3. Simplify a constructor.
Extended testing revealed some additional bugs in shifting. I'll fix these
tomorrow.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34559 91177308-0d34-0410-b5e6-96231b3b80d8
to be safe.
2. Make internal functions that return a carry/borrow return bool instead
of uint64_t because the carry/borrow can only be in range [0,1].
3. Assert that the pointers to KnuthDiv are all different so that the
result and operands can't overlap.
4. Add debug output to KnuthDiv function.
5. Fix a problem with KnuthDiv by separating the b's complement operation
from the subtraction borrow operation. This fixes a wide range of
division problems, but alas, not all of them.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34554 91177308-0d34-0410-b5e6-96231b3b80d8