History log of /llvm-project/llvm/include/llvm/CodeGen/SelectionDAGNodes.h (Results 676 – 700 of 866)
Revision (<<< Hide revision tags) (Show revision tags >>>) Date Author Comments
# 30978078 24-May-2007 Dan Gohman <gohman@apple.com>

Minor comment cleanups.

llvm-svn: 37321


Revision tags: llvmorg-2.0.0
# eca3b671 15-May-2007 Dan Gohman <gohman@apple.com>

Correct a comment.

llvm-svn: 37058


# 8af6d589 22-Apr-2007 Christopher Lamb <christopher.lamb@gmail.com>

PR400 phase 2. Propagate attributed load/store information through DAGs.

llvm-svn: 36356


# 4e91908f 21-Apr-2007 Lauro Ramos Venancio <lauro.venancio@gmail.com>

X86 TLS: Implement review feedback.

llvm-svn: 36318


# 0c1349e6 21-Apr-2007 Reid Spencer <rspencer@reidspencer.com>

Revert Christopher Lamb's load/store alignment changes.

llvm-svn: 36309


# bff50208 21-Apr-2007 Christopher Lamb <christopher.lamb@gmail.com>

add support for alignment attributes on load/store instructions

llvm-svn: 36301


# 25188898 20-Apr-2007 Lauro Ramos Venancio <lauro.venancio@gmail.com>

Implement "general dynamic", "initial exec" and "local exec" TLS models for
X86 32 bits.

llvm-svn: 36283


# 8b7aab00 17-Apr-2007 Anton Korobeynikov <asl@math.spbu.ru>

Implemented correct stack probing on mingw/cygwin for dynamic alloca's.
Also, fixed static case in presence of eax livin. This fixes PR331

PS: Why don't we still have push/pop instructions? :)
llvm-

Implemented correct stack probing on mingw/cygwin for dynamic alloca's.
Also, fixed static case in presence of eax livin. This fixes PR331

PS: Why don't we still have push/pop instructions? :)
llvm-svn: 36195

show more ...


# c7c4880a 01-Apr-2007 Reid Spencer <rspencer@reidspencer.com>

For PR1297:
Intrinsic functions might need to be generated (if they are overloaded) so
returning a const char* is no longer possible. Change getOperationName to
return a std::string instead.

llvm-sv

For PR1297:
Intrinsic functions might need to be generated (if they are overloaded) so
returning a const char* is no longer possible. Change getOperationName to
return a std::string instead.

llvm-svn: 35543

show more ...


# ac8fe430 23-Mar-2007 Dan Gohman <gohman@apple.com>

Add the 'explicit' keyword to several constructors that accept one
argument that don't appear intended as implicit-conversion operators.

llvm-svn: 35280


# ed4b303c 07-Mar-2007 Anton Korobeynikov <asl@math.spbu.ru>

Refactoring of formal parameter flags. Enable properly use of
zext/sext/aext stuff.

llvm-svn: 35008


# b127b0bc 07-Mar-2007 Evan Cheng <evan.cheng@apple.com>

Add a utility function to test whether a load is unindexed.

llvm-svn: 35004


# dc3d4406 04-Mar-2007 Chris Lattner <sabre@nondot.org>

add a new node

llvm-svn: 34912


# e0c549ee 21-Feb-2007 Jim Laskey <jlaskey@mac.com>

Add new instructions for handling data passed into eh landing pad.

llvm-svn: 34472


# 09575bac 15-Feb-2007 Reid Spencer <rspencer@reidspencer.com>

For PR1195:
Change use of "packed" term to "vector" in comments, strings, variable
names, etc.

llvm-svn: 34300


# de9b8d05 13-Feb-2007 Lauro Ramos Venancio <lauro.venancio@gmail.com>

Add comment about original function argument alignment.

llvm-svn: 34243


# 9af2c86b 04-Feb-2007 Chris Lattner <sabre@nondot.org>

Introduce new UnarySDNode/BinarySDNode/TernarySDNode nodes, which coallocate
their operands with the node itself. This reduces malloc traffic for operand
lists. This reduces isel time on kc++ from

Introduce new UnarySDNode/BinarySDNode/TernarySDNode nodes, which coallocate
their operands with the node itself. This reduces malloc traffic for operand
lists. This reduces isel time on kc++ from 2.6164 to 2.5570s, about 2.3%.

llvm-svn: 33878

show more ...


# 69b02ab5 04-Feb-2007 Chris Lattner <sabre@nondot.org>

Add a new SDNode ctor and InitOperands method. This allows the operands for
an SDNode to be allocated as part of the node itself, instead of being a
separate new[]'d object. Switch HandleSDNode, Lo

Add a new SDNode ctor and InitOperands method. This allows the operands for
an SDNode to be allocated as part of the node itself, instead of being a
separate new[]'d object. Switch HandleSDNode, LoadSDNode, and StoreSDNode
to use this mechanism. This saves one heap allocation and free for each node
of this type that is allocated. This reduces isel time from 2.7638 to 2.6164s
on kc++, which is a 5.6% speedup.

llvm-svn: 33877

show more ...


# 22639f3d 04-Feb-2007 Chris Lattner <sabre@nondot.org>

eliminate the SDNode::setValueTypes method.

llvm-svn: 33876


# f17b4222 04-Feb-2007 Chris Lattner <sabre@nondot.org>

eliminate a bunch of duplicate ctors and helper functions.

llvm-svn: 33875


# edfc7e5f 04-Feb-2007 Chris Lattner <sabre@nondot.org>

move MorphNode to out of line and merge setNodeOperands into it. There is
no behavior or performance change here.

llvm-svn: 33869


# 3bf17b6f 04-Feb-2007 Chris Lattner <sabre@nondot.org>

simplify MorphNodeTo to take a VTList operand.

llvm-svn: 33868


# 486edfbc 04-Feb-2007 Chris Lattner <sabre@nondot.org>

eliminate some extraneous methods in SDNode

llvm-svn: 33867


# 20754cc5 04-Feb-2007 Chris Lattner <sabre@nondot.org>

Give each selectiondag node class a home for it's vtable and rtti info

llvm-svn: 33866


# eda5997c 29-Jan-2007 Nate Begeman <natebegeman@mac.com>

Finish off bug 680, allowing targets to custom lower frame and return
address nodes.

llvm-svn: 33636


1...<<21222324252627282930>>...35