- Dec 29, 2009
-
-
Chris Lattner authored
I asked Devang to do back on Sep 27. Instead of going through the MetadataContext class with methods like getMD() and getMDs(), just ask the instruction directly for its metadata with getMetadata() and getAllMetadata(). This includes a variety of other fixes and improvements: previously all Value*'s were bloated because the HasMetadata bit was thrown into value, adding a 9th bit to a byte. Now this is properly sunk down to the Instruction class (the only place where it makes sense) and it will be folded away somewhere soon. This also fixes some confusion in getMDs and its clients about whether the returned list is indexed by the MDID or densely packed. This is now returned sorted and densely packed and the comments make this clear. This introduces a number of fixme's which I'll follow up on. llvm-svn: 92235
-
- Dec 28, 2009
-
-
Chris Lattner authored
llvm-svn: 92234
-
Chris Lattner authored
llvm-svn: 92233
-
Chris Lattner authored
llvm-svn: 92232
-
Chris Lattner authored
llvm-svn: 92231
-
Chris Lattner authored
to go away from IRBuilder.h llvm-svn: 92230
-
Chris Lattner authored
are about to not come in implicitly. llvm-svn: 92228
-
Chris Lattner authored
non-templated IRBuilderBase class. Move that large CreateGlobalString out of line, eliminating the need to #include GlobalVariable.h in IRBuilder.h llvm-svn: 92227
-
Chris Lattner authored
llvm-svn: 92226
-
Chris Lattner authored
doesn't exist already, eliminate registerMDKind. Tidy up a bunch of random stuff. llvm-svn: 92225
-
Chris Lattner authored
and simplify all the clients that use it. llvm-svn: 92224
-
Chris Lattner authored
llvm-svn: 92223
-
Benjamin Kramer authored
llvm-svn: 92222
-
Chris Lattner authored
llvm-svn: 92221
-
Chris Lattner authored
eliminate the temporary smallvector, and only do FindNodeOrInsertPos twice if the first one succeeds and we delete a node. llvm-svn: 92220
-
Chris Lattner authored
llvm-svn: 92219
-
Chris Lattner authored
llvm-svn: 92218
-
Chris Lattner authored
Among other benefits, this doesn't leak the SmallPtrSet, has the verifier code in the verifier pass, actually does the verification at the end, and is considerably simpler. llvm-svn: 92217
-
Chris Lattner authored
llvm-svn: 92216
-
Chris Lattner authored
llvm-svn: 92214
-
Chris Lattner authored
llvm-svn: 92213
-
Chris Lattner authored
llvm-svn: 92212
-
Chris Lattner authored
llvm-svn: 92211
-
Chris Lattner authored
irrelevant internal implementation details to clients. llvm-svn: 92210
-
Chris Lattner authored
of making it a declared part of the value. llvm-svn: 92209
-
Chris Lattner authored
llvm-svn: 92208
-
Chris Lattner authored
other things, this avoids vtable and rtti data for it being splatted in every translation unit that uses it. llvm-svn: 92207
-
Chris Lattner authored
llvm-svn: 92206
-
Zhongxing Xu authored
llvm-svn: 92205
-
Zhongxing Xu authored
llvm-svn: 92204
-
Chris Lattner authored
this speeds up Eonly on the testcase in PR5888 from 30.5s to 0.85s llvm-svn: 92203
-
Chris Lattner authored
not a token number. Fix the reserve logic to get the right amount of space. llvm-svn: 92202
-
Sanjiv Gupta authored
llvm-svn: 92201
-
Sam Weinig authored
Fix for PR5871. Make __PRETTY_FUNCTION__ work for member functions defined in a class local to a function. llvm-svn: 92200
-
Sanjiv Gupta authored
Allow targets to specify the return type of libcalls that are generated for floating point comparisons, rather than hard-coding them as i32. llvm-svn: 92199
-
Bill Wendling authored
llvm-svn: 92198
-
Bill Wendling authored
llvm-svn: 92197
-
Bill Wendling authored
llvm-svn: 92196
-
Bill Wendling authored
llvm-svn: 92195
-
Bill Wendling authored
llvm-svn: 92194
-