summaryrefslogtreecommitdiffstats
path: root/llvm/lib/Target/X86
Commit message (Collapse)AuthorAgeFilesLines
* - Refactor the code that resolve basic block references to a TargetJITInfoEvan Cheng2006-07-256-19/+26
| | | | | | | | | | method. - Added synchronizeICache() to TargetJITInfo. It is called after each block of code is emitted to flush the icache. This ensures correct execution on targets that have separate dcache and icache. - Added PPC / Mac OS X specific code to do icache flushing. llvm-svn: 29276
* Can't commute shufps. The high / low parts elements come from different vectors.Evan Cheng2006-07-252-20/+1
| | | | llvm-svn: 29275
* Done.Evan Cheng2006-07-211-5/+0
| | | | llvm-svn: 29262
* This opt is now handled in DAG combine.Evan Cheng2006-07-211-2/+0
| | | | llvm-svn: 29243
* A splat of a vector constant of all zero or all one is the vector constant.Evan Cheng2006-07-201-0/+2
| | | | llvm-svn: 29234
* Missing a space.Evan Cheng2006-07-201-1/+1
| | | | llvm-svn: 29233
* Clean up.Evan Cheng2006-07-201-3/+3
| | | | llvm-svn: 29228
* New entry.Evan Cheng2006-07-191-0/+25
| | | | llvm-svn: 29215
* Do once flag never set to true.Jim Laskey2006-07-191-1/+3
| | | | llvm-svn: 29214
* Tidy up a few things.Jim Laskey2006-07-191-20/+36
| | | | llvm-svn: 29213
* Reduce size of routine. Shrinks .o by 37%.Jim Laskey2006-07-191-421/+489
| | | | llvm-svn: 29210
* Bug#834 ICE (crash in code generator?) when building PCH .Jim Laskey2006-07-191-3/+5
| | | | | | Missing Darwin check in Intel ATT ASM printer. llvm-svn: 29204
* Misc. new entry.Evan Cheng2006-07-191-0/+5
| | | | llvm-svn: 29202
* INC / DEC instructions have shorter code size than ADD32ri8, etc.Evan Cheng2006-07-192-7/+13
| | | | llvm-svn: 29194
* Add an out-of-line virtual method for X86DwarfWriter to give it a home.Chris Lattner2006-07-142-19/+21
| | | | llvm-svn: 29153
* Add information preventing several register class constraints from working.Chris Lattner2006-07-121-1/+9
| | | | | | This implements PR828 and CodeGen/X86/2006-07-12-InlineAsmQConstraint.ll llvm-svn: 29118
* Emit inc / dec of registers as one byte instruction.Evan Cheng2006-07-111-4/+4
| | | | llvm-svn: 29110
* Implement the inline asm 'A' constraint. This implements PR825 andChris Lattner2006-07-112-1/+17
| | | | | | CodeGen/X86/2006-07-10-InlineAsmAConstraint.ll llvm-svn: 29101
* New entry.Evan Cheng2006-07-101-0/+3
| | | | llvm-svn: 29091
* Fixed stack objects do not specify alignments, but their offsets are known.Evan Cheng2006-07-101-5/+12
| | | | | | | Use that information when doing the transformation to merge multiple loads into a 128-bit load. llvm-svn: 29090
* Mark internal function staticChris Lattner2006-07-101-1/+1
| | | | llvm-svn: 29085
* X86 target specific DAG combine: turn build_vector (load x), (load x+4),Evan Cheng2006-07-073-0/+164
| | | | | | | | | | | | | | | | | (load x+8), (load x+12), <0, 1, 2, 3> to a single 128-bit load (aligned and unaligned). e.g. __m128 test(float a, float b, float c, float d) { return _mm_set_ps(d, c, b, a); } _test: movups 4(%esp), %xmm0 ret llvm-svn: 29042
* Reorg. No functionality change.Evan Cheng2006-07-051-183/+195
| | | | llvm-svn: 28999
* Fix JIT on non MacOS X i386 systems.Evan Cheng2006-07-051-4/+0
| | | | llvm-svn: 28992
* Should just use xorps to clear XMM registers for all data types. pxor is ↵Evan Cheng2006-06-291-14/+10
| | | | | | also one byte longer. llvm-svn: 28984
* Let X86CompilationCallback pass previous frame and return address to ↵Evan Cheng2006-06-291-10/+7
| | | | | | X86CompilationCallback2. Remove alloca hack. llvm-svn: 28982
* Add shift and rotate by 1 instructions / patterns.Evan Cheng2006-06-292-0/+128
| | | | llvm-svn: 28980
* Always use xorps to clear XMM registers.Evan Cheng2006-06-291-1/+1
| | | | llvm-svn: 28979
* Move .literal4 and .literal8 support into AsmPrinter.cppEvan Cheng2006-06-292-57/+2
| | | | llvm-svn: 28978
* Hide x86 symbolsChris Lattner2006-06-284-4/+8
| | | | llvm-svn: 28976
* Doh.Evan Cheng2006-06-281-1/+1
| | | | llvm-svn: 28963
* Oops. Need to keep CP index.Evan Cheng2006-06-282-17/+16
| | | | llvm-svn: 28958
* Darwin puts float and double literal constants into literal4 and literal8 ↵Evan Cheng2006-06-282-0/+58
| | | | | | sections. llvm-svn: 28957
* Remove dead code.Evan Cheng2006-06-271-6/+0
| | | | llvm-svn: 28938
* Simplify X86CompilationCallback: always align to 16-byte boundary; don't ↵Evan Cheng2006-06-243-34/+44
| | | | | | save EAX/EDX if unnecessary. llvm-svn: 28910
* Add and sort "sections" in debug lines. This always stepping throughJim Laskey2006-06-231-4/+1
| | | | | | | code in sections other than ".text", including weak sections like ctors and dtors. llvm-svn: 28909
* Eliminate unneeded parameter.Evan Cheng2006-06-221-8/+4
| | | | llvm-svn: 28907
* variable_ops instructions such as call can have any number of operands.Evan Cheng2006-06-211-1/+1
| | | | llvm-svn: 28906
* Remove some ugly now-redundant casts.Chris Lattner2006-06-201-54/+54
| | | | llvm-svn: 28864
* Fix some mismatched type constraintsChris Lattner2006-06-201-6/+6
| | | | llvm-svn: 28862
* Minor clean up.Evan Cheng2006-06-191-6/+3
| | | | llvm-svn: 28860
* A new entry.Evan Cheng2006-06-171-0/+2
| | | | llvm-svn: 28848
* Later models likely to have Yonah like attributes.Evan Cheng2006-06-161-1/+2
| | | | llvm-svn: 28843
* Don't pass target name into TargetData anymore, it is never used or needed.Chris Lattner2006-06-161-1/+1
| | | | | | | Remove explicit casts to std::string now that there is no overload resolution issues in the TargetData ctors. llvm-svn: 28830
* Add a note that Nate noticed.Chris Lattner2006-06-151-0/+34
| | | | llvm-svn: 28808
* Type of vector extract / insert index operand should be iPTR.Evan Cheng2006-06-152-18/+18
| | | | llvm-svn: 28796
* X86 call instructions can take variable number of operands. Parameters ofEvan Cheng2006-06-141-5/+6
| | | | | | vector types are passed via XMM registers. llvm-svn: 28789
* add a noteChris Lattner2006-06-141-0/+29
| | | | llvm-svn: 28787
* Add argument registers to the end of call operand list (partial fix).Evan Cheng2006-06-141-1/+15
| | | | llvm-svn: 28783
* Place dwarf headers at earliest possible point. Well behaved when skippingJim Laskey2006-06-141-1/+4
| | | | | | functions. llvm-svn: 28781
OpenPOWER on IntegriCloud