RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-25 20:23:11 +01:00
Code Issues Projects Releases Wiki Activity
46,471 Commits 4 Branches 5 Tags 704 MiB
1de5ac9269
Commit Graph

46471 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
Eli Friedman
69845a85a5 Remove a completed optimization. Add a potential optimization I ran 
into.

llvm-svn: 71352
 2009-05-09 08:40:15 +00:00
Duncan Sands
f7af13b2d4 Rename PaddedSize to AllocSize, in the hope that this 
will make it more obvious what it represents, and stop
it being confused with the StoreSize.

llvm-svn: 71349
 2009-05-09 07:06:46 +00:00
Sanjiv Gupta
65a3fde6a5 Use 16 bit arithmetic while retrieving the address of callee's frame during indirect function calls, and set pclath before every call to retrieve the frame address. 
llvm-svn: 71323
 2009-05-09 05:11:19 +00:00
Evan Cheng
bf67b0edef Factor out code that optimize loop terminating condition. 
llvm-svn: 71305
 2009-05-09 01:08:24 +00:00
Dan Gohman
3de703ce43 Don't attempt to handle unsized types in ScalarEvolution's GEP analyzer. 
llvm-svn: 71302
 2009-05-09 00:14:52 +00:00
Evan Cheng
06b0d3879e Enable loop bb placement optimization. 
llvm-svn: 71291
 2009-05-08 23:35:49 +00:00
Dan Gohman
141989d3c2 Fix bogus overflow checks by replacing them with actual 
overflow checks.

llvm-svn: 71284
 2009-05-08 23:11:16 +00:00
Evan Cheng
8e26879521 PPC::B and PPC::BCC's target operand may be an immediate. 
llvm-svn: 71282
 2009-05-08 23:09:25 +00:00
Mike Stump
5deb77232a In non-pic builds, we can use -mdynamic-no-pic for a little more speed. 
llvm-svn: 71281
 2009-05-08 23:08:58 +00:00
Dan Gohman
98da279d6d Use .td for tablegen files, not .ll. 
llvm-svn: 71277
 2009-05-08 23:01:28 +00:00
Mike Stump
9779d47f7a Avoid warning in release-asserts build. 
llvm-svn: 71275
 2009-05-08 22:53:06 +00:00
Bill Wendling
1dccca6452 Mirror how Fast ISel determines if a region.end intrinsic is the end of an 
inlined function or the end of a function. Before, this was never executing the
"inlined" version of the Record method.

This will become important once the inlined Dwarf writer patch lands.

llvm-svn: 71268
 2009-05-08 21:14:49 +00:00
Dan Gohman
603f022049 Fold trunc casts into add-recurrence expressions, allowing the 
add-recurrence to be exposed. Add a new SCEV folding rule to
help simplify expressions in the presence of these extra truncs.

llvm-svn: 71264
 2009-05-08 21:03:19 +00:00
Bill Wendling
6ad9e22d42 Compute the offsets of the compile units. We need this so that when we emit a 
concrete instance of an inlined function, we can get the actual address of the
abstract instance inside of the compile unit.

This isn't currently used, but will be by a future check-in.

llvm-svn: 71263
 2009-05-08 21:03:15 +00:00
Dan Gohman
6eaf7850d8 Fix another bug in r71252. This code supports GetElementPtr 
constant exprs as well as instructions.

llvm-svn: 71262
 2009-05-08 20:58:38 +00:00
Bill Wendling
3bb11df772 Fix typo. 
llvm-svn: 71260
 2009-05-08 20:49:29 +00:00
Dan Gohman
f3461c73e8 Add memoization for getSCEVAtScope results for instructions 
which are not analyzed with SCEV techniques, which can require
brute-forcing through a large number of instructions. This
fixes a massive compile-time issue on 400.perlbench (in
particular, the loop in MD5Transform).

llvm-svn: 71259
 2009-05-08 20:47:27 +00:00
Dan Gohman
1307f44388 Make the SCEV* form of getSCEVAtScope public, to allow ScalarEvolution 
clients to use it.

llvm-svn: 71258
 2009-05-08 20:38:54 +00:00
Bill Wendling
d7428b0d9c Minor clean ups. No functionality change. 
llvm-svn: 71256
 2009-05-08 20:38:02 +00:00
Dan Gohman
6d21c4110b Fix an error from r71252. 
llvm-svn: 71255
 2009-05-08 20:36:47 +00:00
Bill Wendling
5de4207c68 Print out nicer dump info for DIDescriptor. 
llvm-svn: 71253
 2009-05-08 20:28:06 +00:00
Dan Gohman
afd53ad1a3 Factor out the code for creating SCEVs for GEPs into a 
separate function.

llvm-svn: 71252
 2009-05-08 20:26:55 +00:00
Dan Gohman
34a6d38d54 Implement several new SCEV folding rules for UDiv SCEVs. 
This fixes an old FIXME, and is needed by some upcoming changes.

llvm-svn: 71247
 2009-05-08 20:18:49 +00:00
Dan Gohman
13bdd062ea Add a getExitBlock utility function to LoopInfo which returns the 
exit block of a loop, if there is exactly one, similar to
getExitingBlock.

llvm-svn: 71245
 2009-05-08 20:07:23 +00:00
Evan Cheng
aadb8051c0 Don't align loop header unless the loop back edge is below the header. 
llvm-svn: 71242
 2009-05-08 19:01:44 +00:00
Anton Korobeynikov
b3dc881070 Factor out cycle-finder code and make it generic. 
llvm-svn: 71241
 2009-05-08 18:51:58 +00:00
Anton Korobeynikov
026d2328a6 Do not emit bit tests if target does not support natively left shift 
llvm-svn: 71240
 2009-05-08 18:51:34 +00:00
Anton Korobeynikov
5461d8e07d Allow 8 bit select in custom inserter 
llvm-svn: 71239
 2009-05-08 18:51:21 +00:00
Anton Korobeynikov
353d4609cf Properly expand libcalls for urem / srem. Also make code more straightforward. 
llvm-svn: 71238
 2009-05-08 18:51:08 +00:00
Anton Korobeynikov
aa7f982935 Typo 
llvm-svn: 71237
 2009-05-08 18:50:54 +00:00
Anton Korobeynikov
1be416b310 Expand UREM / SREM into libcalls 
llvm-svn: 71236
 2009-05-08 18:50:41 +00:00
Anton Korobeynikov
8364529d37 Add 8 bit select 
llvm-svn: 71235
 2009-05-08 18:50:26 +00:00
Chris Lattner
7b2dabcac9 Fix PR4152: asm constraint validation happens before dag combine, so we 
need to work a bit to combine things like (x+c1+c2) into x+c3.

llvm-svn: 71232
 2009-05-08 18:23:14 +00:00
Duncan Sands
8566ed7a4f Add some examples to show the difference between 
the various notions of type size used in LLVM.

llvm-svn: 71230
 2009-05-08 17:49:48 +00:00
Chris Lattner
5e3bb54910 Change 'make install' to install tblgen, for better support of out-of-tree targets, 
patch by Mikael Lepistö!

llvm-svn: 71226
 2009-05-08 17:32:47 +00:00
Chris Lattner
0fd5aea274 fix RewriteStoreUserOfWholeAlloca to use the correct type size 
method, fixing a crash on PR4146.  While the store will 
ultimately overwrite the "padded size" number of bits in memory,
the stored value may be a subset of this size.  This function
only wants to handle the case where all bits are stored.

llvm-svn: 71224
 2009-05-08 15:54:41 +00:00
Evan Cheng
10038ab095 Reverse branch condition only when there is a conditional branch. 
llvm-svn: 71214
 2009-05-08 09:35:53 +00:00
Nick Lewycky
a8f179d44b Add explicit braces to disambiguate nested if/else. Removes a warning. 
llvm-svn: 71211
 2009-05-08 06:57:41 +00:00
Nick Lewycky
8d9ef77adc This transform requires valid TargetData info. Wrap it in 'if (TD)' in 
preparation for the day we use null TargetData when no target is specified.

llvm-svn: 71210
 2009-05-08 06:47:37 +00:00
Evan Cheng
2a1d20b0fb Optimize code placement in loop to eliminate unconditional branches or move unconditional branch to the outside of the loop. e.g. 
///       A:                                                                                                                                                                 
///       ...                                                                                                                                                                
///       <fallthrough to B>                                                                                                                                                 
///                                                                                                                                                                          
///       B:  --> loop header                                                                                                                                                
///       ...                                                                                                                                                                
///       jcc <cond> C, [exit]                                                                                                                                               
///                                                                                                                                                                          
///       C:                                                                                                                                                                 
///       ...                                                                                                                                                                
///       jmp B                                                                                                                                                              
///                                                                                                                                                                          
/// ==>                                                                                                                                                                      
///                                                                                                                                                                          
///       A:                                                                                                                                                                 
///       ...                                                                                                                                                                
///       jmp B                                                                                                                                                              
///                                                                                                                                                                          
///       C:  --> new loop header                                                                                                                                            
///       ...                                                                                                                                                                
///       <fallthough to B>                                                                                                                                                  
///                                                                                                                                                                          
///       B:                                                                                                                                                                 
///       ...                                                                                                                                                                
///       jcc <cond> C, [exit] 

llvm-svn: 71209
 2009-05-08 06:34:09 +00:00
Nick Lewycky
575cf1d259 Add missing #include for "strlen" which is used inline in this header. Fixes 
build under gcc 4.3.

llvm-svn: 71208
 2009-05-08 06:22:25 +00:00
Sanjiv Gupta
3a905e7586 Moved pic16 naming functions to correct place. 
No functionality change.

llvm-svn: 71207
 2009-05-08 04:50:14 +00:00
Eli Friedman
a280375b23 PR4123: don't crash when inlining a call which uses its own result. 
llvm-svn: 71199
 2009-05-08 00:22:04 +00:00
Bob Wilson
d61f4e70d8 Fix pr4100. Do not remove no-op copies when they are dead. The register 
scavenger gets confused about register liveness if it doesn't see them.
I'm not thrilled with this solution, but it only comes up when there are dead
copies in the code, which is something that hopefully doesn't happen much.

Here is what happens in pr4100: As shown in the following excerpt from the
debug output of llc, the source of a move gets reloaded from the stack,
inserting a new load instruction before the move.  Since that source operand
is a kill, the physical register is free to be reused for the destination
of the move.  The move ends up being a no-op, copying R3 to R3, so it is
deleted.  But, it leaves behind the load to reload %reg1028 into R3, and
that load is not updated to show that it's destination operand (R3) is dead.
The scavenger gets confused by that load because it thinks that R3 is live.

Starting RegAlloc of: %reg1025<def,dead> = MOVr %reg1028<kill>, 14, %reg0, %reg0
  Regs have values: 
  Reloading %reg1028 into R3
  Last use of R3[%reg1028], removing it from live set
  Assigning R3 to %reg1025
  Register R3 [%reg1025] is never used, removing it from live set

Alternative solutions might be either marking the load as dead, or zapping
the load along with the no-op copy.  I couldn't see an easy way to do
either of those, though.

llvm-svn: 71196
 2009-05-07 23:47:03 +00:00
Bob Wilson
a46384485b Fix a comment (again). 
llvm-svn: 71180
 2009-05-07 21:20:42 +00:00
Bob Wilson
8028930294 Fix a comment. 
llvm-svn: 71179
 2009-05-07 21:19:45 +00:00
Dan Gohman
ebacd61d7d Revert 71165. It did more than just revert 71158 and it introduced 
several regressions. The problem due to 71158 is now fixed.

llvm-svn: 71176
 2009-05-07 19:46:24 +00:00
Dan Gohman
679947f6ea Perform constant folding on operands of instructions with non-void 
types, such as loads and calls.

llvm-svn: 71175
 2009-05-07 19:43:39 +00:00
Dan Gohman
93c518ffe9 SCEVComplexityCompare's new code was missing SCEVUDivExpr. Implement 
the SCEVUDivExpr case.

llvm-svn: 71173
 2009-05-07 19:23:21 +00:00
Duncan Sands
e90202e388 Revert r70876 and add a testcase (@c7) showing the problem: 
bits captured, but the pointer marked nocapture.  In fact
I now recall that this problem is why only readnone functions
returning void were considered before!  However keep a small
fix that was also in r70876: a readnone function returning
void can result in bits being captured if it unwinds, so
test for this.

llvm-svn: 71168
 2009-05-07 18:08:34 +00:00