Chris Lattner [Mon, 9 Feb 2004 04:36:50 +0000 (04:36 +0000)]
Increase encapsulation of the StructType class, eliminating the getElementTypes() member
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11227
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 9 Feb 2004 04:14:46 +0000 (04:14 +0000)]
Method is gone
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11226
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 9 Feb 2004 04:14:01 +0000 (04:14 +0000)]
Start using the new and improve interface to FunctionType arguments
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11224
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 9 Feb 2004 04:12:57 +0000 (04:12 +0000)]
Improve encapsulation in the FunctionType class, by adding param_(iterator/begin/end)
members, and eliminating the getParamTypes() method, and the associated typedef.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11223
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 9 Feb 2004 03:22:32 +0000 (03:22 +0000)]
This #include is not needed, it should have been removed with the last patch
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11222
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 9 Feb 2004 03:19:29 +0000 (03:19 +0000)]
Instead of searching the entire type graph for a type to determine if it
contains the type we are looking for, just search the immediately used types.
We can only do this because we keep the "current" type in the nesting level
as we decrement upreferences.
This change speeds up the testcase in PR224 from 50.4s to 22.08s, not
too shabby.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11221
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 9 Feb 2004 03:03:10 +0000 (03:03 +0000)]
Upreferences are always OpaqueTypes, meaning that it is impossible for a non-abstract
type from containing one. This speeds up the asmparser on the testcase in PR224 from
61->50s.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11220
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 9 Feb 2004 02:12:04 +0000 (02:12 +0000)]
Another nice speedup for the register allocator. This time, we replace
the Virt2PhysRegMap std::map with an std::vector. This speeds up the
register allocator another (almost) 40%, from .72->.45s in a release build
of LLC on 253.perlbmk.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11219
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 9 Feb 2004 01:47:10 +0000 (01:47 +0000)]
Add a new (hidden) option that is useful for profiling.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11218
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 9 Feb 2004 01:43:23 +0000 (01:43 +0000)]
Ugh, perform an optimization that GCC should be able to do itself. This
speeds up livevar from .48/.32s -> .45/.31s in LLC on perlbmk
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11217
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 9 Feb 2004 01:35:21 +0000 (01:35 +0000)]
Only do stuff for the REAL number of physical registers we have, not 1024.
This speeds up live variables a lot, from .60/.39s -> .47/.26s in LLC, for
the first/second pass respectively.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11216
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 9 Feb 2004 01:26:13 +0000 (01:26 +0000)]
Change the PhysRegsUsed map into a dense array. Seeing that this is a mapping
from physical registers, and they are always dense, it makes sense to not have
a ton of RBtree overhead. This change speeds up regalloclocal about ~30% on
253.perlbmk, from .35s -> .27s in the JIT (in LLC, it goes from .74 -> .55).
Now live variable analysis is the slowest codegen pass. Of course it doesn't
help that we have to run it twice, because regalloclocal doesn't update it,
but even if it did it would be the slowest pass (now it's just the 2x slowest
pass :(
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11215
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 9 Feb 2004 00:59:07 +0000 (00:59 +0000)]
Two problems with these lines of code:
1. The "work" was not in the assert, so it was punishing the optimized release
2. getNamedFunction is _very_ expensive in large programs. It is not designed
to be used like this, and was taking 7% of the execution time of the code
generator on perlbmk.
Since the assert "can never fail", I'm just killing it.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11214
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 9 Feb 2004 00:20:55 +0000 (00:20 +0000)]
The ConstantExpr::getCast call can cause a CPR to be generated. If so,
strip it off.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11213
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 9 Feb 2004 00:15:41 +0000 (00:15 +0000)]
Fix PR215: [bcwriter] Problem compactifying ConstantPointerRefs
Have I ever mentioned how much I _hate_ constantpointerrefs?
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11212
91177308-0d34-0410-b5e6-
96231b3b80d8
Misha Brukman [Sun, 8 Feb 2004 22:27:33 +0000 (22:27 +0000)]
Fix grammar-o.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11210
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 8 Feb 2004 22:23:33 +0000 (22:23 +0000)]
All of spec, including 253.perlbmk should now work. There is still the 254.gap
regression, but that will definitely be fixed by 1.2
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11209
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 8 Feb 2004 22:14:44 +0000 (22:14 +0000)]
Improve compatibility with programs that already have a prototype for 'write',
even if it is wierd in some way.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11207
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 8 Feb 2004 21:52:30 +0000 (21:52 +0000)]
vi failed me again. :)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11206
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 8 Feb 2004 21:52:04 +0000 (21:52 +0000)]
Rename the invoke 'except' destination to the 'unwind' destination
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11205
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 8 Feb 2004 21:48:25 +0000 (21:48 +0000)]
Change the 'exception' destination to the 'unwind' destination. We will always
allow 'except' instead of 'unwind' here though.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11203
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 8 Feb 2004 21:44:31 +0000 (21:44 +0000)]
rename the "exceptional" destination of an invoke instruction to the 'unwind' dest
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11202
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 8 Feb 2004 21:20:42 +0000 (21:20 +0000)]
Add new bug, add note about llvm-db
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11201
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 8 Feb 2004 21:15:59 +0000 (21:15 +0000)]
Fix PR225: [pruneeh] -pruneeh pass removes invoke instructions it shouldn't
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11200
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 8 Feb 2004 20:49:07 +0000 (20:49 +0000)]
splitBasicBlock "does the right thing" now, no reason to reposition it.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11199
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 8 Feb 2004 19:59:05 +0000 (19:59 +0000)]
Bug fixed
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11198
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 8 Feb 2004 19:53:56 +0000 (19:53 +0000)]
Implement proper invoke/unwind lowering.
This fixed PR16 "[lowerinvoke] The -lowerinvoke pass does not insert calls to setjmp/longjmp"
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11195
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 8 Feb 2004 19:40:58 +0000 (19:40 +0000)]
New testcase, code generators should support invoke/unwind
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11192
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 8 Feb 2004 19:33:23 +0000 (19:33 +0000)]
Print out all globals as they are emitted, not just those emitted from
emitGlobals
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11191
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 8 Feb 2004 19:33:07 +0000 (19:33 +0000)]
There is no reason to #define fd
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11190
91177308-0d34-0410-b5e6-
96231b3b80d8
Brian Gaeke [Sun, 8 Feb 2004 18:49:13 +0000 (18:49 +0000)]
add pr105
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11189
91177308-0d34-0410-b5e6-
96231b3b80d8
Brian Gaeke [Sun, 8 Feb 2004 08:01:00 +0000 (08:01 +0000)]
Fix unterminated-quote typo in echo command
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11188
91177308-0d34-0410-b5e6-
96231b3b80d8
Brian Gaeke [Sun, 8 Feb 2004 07:49:04 +0000 (07:49 +0000)]
Document automake as a build dependency because we use aclocal
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11187
91177308-0d34-0410-b5e6-
96231b3b80d8
Brian Gaeke [Sun, 8 Feb 2004 07:44:48 +0000 (07:44 +0000)]
Move rules for regenerating autoconf files into this shell script.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11186
91177308-0d34-0410-b5e6-
96231b3b80d8
Brian Gaeke [Sun, 8 Feb 2004 07:44:30 +0000 (07:44 +0000)]
Remove rules for autoconf'ing files from top-level Makefile
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11185
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 8 Feb 2004 07:30:29 +0000 (07:30 +0000)]
Add a call to 'write' right before the call to abort() in the unwind path.
This causes the JIT, or LLC'd program to print out a nice message, explaining
WHY the program aborted.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11184
91177308-0d34-0410-b5e6-
96231b3b80d8
Brian Gaeke [Sun, 8 Feb 2004 05:49:29 +0000 (05:49 +0000)]
Fix indentation of selectiondag.
I don't know why its indentation has been bugging me, but it has.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11180
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 8 Feb 2004 01:53:10 +0000 (01:53 +0000)]
Add one that I missed
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11179
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 8 Feb 2004 01:51:48 +0000 (01:51 +0000)]
Instead of callign removeTriviallyDeadNodes on the global graph every time
removeDeadNodes is called, only call it at the end of the pass being run.
This saves 1.3 seconds running DSA on 177.mesa (5.3->4.0s), which is
pretty big. This is only possible because of the automatic garbage
collection done on forwarding nodes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11178
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 8 Feb 2004 01:40:40 +0000 (01:40 +0000)]
Remove another unneeded call.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11177
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 8 Feb 2004 01:38:34 +0000 (01:38 +0000)]
This call is no longer needed now that merging does not produce garbage
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11176
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 8 Feb 2004 01:27:18 +0000 (01:27 +0000)]
Substantially improve the DSA code by removing 'forwarding' nodes from
DSGraphs while they are forwarding. When the last reference to the forwarding
node is dropped, the forwarding node is autodeleted. This should simplify
removeTriviallyDead nodes, and is only (efficiently) possible because we are
using an ilist of dsnodes now.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11175
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 8 Feb 2004 01:05:37 +0000 (01:05 +0000)]
Bugfix for ilist conversion. The ilist wants to make an 'end' node which has
G == 0
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11174
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 8 Feb 2004 00:53:26 +0000 (00:53 +0000)]
Switch the Nodes list from being an std::vector<DSNode*> to an ilist<DSNode>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11173
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 8 Feb 2004 00:51:31 +0000 (00:51 +0000)]
One of the 'annoying' things about ilists is that the iterators don't behave
quite the same as for non-intrusive lists of pointers to nodes. To support
transitioning code bases, add a new 'compatibility' iterator.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11172
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 8 Feb 2004 00:23:16 +0000 (00:23 +0000)]
Change to use node_iterators instead of direct access to Nodes
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11171
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 8 Feb 2004 00:22:41 +0000 (00:22 +0000)]
Change to use iterators instead of direct access
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11170
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 8 Feb 2004 00:06:20 +0000 (00:06 +0000)]
Minor change to breakpoint (lack of) support
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11169
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sat, 7 Feb 2004 23:58:05 +0000 (23:58 +0000)]
getNodes() is gone, use node_begin/end instead
Rename stats from dsnode -> dsa
Add a new stat
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11167
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sat, 7 Feb 2004 23:57:26 +0000 (23:57 +0000)]
getNodes() is gone
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11166
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sat, 7 Feb 2004 23:57:09 +0000 (23:57 +0000)]
Abstract out the Nodes collection. Instead of providing a getNodes() method,
provide node_begin/end iterators, which are only guaranteed to be
bidirectional, not random access.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11165
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sat, 7 Feb 2004 23:02:32 +0000 (23:02 +0000)]
Actually USE isForwarding method
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11160
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sat, 7 Feb 2004 22:54:19 +0000 (22:54 +0000)]
As Alkis pointed out to me, I forgot to commit this... :(
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11159
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sat, 7 Feb 2004 22:00:03 +0000 (22:00 +0000)]
There is no need to clone over nodes that are going to be dead anyway
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11157
91177308-0d34-0410-b5e6-
96231b3b80d8
Brian Gaeke [Fri, 6 Feb 2004 22:33:17 +0000 (22:33 +0000)]
Use autoconf answers from config.h (FIXME, should autoconf this file
directly instead).
Fix LockHolder/MutexLocker typo.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11156
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Fri, 6 Feb 2004 20:27:41 +0000 (20:27 +0000)]
Add a missing link
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11154
91177308-0d34-0410-b5e6-
96231b3b80d8
Misha Brukman [Fri, 6 Feb 2004 18:40:35 +0000 (18:40 +0000)]
Fix grammar.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11153
91177308-0d34-0410-b5e6-
96231b3b80d8
Alkis Evlogimenos [Fri, 6 Feb 2004 18:08:18 +0000 (18:08 +0000)]
Increase code clarity.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11151
91177308-0d34-0410-b5e6-
96231b3b80d8
John Criswell [Fri, 6 Feb 2004 16:37:53 +0000 (16:37 +0000)]
Initial checking of the libpng library.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11149
91177308-0d34-0410-b5e6-
96231b3b80d8
John Criswell [Fri, 6 Feb 2004 16:36:42 +0000 (16:36 +0000)]
Initial checking of the zlib library.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11148
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Fri, 6 Feb 2004 06:37:00 +0000 (06:37 +0000)]
Add information about the piece I forgot to write: parameterized tablegen classes
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11147
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Fri, 6 Feb 2004 06:04:25 +0000 (06:04 +0000)]
Syntax hilight examples and add note about emacs/vim mode files
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11146
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Fri, 6 Feb 2004 05:42:53 +0000 (05:42 +0000)]
Add a new document describing TableGen
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11145
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Fri, 6 Feb 2004 03:19:17 +0000 (03:19 +0000)]
Print the record NAME not the record ADDRESS
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11144
91177308-0d34-0410-b5e6-
96231b3b80d8
Alkis Evlogimenos [Fri, 6 Feb 2004 03:15:40 +0000 (03:15 +0000)]
Eliminate uneeded lookups by passing a Virt2PhysMap::iterator instead
of the virtual register to certain functions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11143
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Thu, 5 Feb 2004 23:20:59 +0000 (23:20 +0000)]
Fix another dominator update bug. These bugs keep getting exposed because GCSE
keeps finding more code motion opportunities now that the dominators are correct!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11142
91177308-0d34-0410-b5e6-
96231b3b80d8
Alkis Evlogimenos [Thu, 5 Feb 2004 22:55:25 +0000 (22:55 +0000)]
Change live interval representation. Machine instructions now have two
slots each. As a concequence they get numbered as 0, 2, 4 and so
on. The first slot is used for operand uses and the second for
defs. Here's an example:
0: A = ...
2: B = ...
4: C = A + B ;; last use of A
The live intervals should look like:
A = [1, 5)
B = [3, x)
C = [5, y)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11141
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Thu, 5 Feb 2004 22:33:26 +0000 (22:33 +0000)]
Fix bug updating dominators
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11140
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Thu, 5 Feb 2004 22:33:19 +0000 (22:33 +0000)]
Add debug output
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11139
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Thu, 5 Feb 2004 21:16:42 +0000 (21:16 +0000)]
Bug fixed
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11138
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Thu, 5 Feb 2004 21:12:24 +0000 (21:12 +0000)]
Fix PR223: Loopsimplify incorrectly updates dominator information
The problem is that the dominator update code didn't "realize" that it's
possible for the newly inserted basic block to dominate anything. Because
it IS possible, stuff was getting updated wrong.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11137
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Thu, 5 Feb 2004 21:11:38 +0000 (21:11 +0000)]
New testcase for PR223: Loopsimplify incorrectly updates dominator information
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11136
91177308-0d34-0410-b5e6-
96231b3b80d8
Alkis Evlogimenos [Thu, 5 Feb 2004 20:45:40 +0000 (20:45 +0000)]
We don't need to scan the blocks that we are live-in on every
access. Rather we only have to do it on the creation of the interval.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11135
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Thu, 5 Feb 2004 17:31:37 +0000 (17:31 +0000)]
s/gcse/scalarrepl
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11133
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Thu, 5 Feb 2004 17:20:00 +0000 (17:20 +0000)]
In a "seeing the forest through the trees" kinda situation, I realized that a
complete rewrite of load-vn will make it a bit faster. This changes speeds up
the gcse pass (which uses load-vn) from 25.45s to 0.42s on the testcase in
PR209.
I've also verified that this gives the exact same results as the old one.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11132
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Thu, 5 Feb 2004 05:56:23 +0000 (05:56 +0000)]
This is a big diff with no functionality change. We just reorder some code,
which causes big reindentation. While I'm at it, I fix the fixme by removing
some dead code.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11131
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Thu, 5 Feb 2004 05:51:40 +0000 (05:51 +0000)]
finegrainify namespacification
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11130
91177308-0d34-0410-b5e6-
96231b3b80d8
Tanya Lattner [Thu, 5 Feb 2004 05:04:39 +0000 (05:04 +0000)]
Added missing include.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11129
91177308-0d34-0410-b5e6-
96231b3b80d8
Tanya Lattner [Thu, 5 Feb 2004 04:45:21 +0000 (04:45 +0000)]
Fixed Chris' typo.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11128
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Thu, 5 Feb 2004 00:48:41 +0000 (00:48 +0000)]
qoi bug fixed
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11127
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Thu, 5 Feb 2004 00:36:43 +0000 (00:36 +0000)]
Implement optimizations for handling large basic blocks.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11126
91177308-0d34-0410-b5e6-
96231b3b80d8
Alkis Evlogimenos [Wed, 4 Feb 2004 22:17:40 +0000 (22:17 +0000)]
Modify the two address instruction pass to remove the duplicate
operand of the instruction and thus simplify the register allocation.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11124
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 4 Feb 2004 21:44:26 +0000 (21:44 +0000)]
Minor speedup, don't query ValueMap each time through the loop
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11123
91177308-0d34-0410-b5e6-
96231b3b80d8
Brian Gaeke [Wed, 4 Feb 2004 21:41:23 +0000 (21:41 +0000)]
Always replace instead of appending when creating archive files. It may be
slightly slower, but I think we can handle it, especially if it means
BytecodeLibs are correctly regenerated.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11122
91177308-0d34-0410-b5e6-
96231b3b80d8
Brian Gaeke [Wed, 4 Feb 2004 21:41:10 +0000 (21:41 +0000)]
Include <iosfwd> and <string> instead of <iostream>.
Take away the default iostream argument of createMachineFunctionPrinterPass(),
at Chris's request.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11121
91177308-0d34-0410-b5e6-
96231b3b80d8
Brian Gaeke [Wed, 4 Feb 2004 21:41:01 +0000 (21:41 +0000)]
Take away the default iostream argument of createMachineFunctionPrinterPass(),
at Chris's request.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11120
91177308-0d34-0410-b5e6-
96231b3b80d8
Brian Gaeke [Wed, 4 Feb 2004 21:40:49 +0000 (21:40 +0000)]
Clean out DESTLIBBYTECODE when making clean in runtime.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11119
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 4 Feb 2004 21:33:42 +0000 (21:33 +0000)]
Two changes:
1. Don't scan to the end of alloca instructions in the caller function to
insert inlined allocas, just insert at the top. This saves a lot of
time inlining into functions with a lot of allocas.
2. Use splice to move the alloca instructions over, instead of remove/insert.
This allows us to transfer a block at a time, and eliminates a bunch of
silly symbol table manipulations.
This speeds up the inliner on the testcase in PR209 from 1.73s -> 1.04s (67%)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11118
91177308-0d34-0410-b5e6-
96231b3b80d8
Alkis Evlogimenos [Wed, 4 Feb 2004 17:21:04 +0000 (17:21 +0000)]
IMULri* instructions do not require their first two registers operands
to be the same (IOW they are not two address instructions).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11117
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 4 Feb 2004 04:17:06 +0000 (04:17 +0000)]
Optimize the case where we are inlining a function that contains only one basic block,
and that basic block ends with a return instruction. In this case, we can just splice
the cloned "body" of the function directly into the source basic block, avoiding a lot
of rearrangement and splitBasicBlock's linear scan over the split block. This speeds up
the inliner on the testcase in PR209 from 2.3s to 1.7s, a 35% reduction.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11116
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 4 Feb 2004 03:59:08 +0000 (03:59 +0000)]
Check in header file I forgot before.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11115
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 4 Feb 2004 03:58:28 +0000 (03:58 +0000)]
Adjust to the new BasicBlock ctor, which requires a function parameter
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11114
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 4 Feb 2004 03:57:50 +0000 (03:57 +0000)]
Adjust to the new BB ctor
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11113
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 4 Feb 2004 03:57:34 +0000 (03:57 +0000)]
Delete the BasicBlock ctor that only takes a BasicBlock to insert before. This
fails when the basic block points to the function->end. Instead, require that
the client pass in the function AND the basicblock to insert into.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11112
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 4 Feb 2004 03:21:51 +0000 (03:21 +0000)]
Remove unneeded code now that splitBasicBlock does the "right thing"
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11111
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 4 Feb 2004 03:21:31 +0000 (03:21 +0000)]
When splitting a basic block, insert the new half immediately after the first
half.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11110
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 4 Feb 2004 02:51:48 +0000 (02:51 +0000)]
More refactoring. Move alloca instructions and handle invoke instructions
before we delete the original call site, allowing slight simplifications of
code, but nothing exciting.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11109
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 4 Feb 2004 01:41:09 +0000 (01:41 +0000)]
Move the cloning of the function body much earlier in the inlinefunction
process. The only optimization we did so far is to avoid creating a
PHI node, then immediately destroying it in the common case where the
callee has one return statement. Instead, we just don't create the return
value. This has no noticable performance impact, but paves the way for
future improvements.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11108
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 4 Feb 2004 01:19:43 +0000 (01:19 +0000)]
Give CloneBasicBlock an optional function argument to specify which function
to add the cloned block to. This allows the block to be added to the function
immediately, and all of the instructions to be immediately added to the function
symbol table, which speeds up the inliner from 3.7 -> 3.38s on the PR209.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11107
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Wed, 4 Feb 2004 01:06:38 +0000 (01:06 +0000)]
If changing a parent, don't add then remove the object from the leak detector
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11106
91177308-0d34-0410-b5e6-
96231b3b80d8