Commit Graph

119 Commits

Author SHA1 Message Date
Andrew Lenharth a565dfda8e Do partial inlining in BU. This resolves more call sites. Also add options to merge in globals during recursion and to back annotate DSNodes when function pointers are resolved. This makes PA work for a whole lot more things (unresolved call sites being what has been killing various DSA based passes)
llvm-svn: 28859
2006-06-19 18:23:36 +00:00
Andrew Lenharth 7f2cee3d3e friendlier error message
llvm-svn: 27829
2006-04-19 15:33:35 +00:00
Chris Lattner 6806c02380 Add explicit iostream #includes
llvm-svn: 25513
2006-01-22 23:19:18 +00:00
Jeff Cohen 5f4ef3c5a8 Eliminate all remaining tabs and trailing spaces.
llvm-svn: 22523
2005-07-27 06:12:32 +00:00
Chris Lattner ecac782786 Correctly handle global-argument aliases induced in main
llvm-svn: 21537
2005-04-25 19:16:31 +00:00
Misha Brukman 01808caded Remove trailing whitespace
llvm-svn: 21416
2005-04-21 21:13:18 +00:00
Chris Lattner 4c5c96f35f add support for taking and resolving the address of free.
llvm-svn: 21396
2005-04-21 16:09:43 +00:00
Chris Lattner 5ecec08ab5 Grow the EQ classes for globals at the end of the BU pass. This shrinks
memory usage in the TD pass for 254.gap from 31.3MB to 3.9MB.

llvm-svn: 20834
2005-03-25 16:45:43 +00:00
Chris Lattner 3322e28f75 remove a debugging timer.
llvm-svn: 20827
2005-03-25 00:06:09 +00:00
Chris Lattner 278bac843c Two changes here:
1. Instead of copying Local graphs to the BU graphs to start with, use
     spliceFrom to do the job (which is constant time in this case).  On
     176.gcc, this chops off .17s from the bu pass.
  2. When building SCC graphs, simplify the logic and use spliceFrom to
     do the heavy lifting, instead of cloneInto/delete.  This slices
     another .14s off 176.gcc.

llvm-svn: 20826
2005-03-25 00:05:04 +00:00
Chris Lattner 923d925953 don't bother |'ing in 0's
llvm-svn: 20815
2005-03-24 18:42:51 +00:00
Chris Lattner 19d7f2883d Fix a crash while promoting a value out of a loop from a global variable
when using ds-aa

llvm-svn: 20802
2005-03-24 04:22:04 +00:00
Chris Lattner 9f9da538d6 a hack to allow count-aa to work with ds-aa :(
llvm-svn: 20791
2005-03-23 21:59:34 +00:00
Chris Lattner ca174306f0 Mark external globals incomplete in the BU Globals graph, fixing
Analysis/DSGraph/2005-03-22-IncompleteGlobal.ll

llvm-svn: 20773
2005-03-22 22:10:22 +00:00
Chris Lattner fcae88f280 Now that the dead ctor is gone, nothing uses the old node mapping exported by
cloneInto: make it an internally used mapping.

llvm-svn: 20760
2005-03-22 00:36:51 +00:00
Chris Lattner ea2c02bcca now that the second argument is always this->ReturnNodes, don't bother passing it.
llvm-svn: 20758
2005-03-22 00:29:44 +00:00
Chris Lattner 19944fade3 remove the second argument to cloneInto
llvm-svn: 20754
2005-03-22 00:21:05 +00:00
Chris Lattner 040426aee0 Fix a major problem where we didn't add call graph edges for call sites with
more than 1 callee.  This fixes Analysis/DSGraph/FunctionPointerTable-const.ll

llvm-svn: 20740
2005-03-21 20:20:49 +00:00
Chris Lattner b1cabb6541 remove some pointless asserts
llvm-svn: 20713
2005-03-20 04:29:39 +00:00
Chris Lattner 5ae215248f Transform BU pass to not use the horrible DSCallSiteIterator class.
llvm-svn: 20708
2005-03-20 02:42:07 +00:00
Chris Lattner 9379e69af9 Create an equivalence class of global variables that DSA will never be able
to tell apart anyway, and only track the leader for of these equivalence
classes in our graphs.

This dramatically reduces the number of GlobalValue*'s that appear in scalar
maps, which A) reduces memory usage, by eliminating many many scalarmap entries
and B) reduces time for operations that need to execute an operation for each
global in the scalar map.

As an example, this reduces the memory used to analyze 176.gcc from 1GB to
511MB, which (while it's still way too much) is better because it doesn't hit
swap anymore.  On eon, this shrinks the local graphs from 14MB to 6.8MB,
shrinks the bu+td graphs of povray from 50M to 40M, shrinks the TD graphs of
130.li from 8.8M to 3.6M, etc.

This change also speeds up DSA on large programs where this makes a big
difference.  For example, 130.li goes from 1.17s -> 0.56s, 134.perl goes
from 2.14 -> 0.93s, povray goes from 15.63s->7.99s (!!!).

This also apparently either fixes the problem that caused DSA to crash on
perlbmk and gcc, or it hides it, because DSA now works on these.  These
both take entirely too much time in the TD pass (147s for perl, 538s for
gcc, vs 7.67/5.9s in the bu pass for either one), but this is a known
problem that I'll deal with later.

llvm-svn: 20696
2005-03-19 22:23:45 +00:00
Chris Lattner bed1c18a44 do not bother inlining nullary functions without return values. The only
effect these calls can have is due to global variables, and these passes
all use the globals graph to capture their effect anyway.  This speeds up
the BU pass very slightly on perlbmk, reducing the number of dsnodes
allocated from 98913 to 96423.

llvm-svn: 20676
2005-03-18 23:19:47 +00:00
Chris Lattner 21a79edb27 fix crashes when we only have a prototype for main.
llvm-svn: 20627
2005-03-15 22:10:04 +00:00
Chris Lattner 92d0c1c1ba Start using retnodes_* for iteration.
llvm-svn: 20618
2005-03-15 16:55:04 +00:00
Chris Lattner b9e2a4dec0 Make sure to remove incomplete markers before we add to them! :)
llvm-svn: 20585
2005-03-13 20:32:26 +00:00
Chris Lattner 2eff970892 After finishing BU analysis, move all global variables from the globals
graph into main and mark them complete.

llvm-svn: 20583
2005-03-13 20:15:06 +00:00
Chris Lattner 863d9c9068 IndCallGraphMap is now a pointer to a new'd map.
llvm-svn: 20065
2005-02-07 16:09:15 +00:00
Chris Lattner 4585abd549 If we have an indirect call site that calls N functions, inline the N functions
into a temporary graph, remember it for later, then inline the tmp graph into
the call site.

In the case where there are other call sites to the same set of functions, this
permits us to just inline the temporary graph instead of all of the callees.

This turns N*M inlining situations into an N+M inlining situation.

llvm-svn: 20036
2005-02-04 19:59:49 +00:00
Chris Lattner 16a53243c6 Eliminate some duplicated debug code
llvm-svn: 19980
2005-02-01 21:55:40 +00:00
Chris Lattner 5ce51e83af Eliminate self-recursion as a special case.
llvm-svn: 19979
2005-02-01 21:49:43 +00:00
Chris Lattner 50cebc6b9b Eliminate use of DSCallSiteIterator in key loop. This is a half step to
a tasty speedup.

llvm-svn: 19978
2005-02-01 21:37:27 +00:00
Chris Lattner d6106a41fe Do not revisit nodes in the SCC traversal. This speeds up the BU pass a bit.
llvm-svn: 19968
2005-02-01 17:35:52 +00:00
Chris Lattner 2977b857c2 Fix some scary bugs that VC++ detected.
llvm-svn: 19941
2005-01-31 00:10:45 +00:00
Chris Lattner a1b39fa5ad * Make some methods more const correct.
* Change the FunctionCalls and AuxFunctionCalls vectors into std::lists.
  This makes many operations on these lists much more natural, and avoids
  *exteremely* expensive copying of DSCallSites (e.g. moving nodes around
  between lists, erasing a node from not the end of the vector, etc).

With a profile build of analyze, this speeds up BU DS from 25.14s to
12.59s on 176.gcc.  I expect that it would help TD even more, but I don't
have data for it.

This effectively eliminates removeIdenticalCalls and children from the
profile, going from 6.53 to 0.27s.

llvm-svn: 19939
2005-01-30 23:51:02 +00:00
Chris Lattner 2e2edef9c6 Make -ds-aa more useful, allowing it to be updated as xforms hack on the program.
llvm-svn: 19818
2005-01-24 20:00:14 +00:00
Chris Lattner 068c0cf56a Fix a bug that was preventing povray and namd from pool allocating correctly.
llvm-svn: 17632
2004-11-08 21:08:46 +00:00
Chris Lattner e3c00e9eee Fix comment
llvm-svn: 17377
2004-10-31 21:54:51 +00:00
Chris Lattner 4f2cf030e8 'Pass' should now not be derived from by clients. Instead, they should derive
from ModulePass.  Instead of implementing Pass::run, then should implement
ModulePass::runOnModule.

llvm-svn: 16436
2004-09-20 04:48:05 +00:00
Reid Spencer 7c16caa336 Changes For Bug 352
Move include/Config and include/Support into include/llvm/Config,
include/llvm/ADT and include/llvm/Support. From here on out, all LLVM
public header files must be under include/llvm/.

llvm-svn: 16137
2004-09-01 22:55:40 +00:00
Chris Lattner 97cf20e1b8 Headers moved
llvm-svn: 14665
2004-07-07 06:35:22 +00:00
Chris Lattner 8db52980b7 Minor changes, remove some debugging code that got checked in somehow.
Make sure to scope the NodeMap passed into cloneInto so that it doesn't point
to nodes that are deleted.  Add some FIXME's for future performance enhancements.

llvm-svn: 12115
2004-03-04 17:05:28 +00:00
Chris Lattner 06a573f63d Only clone global nodes between graphs if both graphs have the global.
llvm-svn: 11928
2004-02-27 20:05:15 +00:00
Chris Lattner ab66f3d143 Instead of cloning the globals for main into the globals graph at the end of
BU propagation, clone the globals into the GG of EACH FUNCTION that finishes
processing!  The GlobalsGraph *must* include all globals and effects from
all functions in the program.  Fixing this makes pool allocation work better
on 175.vpr, but it still ultimately crashes.

llvm-svn: 11686
2004-02-21 00:30:28 +00:00
Chris Lattner 6a4e39677e There is no need to merge the globals graph into the function graphs at the
end of the BU and CBU passes.  The globals will be marked incomplete, so it
doesn't matter if they are missing some info, and merging isn't guaranteed
to bring everything in anyway!

llvm-svn: 11684
2004-02-20 23:52:15 +00:00
Chris Lattner 2a6802ff41 When we complete the bottom-up pass, make sure to merge the globals in 'main' into
the globals graph.

llvm-svn: 11562
2004-02-17 19:06:47 +00:00
Chris Lattner 9ab85275ef 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.

llvm-svn: 11178
2004-02-08 01:51:48 +00:00
Chris Lattner 2f58ffc571 Avoid referencing deleted DSgraphs when merging an SCC into a larger SCC. This
fixes the crash in 176.gcc.

llvm-svn: 11033
2004-01-31 21:02:18 +00:00
Chris Lattner b529730af0 Get clone flags right, so we don't build InlinedGlobals only to clear them
llvm-svn: 10984
2004-01-27 21:50:41 +00:00
Chris Lattner 78c36d2112 Initial support for implementing clonePartiallyInto in terms of cloneReachableSubgraph, though this support is currently disabled.
llvm-svn: 10970
2004-01-23 01:44:53 +00:00
Chris Lattner 71d3377010 Minor code cleanup
llvm-svn: 9962
2003-11-13 05:04:19 +00:00