Commit Graph

125 Commits

Author SHA1 Message Date
Duncan Sands
2914ba6ec7 Fix PR7272: when inlining through a callsite with byval arguments,
the newly created allocas may be used by inlined calls, so these
need to have their tail call flags cleared.  Fixes PR7272.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@105255 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-31 21:00:26 +00:00
Nick Lewycky
eb7d818969 Actually run the test. Thanks Daniel Dunbar!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103720 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-13 17:41:06 +00:00
Nick Lewycky
12b927bad0 Add testcase for r103653.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103699 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-13 06:00:14 +00:00
Chris Lattner
83f66fe614 revert r102831. We already delete dead readonly calls in
other places, killing a valid transformation is not the right
answer.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102850 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-01 17:19:38 +00:00
Owen Anderson
1b4a38646f Disable the call-deletion transformation introduced in r86975. Without
halting analysis, it is illegal to delete a call to a read-only function.
The correct solution is almost certainly to add a "must halt" attribute and
only allow deletions in its presence.

XFAIL the relevant testcase for now.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102831 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-01 08:34:28 +00:00
Chris Lattner
bccb41afc8 fix PR5009 by making CGSCCPM realize that a call was devirtualized
if an indirect call site was removed and a direct one was added, not
just if an indirect call site was modified to be direct.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102830 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-01 06:38:43 +00:00
Chris Lattner
1951e108be rename test
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102829 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-01 06:34:13 +00:00
Chris Lattner
6da12e6767 Implement rdar://6295824 and PR6724 with two tiny changes
that can have a big effect :).  The first is to enable the
iterative SCC passmanager juice that kicks in when the
scc passmgr detects that a function pass has devirtualized
a call.  In this case, it will rerun all the passes it 
manages on the SCC, up to the iteration count limit (4). This
is useful because a function pass may devirualize a call, and
we want the inliner to inline it, or pruneeh to infer stuff
about it, etc.

The second patch is to add *all* call sites to the 
DevirtualizedCalls list the inliner uses.  This list is
about to get renamed, but the jist of this is that the 
inliner now reconsiders *all* inlined call sites as candidates
for further inlining.  The intuition is this that in cases 
like this:

f() { g(1); }     g(int x) { h(x); }

We analyze this bottom up, and may decide that it isn't 
profitable to inline H into G.  Next step, we decide that it is
profitable to inline G into F, and do so, which means that F 
now calls H.  Even though the call from G -> H may not have been
profitable to inline, the call from F -> H may be (in this case
because a constant allows folding etc).

In my spot checks, this doesn't have a big impact on code.  For
example, the LLC output for 252.eon grew from 0.02% (from
317252 to 317308) and 176.gcc actually shrunk by .3% (from 1525612
to 1520964 bytes).  252.eon never iterated in the SCC Passmgr,
176.gcc iterated at most 1 time.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102823 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-01 01:15:56 +00:00
Chris Lattner
159528702a The inliner has traditionally not considered call sites
that appear due to inlining a callee as candidates for
futher inlining, but a recent patch made it do this if
those call sites were indirect and became direct.

Unfortunately, in bizarre cases (see testcase) doing this
can cause us to infinitely inline mutually recursive
functions into callers not in the cycle.  Fix this by
keeping track of the inline history from which callsite
inline candidates got inlined from.

This shouldn't affect any "real world" code, but is required
for a follow on patch that is coming up next.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102822 91177308-0d34-0410-b5e6-96231b3b80d8
2010-05-01 01:05:10 +00:00
Chris Lattner
4b7b42c831 Dan recently disabled recursive inlining within a function, but we
were still inlining self-recursive functions into other functions.

Inlining a recursive function into itself has the potential to
reduce recursion depth by a factor of 2, inlining a recursive
function into something else reduces recursion depth by exactly 
1.  Since inlining a recursive function into something else is a
weird form of loop peeling, turn this off.

The deleted testcase was added by Dale in r62107, since then
we're leaning towards not inlining recursive stuff ever.  In any
case, if we like inlining recursive stuff, it should be done 
within the recursive function itself to get the algorithm 
recursion depth win.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102798 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-30 22:37:22 +00:00
Chris Lattner
d46316e7ae no longer xfail
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102220 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-23 22:39:33 +00:00
Chris Lattner
a56c1c5d4c testcase for the bug that required a patch to be reverted.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102195 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-23 18:31:01 +00:00
Chris Lattner
62cc838b90 disable my previous inliner patch, it appears to be busting self-host.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102153 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-23 00:41:03 +00:00
Chris Lattner
fe9af3b1f7 The inliner was choosing to not consider call sites
that appear in the SCC as a result of inlining as candidates
for inlining.  Change this so that it *does* consider call 
sites that change from being indirect to being direct as a
result of inlining.  This allows it to completely 
"devirtualize" the testcase.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102146 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-22 23:37:35 +00:00
Chris Lattner
b61789d4dd add a DEBUG call so that -debug lists when CGSCCPM iterates.
Fix RefreshCallGraph to use CGN->replaceCallEdge instead of hand
rolling its own loop.  replaceCallEdge properly maintains the
reference counts of the nodes, fixing a crash exposed by the
iterative callgraph stuff.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102120 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-22 20:42:33 +00:00
Chris Lattner
08e322db8a Implement (but don't enable) PR6724 and rdar://6295824. In short,
we have RefreshCallGraph detect when a function pass devirtualizes
a call, and have CGSCCPassMgr iterate (up to a count) when this 
happens.  This allows (in the example) GVN to devirtualize the 
call in foo, then the inliner to inline it away.

This is not currently enabled because I haven't done any analysis
on the (potentially substantial) code size or performance impact of
doing this, and guess what, it exposes callgraph updating bugs in
various passes.  This is progress though, and you can play with it
by passing -max-cg-scc-iterations=5 to opt.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@101973 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-21 00:47:40 +00:00
Dan Gohman
d217cfcf46 Revert r101471. For tight recursive functions which have multiple
recursive callsites, inlining can reduce the number of calls by
exponential factors, as it does in
MultiSource/Benchmarks/Olden/treeadd. More involved heuristics
will be needed.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@101969 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-21 00:43:30 +00:00
Dan Gohman
b391bb8947 Disable inlining of recursive calls. It can complicate tailcallelim and
dependent analyses, and increase code size, so doing it profitably would
require more complex heuristics.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@101471 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-16 16:01:18 +00:00
Chris Lattner
d2075586c8 add newlines at the end of files.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@100705 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-07 22:53:17 +00:00
Eric Christopher
f27e6088a3 Reapply r99451 with a fix to move the NoInline check to the cost functions
instead of InlineFunction.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99483 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-25 04:49:10 +00:00
Eric Christopher
0623e90398 Temporarily revert this, it's causing an issue with an internal project.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99451 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-24 23:35:21 +00:00
Chris Lattner
a54934ae9d add some accessors to callsite/callinst/invokeinst to check
for the noinline attribute, and make the inliner refuse to
inline a call site when the call site is marked noinline even
if the callee isn't.  This fixes PR6682.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99341 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-23 22:59:07 +00:00
Dan Gohman
aceba31b7a Delete useless trailing semicolons.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@92740 91177308-0d34-0410-b5e6-96231b3b80d8
2010-01-05 17:55:26 +00:00
Chris Lattner
dbab4dc942 implement a nice little efficiency hack in the inliner. Since we're now
running IPSCCP early, and we run functionattrs interlaced with the inliner,
we often (particularly for small or noop functions) completely propagate
all of the information about a call to its call site in IPSSCP (making a call
dead) and functionattrs is smart enough to realize that the function is
readonly (because it is interlaced with inliner).

To improve compile time and make the inliner threshold more accurate, realize
that we don't have to inline dead readonly function calls.  Instead, just 
delete the call.  This happens all the time for C++ codes, here are some
counters from opt/llvm-ld counting the number of times calls were deleted vs
inlined on various apps:

Tramp3d opt:
  5033 inline                - Number of call sites deleted, not inlined
 24596 inline                - Number of functions inlined
llvm-ld:
  667 inline           - Number of functions deleted because all callers found
  699 inline           - Number of functions inlined

483.xalancbmk opt:
  8096 inline                - Number of call sites deleted, not inlined
 62528 inline                - Number of functions inlined
llvm-ld:
   217 inline           - Number of allocas merged together
  2158 inline           - Number of functions inlined

471.omnetpp:
  331 inline                - Number of call sites deleted, not inlined
 8981 inline                - Number of functions inlined
llvm-ld:
  171 inline           - Number of functions deleted because all callers found
  629 inline           - Number of functions inlined


Deleting a call is much faster than inlining it, and is insensitive to the
size of the callee. :)



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86975 91177308-0d34-0410-b5e6-96231b3b80d8
2009-11-12 07:56:08 +00:00
Kenneth Uildriks
b908f8ad6a Make opt default to not adding a target data string and update tests that depend on target data to supply it within the test
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@85900 91177308-0d34-0410-b5e6-96231b3b80d8
2009-11-03 15:29:06 +00:00
Chris Lattner
c581acbbba Fix a pretty serious misfeature of the inliner: if it inlines a function
with multiple return values it inserts a PHI to merge them all together.
However, if the return values are all the same, it ends up with a pointless
PHI and this pointless PHI happens to really block SRoA from happening in 
at least a silly C++ example written by Doug, but probably others.  This 
fixes rdar://7339069.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@85206 91177308-0d34-0410-b5e6-96231b3b80d8
2009-10-27 05:39:41 +00:00
Chris Lattner
744f19aa15 convert to filecheck.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@85205 91177308-0d34-0410-b5e6-96231b3b80d8
2009-10-27 05:35:35 +00:00
Dan Gohman
2b110caabf Make these tests more interesting by using
-verify-dom-info and -verify-loop-info, which enable additional
(expensive) consistency checks.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@85017 91177308-0d34-0410-b5e6-96231b3b80d8
2009-10-24 23:23:04 +00:00
Chris Lattner
6128df5255 Simplify some code (first hunk) and fix PR5208 (second hunk) by
updating the callgraph when introducing a call.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@84310 91177308-0d34-0410-b5e6-96231b3b80d8
2009-10-17 05:39:39 +00:00
Dale Johannesen
e91b9a3b59 When considering whether to inline Callee into Caller,
and that will make Caller too big to inline, see if it
might be better to inline Caller into its callers instead.
This situation is described in PR 2973, although I haven't
tried the specific case in SPASS.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@83602 91177308-0d34-0410-b5e6-96231b3b80d8
2009-10-09 00:11:32 +00:00
Dan Gohman
cabfea2a58 Eliminate more redundant llvm-as calls.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81540 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-11 18:17:12 +00:00
Dan Gohman
f2f6ce65b7 Change tests from "opt %s" to "opt < %s" so that opt doesn't see the
input filename so that opt doesn't print the input filename in the
output so that grep lines in the tests don't unintentionally match
strings in the input filename.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81537 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-11 18:01:28 +00:00
Dan Gohman
fce288fc91 Eliminate more uses of llvm-as and llvm-dis.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81293 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-09 00:09:15 +00:00
Dan Gohman
3e054fe9ef Use opt -S instead of piping bitcode output through llvm-dis.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81257 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-08 22:34:10 +00:00
Owen Anderson
a0ecbe7821 Fix PR4909, patch by Jakub Staszak.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81250 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-08 19:53:15 +00:00
Dan Gohman
b1e1e82c54 Change these tests to feed the assembly files to opt directly, instead
of using llvm-as, now that opt supports this.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81226 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-08 16:50:01 +00:00
Daniel Dunbar
31ab6e3364 Eliminate uses of %prcontext.
- I'd appreciate it if someone else eyeballs my changes to make sure I captured
   the intent of the test.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81083 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-05 11:35:16 +00:00
Chris Lattner
9535b31483 testcase for PR3601
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@80664 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-01 06:33:49 +00:00
Chris Lattner
46ca76f6bb fix a crash building SPASS by tolerating a callsite that doesn't exist
in the callgraph, see the big comment at the top of the testcase.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@80541 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-31 05:46:59 +00:00
Chris Lattner
b374b90e81 Fix PR4834, a tricky case where the inliner would resolve an
indirect function pointer, inline it, then go to delete the body.
The problem is that the callgraph had other references to the function,
though the inliner had no way to know it, so we got a dangling pointer
and an invalid iterator out of the deal.

The fix to this is pretty simple: stop the inliner from deleting the
function by knowing that there are references to it.  Do this by making
CallGraphNodes contain a refcount.  This requires moving deletion of 
available_externally functions to the module-level cleanup sweep where
it belongs.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@80533 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-31 03:15:49 +00:00
Chris Lattner
199ba42cbf Implement a new optimization in the inliner: if inlining multiple
calls into a function and if the calls bring in arrays, try to merge
them together to reduce stack size.  For example, in the testcase
we'd previously end up with 4 allocas, now we end up with 2 allocas.

As described in the comments, this is not really the ideal solution
to this problem, but it is surprisingly effective.  For example, on
176.gcc, we end up eliminating 67 arrays at "gccas" time and another
24 at "llvm-ld" time.

One piece of concern that I didn't look into: at -O0 -g with
forced inlining this will almost certainly result in worse debug
info.  I think this is acceptable though given that this is a case
of "debugging optimized code", and we don't want debug info to
prevent the optimizer from doing things anyway.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@80215 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-27 06:29:33 +00:00
Chris Lattner
ff517d1ecd the inliner shouldn't crash on this.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@80214 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-27 06:20:45 +00:00
Dan Gohman
fea1dd0804 Remove obsolete -f flags.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79992 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-25 15:38:29 +00:00
Eli Friedman
5877ad7e90 PR4123: don't crash when inlining a call which uses its own result.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@71199 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-08 00:22:04 +00:00
Devang Patel
517576d6f9 While inlining, clone llvm.dbg.func.start intrinsic and adjust
llvm.dbg.region.end instrinsic. This nested llvm.dbg.func.start/llvm.dbg.region.end pair now enables DW_TAG_inlined_subroutine support in code generator.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@69118 91177308-0d34-0410-b5e6-96231b3b80d8
2009-04-15 00:17:06 +00:00
Chris Lattner
266c7bbbbc Add a new "available_externally" linkage type. This is intended
to support C99 inline, GNU extern inline, etc.  Related bugzilla's
include PR3517, PR3100, & PR2933.  Nothing uses this yet, but it
appears to work.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@68940 91177308-0d34-0410-b5e6-96231b3b80d8
2009-04-13 05:44:34 +00:00
Evan Cheng
8c7848f17f If a function is marked alwaysinline, it must be inlined (possibly for correctness). Do so even if the callee has dynamic alloca and the caller doesn't.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@66539 91177308-0d34-0410-b5e6-96231b3b80d8
2009-03-10 07:57:50 +00:00
Chris Lattner
af9985c6b9 Fix a nasty bug (PR3550) where the inline pass could incorrectly mark
calls with the tail marker when inlining them through an invoke.  Patch,
testcase, and perfect analysis by Jay Foad!


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@64364 91177308-0d34-0410-b5e6-96231b3b80d8
2009-02-12 07:06:42 +00:00
Devang Patel
f66d7b5a51 Do not clone llvm.dbg.func.start and corresponding llvm.dbg.region.end during inlining.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@64209 91177308-0d34-0410-b5e6-96231b3b80d8
2009-02-10 07:48:18 +00:00
Nick Lewycky
2eafb8b404 Revert r63600. It didn't fix the bug, it just moved it a bit.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@63618 91177308-0d34-0410-b5e6-96231b3b80d8
2009-02-03 06:30:37 +00:00