Commit Graph

43696 Commits

Author SHA1 Message Date
Dale Johannesen
e6ec25543f Fix the time regression I introduced in 464.h264ref with
my last patch to this file.

The issue there was that all uses of an IV inside a loop
are actually references to Base[IV*2], and there was one
use outside that was the same but LSR didn't see the base
or the scaling because it didn't recurse into uses outside
the loop; thus, it used base+IV*scale mode inside the loop
instead of pulling base out of the loop.  This was extra bad
because register pressure later forced both base and IV into
memory.  Doing that recursion, at least enough
to figure out addressing modes, is a good idea in general;
the change in AddUsersIfInteresting does this.  However,
there were side effects....

It is also possible for recursing outside the loop to
introduce another IV where there was only 1 before (if
the refs inside are not scaled and the ref outside is).
I don't think this is a common case, but it's in the testsuite.
It is right to be very aggressive about getting rid of
such introduced IVs (CheckForIVReuse and the handling of
nonzero RewriteFactor in StrengthReduceStridedIVUsers).
In the testcase in question the new IV produced this way
has both a nonconstant stride and a nonzero base, neither
of which was handled before.  And when inserting 
new code that feeds into a PHI, it's right to put such 
code at the original location rather than in the PHI's 
immediate predecessor(s) when the original location is outside 
the loop (a case that couldn't happen before)
(RewriteInstructionToUseNewBase); better to avoid making
multiple copies of it in this case.

Also, the mechanism for keeping SCEV's corresponding to GEP's
no longer works, as the GEP might change after its SCEV
is remembered, invalidating the SCEV, and we might get a bad
SCEV value when looking up the GEP again for a later loop.  
This also couldn't happen before, as we weren't recursing
into GEP's outside the loop.

I owe some testcases for this, want to get it in for nightly runs.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61362 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-23 02:12:52 +00:00
Dale Johannesen
58e39b0200 One more permutation of subtracting off a base value.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61361 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-23 01:59:54 +00:00
Owen Anderson
ae53c932b9 Don't forget to remove phi nodes from the value numbering table after we collapse them.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61358 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-23 00:49:51 +00:00
Dan Gohman
15a3a49e19 Make the fuse-failed debug output human-readable.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61356 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-23 00:19:20 +00:00
Bill Wendling
30788b828a Comment clean-ups. No functionality change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61354 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-22 22:32:22 +00:00
Bill Wendling
6d463f2efc Check that the instruction isn't in the value numbering scope.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61353 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-22 22:28:56 +00:00
Bill Wendling
75f02ee771 Simplification: Negate the operator== method instead of implementing a full operator!= method.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61352 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-22 22:16:31 +00:00
Bill Wendling
70ded19b3f Add verification that deleted instruction isn't hiding in the PHI map.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61350 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-22 22:14:07 +00:00
Bill Wendling
ec40d50aa4 Verify removed in a few more places.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61349 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-22 21:57:30 +00:00
Bill Wendling
246dbbb8be Add verification functions to GVN which check to see that an instruction was
truely deleted. These will be expanded with further checks of all of the data
structures.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61347 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-22 21:36:08 +00:00
Dan Gohman
00d448a341 Refactor a bunch of code out of AsmPrinter::EmitGlobalConstant into separate
functions.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61345 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-22 21:14:27 +00:00
Dan Gohman
8044e9b3af Optimize setDepthDirty and setHeightDirty a little, as they showed
up on a profile.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61344 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-22 21:11:33 +00:00
Dan Gohman
361c31da5e Add an assertion to the ScheduleDAGInstrs class to catch SUnits
reallocations. We don't do cloning on MachineInstr schedule DAGs,
but this is a worthwhile sanity check regardless.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61343 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-22 21:08:08 +00:00
Dan Gohman
fb8a1356b2 Add an accesor for the isNormalMemory field in the SDep class.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61342 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-22 21:06:56 +00:00
Dan Gohman
983bbbaf36 Add an assertion to catch SUnits reallocations. And add a doxygen
comment for the ScheduleDAGSDNodes class.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61341 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-22 21:06:20 +00:00
Dan Gohman
b5544940c1 Clarify a comment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61338 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-22 19:44:39 +00:00
Mikhail Glushenkov
a01ce0091a Use ignore & grep instead of XFAIL.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61307 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-21 07:47:49 +00:00
Nick Lewycky
13a09e298c Turn strcmp into memcmp, such as strcmp(P, "x") --> memcmp(P, "x", 2).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61297 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-21 00:19:21 +00:00
Dan Gohman
b12b1a27f5 Fix fast-isel to not emit invalid assembly when presented with a
constant shift count that doesn't fit in the shift instruction's
immediate field. This fixes PR3242.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61281 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-20 17:19:40 +00:00
Dan Gohman
edf22b97d0 Reword the comment for ConstantInt's getLimitedValue.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61280 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-20 17:06:39 +00:00
Nick Lewycky
9dce873c1a Remove redundant test for vector-nature. Scan the vector first to see whether
our optz'n will apply to it, then build the replacement vector only if needed.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61279 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-20 16:48:00 +00:00
Dan Gohman
fce4028718 LiveInterval::removeKills and isKill don't need 'this' and
can be static member functions.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61278 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-20 16:44:40 +00:00
Dan Gohman
e19c6362d2 Use SmallVector's pop_back_val.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61277 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-20 16:42:33 +00:00
Dan Gohman
f89e6e6577 Use the correct Preds and Succs lists in setHeightDirty()
and setDepthDirty(), respectively. This fixes PR3241.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61276 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-20 16:34:57 +00:00
Bill Wendling
b8fd41985f More precise XFAIL.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61265 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-19 22:28:23 +00:00
Bill Wendling
c6acb579f9 Un-XFAIL this test because it's passing and John doesn't seem interested in un-XFAILing it.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61264 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-19 22:25:01 +00:00
Dan Gohman
6c3643c41b Use ~0u instead of -1u as the special value, to hopefully avoid
warnings on compilers that warn about such things.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61263 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-19 22:23:43 +00:00
Evan Cheng
5379f412bc Fix PR3149. If an early clobber def is a physical register and it is tied to an input operand, it effectively extends the live range of the physical register. Currently we do not have a good way to represent this.
172     %ECX<def> = MOV32rr %reg1039<kill>
180     INLINEASM <es:subl $5,$1
        sbbl $3,$0>, 10, %EAX<def>, 14, %ECX<earlyclobber,def>, 9, %EAX<kill>,
36, <fi#0>, 1, %reg0, 0, 9, %ECX<kill>, 36, <fi#1>, 1, %reg0, 0
188     %EAX<def> = MOV32rr %EAX<kill>
196     %ECX<def> = MOV32rr %ECX<kill>
204     %ECX<def> = MOV32rr %ECX<kill>
212     %EAX<def> = MOV32rr %EAX<kill>
220     %EAX<def> = MOV32rr %EAX
228     %reg1039<def> = MOV32rr %ECX<kill>

The early clobber operand ties ECX input to the ECX def.

The live interval of ECX is represented as this:
%reg20,inf = [46,47:1)[174,230:0)  0@174-(230) 1@46-(47)

The right way to represent this is something like
%reg20,inf = [46,47:2)[174,182:1)[181:230:0)  0@174-(182) 1@181-230 @2@46-(47)

Of course that won't work since that means overlapping live ranges defined by two val#.

The workaround for now is to add a bit to val# which says the val# is redefined by a early clobber def somewhere. This prevents the move at 228 from being optimized away by SimpleRegisterCoalescing::AdjustCopiesBackFrom.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61259 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-19 20:58:01 +00:00
John Criswell
de8c611c08 The fields for the stoppoint debug intrinsic have not changed, so update the
version number assertions.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61257 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-19 19:56:36 +00:00
Bill Wendling
94920c123a This test works again for Darwin because a patch was reverted.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61254 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-19 19:08:13 +00:00
Gordon Henriksen
95ca7314fd Add dyn_cast_or_null bindings for some additional classes missed in r61252.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61253 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-19 18:51:17 +00:00
Gordon Henriksen
2a9c671a51 C bindings for dyn_cast_or_null.
This operation can be used to build dyn_cast, isa, and cast.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61252 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-19 18:39:45 +00:00
Chris Lattner
38e7721455 Add support for writing LLVM IR to a specified BitstreamWriter.
Patch by Lukasz Janyst!


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61251 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-19 18:37:59 +00:00
Dan Gohman
018a34c576 Move the patterns which have i8 immediates before the patterns
that have i32 immediates so that they get selected first. This
currently only matters in the JIT, as assemblers will
automatically use the smallest encoding.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61250 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-19 18:25:21 +00:00
Dan Gohman
0b53d98c29 Use dyn_cast intead of isa + cast in the generated DAGISel code. This
reduces the amount of code slightly when assertions are enabled.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61249 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-19 18:13:39 +00:00
Evan Cheng
ab63152871 - CodeGenPrepare does not split loop back edges but it only knows about back edges of single block loops. It now does a DFS walk to find loop back edges.
- Use SplitBlockPredecessors to factor out common predecessors of the critical edge destination. This is disabled for now due to some regressions.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61248 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-19 18:03:11 +00:00
Chris Lattner
a33649e98c Fix some release-assert warnings
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61244 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-19 17:03:38 +00:00
Rafael Espindola
2f6fea90a5 Fix bug 3202.
The EH_frame and .eh symbols are now private, except for darwin9 and earlier.
The patch also fixes the definition of PrivateGlobalPrefix on pcc linux.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61242 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-19 10:55:56 +00:00
Nick Lewycky
d802be39d7 Update the .cvs files for nocapture.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61241 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-19 09:41:54 +00:00
Nick Lewycky
73ddd4f00d Commit missed files from nocapture change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61240 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-19 09:38:31 +00:00
Nick Lewycky
f23d0d392a Resubmit support for the 'nocapture' attribute.
The problematic part of this patch is that we were out of attribute bits,
requiring some fancy bit hacking to make it fit (by shrinking alignment)
without breaking existing users or the file format.

This change will require users to rebuild llvm-gcc to match llvm.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61239 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-19 06:39:12 +00:00
Bill Wendling
bebbded574 Perform this loop only when the -debug flag is specified.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61238 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-19 02:09:57 +00:00
Dan Gohman
1385758215 Initialize the ImplicitDefed member, to avoid getting stale
data from a previous block.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61237 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-19 00:46:20 +00:00
Dan Gohman
5b200d8a13 Fix RegScavenger::forward() to work on basic blocks containing exactly
one instruction.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61236 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-19 00:45:13 +00:00
Dan Gohman
9755faab60 Delete the RegScavenging constructor that takes a MachineBasicBlock
argument. Nothing was using it, and it set the MBB member without
calling enterBasicBlock, which was problematic.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61234 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-19 00:34:32 +00:00
Mon P Wang
7538454f9f Fix test to account for generating some vector code for mul v2i64 instead
of incorrectly generating pmuldq


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61228 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-18 23:42:37 +00:00
Bill Wendling
e533f4e343 Didn't mean to commit this.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61222 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-18 22:19:50 +00:00
Dan Gohman
a5b2fee757 Teach LowerSubregs to preserve kill/dead information when lowering
subreg instructions.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61220 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-18 22:14:08 +00:00
Bill Wendling
82c19ae10a Re-XFAIL this test until debug stuff settles down.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61219 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-18 22:13:31 +00:00
Dan Gohman
98c206921a Make LowerSubregs' debug output for EXTRACT_SUBREG consistent with
that of INSERT_SUBREG and SUBREG_TO_REG.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61218 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-18 22:11:34 +00:00