Daniel Dunbar [Tue, 18 May 2010 17:22:19 +0000 (17:22 +0000)]
llc (et al): Add support for --show-encoding and --show-inst.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@104029
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Tue, 18 May 2010 15:25:14 +0000 (15:25 +0000)]
Usage of O_NONBLOCK in bjam is now confirmed as a bug and fixed upstream.
Update the comment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@104021
91177308-0d34-0410-b5e6-
96231b3b80d8
Benjamin Kramer [Tue, 18 May 2010 12:15:34 +0000 (12:15 +0000)]
Simplify MCContext::(Next|Get)Instance
- Allocate MCLabels in the context so they don't leak.
- Avoid duplicated densemap lookup.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@104020
91177308-0d34-0410-b5e6-
96231b3b80d8
Sean Hunt [Tue, 18 May 2010 06:22:50 +0000 (06:22 +0000)]
Replace FIRST_* and LAST_* macros with a generic STMT_RANGE macro
Also rename ABSTRACT to ABSTRACT_STMT
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@104018
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Tue, 18 May 2010 00:03:40 +0000 (00:03 +0000)]
FIX PR7158. SimplifyVBinOp was asserting when it fails to constant fold (op (build_vector), (build_vector)).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@104004
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Mon, 17 May 2010 23:29:23 +0000 (23:29 +0000)]
ARMBaseRegisterInfo::estimateRSStackSizeLimit() could return prematurely with a
too large limit.
The function would return immediately when finding an addrmode 3/5 instruction.
It needs to keep scanning in case there is an addrmode 6 instruction which drops
the limit to 0.
A test case is very difficult to produce because it will only fail when the
scavenger is used.
rdar://problem/
7894847
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103995
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Mon, 17 May 2010 23:24:12 +0000 (23:24 +0000)]
Fix PR7175. Insert copies of a REG_SEQUENCE source if it is used by other REG_SEQUENCE instructions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103994
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Mon, 17 May 2010 23:09:50 +0000 (23:09 +0000)]
- Set the "HasCalls" flag after instruction selection is finished.
- Change the logic DisableFramePointerElim() to check for the
-disable-non-leaf-fp-elim before -disable-fp-elim.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103990
91177308-0d34-0410-b5e6-
96231b3b80d8
Kevin Enderby [Mon, 17 May 2010 23:08:19 +0000 (23:08 +0000)]
Added support in MC for Directional Local Labels.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103989
91177308-0d34-0410-b5e6-
96231b3b80d8
Eric Christopher [Mon, 17 May 2010 22:53:55 +0000 (22:53 +0000)]
More data/parsing support for tls directives. Add a few more testcases
and cleanup comments as well.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103985
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Mon, 17 May 2010 22:09:49 +0000 (22:09 +0000)]
Fix PR7156. If the sources of a REG_SEQUENCE are all IMPLICIT_DEF's. Replace it with an IMPLICIT_DEF rather than deleting it or else it would be left without a def.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103984
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Mon, 17 May 2010 21:54:50 +0000 (21:54 +0000)]
vmov of immediates are trivially re-materializable.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103982
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Dunbar [Mon, 17 May 2010 21:54:30 +0000 (21:54 +0000)]
MC/Mach-O/x86: Optimal nop sequences should only be used for the .text sections, not all sections in the text segment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103981
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Dunbar [Mon, 17 May 2010 21:54:26 +0000 (21:54 +0000)]
MC: Add dyn_cast support to MCSection.
- Of questionable utility, since in general anything which wants to do this should probably be within a target specific hook, which can rely on the sections being of the appropriate type. However, it can be useful for short term hacks.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103980
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Dunbar [Mon, 17 May 2010 21:19:59 +0000 (21:19 +0000)]
MC/Mach-O: Reverse order of SymbolData scanning when emitting instructions.
- This fixes a string table mismatch with 'as' when two new symbols are defined
in a single instruction.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103979
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Mon, 17 May 2010 21:02:08 +0000 (21:02 +0000)]
Pull the UsedInInstr.test() calls into calcSpillCost() and remember aliases.
This fixes the miscompilations of MultiSource/Applications/JM/l{en,de}cod.
Clang now successfully self hosts in a debug build with the fast register allocator.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103975
91177308-0d34-0410-b5e6-
96231b3b80d8
Eric Christopher [Mon, 17 May 2010 21:02:07 +0000 (21:02 +0000)]
Add some section and constant support for darwin TLS.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103974
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Mon, 17 May 2010 20:57:12 +0000 (20:57 +0000)]
Careful with reg_sequence coalescing to not to overwrite sub-register indices.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103971
91177308-0d34-0410-b5e6-
96231b3b80d8
Bob Wilson [Mon, 17 May 2010 20:31:13 +0000 (20:31 +0000)]
Fix a regression in 464.h264 for thumb1 and thumb2 nightly tests.
Obvious in retrospect but not fun to debug.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103969
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Dunbar [Mon, 17 May 2010 20:12:31 +0000 (20:12 +0000)]
MC/Mach-O: Fix some differences in symbol flag handling.
- Don't clear weak reference flag, 'as' was only "trying" to do this, it wasn't
actually succeeding.
- Clear the "lazy bound" bit when we mark something external. This corresponds
roughly to the lazy clearing of the bit that 'as' implements in
symbol_table_lookup.
- The exact meaning of these flags appears pretty loose, since 'as' isn't very
consistent. For now we just try to match 'as', we will clean this up one day
hopefully.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103964
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Mon, 17 May 2010 20:01:24 +0000 (20:01 +0000)]
Optimize empty DenseMap iteration.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103962
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Mon, 17 May 2010 20:01:22 +0000 (20:01 +0000)]
Remove debug option. Add comment on spill order determinism.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103961
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Mon, 17 May 2010 19:51:20 +0000 (19:51 +0000)]
Turn on -neon-reg-sequence by default.
Using NEON load / store multiple instructions will no longer create gobs of vmov of D registers!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103960
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Dunbar [Mon, 17 May 2010 17:46:23 +0000 (17:46 +0000)]
llvm-mc: Support reassignment of variables in one special case, when the
variable has not yet been used in an expression. This allows us to support a few
cases that show up in real code (mostly because gcc generates it for Objective-C
on Darwin), without giving up a reasonable semantic model for assignment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103950
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Mon, 17 May 2010 17:18:59 +0000 (17:18 +0000)]
Avoid allocating the same physreg to multiple virtregs in one instruction.
While that approach works wonders for register pressure, it tends to break
everything.
This should unbreak the arm-linux builder and fix a number of miscompilations.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103946
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Mon, 17 May 2010 15:30:37 +0000 (15:30 +0000)]
Minor optimizations. DenseMap::begin() is surprisingly slow on an empty map.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103940
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Mon, 17 May 2010 15:30:32 +0000 (15:30 +0000)]
Extract spill cost calculation to a new method, and use definePhysReg() to clear
out aliases when allocating. Clean up allocVirtReg().
Use calcSpillCost() to allow more aggressive hinting. Now the hint is always
taken unless blocked by a reserved register. This leads to more coalescing,
lower register pressure, and less spilling.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103939
91177308-0d34-0410-b5e6-
96231b3b80d8
Zhongxing Xu [Mon, 17 May 2010 09:47:55 +0000 (09:47 +0000)]
Remove unused member variable.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103936
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Mon, 17 May 2010 04:50:57 +0000 (04:50 +0000)]
Only use clairvoyance when defining a register, and then only if it has one use.
This makes allocation independent on the ordering of use-def chains.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103935
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Mon, 17 May 2010 03:26:09 +0000 (03:26 +0000)]
Eliminate a hash table probe when killing virtual registers.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103934
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Mon, 17 May 2010 03:26:06 +0000 (03:26 +0000)]
Execute virtreg kills immediately instead of after processing all uses.
This is safe to do because the physreg has been marked UsedInInstr and the kill flag will be set on the last operand using the virtreg if there are more then one.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103933
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Mon, 17 May 2010 02:49:21 +0000 (02:49 +0000)]
Sprinkle superregister <imp-def> and <imp-kill> operands when dealing with subregister indices.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103931
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Mon, 17 May 2010 02:49:18 +0000 (02:49 +0000)]
Now that we don't keep live registers across calls, there is not reason to go
through the very long list of call-clobbered registers. We just assume all
registers are clobbered.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103930
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Mon, 17 May 2010 02:49:15 +0000 (02:49 +0000)]
Boldly attempt consistent capitalization. Functional changes unintended.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103929
91177308-0d34-0410-b5e6-
96231b3b80d8
Eric Christopher [Mon, 17 May 2010 02:13:02 +0000 (02:13 +0000)]
Assume that we'll handle mangling the symbols earlier and just put the
symbol to the file as we have it. Simplifies out tbss handling.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103928
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Mon, 17 May 2010 02:07:32 +0000 (02:07 +0000)]
Spill and kill all virtual registers across a call.
Debug code doesn't use callee saved registers anyway, and the code is simpler this way. Now spillVirtReg always kills, and the isKill parameter is not needed.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103927
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Mon, 17 May 2010 02:07:29 +0000 (02:07 +0000)]
Reduce hashtable probes by using DenseMap::insert() for lookup.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103926
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Mon, 17 May 2010 02:07:22 +0000 (02:07 +0000)]
Make MBB a class member instead of passing it around everywhere.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103925
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Mon, 17 May 2010 01:47:47 +0000 (01:47 +0000)]
Yes, if the redef is a copy, update the old val# with the copy. But make sure to clear the copy field if the redef is not a copy.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103922
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Mon, 17 May 2010 01:11:46 +0000 (01:11 +0000)]
No reason not to run the NEON domain croassing fix up pass in thumb2 mode.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103917
91177308-0d34-0410-b5e6-
96231b3b80d8
Dale Johannesen [Sun, 16 May 2010 20:19:40 +0000 (20:19 +0000)]
Removing as part of previous reversion.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103915
91177308-0d34-0410-b5e6-
96231b3b80d8
Dale Johannesen [Sun, 16 May 2010 20:19:04 +0000 (20:19 +0000)]
Revert 103911; it broke a test that expects bitconvert
<1xi64> -> i64 to work in MMX registers on hosts where -no-sse
is the default (not mine). The right thing is
to accept this and make i64->f64 conversions go through memory,
but I don't have time right now.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103914
91177308-0d34-0410-b5e6-
96231b3b80d8
Dale Johannesen [Sun, 16 May 2010 18:22:38 +0000 (18:22 +0000)]
Make x86-64 64-bit bitconvert work when SSE is not available.
(This worked as of about 6 months ago and I didn't track down
exactly what broke it; I think this fix is appropriate.)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103911
91177308-0d34-0410-b5e6-
96231b3b80d8
Anton Korobeynikov [Sun, 16 May 2010 09:15:36 +0000 (09:15 +0000)]
Chris said that the comment char should be escaped. Fix all the occurences of "@" in *.td
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103903
91177308-0d34-0410-b5e6-
96231b3b80d8
Anton Korobeynikov [Sun, 16 May 2010 09:08:45 +0000 (09:08 +0000)]
Add support for thiscall calling convention.
Patch by Charles Davis and Steven Watanabe!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103902
91177308-0d34-0410-b5e6-
96231b3b80d8
Anton Korobeynikov [Sun, 16 May 2010 08:54:20 +0000 (08:54 +0000)]
Generalize the ARM DAG combiner of mul with constants to all power-of-two cases.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103901
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Sun, 16 May 2010 03:27:48 +0000 (03:27 +0000)]
Model vst lane instructions with REG_SEQUENCE.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103898
91177308-0d34-0410-b5e6-
96231b3b80d8
Rafael Espindola [Sun, 16 May 2010 03:13:23 +0000 (03:13 +0000)]
Avoid renaming loadable modules at install time. Now the gold plugin is named
LLVMgold.so both in both the build and install directories.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103897
91177308-0d34-0410-b5e6-
96231b3b80d8
Rafael Espindola [Sun, 16 May 2010 03:05:14 +0000 (03:05 +0000)]
Use $ORIGIN in the rpath of libraries.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103896
91177308-0d34-0410-b5e6-
96231b3b80d8
Dale Johannesen [Sat, 15 May 2010 18:51:12 +0000 (18:51 +0000)]
Fix uint64->{float, double} conversion to do rounding correctly in 32-bit.
The implementation in LegalizeIntegerTypes to handle this as
sint64->float + appropriate power of 2 is subject to double rounding,
considered incorrect by numerics people. Use this implementation only
when it is safe. This leads to using library calls in some cases
that produced inline code before, but it's correct now.
(EVTToAPFloatSemantics belongs somewhere else, any suggestions?)
Add a correctly rounding (though not particularly fast) conversion
that uses X87 80-bit computations for x86-32.
7885399,
5901940. This shows up in gcc.c-torture/execute/ieee/rbug.c
in the gcc testsuite on some platforms.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103883
91177308-0d34-0410-b5e6-
96231b3b80d8
Dale Johannesen [Sat, 15 May 2010 18:38:02 +0000 (18:38 +0000)]
Improve assertion messages.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103882
91177308-0d34-0410-b5e6-
96231b3b80d8
Anton Korobeynikov [Sat, 15 May 2010 18:16:59 +0000 (18:16 +0000)]
Some cheap DAG combine goodness for multiplication with a particular constant.
This can be extended later on to handle more "complex" constants.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103881
91177308-0d34-0410-b5e6-
96231b3b80d8
Anton Korobeynikov [Sat, 15 May 2010 17:19:20 +0000 (17:19 +0000)]
"trap" pseudo-op turned out to be apple-local.
Temporary emit it as raw bytes until it will be added to binutils as well.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103878
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sat, 15 May 2010 17:11:55 +0000 (17:11 +0000)]
improve portability to systems that don't have round, patch by
Evzen Muller!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103877
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sat, 15 May 2010 17:10:24 +0000 (17:10 +0000)]
improve portability to systems that don't have powf/modf (e.g. solaris 9)
patch by Evzen Muller!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103876
91177308-0d34-0410-b5e6-
96231b3b80d8
Chandler Carruth [Sat, 15 May 2010 10:23:23 +0000 (10:23 +0000)]
Fix an GCC warning that seems to have actually caught a bug (!!!) in
a condition's grouping. Every other use of Allocatable.test(Hint) groups it the
same way as it is indented, so move the parentheses to agree with that
grouping.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103869
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Sat, 15 May 2010 07:53:37 +0000 (07:53 +0000)]
Model 128-bit vld lane with REG_SEQUENCE.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103868
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Sat, 15 May 2010 06:09:08 +0000 (06:09 +0000)]
Calculate liveness on the fly for local registers.
When working top-down in a basic block, substituting physregs for virtregs, the use-def chains are kept up to date. That means we can recognize a virtreg kill by the use-def chain becoming empty.
This makes the fast allocator independent of incoming kill flags.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103866
91177308-0d34-0410-b5e6-
96231b3b80d8
Nick Lewycky [Sat, 15 May 2010 04:26:25 +0000 (04:26 +0000)]
Teach the always inliner to release its inline cost estimates, like the basic
inliner did in r103653. Why does the always inliner even bother with cost
estimates anyways?
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103858
91177308-0d34-0410-b5e6-
96231b3b80d8
Nick Lewycky [Sat, 15 May 2010 03:41:58 +0000 (03:41 +0000)]
Clean up, no functional change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103857
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Sat, 15 May 2010 02:52:58 +0000 (02:52 +0000)]
Add reg_nodbg_iterator
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103856
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Sat, 15 May 2010 02:20:21 +0000 (02:20 +0000)]
v4i64 and v8i64 are only synthesizable when NEON is available.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103855
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Sat, 15 May 2010 02:18:07 +0000 (02:18 +0000)]
Allow TargetLowering::getRegClassFor() to be called on illegal types. Also
allow target to override it in order to map register classes to illegal
but synthesizable types. e.g. v4i64, v8i64 for ARM / NEON.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103854
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Sat, 15 May 2010 01:36:29 +0000 (01:36 +0000)]
Model 64-bit lane vld with REG_SEQUENCE.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103851
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Sat, 15 May 2010 01:35:44 +0000 (01:35 +0000)]
A partial re-def instruction may be a copy.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103850
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Fri, 14 May 2010 23:21:14 +0000 (23:21 +0000)]
Teach two-address pass to do some coalescing while eliminating REG_SEQUENCE
instructions.
e.g.
%reg1026<def> = VLDMQ %reg1025<kill>, 260, pred:14, pred:%reg0
%reg1027<def> = EXTRACT_SUBREG %reg1026, 6
%reg1028<def> = EXTRACT_SUBREG %reg1026<kill>, 5
...
%reg1029<def> = REG_SEQUENCE %reg1028<kill>, 5, %reg1027<kill>, 6, %reg1028, 7, %reg1027, 8, %reg1028, 9, %reg1027, 10, %reg1030<kill>, 11, %reg1032<kill>, 12
After REG_SEQUENCE is eliminated, we are left with:
%reg1026<def> = VLDMQ %reg1025<kill>, 260, pred:14, pred:%reg0
%reg1029:6<def> = EXTRACT_SUBREG %reg1026, 6
%reg1029:5<def> = EXTRACT_SUBREG %reg1026<kill>, 5
The regular coalescer will not be able to coalesce reg1026 and reg1029 because it doesn't
know how to combine sub-register indices 5 and 6. Now 2-address pass will consult the
target whether sub-registers 5 and 6 of reg1026 can be combined to into a larger
sub-register (or combined to be reg1026 itself as is the case here). If it is possible,
it will be able to replace references of reg1026 with reg1029 + the larger sub-register
index.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103835
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Fri, 14 May 2010 23:01:02 +0000 (23:01 +0000)]
Add some comments about undef.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103834
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Fri, 14 May 2010 22:54:52 +0000 (22:54 +0000)]
Model VST*_UPD and VST*oddUPD pair with REG_SEQUENCE.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103833
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Fri, 14 May 2010 22:53:18 +0000 (22:53 +0000)]
Fast ISel trivially coalesces away no-op casts, so check for this when
setting kill flags.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103832
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Fri, 14 May 2010 22:40:43 +0000 (22:40 +0000)]
Don't bother spilling before a return
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103831
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Fri, 14 May 2010 22:40:40 +0000 (22:40 +0000)]
RegAllocLocal can count copies too
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103830
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Fri, 14 May 2010 22:17:42 +0000 (22:17 +0000)]
SystemZ really does mean "has calls" and not just "adjusts stack." Go ahead and
replace the check with the appropriate predicate. Modify the testcase to reflect
the correct code. (It should be saving callee-saved registers on the stack
allocated by the calling fuction.)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103829
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Fri, 14 May 2010 22:02:56 +0000 (22:02 +0000)]
Track allocatable instead of reserved regs, and never take an unallocatable hint.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103828
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Fri, 14 May 2010 22:01:14 +0000 (22:01 +0000)]
Don't set kill flags for instructions which the scheduler has cloned.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103827
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Fri, 14 May 2010 22:00:27 +0000 (22:00 +0000)]
BR is a barrier.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103826
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Fri, 14 May 2010 21:58:35 +0000 (21:58 +0000)]
Several tail call tests apparently rely upon this being "adjusts stack" instead
of "has calls". That's probably wrong, but it needs further
investigation. Revert to the original behavior until this is settled.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103824
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Fri, 14 May 2010 21:55:52 +0000 (21:55 +0000)]
Avoid scanning the long tail of physreg operands on calls
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103823
91177308-0d34-0410-b5e6-
96231b3b80d8
Devang Patel [Fri, 14 May 2010 21:55:50 +0000 (21:55 +0000)]
Do not forget to mark prcessed arguments.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103822
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Fri, 14 May 2010 21:55:50 +0000 (21:55 +0000)]
Count coalesced copies
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103821
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Fri, 14 May 2010 21:55:44 +0000 (21:55 +0000)]
Allow virtreg redefines when verifying for RegAllocFast
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103820
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Fri, 14 May 2010 21:38:44 +0000 (21:38 +0000)]
This should happen if there are no calls, not if it just doesn't adjust the
stack.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103813
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Fri, 14 May 2010 21:28:24 +0000 (21:28 +0000)]
Revert r103804. The comment is correct.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103808
91177308-0d34-0410-b5e6-
96231b3b80d8
Jim Grosbach [Fri, 14 May 2010 21:20:46 +0000 (21:20 +0000)]
Remove trailing whitespace
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103807
91177308-0d34-0410-b5e6-
96231b3b80d8
Jim Grosbach [Fri, 14 May 2010 21:19:48 +0000 (21:19 +0000)]
80 column and trailing whitespace cleanup
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103806
91177308-0d34-0410-b5e6-
96231b3b80d8
Jim Grosbach [Fri, 14 May 2010 21:18:04 +0000 (21:18 +0000)]
add cmd line option to leave dbgvalues in during post-RA sceduling. Useful
while debugging what's mishandled about them in the post-RA pass.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103805
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Fri, 14 May 2010 21:17:29 +0000 (21:17 +0000)]
Fix comment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103804
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Fri, 14 May 2010 21:14:32 +0000 (21:14 +0000)]
Rename "HasCalls" in MachineFrameInfo to "AdjustsStack" to better describe what
the variable actually tracks.
N.B., several back-ends are using "HasCalls" as being synonymous for something
that adjusts the stack. This isn't 100% correct and should be looked into.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103802
91177308-0d34-0410-b5e6-
96231b3b80d8
Devang Patel [Fri, 14 May 2010 21:04:45 +0000 (21:04 +0000)]
Test case for r103800.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103801
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Fri, 14 May 2010 21:01:44 +0000 (21:01 +0000)]
Lowering of atomic instructions can result in operands being
used more than once. If ISel had put a kill flag on one of them,
it's not valid to transfer the kill flag to each new instance.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103799
91177308-0d34-0410-b5e6-
96231b3b80d8
Devang Patel [Fri, 14 May 2010 21:01:35 +0000 (21:01 +0000)]
Add support to preserve type info for the variables that are removed by the optimizer.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103798
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Fri, 14 May 2010 20:28:32 +0000 (20:28 +0000)]
When verifying two-address instructions, check the following:
- Kill is implicit when use and def registers are identical.
- Only virtual registers can differ.
Add a -verify-fast-regalloc to run the verifier before the fast allocator.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103797
91177308-0d34-0410-b5e6-
96231b3b80d8
Kevin Enderby [Fri, 14 May 2010 19:16:02 +0000 (19:16 +0000)]
Fix so "int3" is correctly accepted, added "into" and fixed "int" with an
argument, like "int $4", to not get an Assertion error.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103791
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Fri, 14 May 2010 18:54:59 +0000 (18:54 +0000)]
Model VLD*_UPD and VLD*odd_UPD pair with REG_SEQUENCE.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103790
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Dunbar [Fri, 14 May 2010 18:53:40 +0000 (18:53 +0000)]
MC/Mach-O/x86_64: Darwin's special "signed_N" relocation types should only be
used to replace a normal relocation, not a reference to a GOT entry.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103789
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Fri, 14 May 2010 18:03:25 +0000 (18:03 +0000)]
Simplify the handling of physreg defs and uses in RegAllocFast.
This adds extra security against using clobbered physregs, and it adds kill
markers to physreg uses.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103784
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Fri, 14 May 2010 16:46:02 +0000 (16:46 +0000)]
Set isTerminator on TRAP instructions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103778
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Fri, 14 May 2010 16:42:16 +0000 (16:42 +0000)]
Don't use isBarrier for the PowerPC sync instruction. isBarrier is for
control barriers, not memory ordering barriers.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103777
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Fri, 14 May 2010 16:34:55 +0000 (16:34 +0000)]
Add mayLoad and mayStore flags to instructions which missed them.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103776
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Fri, 14 May 2010 15:36:54 +0000 (15:36 +0000)]
Use regular PassManager instead of FunctionPassManager in opt, since it
isn't doing lazy streaming. This also fixes a missing doFinalization call.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103774
91177308-0d34-0410-b5e6-
96231b3b80d8
Dan Gohman [Fri, 14 May 2010 15:35:10 +0000 (15:35 +0000)]
Use PassManagerBase, to give clients the option of using either
FunctionPassManager or regular PassManager.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103773
91177308-0d34-0410-b5e6-
96231b3b80d8