Commit Graph

52576 Commits

Author SHA1 Message Date
Bob Wilson
c8ce2d4cb9 Add some comments to clarify things that I discovered this week.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82773 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-25 16:34:46 +00:00
Daniel Dunbar
42985bbe38 Make llvm-bcanalyzer percentages more readable.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82772 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-25 16:04:21 +00:00
Daniel Dunbar
80ba3d2c0b Strip trailing whitespace.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82771 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-25 16:03:57 +00:00
Bob Wilson
0035f9c3b9 pr4926: ARM requires the stack pointer to be aligned, even for leaf functions.
For the AAPCS ABI, SP must always be 4-byte aligned, and at any "public
interface" it must be 8-byte aligned.  For the older ARM APCS ABI, the stack
alignment is just always 4 bytes.  For X86, we currently align SP at
entry to a function (e.g., to 16 bytes for Darwin), but no stack alignment
is needed at other times, such as for a leaf function.

After discussing this with Dan, I decided to go with the approach of adding
a new "TransientStackAlignment" field to TargetFrameInfo.  This value
specifies the stack alignment that must be maintained even in between calls.
It defaults to 1 except for ARM, where it is 4.  (Some other targets may
also want to set this if they have similar stack requirements. It's not
currently required for PPC because it sets targetHandlesStackFrameRounding
and handles the alignment in target-specific code.) The existing StackAlignment
value specifies the alignment upon entry to a function, which is how we've
been using it anyway.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82767 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-25 14:41:49 +00:00
Duncan Sands
0cad4e3f0b Print INST_INBOUNDS_GEP rather than UnknownCode30.
Likewise for constant inbounds GEP.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82763 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-25 12:28:37 +00:00
Chris Lattner
5d7c79d9c6 convert testcases to filecheck.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82759 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-25 06:49:41 +00:00
Chris Lattner
d9485ddfdd filecheck should not match a \n with a .
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82758 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-25 06:47:09 +00:00
Chris Lattner
0f427ac24c remove a large unreduced testcase
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82756 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-25 06:37:22 +00:00
Chris Lattner
bfa2eed139 turn a std::pair into a real class.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82754 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-25 06:32:47 +00:00
Nate Begeman
cc66cdd79c Fix combiner-aa issue with bases which are different, but can alias.
Previously, it treated GV+28 GV+0 as different bases, and assumed they could
not alias.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82753 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-25 06:05:26 +00:00
Bob Wilson
6263778d2c Update the description of MachineFrameInfo's OffsetAdjustment. The value of
this adjustment does not change the direction or the signs of the object
offsets, and the details of the offset calculations can be target-specific.
Also mention that for most targets this value is only used to generate debug
info.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82750 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-25 05:30:55 +00:00
Dan Gohman
6acaaa8f32 Don't try to use pre-indexed addressing with sthbrx/stwbrx
instructions. This fixes a PowerPC bug exposed by some unrelated
changes I'm working on.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82743 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-25 00:57:30 +00:00
Dan Gohman
8fc13cb4f7 Add a version of dumpr() that has a SelectionDAG* argument.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82742 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-25 00:34:34 +00:00
Jim Grosbach
3d6cb88a64 Start of revamping the register scavenging in PEI. ARM Thumb1 is the driving
interest for this, as it currently reserves a register rather than using
the scavenger for matierializing constants as needed.

Instead of scavenging registers on the fly while eliminating frame indices,
new virtual registers are created, and then a scavenged collectively in a
post-pass over the function. This isolates the bits that need to interact
with the scavenger, and sets the stage for more intelligent use, and reuse,
of scavenged registers.

For the time being, this is disabled by default. Once the bugs are worked out,
the current scavenging calls in replaceFrameIndices() will be removed and
the post-pass scavenging will be the default. Until then,
-enable-frame-index-scavenging enables the new code. Currently, only the
Thumb1 back end is set up to use it.




git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82734 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-24 23:52:18 +00:00
Dale Johannesen
2bbcd33feb Reenable sqrt IR generation test.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82731 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-24 23:37:40 +00:00
Mike Stump
7e3720da38 Delete space after function name, before (, reflow a comment and
delete a few blank lines.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82729 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-24 23:21:26 +00:00
Mike Stump
ee4b8a7781 Fix spacing.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82727 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-24 23:11:08 +00:00
Chris Lattner
528700863a add and document regex support for FileCheck. You can now do stuff like:
; CHECK: movl {{%e[a-z][xi]}}, %eax

or whatever.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82717 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-24 21:47:32 +00:00
Chris Lattner
fe03dbeaa3 unconditionally request MMI
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82716 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-24 21:45:57 +00:00
Chris Lattner
ad78500712 wrap long lines.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82715 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-24 21:44:20 +00:00
Chris Lattner
adea46ed61 Use CanonicalizeInputFile to canonicalize the entire buffer containing the
CHECK strings, instead of canonicalizing the patterns directly.  This allows
Pattern to just contain a StringRef instead of std::string.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82713 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-24 20:45:07 +00:00
Chris Lattner
a29703e842 change 'not' matching to use Pattern, move pattern parsing logic into
the Pattern class.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82712 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-24 20:39:13 +00:00
Chris Lattner
9fc6678bea refactor out the match string into its own Pattern class.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82711 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-24 20:25:55 +00:00
Bob Wilson
08ae76c042 Convert to FileCheck.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82710 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-24 20:23:02 +00:00
David Goodwin
1f52895692 Make the end-of-itinerary mark explicit. Some cleanup.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82709 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-24 20:22:50 +00:00
Dan Gohman
3c98fefa35 Add nounwind to this test.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82708 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-24 20:20:08 +00:00
Chris Lattner
48ba9ff3c4 tidy up, fix a memory leak in Regex::isValid
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82707 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-24 20:15:51 +00:00
Dale Johannesen
07de8d1acf Clarify that llvm attaches C language semantics to
functions with names that match the C library.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82701 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-24 18:38:21 +00:00
Torok Edwin
c338499525 Constant propagating byval pointer is safe if function is readonly.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82700 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-24 18:33:42 +00:00
Victor Hernandez
96b930ddc7 Auto-upgrade malloc instructions to malloc calls.
Reviewed by Devang Patel.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82694 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-24 17:47:49 +00:00
Bob Wilson
c34666ee18 Fix a hypothetical problem for targets with StackGrowsUp and a non-zero
LocalAreaOffset.  (We don't have any of those right now.)
PEI::calculateFrameObjectOffsets includes the absolute value of the
LocalAreaOffset in the cumulative offset value used to calculate the
stack frame size.  It then adds the raw value of the LocalAreaOffset
to the stack size.  For a StackGrowsDown target, that raw value is negative
and has the effect of cancelling out the absolute value that was added
earlier, but that obviously won't work for a StackGrowsUp target.  Change
to subtract the absolute value of the LocalAreaOffset.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82693 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-24 16:42:27 +00:00
Devang Patel
f9e0a220fc Move parent assertion check before metadata deletion.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82692 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-24 16:19:11 +00:00
Torok Edwin
30a94e3b42 Don't constant propagate byval pointers, since they are not really pointers, but
rather structs passed by value.
This fixes PR5038.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82689 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-24 09:47:18 +00:00
Daniel Dunbar
48f7ce88a6 Add count/not tools as executables.
- Apparently, I'm willing to do incredibly stupid things in the name of portability.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82685 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-24 06:23:57 +00:00
Chris Lattner
b55e068e53 unconditionally compute MMI even if the target doesn't support EH or Debug info, because the target may use it for other things, this fixes PR5036
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82684 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-24 05:44:53 +00:00
Chris Lattner
63e944b05e reapply r82348 with a fix, thanks Jeffrey.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82683 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-24 05:38:36 +00:00
Evan Cheng
393e277ecd Fix PR5024 with a big hammer: disable the double-def assertion in the scavenger.
LiveVariables add implicit kills to correctly track partial register kills. This works well enough and is fairly accurate. But coalescer can make it impossible to maintain these markers. e.g.

        BL <ga:sss1>, %R0<kill,undef>, %S0<kill>, %R0<imp-def>, %R1<imp-def,dead>, %R2<imp-def,dead>, %R3<imp-def,dead>, %R12<imp-def,dead>, %LR<imp-def,dead>, %D0<imp-def>, ...
...
	%reg1031<def> = FLDS <cp#1>, 0, 14, %reg0, Mem:LD4[ConstantPool]
...
   	%S0<def> = FCPYS %reg1031<kill>, 14, %reg0, %D0<imp-use,kill>

When reg1031 and S0 are coalesced, the copy (FCPYS) will be eliminated the the implicit-kill of D0 is lost. In this case it's possible to move the marker to the FLDS. But in many cases, this is not possible. Suppose

	%reg1031<def> = FOO <cp#1>, %D0<imp-def>
...
   	%S0<def> = FCPYS %reg1031<kill>, 14, %reg0, %D0<imp-use,kill>

When FCPYS goes away, the definition of S0 is the "FOO" instruction. However, transferring the D0 implicit-kill to FOO doesn't work since it is the def of D0 itself. We need to fix this in another time by introducing a "kill" pseudo instruction to track liveness.

Disabling the assertion is not ideal, but machine verifier is doing that job now. It's important to know double-def is not a miscomputation since it means a register should be free but it's not tracked as free. It's a performance issue instead.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82677 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-24 02:27:09 +00:00
Evan Cheng
ad934b821c Clean up LiveVariables and change how it deals with partial updates and kills. This also eliminate the horrible check which scan forward to the end of the basic block. It should be faster and more accurate.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82676 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-24 02:15:22 +00:00
Jeffrey Yasskin
1d75d3a8ae Roll back r82348, which introduced an infinite loop in ParseCStringVector() that
a trivial unittest would have caught.  This revision also adds the trivial
unittest.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82675 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-24 01:14:07 +00:00
Dale Johannesen
d5405a67d5 A minor improvment in accuracy to inline cost
computation, and some cosmetics.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82660 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-23 22:05:24 +00:00
Dale Johannesen
30f290dac3 Disable test; what it's testing for is wrong.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82658 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-23 21:46:36 +00:00
David Goodwin
127221fbdc Checkpoint NEON scheduling itineraries.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82657 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-23 21:38:08 +00:00
Dan Gohman
4e918b2c8c Use getStoreSize() instead of getStoreSizeInBits()/8.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82656 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-23 21:07:02 +00:00
Dan Gohman
d1f3cee976 Spruce up some comments.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82655 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-23 21:06:36 +00:00
Dan Gohman
8a55ce4a39 Rename several variables from EVT to more descriptive names, now that EVT
is also the name of their type, as declarations like "EVT EVT" look
really odd.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82654 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-23 21:02:20 +00:00
Dan Gohman
54d4a53c42 Add an EVT::getStoreSize function, like getStoreSizeInBits but in bytes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82653 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-23 20:59:10 +00:00
Jakob Stoklund Olesen
39523e2cfb Fix verification of explicit operands.
The machine code verifier did not check for explicit operands correctly. It
used MachineInstr::getNumExplicitOperands, but that method may cheat and use
the declared count in the TargetInstrDesc.

Now we check the explicit operands one at a time in visitMachineOperand.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82652 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-23 20:57:55 +00:00
Bob Wilson
959b002c4d Remove BlackfinRegisterInfo::getFrameIndexOffset since it is the same as the
default implementation.  Update comment on the default version, which made it
sound like most targets override it.  Currently only X86 and SystemZ override
this method.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82651 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-23 20:57:02 +00:00
Dan Gohman
9f765dd0ee Correct a comment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82648 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-23 20:40:16 +00:00
Devang Patel
ffa363c94b Delete attached metadata when an instruction is deleted.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82647 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-23 20:33:51 +00:00