Commit Graph

131 Commits

Author SHA1 Message Date
Dan Gohman f9bbcd1afd Major calling convention code refactoring.
Instead of awkwardly encoding calling-convention information with ISD::CALL,
ISD::FORMAL_ARGUMENTS, ISD::RET, and ISD::ARG_FLAGS nodes, TargetLowering
provides three virtual functions for targets to override:
LowerFormalArguments, LowerCall, and LowerRet, which replace the custom
lowering done on the special nodes. They provide the same information, but
in a more immediately usable format.

This also reworks much of the target-independent tail call logic. The
decision of whether or not to perform a tail call is now cleanly split
between target-independent portions, and the target dependent portion
in IsEligibleForTailCallOptimization.

This also synchronizes all in-tree targets, to help enable future
refactoring and feature work.

llvm-svn: 78142
2009-08-05 01:29:28 +00:00
Daniel Dunbar c3719c36e6 Move most targets TargetMachine constructor to only taking a target triple.
- The C, C++, MSIL, and Mips backends still need the module.

llvm-svn: 77927
2009-08-02 23:37:13 +00:00
Richard Osborne bbb772ace9 Add extra SEXT pattern.
llvm-svn: 77920
2009-08-02 22:45:24 +00:00
Daniel Dunbar 31b44e8f6c Normalize Subtarget constructors to take a target triple string instead of
Module*.

Also, dropped uses of TargetMachine where unnecessary. The only target which
still takes a TargetMachine& is Mips, I would appreciate it if someone would
normalize this to match other targets.

llvm-svn: 77918
2009-08-02 22:11:08 +00:00
Chris Lattner 1472cf5b3f move dwarf debug info section selection stuff from TAI to
TLOF, unifying all the dwarf targets at the same time.

llvm-svn: 77889
2009-08-02 07:24:22 +00:00
Chris Lattner 2be66ce420 xcore TAI doesn't need TM.
llvm-svn: 77868
2009-08-02 04:42:09 +00:00
Chris Lattner b25afe081c remove TargetAsmInfo::TM, which is now dead. The basic TAI class now
no longer depends on TM!

llvm-svn: 77863
2009-08-02 04:27:24 +00:00
Chris Lattner 73d577c933 Make SectionKind::get() private.
llvm-svn: 77835
2009-08-02 00:02:44 +00:00
Chris Lattner 0c40266b5a Remove "JumpTableDataSection" from TAI, instead, have AsmPrinter
compute it based on what it knows.  As part of this, rename getSectionForMergeableConstant
to getSectionForConstant because it works for non-mergable constants also.

The only functionality change from this is that Xcore will start dropping
its jump tables into readonly section instead of data section in -static mode.
This should be fine as the linker resolves the relocations.  If this is a
problem, let me know and we'll come up with another solution.

llvm-svn: 77833
2009-08-01 23:46:12 +00:00
Chris Lattner b2bbb61f33 REmove dead fields of TAI.
llvm-svn: 77820
2009-08-01 22:40:22 +00:00
Chris Lattner 5aa4952625 update for rename
llvm-svn: 77817
2009-08-01 22:06:53 +00:00
Chris Lattner c9c277ba0f Change SectionKind to be a property that is true of a *section*, it
should have no state that is specific to particular globals in the
section.  In this case, it means the removal of the "isWeak" and
"ExplicitSection" bits.  MCSection uses the new form of SectionKind.

To handle isWeak, I introduced a new SectionInfo class, which is
SectionKind + isWeak, and it is used by the part of the code generator
that does classification of a specific global.

The ExplicitSection disappears.  It is moved onto MCSection as a new
"IsDirective" bit.  Since the Name of a section is either a section
or directive, it makes sense to keep this bit in MCSection.  Ultimately
the creator of MCSection should canonicalize (e.g.) .text to whatever
the actual section is.

llvm-svn: 77803
2009-08-01 21:11:14 +00:00
Nick Lewycky 04e3a30a21 Add newline at end of file to remove gcc warning.
llvm-svn: 77791
2009-08-01 19:09:44 +00:00
Chris Lattner 4d2c0f9008 switch off of 'Section' onto MCSection. We're not properly using
MCSection subclasses yet, but this is a step in the right direction.

llvm-svn: 77708
2009-07-31 18:48:30 +00:00
Owen Anderson 4056ca9568 Move types back to the 2.5 API.
llvm-svn: 77516
2009-07-29 22:17:13 +00:00
Chris Lattner 5034329f8d pass the mangler down into the various SectionForGlobal methods.
No functionality change.

llvm-svn: 77432
2009-07-29 05:09:30 +00:00
Owen Anderson c2c7932c64 Change ConstantArray to 2.5 API.
llvm-svn: 77347
2009-07-28 18:32:17 +00:00
Chris Lattner 5e693ed07b Rip all of the global variable lowering logic out of TargetAsmInfo. Since
it is highly specific to the object file that will be generated in the end,
this introduces a new TargetLoweringObjectFile interface that is implemented
for each of ELF/MachO/COFF/Alpha/PIC16 and XCore.

Though still is still a brutal and ugly refactoring, this is a major step
towards goodness.

This patch also:
1. fixes a bunch of dangling pointer problems in the PIC16 backend.
2. disables the TargetLowering copy ctor which PIC16 was accidentally using.
3. gets us closer to xcore having its own crazy target section flags and
   pic16 not having to shadow sections with its own objects.
4. fixes wierdness where ELF targets would set CStringSection but not
   CStringSection_.  Factor the code better.
5. fixes some bugs in string lowering on ELF targets.

llvm-svn: 77294
2009-07-28 03:13:23 +00:00
Chris Lattner 86b7255776 Eliminate getNamed/getUnnamedSection, adding a new and unified getOrCreateSection
instead.

llvm-svn: 77186
2009-07-27 06:17:14 +00:00
Chris Lattner 149465ea06 Eliminate SectionFlags, just embed a SectionKind into Section
instead and drive things based off of that.

llvm-svn: 77184
2009-07-27 05:32:16 +00:00
Daniel Dunbar ee01b242e8 Factor commonality in triple match routines into helper template for registering
classes, and migrate existing targets over.

llvm-svn: 77126
2009-07-26 05:03:33 +00:00
Daniel Dunbar bc981d8efa Kill Target specific ModuleMatchQuality stuff.
- This was overkill and inconsistently implemented.

llvm-svn: 77114
2009-07-26 02:22:58 +00:00
Chris Lattner 60f3b73e11 this is (unfortunately) several changes mixed together:
1. Spell SectionFlags::Writeable as "Writable".
2. Add predicates for deriving SectionFlags from SectionKinds.
3. Sink ELF-specific getSectionPrefixForUniqueGlobal impl into
   ELFTargetAsmInfo.
4. Fix SectionFlagsForGlobal to know that BSS/ThreadBSS has the
   BSS bit set (the real fix for PR4619).
5. Fix isSuitableForBSS to not put globals with explicit sections
   set in BSS (which was the reason  wasn't fixed earlier).
6. Remove my previous hack for PR4619.

llvm-svn: 77085
2009-07-25 18:57:34 +00:00
Daniel Dunbar 691a4784db Simplify JIT target selection.
- Instead of requiring targets to define a JIT quality match function, we just
   have them specify if they support a JIT.

 - Target selection for the JIT just gets the host triple and looks for the best
   target which matches the triple and has a JIT.

llvm-svn: 77060
2009-07-25 10:09:50 +00:00
Daniel Dunbar 5680b4f285 Add new helpers for registering targets.
- Less boilerplate == good.

llvm-svn: 77052
2009-07-25 06:49:55 +00:00
Daniel Dunbar 0dd5e1ed39 More migration to raw_ostream, the water has dried up around the iostream hole.
- Some clients which used DOUT have moved to DEBUG. We are deprecating the
   "magic" DOUT behavior which avoided calling printing functions when the
   statement was disabled. In addition to being unnecessary magic, it had the
   downside of leaving code in -Asserts builds, and of hiding potentially
   unnecessary computations.

llvm-svn: 77019
2009-07-25 00:23:56 +00:00
Owen Anderson edb4a70325 Revert the ConstantInt constructors back to their 2.5 forms where possible, thanks to contexts-on-types. More to come.
llvm-svn: 77011
2009-07-24 23:12:02 +00:00
Eli Friedman 95fc6ee51a Remove unused member functions.
llvm-svn: 76960
2009-07-24 07:43:59 +00:00
Chris Lattner d2aeef0703 remove SectionFlags::Small: it is only used on Xcore, and we'll find
a better solution for it in the future.

llvm-svn: 76818
2009-07-22 23:27:22 +00:00
Richard Osborne 4f7f19a7c0 Thread local globals don't require special handling by the linker and so can
be placed in the standard data / bss sections.

llvm-svn: 76735
2009-07-22 11:01:00 +00:00
Owen Anderson 47db941fd3 Get rid of the Pass+Context magic.
llvm-svn: 76702
2009-07-22 00:24:57 +00:00
Chris Lattner 0ecf7b6f4a remove the last bits of SectionFlagsForGlobal. There is some flag here that
depends on XS1A, but I think the ReadOnlySection is already set up for this
and there is no testcase that this breaks.  If this is really needed, we can
add the appropriate parameterization to TargetAsmInfo in the future to support
this.

llvm-svn: 76667
2009-07-21 22:42:37 +00:00
Chris Lattner bf4bc25236 don't mask out the small flag and then reapply it later.
llvm-svn: 76666
2009-07-21 22:39:28 +00:00
Chris Lattner 684d5fe95d if Xcore doesn't support TLS, it doesn't have to worry about thread local LLVM IR, it should be rejected by a front-end.
llvm-svn: 76665
2009-07-21 22:37:38 +00:00
Chris Lattner 43893573f4 remove the Xcore implementation of SelectSectionForGlobal. While you have
to twist your brain to see it, I believe it is the same as ELFTargetAsmInfo::SelectSectionForGlobal.

llvm-svn: 76664
2009-07-21 22:36:53 +00:00
Chris Lattner aa30d25bb4 Remove the XCore custom implementation of MergeableConstSection, relying on
the generic ELF version instead.  This will result in its mergable constant
sections getting named ".rodata.cst4" instead of ".cp.const4", but the
linker looks at the section flags, not the name of the section AFAICT.

llvm-svn: 76659
2009-07-21 22:25:52 +00:00
Chris Lattner fd047e11a1 Remove some overridden functions in XCoreTargetAsmInfo that are
implemented exactly the same way as its ELFTargetAsmInfo subclass 
has them.

llvm-svn: 76653
2009-07-21 21:26:32 +00:00
Eli Friedman da9eda8ef6 Remove shift amount flavor. It isn't actually complete enough to
be useful, and it's currently unused.  (Some issues: it isn't actually 
rich enough to capture the semantics on many architectures, and
semantics can vary depending on the type being shifted.)

llvm-svn: 76633
2009-07-21 20:12:16 +00:00
Chris Lattner 100865e59d make AsmPrinter::doFinalization iterate over the global variables
and call PrintGlobalVariable, allowing elimination and simplification
of various targets.

llvm-svn: 76604
2009-07-21 18:38:57 +00:00
Chris Lattner f5f5880c5a reduce indentation by using an early exit.
llvm-svn: 76596
2009-07-21 17:59:36 +00:00
Bill Wendling a3c6f6bffa Add plumbing for the `linker_private' linkage type. This type is meant for
"private" symbols which the assember shouldn't strip, but which the linker may
remove after evaluation. This is mostly useful for Objective-C metadata.

This is plumbing, so we don't have a use of it yet. More to come, etc.

llvm-svn: 76385
2009-07-20 01:03:30 +00:00
Daniel Dunbar 7eaf6dd84a Tweak cmake files for the four targets that don't split CodeGen out.
- We should canonicalize this and get rid of the cmake and llvm-config hacks to
   support both variants.

llvm-svn: 76350
2009-07-19 00:26:46 +00:00
Daniel Dunbar 83779208cc Add dependencies from TargetInfo onto .td generation.
- Shouldn't really be necessary, but currently .inc files get included into
   some main target headers.

llvm-svn: 76349
2009-07-19 00:21:12 +00:00
Daniel Dunbar 67038c1333 Put Target definitions inside Target specific header, and llvm namespace.
llvm-svn: 76344
2009-07-18 23:03:22 +00:00
Eli Friedman efa2215e30 Handle void in XCoreTargetLowering::isLegalAddressingMode. Triggers in
test/CodeGen/Generic.

llvm-svn: 76162
2009-07-17 07:16:38 +00:00
Richard Osborne 0cceec520c Combine an unaligned store of unaligned load into a memmove.
llvm-svn: 75908
2009-07-16 12:50:48 +00:00
Richard Osborne efe3f9c9f4 Lower the threshold at which memcpy / memmove / memset stop being expanded
inline in the XCore.

llvm-svn: 75906
2009-07-16 12:41:34 +00:00
Richard Osborne 5092f7ee93 Fix typo in last commit on expansion of unaligned loads.
llvm-svn: 75903
2009-07-16 10:48:47 +00:00
Richard Osborne bfdc557c8a Expand unaligned 32 bit loads from an address which is a constant
offset from a 32 bit aligned base as follows:

  ldw low, base[offset >> 2]
  ldw high, base[(offset >> 2) + 1]
  shr low_shifted, low, (offset & 0x3) * 8
  shl high_shifted, high, 32 - (offset & 0x3) * 8
  or result, low_shifted, high_shifted

Expand 32 bit loads / stores with 16 bit alignment into two 16 bit
loads / stores.

llvm-svn: 75902
2009-07-16 10:42:35 +00:00
Richard Osborne 25b33cb035 Custom lower unaligned 32 bit stores and loads into libcalls. This is
a big code size win since before they were expanding to upto 16
instructions.

llvm-svn: 75901
2009-07-16 10:21:18 +00:00