really horrible extensions that are disabled by default but that can
be accepted by -fheinous-gnu-extensions (but which always emit a 
warning when enabled).

As our first instance of this, implement PR3788/PR3794, which allows
non-lvalues in inline asms in contexts where lvalues are required. bleh.

llvm-svn: 66910

llvm-svn: 66598

code where Sema can get to it.  No functionality change.

llvm-svn: 66596

done in sema, and is reflected by the existing PR3258.  In the meantime,
fix PR3682 by disabling a bogus assertion (which doesn't account for + 
operands).

llvm-svn: 66533

add support for modifiers on named references, like %c[foo].

llvm-svn: 66532

time handle + operands in operand counting, fixing asm.c:t7 to
expand into $2 instead of $1.

llvm-svn: 66531

temporary std::string to fix a fixme.

llvm-svn: 66530

asm.  This allows us to properly handle the case when an optimizer duplicates
the asm, such as here:

void bar() {
  int i;
  for (i = 0; i < 3; ++i)
    asm("foo %=" : : "r"(0));
}

we now produce:

_bar:
	xorl	%eax, %eax
	## InlineAsm Start
	foo 0
	## InlineAsm End
	## InlineAsm Start
	foo 1
	## InlineAsm End
	## InlineAsm Start
	foo 2
	## InlineAsm End
	ret

instead of:

_bar:
	xorl	%eax, %eax
	## InlineAsm Start
	foo 1
	## InlineAsm End
	## InlineAsm Start
	foo 1
	## InlineAsm End
	## InlineAsm Start
	foo 1
	## InlineAsm End
	ret

This also fixes a fixme by eliminating a static.

llvm-svn: 66528

llvm-svn: 66521

ConvertAsmString and shrink it a bit.  No functionality change.

llvm-svn: 66520

it in the stack trace, giving us stuff like:

Stack dump:
0.	Program arguments: clang t.c -emit-llvm 
1.	<eof> parser at end of file
2.	t.c:1:5: LLVM IR generation of declaration 'a'
3.	t.c:1:9: LLVM IR generation of compound statement ('{}')
4.	t.c:2:3: LLVM IR generation of compound statement ('{}')
Abort

llvm-svn: 66154

multiple sequential cases to a) not create tons of fall-through basic blocks
and b) not recurse deeply.  This fixes codegen on 100K deep cases, and improves
codegen on moderate cases from this:

        switch i32 %tmp, label %sw.epilog [
                i32 1000, label %sw.bb
                i32 1001, label %sw.bb1
                i32 1002, label %sw.bb2
                i32 1003, label %sw.bb3
                i32 1004, label %sw.bb4
...
sw.bb:          ; preds = %entry
        br label %sw.bb1

sw.bb1:         ; preds = %entry, %sw.bb
        br label %sw.bb2

sw.bb2:         ; preds = %entry, %sw.bb1
        br label %sw.bb3

sw.bb3:         ; preds = %entry, %sw.bb2
        br label %sw.bb4

to:

        switch i32 %tmp, label %sw.epilog [
                i32 1000, label %sw.bb
                i32 1001, label %sw.bb
                i32 1002, label %sw.bb
                i32 1003, label %sw.bb
                i32 1004, label %sw.bb
sw.bb:  ;; many preds

llvm-svn: 66015

llvm-svn: 65866

Make sure to reset the DidCallStackSave variable before emitting a compound statement. Fixes PR3649.

llvm-svn: 65291

llvm-svn: 64445

llvm-svn: 64252

llvm-svn: 64205

llvm-svn: 64203

llvm-svn: 64159

llvm-svn: 64157

llvm-svn: 64095

If people could beat on it and let me know if there are any new
semantics required by newer language standards or DRs or any little
details I goofed on, I'd be happy to fix any issues found.

llvm-svn: 64079

llvm-svn: 64053

Add support for emitting cleanup blocks. Make EmitCompoundStatement emit cleanup blocks if necessary

llvm-svn: 64051

correctly.  This should lay the ground work to throw the big switch
and start code gening break and continue in the presense of vlas.

llvm-svn: 64046

llvm-svn: 64030

llvm-svn: 64021

llvm-svn: 64020

llvm-svn: 64014

If an input constraint refers to an output constraint, it should have the same constraint info as the output constraint. Fixes PR3417

llvm-svn: 63127

llvm-svn: 62674

llvm-svn: 62444

llvm-svn: 62441

Change TargetInfo::validateInputConstraint to take begin/end name iterators instead of the number of outputs. No functionality change.

llvm-svn: 62433

llvm-svn: 62069

llvm-svn: 62049

llvm-svn: 62043

llvm-svn: 62041

string.

That said, we should probably try and track down the correct clobber 
lists for the targets that don't have them (PPC, ARM, and Sparc), 
so that we can generate correct code.

llvm-svn: 61298

warning by using an unsigned index.

llvm-svn: 61292