Fwd: [RFC, PowerPC, Darwin] Clean up generation of ha16() / lo16() markers

[David Fang]

Hal, Ulrich,

Full llvm tests finished, and I did find 3 new failures:

FAIL: LLVM :: DebugInfo/dwarf-public-names.ll (4448 of 9018)
******************** TEST 'LLVM :: DebugInfo/dwarf-public-names.ll' FAILED 
*****
***************
Script:
--
gtimeout 5m  /Users/fang/local/src/LLVM-svn/gcc40-cmake-build/bin/./llc 
-generat
e-dwarf-pubnames -filetype=obj -o 
/Volumes/Isolde/builds/LLVM/gcc40-cmake-build/
test/DebugInfo/Output/dwarf-public-names.ll.tmp.o < 
/Volumes/Isolde/sources/LLVM
-svn/llvm.git/test/DebugInfo/dwarf-public-names.ll
gtimeout 5m 
/Users/fang/local/src/LLVM-svn/gcc40-cmake-build/bin/./llvm-dwarfdu
mp -debug-dump=pubnames 
/Volumes/Isolde/builds/LLVM/gcc40-cmake-build/test/Debug
Info/Output/dwarf-public-names.ll.tmp.o | 
/Users/fang/local/src/LLVM-svn/gcc40-c
make-build/bin/./FileCheck 
/Volumes/Isolde/sources/LLVM-svn/llvm.git/test/DebugI
nfo/dwarf-public-names.ll
--
Exit Code: 1
Command Output (stderr):
--
LLVM ERROR: expected relocatable expression
******************** TEST 'LLVM :: DebugInfo/namespace.ll' FAILED 
**************
******
Script:
--
gtimeout 5m  /Users/fang/local/src/LLVM-svn/gcc40-cmake-build/bin/./llc 
-O0 -fil
etype=obj < 
/Volumes/Isolde/sources/LLVM-svn/llvm.git/test/DebugInfo/namespace.l
l > 
/Volumes/Isolde/builds/LLVM/gcc40-cmake-build/test/DebugInfo/Output/namespac
e.ll.tmp
gtimeout 5m 
/Users/fang/local/src/LLVM-svn/gcc40-cmake-build/bin/./llvm-dwarfdu
mp 
/Volumes/Isolde/builds/LLVM/gcc40-cmake-build/test/DebugInfo/Output/namespace
.ll.tmp | /Users/fang/local/src/LLVM-svn/gcc40-cmake-build/bin/./FileCheck 
/Volu
mes/Isolde/sources/LLVM-svn/llvm.git/test/DebugInfo/namespace.ll
--
Exit Code: 1
Command Output (stderr):
--
LLVM ERROR: expected relocatable expression
--

********************
FAIL: LLVM :: DebugInfo/two-cus-from-same-file.ll (4462 of 9018)
******************** TEST 'LLVM :: DebugInfo/two-cus-from-same-file.ll' 
FAILED *
*******************
Script:
--
gtimeout 5m  /Users/fang/local/src/LLVM-svn/gcc40-cmake-build/bin/./llc 
/Volumes
/Isolde/sources/LLVM-svn/llvm.git/test/DebugInfo/two-cus-from-same-file.ll 
-o /V
olumes/Isolde/builds/LLVM/gcc40-cmake-build/test/DebugInfo/Output/two-cus-from-s
ame-file.ll.tmp -filetype=obj -O0
gtimeout 5m 
/Users/fang/local/src/LLVM-svn/gcc40-cmake-build/bin/./llvm-dwarfdu
mp -debug-dump=info 
/Volumes/Isolde/builds/LLVM/gcc40-cmake-build/test/DebugInfo
/Output/two-cus-from-same-file.ll.tmp | 
/Users/fang/local/src/LLVM-svn/gcc40-cma
ke-build/bin/./FileCheck 
/Volumes/Isolde/sources/LLVM-svn/llvm.git/test/DebugInf
o/two-cus-from-same-file.ll
--
Exit Code: 1
Command Output (stderr):
--
LLVM ERROR: expected relocatable expression
--

********************

My branch is currently sync'd to r181825.
Is there something newer on trunk that might address this?
If so, what's the minumum revision I should re-merge from?

Fang


> Results sooner than expected:
> Applying Ulrich's patches against the powerpc-darwin8 branch, the 
> CodeGen/PowerPC tests (375) came clean, as did MC/PowerPC.
>
> % bin/llvm-lit -s -v test/CodeGen/PowerPC
> Testing Time: 156.69s
>   Expected Passes    : 372
>   Expected Failures  : 3
>
> % bin/llvm-lit -s -v test/MC/PowerPC
> Testing Time: 3.86s
>   Expected Passes    : 11
>
> Is that good enough, or do you want a full test run?  (I'll start one now.)
>
> Fang
>
>>  Can you please test this on your Darwin system?
>>
>>  Thanks again,
>>  Hal
>>
>>  ----- Forwarded Message -----
>>  From: "Ulrich Weigand" <Ulrich.Weigand at de.ibm.com>
>>  To: llvm-commits at cs.uiuc.edu
>>  Cc: "Hal Finkel" <hfinkel at anl.gov>
>>  Sent: Wednesday, May 15, 2013 11:01:58 AM
>>  Subject: [RFC, PowerPC, Darwin] Clean up generation of ha16() / lo16()
>>  markers
>> 
>>
>>  Hello,
>>
>>  when targeting the Darwin assembler, we need to generate markers ha16()
>>  and
>>  lo16() to designate the high and low parts of a (symbolic) immediate.
>>  This
>>  is necessary not just for plain symbols, but also for certain symbolic
>>  expression, typically along the lines of ha16(A - B).  The latter doesn't
>>  work when simply using VariantKind flags on the symbol reference.
>>
>>  This is why the current back-end used hacks (explicitly called out as such
>>  via multiple FIXMEs) in the symbolLo/symbolHi print methods.  I'd like to
>>  get rid of this special treatment as a first step towards getting rid of
>>  symbolLo/symbolHi itself.  This is interesting because it will allow
>>  non-standard (but legal) use of the @l and @hi markers, e.g. to do
>>  "lis ..., symbol at l"  (which is valid if unusual assembler, but will not be
>>  handled correctly by the llvm assembler right now).
>>
>>  The attached patch uses target-defined MCExpr codes to represent the
>>  Darwin
>>  ha16/lo16 constructs, following along the lines of the equivalent solution
>>  used by the ARM back end to handle their :upper16: / :lower16: markers.
>>  This allows us to get rid of special handling both in the
>>  symbolLo/symbolHi
>>  print method and in the common code MCExpr::print routine.  Instead, the
>>  ha16 / lo16 markers are printed simply in a custom print routine for the
>>  target MCExpr types.
>>
>>  I don't have a live Darwin system to test on, but the Darwin assembler
>>  tests in the test suite all still pass.
>>
>>  Does this look reasonable?
>>
>>  Bye,
>>  Ulrich
>>
>>  (See attached file: diff-reloc-mcexpr)
>
>

-- 
David Fang
http://www.csl.cornell.edu/~fang/