Fixing one of the problems discovered for bug 154 (internal).

  - Output of 16-bit integer multiplication (IMUL.U16) writes to the entire 32-bit register if a full register (i.e. not half register such as R1L) is specified as the destination.  Before, it was only storing the lower 16-bit of the product into the full register.
  - Modified function header to declare at least 4 predicates in cuobjdump_to_ptxplus.  This will mask out the problem with the predicate registers used before initialized.
  - Modified register value print out: now unsigned integers display their decimal values as well.

[git-p4: depot-paths = "//depot/gpgpu_sim_research/fermi/distribution/": change = 13246]

1 files changed, 20 insertions, 1 deletions

diff --git a/cuobjdump_to_ptxplus/cuobjdumpInst.cc b/cuobjdump_to_ptxplus/cuobjdumpInst.cc
index fac53b4..2bdac6d 100644
--- a/cuobjdump_to_ptxplus/cuobjdumpInst.cc
+++ b/cuobjdump_to_ptxplus/cuobjdumpInst.cc
@@ -721,6 +721,24 @@ void cuobjdumpInst::printCuobjdumpOperands()
 	}
 }
 
+void cuobjdumpInst::printCuobjdumpOutputModifiers(const char* defaultMod)
+{
+	std::list<std::string>::iterator typemod = m_typeModifiers->begin();
+	if (*typemod == ".U16" or *typemod == ".S16") {
+		std::list<std::string>::iterator dest_op = m_operands->begin(); 
+		std::string& destination = *dest_op; 
+		if (destination[destination.length()-1] == 'l') {
+			output(".lo");  // write to the lower 16-bits 
+		} else if (destination[destination.length()-1] == 'h') {
+			output(".hi");  // write to the upper 16-bits 
+		} else {
+			output(".wide");  // write to the whole 32-bits 
+		}
+		return; 
+	}
+	output(defaultMod);  // default output modifier for mul 
+}
+
 std::string int_default_mod () { return ".u32" ;}
 
 
@@ -1309,7 +1327,8 @@ void cuobjdumpInst::printCuobjdumpPtxPlus(std::list<std::string> labelList, std:
 	else if(m_base == "IMUL" || m_base == "IMUL32I")
 	{
 		printCuobjdumpPredicate();
-		output("mul.lo");
+		output("mul");
+      printCuobjdumpOutputModifiers(".lo"); 
 		printCuobjdumpBaseModifiers();
 
 		if(m_typeModifiers->size() == 0)