I noticed tons of warnings associated with CGTARG_Is_Right_Shift_Op() when I compiled a source file in CG with "-Wall -O2".
The patch cleans up the warnings.
Note that the only use for CGTARG_Is_Right_Shift_Op() is the following
code in cgemit.cxx:
if (CGTARG_Is_Right_Shift_Op (op)) {
if (TN_register(OP_opnd(op,0+predicated)) == REGISTER_zero) {
DevWarn ("Redundant shift instruction in %sBB:%d (PC=0x%x)",
BB_rotating_kernel(OP_bb(op)) ? "SWPd " : "",
BB_id(OP_bb(op)), PC);
if (TFile != stdout) { /* only print to .t file */
Print_OP_No_SrcLine (op);
}
}
Since there are no zero registers the X8664 architectures, it didn't
matter that CGTARG_Is_Right_Shift_Op() produced bogus results.
Another approach would be just removing the function from
osprey/be/cg/x8664/cgtarget_arch.h and replacing the above code
snippet with:
#if defined(TARG_MIPS) || defined(TARG_IA64) || defined(TARG_LOONGSON)
<SNIPPET>
#endif
I think either approach is fine, I'll let the reviewers decide which
(or yet another) approach we should use.
Could a gatekeeper take a look at the patch when they have the chance?
Thanks,
Doug
cgtarget_arch.patch
Description: cgtarget_arch.patch
------------------------------------------------------------------------------ All the data continuously generated in your IT infrastructure contains a definitive record of customers, application performance, security threats, fraudulent activity and more. Splunk takes this data and makes sense of it. Business sense. IT sense. Common sense. http://p.sf.net/sfu/splunk-d2d-oct
_______________________________________________ Open64-devel mailing list Open64-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/open64-devel
