[LLDB][MIPS] Getting 0 index for h/w watchpoint is not necessarily an error

[mohit.bhakkad]

Summary

Index for H/W watchpoint regsisters starts from 0, so considering 0 as failure is wrong.

[jaydeep]

Looks good to me.

lldb/trunk/source/Plugins/Process/Linux/NativeRegisterContextLinux_mips64.cpp

if(vacant_watches > 0)
                         return index;
                     }
                     size = addr + size - break_addr;
                     addr = break_addr;
                 }
             }
         }
     }
-    return 0;
+    return LLDB_INVALID_INDEX32;
 }
 
 bool
 NativeRegisterContextLinux_mips64::IsMSA(uint32_t reg_index) const
 {
     return (m_reg_info.first_msa <= reg_index && reg_index <= m_reg_info.last_msa);
 }
 
NativeRegisterContextLinux_mips64::SetHardwareWatchpoint (
 
     // First reading the current state of watch regs
     DoReadWatchPointRegisterValue(m_thread.GetID(), static_cast<void *>(&regs));
 
     // Try if a new watch point fits in this state
     int index = GetVacantWatchIndex (&regs, addr, size, watch_flags, NumSupportedHardwareWatchpoints());
 
     // New watchpoint doesn't fit
-    if (!index)
+    if (index == LLDB_INVALID_INDEX32)
     return LLDB_INVALID_INDEX32;
 
 
     // It fits, so we go ahead with updating the state of watch regs 
     DoWriteWatchPointRegisterValue(m_thread.GetID(), static_cast<void *>(&regs));
 
     // Storing exact address  
     hw_addr_map[index] = addr;