This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: arm_hard_regno_mode_ok update


Nick,

I don't think this code can't be right in general.  I really think a user 
needs to be able to name *any* reserved register in an ASM type statement. 
 Why is the check in there in the first place?  We didn't need it before...

  if (regno <= LAST_ARM_REGNUM)
    /* If the register is a general purpose ARM register we allow
       it only if it not a special register (SP, LR, PC) and only
       if there will be enough (non-special) registers to hold the
       entire value.  */
    return regno < (SP_REGNUM - (unsigned) NUM_REGS (mode));

Why can't the user use LR (r14)?  Other than on a call boundary, that is 
just a general purpose register.

R.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]