This is the mail archive of the binutils@sourceware.org mailing list for the binutils 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: [Patch]Extend GAS arm_feature_set struct to provide more available bits


On Mon, Mar 16, 2015 at 3:42 PM, Alan Modra <amodra@gmail.com> wrote:
> On Mon, Mar 16, 2015 at 02:43:01PM +0800, Terry Guo wrote:
>> I am also curious here. As you can see from below URL, the default
>> mode for gcc 4.7 and 4.8 is still c90 which doesn't support long long.
>
> https://gcc.gnu.org/onlinedocs/gcc/Long-Long.html
> Note "as an extension GCC supports them in C90 mode".
>
> --

Thanks Alan. So in your opinion can we say it is safe to use long long
here? Will there be any other subtle performance issue if we use long
long? I just grep-ed the gas source code and did find some usages of
long long as below:

config/tc-d30v.c:  long long insn;
config/tc-d30v.c:      long long value, /* The value of the bits.  */
config/tc-d30v.c:    long long insn,
config/tc-d30v.c:       long long insn,
config/tc-d30v.c:       long long insn1,
config/tc-d30v.c:       long long insn2,
config/tc-d30v.c:  long long insn = NOP2;
config/tc-d30v.c:  long long      insn;
config/tc-d30v.c:static long long prev_insn = -1;
config/tc-d30v.c:  long long insn;
config/tc-xtensa.c:typedef unsigned long long bitfield;
config/bfin-defs.h:  long long i_value; /* if constant, the value.  */
config/te-vms.c:     long long *cdt,
config/te-vms.c:  unsigned long long create;
config/te-vms.c:  long long gmtoff, secs, nsecs;
config/te-vms.c:  long long cdt;

BR,
Terry


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