[Bug gas/18500] VLDR immediate not using VMOV

[nickc at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=18500

Nick Clifton <nickc at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED
                 CC|                            |nickc at redhat dot com

--- Comment #4 from Nick Clifton <nickc at redhat dot com> ---
Hi Alessandro,

  There are three problems with this patch:

  1. It checks for X_op being O_bignum, but then it uses the value in
X_add_number.  If X_op is O_bignum then the value is in the generic_bignum
array, and the value is only a floating point value if X_add_number is -1.  If
X_add_number is zero (or any value other than -1) then generic_bignum holds a
large integer value, not a large floating point value.

  2. It is not clear to me that we actually want to convert VLDR to VMOV.  If
the same constant is being used multiple times then putting it in the literal
pool will save space.  Ideally we could assume that the producer of the VLDR
instructions (either the compiler or the assembly programmer), knows what they
are doing and not modify their code for them.  But if you still feel that this
conversion is a useful idea then it ought to be controllable via a command line
option.

  3. It does not follow the GNU Coding Standards in terms of the layout of the
code.  (See: http://www.gnu.org/prep/standards/)

Cheers
  Nick

-- 
You are receiving this mail because:
You are on the CC list for the bug.