[Patch, AArch64] Add missing support for hexadecimal-format floating-point literal in the FMOV instruction

[Tristan Gingold]

On Mar 5, 2013, at 3:49 PM, Yufeng Zhang wrote:

> Hi,
> 
> The AArch64 floating-point move (immediate) instruction FMOV can have the floating point constant specified either in decimal notation or as a string beginning "0x" followed by the hexadecimal representation of its IEEE754 encoding.
> 
> This patch adds the missing support for the latter notation.
> 
> OK for the trunk and 2.23 release branch?

Ok for backporting to the 2.23 branch.

Tristan.

> 
> Thanks,
> Yufeng
> 
> 
> gas/
> 
> 2013-03-05  Yufeng Zhang  <yufeng.zhang@arm.com>
> 
> 	* config/tc-aarch64.c (aarch64_imm_float_p): Rename 'e' to 'pattern';
> 	add comment.
> 	(aarch64_double_precision_fmovable): New function.
> 	(parse_aarch64_imm_float): Add parameter 'dp_p'; call the new
> 	function; handle hexadecimal representation of IEEE754 encoding.
> 	(parse_operands): Update the call to parse_aarch64_imm_float.
> 
> gas/testsuite/
> 
> 2013-03-05  Yufeng Zhang  <yufeng.zhang@arm.com>
> 
> 	* gas/aarch64/diagnostic.s: Add test.
> 	* gas/aarch64/diagnostic.l: Update.
> 	* gas/aarch64/illegal.s: Add test.
> 	* gas/aarch64/illegal.l: Update.
> 	* gas/aarch64/fpmov.s: New file.
> 	* gas/aarch64/fpmov.d: New file.<aarch64-fmov-hexadecimal.patch>