Size changes, round 1

Joern Rennecke amylaar@cygnus.co.uk
Tue Feb 22 09:50:00 GMT 2000


In article < 10002200123.AA18561@vlsi1.ultra.nyu.edu > you wrote:
: Since nobody seemed to have any strong opinions on the topic of
: TYPE_SIZE_UNIT, I made a first round of changes to clean the code up.  I
: added a DECL_SIZE_UNIT, BINFO_SIZE_UNIT and CLASSTYPE_SIZE_UNIT, made the
: handling of the size_int and bitsize_int macros more consistent, and made a
: few related changes in areas affected by the above.

: These new fileds do add a small amount of memory usage in tree nodes, but
: make the handling of teh two different sizes more consistent.

: I committed this change:

: Sat Feb 19 18:43:13 2000  Richard Kenner  <kenner@vlsi1.ultra.nyu.edu>

: 	* c-common.c (decl_attributes): Set DECL_SIZE_UNIT.
: 	* c-decl.c (duplicate_decls, finish_enum): Likewise.
: 	(finish_decl): Remove -Wlarger-than code from here.
: 	* flags.h (id_clash_len): Now int.
: 	(larger_than_size): Now HOST_WIDE_INT.
: 	* fold-const.c (size_int_wide): No more HIGH parm; NUMBER is signed.

Why do you declare NUMBER as signed?  It really isn't.


If the objective is to save memory without sacrificing functionality,
we could use a single INTEGER_CST for the size if it is constant,
giving the size in bits, and use a special tree node that contains
separate sizes in bits and units for the variable case.


More information about the Gcc-patches mailing list