This is the mail archive of the
mailing list for the GCC project.
Re: [SH] double precision floating point dwarf fix
Kaz Kojima wrote:
Christian BRUEL <firstname.lastname@example.org> wrote:
This fixes the DWARF information for double precision floating point
values in register location for little endian SHs
The only SH core that I know about that have register numbering correct
is sh-300 when PR=SZ=1. I commented it out in the code for reference, in
case -mfmovd implies SZ=1. But not by default since this is not the ABI.
Could you give a comment for the new function at the just
before the function? The inline comment for sh-300 could
be move into it too. The patch itself looks ok.
I will add the following comment and commit if ok.
/* 64 bit floating points memory transfers are loaded as paired single
precision loads or store. So DWARF information needs fixing in little
endian, (unless PR=SZ=1 in FPSCR). */
>> (TARGET_DWARF_REGISTER_SPAN): Defined.
>> * config/sh/sh-protos.h (sh_dwarf_register_span): Declared.
I guess that the active voice is usual in gcc/ChangeLog for these.