This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: latest i370 patches
- To: rth at cygnus dot com (Richard Henderson)
- Subject: Re: latest i370 patches
- From: linas at linas dot org
- Date: Tue, 20 Jul 1999 00:45:14 -0500 (CDT)
- Cc: linas at linas dot org, egcs-patches at egcs dot cygnus dot com
Hi,
It's been rumoured that Richard Henderson said:
> +/* Some remarks about unsigned_jump_follows_p():
> + gcc is built around the assumption that branches are signed
> + or unsigned, whereas the 370 doesn't care; its the compares that
> + are signed or unsigned.
>
> Look at alpha cmpsi and blt for a typical example of how
> to handle this.
OK, noted; I did not design the unsigned_jump_follows_p() thing,
what is there is how I found it. I added some notes for myself
and/or future readers of the code. I studied the alpha cmpsi breifly,
not sure I understood it exactly; when I get a chance to work
on it, I will do this.
> - ASM_OUTPUT_SHORT (FILE, GEN_INT (0x05E0)); \
> - ASM_OUTPUT_SHORT (FILE, GEN_INT (0x5800 | STATIC_CHAIN_REGNUM << 4)); \
> - ASM_OUTPUT_SHORT (FILE, GEN_INT (0xE00A)); \
> - ASM_OUTPUT_SHORT (FILE, GEN_INT (0x58F0)); \
> - ASM_OUTPUT_SHORT (FILE, GEN_INT (0xE00E)); \
> - ASM_OUTPUT_SHORT (FILE, GEN_INT (0x07FF)); \
> + ASM_OUTPUT_SHORT (FILE, gen_rtx (CONST_INT, VOIDmode, 0x05E0)); \
> + ASM_OUTPUT_SHORT (FILE, gen_rtx (CONST_INT, VOIDmode, 0x5800 | \
> + STATIC_CHAIN_REGNUM << 4)); \
> + ASM_OUTPUT_SHORT (FILE, gen_rtx (CONST_INT, VOIDmode, 0xE00A)); \
> + ASM_OUTPUT_SHORT (FILE, gen_rtx (CONST_INT, VOIDmode, 0x58F0)); \
> + ASM_OUTPUT_SHORT (FILE, gen_rtx (CONST_INT, VOIDmode, 0xE00E)); \
> + ASM_OUTPUT_SHORT (FILE, gen_rtx (CONST_INT, VOIDmode, 0x07FF)); \
>
> This change is exceedingly wrong.
? I made no changes to this code, I don't know why its diffing like this.
This looks like someone else went in and patched this code; this is not
what I have. I guess there have been patches applied to the i370 code
since Sept 98 that came from someone else? I greped changelog for
370, 390, esa, mvs and nothing showed up... If you have any record of
who has been going in there & patching this code, let me know; we are
definitely going to be on a collision course unless we coordinate.
(Is this patched code from someone in IBM Germany maybe?)
--linas