[M3commit] CVS Update: cm3

Jay K jay.krell at cornell.edu
Mon May 10 09:48:48 CEST 2010


diff attached
really it's not that interesting -- all just Solaris configurys/m3makefile/python/script stuff and trivially porting a small change from gcc 4.4.0, just a few lines added in x86/solaris specific file.

 - Jay

----------------------------------------
> Date: Mon, 10 May 2010 09:35:56 +0000
> To: m3commit at elegosoft.com
> From: jkrell at elego.de
> Subject: [M3commit] CVS Update: cm3
>
> CVSROOT: /usr/cvs
> Changes by: jkrell at birch. 10/05/10 09:35:56
>
> Modified files:
> cm3/m3-sys/cminstall/src/config-no-install/: AMD64_SOLARIS
> I386_SOLARIS SOLgnu
> SOLsun
> SPARC64_SOLARIS
> Solaris.common
> cm3/m3-sys/m3cc/gcc/gcc/config/i386/: sol2-10.h
> cm3/m3-sys/m3cc/src/: m3makefile platforms.quake
> cm3/scripts/python/: pylib.py
>
> Log message:
> Use Sun assembler on I386_SOLARIS, AMD64_SOLARIS.
>
> Tweak various Solaris code around C and assembly flags.
> With an eye toward matching what gcc configured
> on Solaris to use Sun assembler would do.
> Including adding the -Qy flag, which isn't important.
> At least for x86, Sparc is still building to see, and Sparc64
> hasn't been tried far enough.
>
> Switch to "small model PIC" by default.
> -xcode=pic13 vs. -xcode=pic32
> Seems no ideal solution here.
>
> Downgrade a bit more from i586 to i686.
> I do wonder though, maybe we should only do i586 where
> users ask -- not all x86 platforms.
> Maybe a "phone home" feature that collects uname -a or such?
>
> Notice that Sun assembler has -K PIC flag, but passing it
> for x86/amd64 gcc output errors, and gcc doesn't pass it.
>
> Notice that though documentation says you can use two part
> GNU configurations like sparc-solaris2.10, I seem to get
> burned repeatedly doing that. Here I change to sparc-sun-solaris2.10
> and that might have been necessary.
> There are supposed to be two sets of names, the user supplied
> possibly shorter one, and the "more canonical" one.
>
> Add -xF switch to Sun cc.
> It sounds like this might enable linker dead stripping.
> I should test out the difference.
>
> Remove the distinction in pylib.py between CC in make.sh vs. Makefile,
> now that I no longer hunt around for GNU as, but just always use
> Sun /usr/bin/ccs/as.
>
> Port this change from gcc 4.4.0:
> http://gcc.gnu.org/viewcvs?view=revision&revision=143758
>
> 2009-01-29 Rainer Orth 
> * config/i386/sol2-10.h [!HAVE_AS_IX86_DIFF_SECT_DELTA]
> (ASM_OUTPUT_DWARF_PCREL): Define.
>
> Which fixes AMD64_SOLARIS -fPIC -funwind-tables with Sun assembler.
>
 		 	   		  
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: 3.txt
URL: <http://m3lists.elegosoft.com/pipermail/m3commit/attachments/20100510/a8ef1084/attachment-0002.txt>


More information about the M3commit mailing list