This is the mail archive of the
binutils@sources.redhat.com
mailing list for the binutils project.
RE: PowerPC no small data build uses .sbss? (but sbss should not be used, right?)
- From: "Dave Korn" <dave dot korn at artimi dot com>
- To: "'William. Rivet'" <William dot Rivet at hs dot utc dot com>,<binutils at sources dot redhat dot com>
- Date: Tue, 7 Jun 2005 16:55:45 +0100
- Subject: RE: PowerPC no small data build uses .sbss? (but sbss should not be used, right?)
----Original Message----
>From: William. Rivet
>Sent: 07 June 2005 14:41
> By using the gcc -mno-sdata it seems the code in fact does not use the
> small data registers, and upon inspection of the assembler listing file,
> it appears data is marked as ".comm zz1, 1, 4" for one example, and
> ".comm zz3,300,4" for another.
>
> So at the assembly level, there appears to be no distinction for small
> data....looking at the map file, I find zz1 is mapped to the .sbss
> section and zz3 is mapped to the COMMON section.
>
> I really would like to avoid explaining to our regulating body why we
> have a small-bss section in our mapfile....is there a way to really
> prevent .sbss sections from being generated? (the small initialized data
> correctly goes into the .data section, not .sdata, so I don't show that
> in my example)
>
> I tried the ld "-d" option and the linker script
> FORCE_COMMON_ALLOCATION, but no luck. I have run out of ideas.
> .sbss 0x00000000 0x8 load address 0xc0100010
> .sbss 0x00000000 0x7 fred.o
> 0x00000000 zz1
> 0x00000004 zz2
> 0x00000006 yy
> .sbss 0x00000007 0x1 joe.o
> 0x00000007 x
I notice all those things are less than 8 bytes. Does "ld --gpsize=0"
help any?
cheers,
DaveK
--
Can't think of a witty .sigline today....