tests in 5.9. cocoon toolchain
peterph at centrum.cz
Mon Mar 3 11:48:20 PST 2008
Robert Connolly wrote:
> On Thursday February 14 2008 10:55:38 am Petr Cerny wrote:
>> AFAIK gcc/config/ia64/linux.h is for Itanium (or whatever name that
>> Intel RISC platform has). x86-64 should be in gcc/config/i386/linux64.h
>> and (maybe but quite unlikely in gcc/config/i386/*64*).
> I'm worried it's miscompiling.
I have managed look a little bit more into it.
I have tried to compile the book example on SuSE x86_64 and it has the
same results. I have also written my own SSP testcase which worked as
expected in all 64/32bit SuSE/own multilib HLFS combinations (i.e.
SIGABRT after first byte of stack being overwritten).
My colleague traced the culprit in the book's original testcase to stack
alignment on a 16 byte boundary
(http://www.x86-64.org/documentation/abi-0.99.pdf - section 3.2.2). So
it seems to be GCC aligning the stack to meet the specification.
I'll try to build the system again with latest SVN to reproduce it and
look more into it...
More information about the hlfs-dev