This is the mail archive of the
gcc-help@gcc.gnu.org
mailing list for the GCC project.
Re: Help working out where GCC decides the size of 0xFFFFFFFF
- From: Oleg Endo <oleg dot endo at t-online dot de>
- To: Alex Brown <alex dot g dot brown at gmail dot com>
- Cc: gcc-help at gcc dot gnu dot org
- Date: Sat, 18 Apr 2015 20:01:55 +0200
- Subject: Re: Help working out where GCC decides the size of 0xFFFFFFFF
- Authentication-results: sourceware.org; auth=none
- References: <CAB-RZVSf=P9OAaqvBGqO-MUiysbuEs5_sLv12h_usaefNPToPQ at mail dot gmail dot com>
Hi,
On Sat, 2015-04-18 at 12:13 +1000, Alex Brown wrote:
> Hi,
>
> I am running GCC 5 on an SH-2 (7058) simulator that I wrote myself.
> The simulator works well enough to (mostly) run GCC, however I have
> encountered what is probably a bug in my simulator that GCC triggers
> and am looking for help to work out where in GCC this bug is
> triggered.
>
> Running GCC as cross compiler targeting my platform seems to produce
> the correct results.
> The bug is that when I run GCC on my target, GCC identifies the
> sizeof(0xFFFFFFFF) as 8 bytes, whereas when I run GCC as a cross
> compiler targeting my simulator it correctly identifies the size as 4
> bytes. The sizeof(0x7FFFFFFF) is identified correctly as 4 bytes on
> both.
I have checked this on a generic sh-elf/newlib cross compiler config
(i686 host) and on a native sh4-linux config (gcc 4.9). In both cases
the result is '4'.
> I am looking for help on narrowing down where GCC makes the decision
> on the size of the 0xFFFFFFFF literal so I can investigate further.
>
> So any pointers as to were to start looking for where GCC calculates
> the size of a number would help me greatly.
How did you configure/build the toolchain that runs natively on SH?
Cheers,
Oleg