This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Unable to bootstrap gcc snapshot of 20001211 on Dynix/ptx
- To: Joe Buck <jbuck at racerx dot synopsys dot com>
- Subject: Re: Unable to bootstrap gcc snapshot of 20001211 on Dynix/ptx
- From: Robert Lipe <robertlipe at usa dot net>
- Date: Mon, 18 Dec 2000 21:31:36 -0600
- Cc: Janis Johnson <janis at us dot ibm dot com>, biswapesh dot chattopadhyay at bt dot com, jsm28 at cam dot ac dot uk, gcc at gcc dot gnu dot org
- References: <OF9F982775.D7379715-ON882569BA.0003873E@LocalDomain> <200012190053.QAA07046@racerx.synopsys.com>
Joe Buck wrote:
> Janis Johnson writes:
>
> > Biswa posted a test case that showed that bcopy() from DYNIX/ptx V4.4.8
> > does not handle overlapping memory. Apparently it was following an
> > earlier standard that did not require it to handle such a case, but
> > DYNIX/ptx V4.6.0 follows UNIX 98 and bcopy() handles overlaps there.
>
> Ouch. I was going to suggest just getting rid of the bcopy calls
> and rely on libiberty to provide memcpy and memmove for systems that
> lack it. But libiberty appears to assume that it can implement memmove, on
> systems that don't have it, as a wrapper around bcopy.
But isn't a system far more likely to have memmove as described in ISO C
library ago than bcopy anyway? Replacing the calls to bcopy still seems
like a better path.
RJL