This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: egcs-19980531, warning patches [part 1/2]
- To: jbuck at synopsys dot com, law at cygnus dot com
- Subject: Re: egcs-19980531, warning patches [part 1/2]
- From: "Kaveh R. Ghazi" <ghazi at caip dot rutgers dot edu>
- Date: Sun, 7 Jun 1998 16:06:43 -0400 (EDT)
- Cc: egcs at cygnus dot com, martin dot kahlert at mchp dot siemens dot de
> From: Jeffrey A Law <law@hurl.cygnus.com>
>
> It just seems silly to add yet another routine that has to be
> provided for embedded systems or kernels just to eliminate a
> warning in libgcc2.
> jeff
BTW, it is also more optimal on linux, as confirmed by
Ulrich. (Not too big a deal here as its done in the __bb_init_prg
function which shouldn't get used more than once.)
Anyway, I'd rather not bottleneck the whole patch on this one issue. :-)
Jeff, would you please let me know if the rest of the patch is okay?
I can install everything but the calloc change and we can hash that
out at our leisure.
On the calloc issue, I've heard ~5 positive (including
meissner, martin.kahlert, jbuck, drepper, etc) but Jeff is opposed.
It's been asserted calloc is a simple couple of line function for
embedded/kernel hackers to code up if its missing. And note that
switching to using it removes the only appearance of bzero in
libgcc2.c. Seems like we would thus be requiring the same _number_ of
functions for embedded targets/kernels, just a different one.
Jeff, if you decide to reject calloc, I think we could eliminate
the warning by using memset instead of bzero. It looks like gcc doesn't
warn about memset even if it isn't prototyped. (Memset is already
required as it is used elsewhere, so it doesn't further burden anyone.)
But we lose the calloc optimization on linux. (No big deal.)
However, we do have the issue reamaining that the existing
bzero does not cover the whole malloc'ed area. I'd like you to
determine if this is intentional or a bug.
Thanks,
--Kaveh
--
Kaveh R. Ghazi Project Manager / Custom Development
ghazi@caip.rutgers.edu Icon CMT Corp.