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: David Edelsohn <dje at watson dot ibm dot com>
- Subject: Re: egcs-19980531, warning patches [part 1/2]
- From: Jeffrey A Law <law at cygnus dot com>
- Date: Sun, 07 Jun 1998 22:31:09 -0600
- cc: "Kaveh R. Ghazi" <ghazi at caip dot rutgers dot edu>, jbuck at synopsys dot com, egcs at cygnus dot com, martin dot kahlert at mchp dot siemens dot de
- Reply-To: law at cygnus dot com
In message <9806080420.AA37828@rios1.watson.ibm.com>you write:
> I don't understand what Jeff's demonstration that the HPUX kernel
> does not contain the symbol calloc proves.
Actually, that was a 4.4BSD kernel, not hpux.
What is shows is that if we start using calloc, and I want to compile
my 4.4 kernel with egcs, then I may now have to start providing calloc
for the kernel to link.
While it is simple to add the routine to the kernel, the fact that
gcc requires such routines to be added to the kernel has been a
source of major complaints for kernel developers for years.
Kernel developers, after protest, have added memcpy, memset and other
routines to their kernels so they can be compiled with gcc. Making
them add more routines does not help us from a public relations
standpoint.
Granted, this will only be needed if one tries to do block profiling
of the kernel, but I don't want to see us starting adding references
to new library routines in libgcc. It's just the wrong thing to do
for what is supposed to be support routines for things the compiler
can't do itself (ie floating point emulation, wide integer operations,
some EH things, etc).
jeff