Re: A Linux/libc 5 patch for egcs 1.1.2

["H.J. Lu" <hjl at varesearch dot com>]

Jeffrey A Law wrote:

>   In message <m10HtKT-000AUZC@shanghai.varesearch.com>you write:
>   > The connection to lucon.org has been down for 5 days now. If you need
>   > a reply, please send to hjl@varesearch.com for now.
> Thanks.  Will do.
>
>   > As I said, libio is very tricky. If you take a look at
>   > libio/config/linuxlibc1.mt, you will find this:
>   >
>   > IO_OBJECTS=iogetc.o ioputc.o iofeof.o ioferror.o \
>   >         filedoalloc.o fileops.o genops.o iofclose.o \
>   >         iovsprintf.o iovsscanf.o strops.o iogetline.o
>   >
>   > That means other C files in libio are not included. We don't care those
>   > weak aliases in files which are not in libstdc++. BTW, you won't get
>   > those weak aliases on any other systems, except for Linux.
> OK.  Thanks for the explanation.
>
> So I see iogetline.o in both linux.mt and linuxlibc1.mt, but no weak
> defs in it.  libc-5.46 defines _IO_getline.
>
> Similarly we don't use weak symbols in filedoalloc, fileops and strops, but
> they define functions which are also found in libc.a for libc-5.46.

That is ok as long as the one file in libio from egcs has more symbols
than the same file in libc 5.4.46.

>
>
> So, is there some reason why we don't have weak definitions for functions
> in those files?

The weak aliases are mainly used to provide the standard stdio functions
in such a way that user can override them while libc still works.


H.J.