glibc fix for /lib/ld.so.1 (with attachments this time)

Hello parisc-linux hackers,

Finally tracked down the glibc problem (with help from tausq and others :)). I have attached the changelog entry, and the dpatch. I am assuming you will push it upstream Carlos?

When working on this problem it caused a gcc ICE. The gcc ICE has been reported at:

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=10215

This is why the ENTRY_POINT macro is uglier than really necessary.

Once this patch is in glibc, then we can remove the RGH_FIX hack from the 2.4 kernels as well.

oh. This patch fixes this assert in glibc:

rbrad@vega:~$ /lib/ld.so.1 --verify /bin/sh
Inconsistency detected by ld.so: rtld.c: 879: dl_main: Assertion
`_rtld_local._dl_rtld_map.l_libname' failed!

Thanks!

- Ryan

Ryan Bradetich <[email protected]>

2003-03-26 Ryan Bradetich <[email protected]>

* sysdeps/hppa/elf/entry.h: New file.

#! /bin/sh -e

# All lines beginning with `# DP:' are a description of the patch. # DP: Description: <your description>
# DP: Author:
# DP: Upstream status: [In CVS | Debian-Specific | Pending | Not submitted ] # DP: Status Details:
# DP: Date:

if [ $# -ne 2 ]; then
echo >&2 "basename $0: script expects -patch|-unpatch as argument" exit 1
fi
case "$1" in

    -patch) patch -d "$2" -f --no-backup-if-mismatch -p1 < $0;;
    -unpatch) patch -d "$2" -f --no-backup-if-mismatch -R -p1 < $0;;
    *)
	echo >&2 "`basename $0`: script expects -patch|-unpatch as argument"
	exit 1

esac
exit 0

# append the patch here and adjust the -p? flag in the patch calls.
--- /dev/null	2003-01-25 02:37:00.000000000 -0700
+++ glibc-2.3.1/sysdeps/hppa/elf/entry.h	2003-03-27 03:55:34.000000000 -0700
@@ -0,0 +1,10 @@
+#ifndef __ASSEMBLY__
+extern void _start (void);
+#endif
+
+/* The function's entry point is stored in the first word of the
+   function descriptor (plabel) of _start().  */
+#define ENTRY_POINT ({long int *tmp = (long int *)((long)_start & ~2); tmp[0];})
+
+/* We have to provide a special declaration.  */
+#define ENTRY_POINT_DECL(class) class void _start (void);

Ryan,

Looks good. It looks like the only real way to solve this since ia64 and powerpc64 have similar entry.h macro's.

Not any uglier than the ia64 and powerpc64 macro's that do similar things ;)

Thank you very much for working on this! :)

c.