--MimeMultipartBoundary
Content-Type: text/plain; charset=us-ascii

Hi,

On Mon, Sep 21, 1998 at 09:58:10AM +0000, I wrote:
> After a bit of investigation, I found out that the crash
> happens after the make stage1, when ./xgcc is used to
> compile opnew.o from egcs/gcc/cp/new1.cc. The crash is
> triggered by (the newly created) cc1plus.

Looking a bit deeper, I located the crash somewhere around
the end of get_run_time() which is in gcc/toplev.c, l. 1238.
The offending get_run_time() call is from the TIMEVAR macro
in rest_of_compilation(), line 3645 of the same file. This
function is called from finish_function(), gcc/cp/decl.c,
when the lineno parameter =88.

get_run_time() is left via
    unlk    a6
    rts
which should normally cause no harm, even if the stack is
corrupted, right?

When debugging with gdb and the pc is at the unlk statement
a "finish" command to gdb leads to the crash while doing
two "stepi" commands (one for unlk and one for rts) returns
the pc normally to the calling function.

Could this be a caching problem?

I'll try to do some more experiments tonight or so, but do
not expect too much.

Regards
    Ingolf
-- 

Ingolf Koch         http://www.minet.uni-jena.de/~ingolf/ 
PGP: 0x7B3B5661  213C 828E 0C92 16B5  05D0 4D5B A324 EC04
--MimeMultipartBoundary--