Re: Synthetic register related: Seg fault

[Fergus Henderson <fjh at cs dot mu dot OZ dot AU>]

On 15-Jan-2003, Andy Walker <ja_walker@earthlink.net> wrote:
> What would be the best approach to hunting down flaws that cause things like 
> this Segmentation fault?

Configure GCC with --enable-checking.
Compile it with `-g' so that you can use gdb.

Compile your test case with `-v -da -Q'.
`-Q' will show which function in the test case is causing it to crash.
`-v' shows how `cc1' was invoked (useful for invoking cc1 manually in
gdb).  `-da' dumps the RTL to a file after each stage.

Use gdb to get a stack trace.
Compile the compiler with debugging enabled (-g),
use the `-v' option to see how cc1 is getting invoked,
and then do

	bash$ gdb cc1
	gdb> run <arguments>
	... (it will stop at the Seg fault)
	gdb> where
	...
	gdb> list

Print out the values of interesting variables (e.g. the ones
in the statement which got a segmentation fault).
You can use the `pt' and `pr' macros from the gdbinit.in file to
display them.  For example, if there is a value of type `tree' named
`t', and a value of type `rtx' named `r', you can use these commands:

	gdb> source .gdbinit
	gdb> print t
	gdb> pt
	gdb> print r
	gdb> pr
	gdb> pt

-- 
Fergus Henderson <fjh@cs.mu.oz.au>  |  "I have always known that the pursuit
The University of Melbourne         |  of excellence is a lethal habit"
WWW: <http://www.cs.mu.oz.au/~fjh>  |     -- the last words of T. S. Garp.