dangling pointer bug and proposed patch

[john]

Hello,

There is a dangling pointer bug in GNU m4, exposed by this example:

$ m4 -dqeat
define(`f',`one')f(undefine(`f')`two')
m4trace: -1- define(`f', `one')
m4trace: -2- undefine(`f')
m4trace: -1- øò·øò·(`two') -> `øò·øò·H'
øò·øò·H

The problem is that undefining a macro while it is in the "collect
arguments" phase leaves a dangling symbol pointer on the stack.  Bug
aside, the GNU m4 info file doesn't seem to specify what *should*
happen in this situation.  I propose that when a macro whose arguments
have just been expanded discovers that its pre-argument-expansion
definition has been deleted, it should use the "current" definition
(i.e. the same definition a new invocation of the macro would use).
If no current definition exists, the macro should use the expansion
$0($@), which treats the macro as effectively undefined, except that
leading whitespace is stripped from the arguments.

I've written up the gory details (mostly to help me make sure I
understand them), including justification of the proposed behaviour,
in the attached PDF.

A patch for the proposed fix against m4-1.4.4 is also attached.
(The branch name m4-1.4.4.1 is just for my own convenience.)

Regards,

John B.

-- 
John Brzustowski - http://www.ProgramOrPerish.com

m4bug.pdf
Description: m4bug.pdf

m4_1.4.4-1.diff.gz
Description: Binary data