On Mon, 24 Jul 2000, Bruce Dawson <[EMAIL PROTECTED]> wrote:
> I'm afraid the problem isn't Netscape as much as its free(3). Free
> doesn't call brk(2) to give the memory back to the system - at least
> I've never seen it do that.
>
> Why not? To maintain backward compatibility with programs that use both
> free() and brk(). Unfortunately, one needs to use a different memory
> management package to actually free and unmap pages AND I'm not aware of
> any generally used package that provides a free() that automatically
> calls brk when a large trailing contiguous chunk becomes free.
I believe I heard that if the process relinquishes (I am not
neccessarily saying `free(3)' here...) a whole page at once then the
linux kernel will reclaim it. Fractions of a page will not be
reclaimed.
Simple experiments calling malloc and free and then waiting around
doing something else suggest the memory is reclaimed... (via ps,
free(1), /proc/meminfo)
But IANAKH :-)
Karl Runge
**********************************************************
To unsubscribe from this list, send mail to
[EMAIL PROTECTED] with the following text in the
*body* (*not* the subject line) of the letter:
unsubscribe gnhlug
**********************************************************
