From:	Hans-Bernhard Broeker <broeker AT physik DOT rwth-aachen DOT de>
Newsgroups:	comp.os.msdos.djgpp
Subject:	Re: one more about memory alloc - free memory of needed amount only
Date:	24 Aug 2000 08:23:17 GMT
Organization:	Aachen University of Technology (RWTH)
Lines:	34
Message-ID:	<8o2m1l$6tb$1@nets3.rz.RWTH-Aachen.DE>
References:	<39A4EB1B DOT 29871 DOT 12C6F4 AT localhost>
NNTP-Posting-Host:	acp3bf.physik.rwth-aachen.de
X-Trace:	nets3.rz.RWTH-Aachen.DE 967105397 7083 137.226.32.75 (24 Aug 2000 08:23:17 GMT)
X-Complaints-To:	abuse AT rwth-aachen DOT de
NNTP-Posting-Date:	24 Aug 2000 08:23:17 GMT
Originator:	broeker@
To:	djgpp AT delorie DOT com
DJ-Gateway:	from newsgroup comp.os.msdos.djgpp
Reply-To:	djgpp AT delorie DOT com

Leon AT caresystems DOT com DOT au wrote:
> Hello sorry to revisit this subejct but

> a day or two ago i asked about possibility of alligning memory so
> that say malloc(25) returns address x and then one modulates this
> address so that it is alligned to a multiple of say 64 bytes. (by
> making malloc(25+63))

You don't modulate it after the fact. You make the original request
larger so that a suitable block of the required length and alignment
is guaranteed to be found inside the block returned by malloc().
I.e. if you want a block of 300 bytes aligned to a 64-byte boundary,
you would:

	unsigned char * realptr = malloc(300+63);
	some_type * ptr_to_use = (some_type *) 
           (realptr + (((unsigned long) realptr) % 64));



> the problem is - how would one free memory that is a remainder of offset?

One wouldn't. You can't do that. The memory sacrificed for alignment
is lost. Which is one of the reasons why this is so rarely used.  The
gains from better alignment often won't compensate the additional
hassles needed to set it up (two pointers to store, loss of
potentially large fractions of memory).

Linux systems have a special function 'memalign' to allocate such
blocks with a particular alignment, which does essentially the same as
described above, but without you having to do it by hand.
-- 
Hans-Bernhard Broeker (broeker AT physik DOT rwth-aachen DOT de)
Even if all the snow were burnt, ashes would remain.

- Raw text -