This is the mail archive of the
java-patches@sourceware.cygnus.com
mailing list for the Java project.
Re: IA-64 Boehm gc patch
- To: Andrew Haley <aph at pasanda dot cygnus dot co dot uk>
- Subject: Re: IA-64 Boehm gc patch
- From: Jeff Sturm <jeff dot sturm at appnet dot com>
- Date: Mon, 19 Jun 2000 16:07:24 -0400
- CC: java-patches at sourceware dot cygnus dot com, Hans Boehm <boehm at acm dot org>
- Organization: AppNet Inc.
- References: <20000619194959.1687.qmail@pasanda.cygnus.co.uk>
Andrew Haley wrote:
>
> > Date: Mon, 19 Jun 2000 15:18:30 -0400
> > From: Jeff Sturm <jeff.sturm@appnet.com>
> >
> > Actually we don't use that read wrapper unless MPROTECT_VDB is defined,
> > and last I checked it was disabled in libgcj due to problems with
> > linuxthreads.
>
> Well, MPROTECT_VDB is unconditionally defined for IA64. I'm guessing
> that there's a good reason for this. I'm not using linuxthreads yet,
> so I get to use syscall().
IIRC linuxthreads has its own read wrapper, and the one in boehm-gc
circumvents this. We had problems earlier linking with MPROTECT_VDB
defined. Hans may be able to shed more light on this.
> Are you saying that we could just as well use readv() as syscall()?
Probably, if readv() is a true system call. There may still be
platforms that implement readv in libc.
--
Jeff Sturm
jeff.sturm@appnet.com