Date: Tue, 29 Dec 1998 20:54:48 +0800 From: Peter Wemm <peter@netplex.com.au> To: Alfred Perlstein <bright@hotjobs.com> Cc: "Jordan K. Hubbard" <jkh@zippy.cdrom.com>, Kelly Yancey <kbyanc@freedomnet.com>, freebsd-hackers@FreeBSD.ORG Subject: Re: (where are linux threads?) Re: pthreads question/problem... Message-ID: <199812291254.UAA79782@spinner.netplex.com.au> In-Reply-To: Your message of "Tue, 29 Dec 1998 03:16:34 EST." <Pine.BSF.4.05.9812290305210.2148-100000@bright.fx.genx.net>
next in thread | previous in thread | raw e-mail | index | archive | help
Alfred Perlstein wrote: > > On Mon, 28 Dec 1998, Jordan K. Hubbard wrote: > > > > Doesn't it seem obvious to anyone other than myself that the current > > > interest in porting Linux's 1-1 threads is a pretty good indication that > > > people want 1-1 threads in FreeBSD? Is the only way we are going to get > > > > I think it seems more obvious that the set of people willing to talk > > about more aggressive threading models and the people willing to > > actually do the work don't often (enough) intersect or there would be > > more tangible development effort going on in this area. > > > > Don't just gripe to us about how disillusioned you are over the fact > > that santa clause was discovered to be a fake and your first sexual > > experience didn't live up to its billing, that is merely extraneous > > information and doesn't convey much more than "I'm really upset!", > > something which won't get you much sympathy around here given that > > we're not real big on maternal instincts around here. If the lack of > > something in FreeBSD annoys you, fix it. Where do you think the > > motivation to fix so much of what was previously broken in FreeBSD > > came from? > > I may be mistaken, but you're both misunderstanding me. from reading > recent posts on -current i could have sworn i saw that _NATIVE_ > "linuxthreads" were now available to freebsd. > > I think you need to compile a kernel and world with: > CFLAGS+= COMPAT_LINUX_THREADS > > to achive this, the only thing that seems to be missing is a FreeBSD > userland interface to it. > > Is there any chance that this will become standard in FreeBSD? (the thread > died off without mentioning if it will become default) what about > userland interface? I hope so.. If we can get the stuff cleaned up a bit in time, yes it should be a go. The main reservations I've heard about it are moving struct sigacts outside of the upages and loosing the swappability. Granted it't not a really big thing yet, but once we go from 32 to (say) 64 signals it doubles. Struct proc itself is around the same size. Because of malloc roundup, that'd be 1/2KB of extra unswappable data per process (and soon maybe 1K). Multiply that by a heap of processes and it adds up in a memory crunch. Cheers, -Peter To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199812291254.UAA79782>