From owner-freebsd-ppc Thu Jun 8 1:51:56 2000 Delivered-To: freebsd-ppc@freebsd.org Received: from haldjas.folklore.ee (Haldjas.folklore.ee [193.40.6.121]) by hub.freebsd.org (Postfix) with ESMTP id 5421137BB0D for ; Thu, 8 Jun 2000 01:51:51 -0700 (PDT) (envelope-from narvi@haldjas.folklore.ee) Received: from localhost (narvi@localhost) by haldjas.folklore.ee (8.9.3/8.9.3) with SMTP id KAA62242 for ; Thu, 8 Jun 2000 10:51:44 +0200 (EET) (envelope-from narvi@haldjas.folklore.ee) Date: Thu, 8 Jun 2000 10:51:44 +0200 (EET) From: Narvi To: freebsd-ppc@FreeBSD.ORG Subject: the abi Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-ppc@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG Or, actually, *THE* ABI. We really do need it, so somebody needs to step forward and say what it will be. And no, I won't do it. We have the chices of: 1) the AIX/PowerOpen ABI 2) the SYSV4 PPC ABI 3) the EABI 4) grow our own 1) Really stinks 2) Stinks, but everybody uses it 3) Is a variation of 2) 4) Need not stink in principle. We need to tell the toolchain what it looks like. 1) is probably the worst, and I don't know the best. This way or other, how about somebody deciding something? Sander To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-ppc" in the body of the message From owner-freebsd-ppc Thu Jun 8 7: 3:37 2000 Delivered-To: freebsd-ppc@freebsd.org Received: from alpo.whistle.com (alpo.whistle.com [207.76.204.38]) by hub.freebsd.org (Postfix) with ESMTP id 5A6A937BFE9 for ; Thu, 8 Jun 2000 07:03:31 -0700 (PDT) (envelope-from mark@whistle.com) Received: from [10.1.10.109] (PBG3.whistle.com [207.76.207.128]) by alpo.whistle.com (8.9.1a/8.9.1) with ESMTP id HAA14422 for ; Thu, 8 Jun 2000 07:03:30 -0700 (PDT) Mime-Version: 1.0 X-Sender: mark@207.76.206.1 Message-Id: In-Reply-To: References: Date: Thu, 8 Jun 2000 07:03:27 -0700 To: freebsd-ppc@FreeBSD.ORG From: Mark Peek Subject: Re: the abi Content-Type: text/plain; charset="us-ascii" Sender: owner-freebsd-ppc@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG At 10:51 AM +0200 6/8/2000, Narvi wrote: >We have the chices of: > 1) the AIX/PowerOpen ABI > 2) the SYSV4 PPC ABI > 3) the EABI > 4) grow our own > >1) Really stinks >2) Stinks, but everybody uses it >3) Is a variation of 2) >4) Need not stink in principle. We need to tell the toolchain what it >looks like. > >1) is probably the worst, and I don't know the best. > >This way or other, how about somebody deciding something? I haven't researched this at all so this may be a duplicate of 1-4. How about: 5) Apple Darwin compatible ABI This would allow sharing of tools and might even have some chance of, should I say, binary compatibility. :-) Besides, Apple might be the largest supplier of FreeBSD technology when they start shipping MacOS X. Being compatible would be a "Good Thing (tm)" while having a different standard would fragment FreeBSD. Mark To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-ppc" in the body of the message From owner-freebsd-ppc Thu Jun 8 7:20:36 2000 Delivered-To: freebsd-ppc@freebsd.org Received: from haldjas.folklore.ee (Haldjas.folklore.ee [193.40.6.121]) by hub.freebsd.org (Postfix) with ESMTP id BD8D037B94F for ; Thu, 8 Jun 2000 07:20:30 -0700 (PDT) (envelope-from narvi@haldjas.folklore.ee) Received: from localhost (narvi@localhost) by haldjas.folklore.ee (8.9.3/8.9.3) with SMTP id QAA88778; Thu, 8 Jun 2000 16:20:22 +0200 (EET) (envelope-from narvi@haldjas.folklore.ee) Date: Thu, 8 Jun 2000 16:20:21 +0200 (EET) From: Narvi To: Mark Peek Cc: freebsd-ppc@FreeBSD.ORG Subject: Re: the abi In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-ppc@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG On Thu, 8 Jun 2000, Mark Peek wrote: > At 10:51 AM +0200 6/8/2000, Narvi wrote: > >We have the chices of: > > 1) the AIX/PowerOpen ABI > > 2) the SYSV4 PPC ABI > > 3) the EABI > > 4) grow our own > > > >1) Really stinks > >2) Stinks, but everybody uses it > >3) Is a variation of 2) > >4) Need not stink in principle. We need to tell the toolchain what it > >looks like. > > > >1) is probably the worst, and I don't know the best. > > > >This way or other, how about somebody deciding something? > > I haven't researched this at all so this may be a duplicate of 1-4. How about: > 5) Apple Darwin compatible ABI > > This would allow sharing of tools and might even have some chance of, > should I say, binary compatibility. :-) Besides, Apple might be the largest > supplier of FreeBSD technology when they start shipping MacOS X. Being > compatible would be a "Good Thing (tm)" while having a different standard > would fragment FreeBSD. > But - what do they use? Esp. as I haven't been able to connect to the darwin site during ~ a week now (and that's why it's not listed). It is probably sysv4, though. Also consider that EABI or similar reduced abi has advantages in itself. And I don't think a different standard would 'fragment' FreeBSD. > Mark > To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-ppc" in the body of the message From owner-freebsd-ppc Thu Jun 8 7:42:21 2000 Delivered-To: freebsd-ppc@freebsd.org Received: from raven.ravenbrook.com (raven.ravenbrook.com [193.82.131.18]) by hub.freebsd.org (Postfix) with ESMTP id 6921D37B8C9 for ; Thu, 8 Jun 2000 07:42:18 -0700 (PDT) (envelope-from Richard.Brooksby@pobox.com) Received: from [193.82.131.28] (skylark.ravenbrook.com [193.82.131.28]) by raven.ravenbrook.com (8.9.3/8.9.3) with ESMTP id PAA26765; Thu, 8 Jun 2000 15:41:40 +0100 (BST) From: Richard.Brooksby@pobox.com Mime-Version: 1.0 X-Sender: (Unverified) Message-Id: In-Reply-To: References: Date: Thu, 8 Jun 2000 15:41:44 +0100 To: Narvi Subject: Re: the abi Cc: Mark Peek , freebsd-ppc@FreeBSD.ORG Content-Type: text/plain; charset="us-ascii" ; format="flowed" Sender: owner-freebsd-ppc@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG At 2000-06-08 16:20 +0200, Narvi wrote: > > I haven't researched this at all so this may be a duplicate of >1-4. How about: > > 5) Apple Darwin compatible ABI >> >> This would allow sharing of tools and might even have some chance of, >> should I say, binary compatibility. :-) Besides, Apple might be the largest >> supplier of FreeBSD technology when they start shipping MacOS X. Being >> compatible would be a "Good Thing (tm)" while having a different standard > > would fragment FreeBSD. > >But - what do they use? Esp. as I haven't been able to connect to the >darwin site during ~ a week now (and that's why it's not listed). It is >probably sysv4, though. I don't think such an important decision should be made according to whether or not a particular host is up or down or whether a network connection happens to be working at the moment. Darwin's ABI should be on the list for consideration. >And I don't think a different standard would 'fragment' FreeBSD. That depends how much development has to be duplicated, and how many parts of the system have to be maintained in parallel. For example, if GDB depends on the ABI then do we have to maintain two variants of GDB: one for FreeBSD/PPC and one for Darwin/PPC? What's the list of FreeBSD components that would have to be fragmented like this? The length of the list tells you something about the cost of the choice of ABI. Personally, I think the ABI should be chosen to minimize porting and maintenance effort and maximize stability. To me that means choosing one that's either already well established or will become well established. Darwin's is a good candidate because Apple will be putting it on every Mac before much longer. Which of the other ABIs you've listed is well established? To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-ppc" in the body of the message From owner-freebsd-ppc Thu Jun 8 7:52:38 2000 Delivered-To: freebsd-ppc@freebsd.org Received: from haldjas.folklore.ee (Haldjas.folklore.ee [193.40.6.121]) by hub.freebsd.org (Postfix) with ESMTP id BFC8B37B8C9 for ; Thu, 8 Jun 2000 07:52:19 -0700 (PDT) (envelope-from narvi@haldjas.folklore.ee) Received: from localhost (narvi@localhost) by haldjas.folklore.ee (8.9.3/8.9.3) with SMTP id QAA91987; Thu, 8 Jun 2000 16:52:06 +0200 (EET) (envelope-from narvi@haldjas.folklore.ee) Date: Thu, 8 Jun 2000 16:52:06 +0200 (EET) From: Narvi To: Richard.Brooksby@pobox.com Cc: Mark Peek , freebsd-ppc@FreeBSD.ORG Subject: Re: the abi In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-ppc@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG On Thu, 8 Jun 2000 Richard.Brooksby@pobox.com wrote: > At 2000-06-08 16:20 +0200, Narvi wrote: > > > > I haven't researched this at all so this may be a duplicate of > >1-4. How about: > > > 5) Apple Darwin compatible ABI > >> > >> This would allow sharing of tools and might even have some chance of, > >> should I say, binary compatibility. :-) Besides, Apple might be the largest > >> supplier of FreeBSD technology when they start shipping MacOS X. Being > >> compatible would be a "Good Thing (tm)" while having a different standard > > > would fragment FreeBSD. > > > >But - what do they use? Esp. as I haven't been able to connect to the > >darwin site during ~ a week now (and that's why it's not listed). It is > >probably sysv4, though. > > I don't think such an important decision should be made according to > whether or not a particular host is up or down or whether a network > connection happens to be working at the moment. Darwin's ABI should > be on the list for consideration. You are readying way too much into what I said. And, just in case anybody *STILL* did not notice it: I will not be making the decision. I am but trying to get somebody with the experience and knowledge (and willingness to step forward) to make such. > > >And I don't think a different standard would 'fragment' FreeBSD. > > That depends how much development has to be duplicated, and how many > parts of the system have to be maintained in parallel. For example, > if GDB depends on the ABI then do we have to maintain two variants of > GDB: one for FreeBSD/PPC and one for Darwin/PPC? > Why would we be maintaining the gdb for darwin? Besides, gdb (and the rest of binutils) already support 1-3. > What's the list of FreeBSD components that would have to be > fragmented like this? The length of the list tells you something > about the cost of the choice of ABI. > Errmmm... > Personally, I think the ABI should be chosen to minimize porting and > maintenance effort and maximize stability. To me that means choosing > one that's either already well established or will become well > established. Darwin's is a good candidate because Apple will be > putting it on every Mac before much longer. > The ABI should be selected for it's technical merrits, imho. > Which of the other ABIs you've listed is well established? > All except for 4 (which doesn't even exist yet). It is clear even with a casual look at the list. Sander To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-ppc" in the body of the message From owner-freebsd-ppc Thu Jun 8 8:11:33 2000 Delivered-To: freebsd-ppc@freebsd.org Received: from raven.ravenbrook.com (raven.ravenbrook.com [193.82.131.18]) by hub.freebsd.org (Postfix) with ESMTP id 3D35B37BFF6 for ; Thu, 8 Jun 2000 08:11:24 -0700 (PDT) (envelope-from Richard.Brooksby@pobox.com) Received: from [193.82.131.28] (skylark.ravenbrook.com [193.82.131.28]) by raven.ravenbrook.com (8.9.3/8.9.3) with ESMTP id QAA26851; Thu, 8 Jun 2000 16:11:01 +0100 (BST) From: Richard.Brooksby@pobox.com Mime-Version: 1.0 X-Sender: (Unverified) Message-Id: In-Reply-To: References: Date: Thu, 8 Jun 2000 16:10:52 +0100 To: Narvi Subject: Re: the abi Cc: Mark Peek , freebsd-ppc@FreeBSD.ORG Content-Type: text/plain; charset="us-ascii" ; format="flowed" Sender: owner-freebsd-ppc@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG At 2000-06-08 16:52 +0200, Narvi wrote: >You are readying way too much into what I said. And, just in case anybody >*STILL* did not notice it: I will not be making the decision. I am but >trying to get somebody with the experience and knowledge (and willingness >to step forward) to make such. I don't intend any offense. By creating a list you could affect the outcome, even if you don't make the decision. > > >And I don't think a different standard would 'fragment' FreeBSD. >> >> That depends how much development has to be duplicated, and how many >> parts of the system have to be maintained in parallel. For example, >> if GDB depends on the ABI then do we have to maintain two variants of > > GDB: one for FreeBSD/PPC and one for Darwin/PPC? > >Why would we be maintaining the gdb for darwin? Besides, gdb (and the >rest of binutils) already support 1-3. There are quite a few people (including me) who would like to have a wide range of Unix development tools available on Darwin, so that they can be used in the combined Mac OS X and Unix environment. GDB (and all the other usual stuff) would be of great benefit. One of my clients will be interested in developing the server component of their application to run on Mac OS X. It will be best to do that in the FreeBSD-like part of Mac OS X, since their system is mostly Unix based. They will need tools. > > Personally, I think the ABI should be chosen to minimize porting and >> maintenance effort and maximize stability. To me that means choosing >> one that's either already well established or will become well >> established. Darwin's is a good candidate because Apple will be > > putting it on every Mac before much longer. > >The ABI should be selected for it's technical merrits, imho. Well, it all depends on what this project is _for_. All decisions should be justifiable in terms of the requirements. Since I have no idea what those are I can't really offer any opinion. In general, given the choice between something that is stable, maintainable, and compatible or something that merely has technical merit I would choose stable, maintainable, and compatible any day, even if it's clunky. I advise anyone else to do the same. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-ppc" in the body of the message From owner-freebsd-ppc Thu Jun 8 11:56:44 2000 Delivered-To: freebsd-ppc@freebsd.org Received: from tortilla.manson.net (cc49460-a.bnapk1.occa.home.com [24.0.214.20]) by hub.freebsd.org (Postfix) with ESMTP id 682D037C171 for ; Thu, 8 Jun 2000 11:56:37 -0700 (PDT) (envelope-from jehamby@anobject.com) Received: from localhost ([10.0.0.103]) by server.manson.net (8.9.3/8.9.3) with SMTP id LAA45708; Thu, 8 Jun 2000 11:40:13 -0700 (PDT) (envelope-from jehamby@anobject.com) Message-Id: <200006081840.LAA45708@server.manson.net> To: Narvi Subject: Re: the abi Cc: Mark Peek , freebsd-ppc@FreeBSD.ORG Date: Thu, 8 Jun 2000 11:40:13 -0700 From: Jake Hamby x-mailer: Apple Mail (2.317) Sender: owner-freebsd-ppc@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG Begin forwarded message: > But - what do they use? Esp. as I haven't been able to connect to the > darwin site during ~ a week now (and that's why it's not listed). It is > probably sysv4, though. Darwin uses Mach-O (the Mach object format). I was going to mention this last night as kind of a joke option, but I guess it's a good thing I didn't, since it looks like not many people on the list would have gotten the joke. :) I'm a big MacOS X supporter right now but I don't think that Mach-O's a very good choice, at least since so far I haven't been able to find ANY documentation about the format, besides Apple's source code mods to GCC and binutils, and their "libmacho", which I don't consider sufficient to base the future of FreeBSD/ppc on. For all I know, it might be a pretty cool format, but unless someone on this list has a better idea where to go to find a solid spec for this format, I don't think we should consider it. It looks like it's just too different. Besides, we wouldn't be able to directly take advantage of Apple's modified GCC and binutils anyway, because we'd have to keep the Mach-O changes while backing out or disabling all of the custom NeXT mods to do things like search /System/Library/*.framework/Headers/ for headers and search /System/Library/{FrameworkName}.framework/{FrameworkName} for shared libs. Also, it's not like their modded GCC is 100% solid yet, as I was able to trip up the C++ compiler in MacOS X with a simple (17-line) test program that gave "Can't emit reloc {- symbol "L1$pb"} @ file address 164" errors. I didn't look into this very far, but my theory is that some patch for better Objective-C support broke something the compiler's doing for C++. Which reminds me: if you want REALLY good Obj-C support as a goal for FreeBSD/ppc, then maybe the Apple toolchain is the way to go. :) Still, even for that, I don't think they'll be open-sourcing all of their tools.. they still have some proprietary pieces to their ObjC runtime that I don't think are ever going to make it into Darwin. For FreeBSD/ppc, I vote for whatever Linux/ppc uses (SVR4, I believe). Not only is this going to provide the most solid (and relevant to our needs) toolchain, but it makes Linux ABI emulation in the kernel a little easier. -Jake To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-ppc" in the body of the message From owner-freebsd-ppc Thu Jun 8 11:58: 9 2000 Delivered-To: freebsd-ppc@freebsd.org Received: from haldjas.folklore.ee (Haldjas.folklore.ee [193.40.6.121]) by hub.freebsd.org (Postfix) with ESMTP id E65D037C155 for ; Thu, 8 Jun 2000 11:58:05 -0700 (PDT) (envelope-from narvi@haldjas.folklore.ee) Received: from localhost (narvi@localhost) by haldjas.folklore.ee (8.9.3/8.9.3) with SMTP id UAA97854; Thu, 8 Jun 2000 20:57:46 +0200 (EET) (envelope-from narvi@haldjas.folklore.ee) Date: Thu, 8 Jun 2000 20:57:46 +0200 (EET) From: Narvi To: Jake Hamby Cc: Mark Peek , freebsd-ppc@FreeBSD.ORG Subject: Re: the abi In-Reply-To: <200006081840.LAA45708@server.manson.net> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-ppc@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG On Thu, 8 Jun 2000, Jake Hamby wrote: > Begin forwarded message: > > > But - what do they use? Esp. as I haven't been able to connect to the > > darwin site during ~ a week now (and that's why it's not listed). It is > > probably sysv4, though. > > Darwin uses Mach-O (the Mach object format). I was going to mention this > last night as kind of a joke option, but I guess it's a good thing I > didn't, since it looks like not many people on the list would have gotten > the joke. :) > 8-) > I'm a big MacOS X supporter right now but I don't think that Mach-O's a > very good choice, at least since so far I haven't been able to find ANY > documentation about the format, besides Apple's source code mods to GCC and > binutils, and their "libmacho", which I don't consider sufficient to base > the future of FreeBSD/ppc on. For all I know, it might be a pretty cool > format, but unless someone on this list has a better idea where to go to > find a solid spec for this format, I don't think we should consider it. It > looks like it's just too different. > I'd vote for forgetting about it for now. > Besides, we wouldn't be able to directly take advantage of Apple's > modified GCC and binutils anyway, because we'd have to keep the Mach-O > changes while backing out or disabling all of the custom NeXT mods to do > things like search /System/Library/*.framework/Headers/ for headers and > search /System/Library/{FrameworkName}.framework/{FrameworkName} for shared > libs. Also, it's not like their modded GCC is 100% solid yet, as I was > able to trip up the C++ compiler in MacOS X with a simple (17-line) test > program that gave "Can't emit reloc {- symbol "L1$pb"} @ file address 164" > errors. I didn't look into this very far, but my theory is that some patch > for better Objective-C support broke something the compiler's doing for > C++. Which reminds me: if you want REALLY good Obj-C support as a goal > for FreeBSD/ppc, then maybe the Apple toolchain is the way to go. :) > Still, even for that, I don't think they'll be open-sourcing all of their > tools.. they still have some proprietary pieces to their ObjC runtime that > I don't think are ever going to make it into Darwin. > Ouch. I don't want to do this. > For FreeBSD/ppc, I vote for whatever Linux/ppc uses (SVR4, I believe). > Not only is this going to provide the most solid (and relevant to our > needs) toolchain, but it makes Linux ABI emulation in the kernel a little > easier. > > -Jake > To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-ppc" in the body of the message From owner-freebsd-ppc Thu Jun 8 12:11:52 2000 Delivered-To: freebsd-ppc@freebsd.org Received: from tortilla.manson.net (cc49460-a.bnapk1.occa.home.com [24.0.214.20]) by hub.freebsd.org (Postfix) with ESMTP id 6727737B59B for ; Thu, 8 Jun 2000 12:11:50 -0700 (PDT) (envelope-from jehamby@anobject.com) Received: from localhost (jehamby@localhost) by tortilla.manson.net (8.9.3/8.9.3) with ESMTP id MAA45798 for ; Thu, 8 Jun 2000 12:11:46 -0700 (PDT) (envelope-from jehamby@anobject.com) X-Authentication-Warning: tortilla.manson.net: jehamby owned process doing -bs Date: Thu, 8 Jun 2000 12:11:46 -0700 (PDT) From: Jake Hamby X-Sender: jehamby@tortilla.manson.net To: freebsd-ppc@freebsd.org Subject: a more Darwin-centric approach? Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-ppc@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG I have an idea: Since AFAIK nobody has done a whole lot of work yet in getting the FreeBSD kernel up and running on PPC (please correct me if I'm wrong!), why don't we focus first on getting the stuff from userland, including the ports collection, brought over to Darwin/MacOSX? At WWDC, Apple demoed XFree86 running on Darwin, so it's not a totally uncomfortable environment for UNIX development. This helps Darwin and MacOS X users as well as FreeBSD users (sending goodwill to both communities), and it also helps with the toolchain approach, as if we choose, for example, the SVR4 ABI, we can build a GCC on Darwin that generates this format, and then test it out by adding ELF support to the Darwin kernel. This doesn't have to be an either/or thing, of course. I imagine most of us in this list are going to fit primarily in one of two categories: kernel hacker or non-kernel hacker. As a primarily non-kernel hacker, I'd be much happier helping to bring up the ports collection under Darwin, and also helping with the GCC toolchain stuff to the extent of my time and abilities. This would give us a comfortable base of operations to bootstrap the real FreeBSD kernel and ideally set up a situation where a single executable could run on FreeBSD, Darwin, and (with the help of a loadable module), MacOS X. -Jake To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-ppc" in the body of the message From owner-freebsd-ppc Thu Jun 8 12:21:34 2000 Delivered-To: freebsd-ppc@freebsd.org Received: from haldjas.folklore.ee (Haldjas.folklore.ee [193.40.6.121]) by hub.freebsd.org (Postfix) with ESMTP id 67C2E37C12B for ; Thu, 8 Jun 2000 12:21:28 -0700 (PDT) (envelope-from narvi@haldjas.folklore.ee) Received: from localhost (narvi@localhost) by haldjas.folklore.ee (8.9.3/8.9.3) with SMTP id VAA98135; Thu, 8 Jun 2000 21:21:19 +0200 (EET) (envelope-from narvi@haldjas.folklore.ee) Date: Thu, 8 Jun 2000 21:21:19 +0200 (EET) From: Narvi To: Jake Hamby Cc: freebsd-ppc@FreeBSD.ORG Subject: Re: a more Darwin-centric approach? In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-ppc@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG On Thu, 8 Jun 2000, Jake Hamby wrote: > I have an idea: Since AFAIK nobody has done a whole lot of work yet in > getting the FreeBSD kernel up and running on PPC (please correct me if I'm > wrong!), why don't we focus first on getting the stuff from userland, > including the ports collection, brought over to Darwin/MacOSX? At WWDC, > Apple demoed XFree86 running on Darwin, so it's not a totally > uncomfortable environment for UNIX development. > I don't know about kernel work - I have been doing some work on userland, but from the perspective of Netbsd/ppc. Not terribly much, though 8-( [snip] > -Jake > To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-ppc" in the body of the message From owner-freebsd-ppc Thu Jun 8 16:23:57 2000 Delivered-To: freebsd-ppc@freebsd.org Received: from pobox.rwwa.com (pobox.rwwa.com [216.254.75.81]) by hub.freebsd.org (Postfix) with ESMTP id 8E54F37B959 for ; Thu, 8 Jun 2000 16:23:53 -0700 (PDT) (envelope-from witr@rwwa.com) Received: from rwwa.com (spooky.rwwa.com [192.124.97.13]) by pobox.rwwa.com (8.9.3/8.9.3) with ESMTP id TAA58251; Thu, 8 Jun 2000 19:30:28 -0400 (EDT) (envelope-from witr@rwwa.com) Message-Id: <200006082330.TAA58251@pobox.rwwa.com> X-Mailer: exmh version 2.1.1 10/15/1999 To: Narvi Cc: freebsd-ppc@FreeBSD.ORG Subject: Re: the abi In-Reply-To: Your message of "Thu, 08 Jun 2000 10:51:44 +0200." Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Thu, 08 Jun 2000 19:24:10 -0400 From: Robert Withrow Sender: owner-freebsd-ppc@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG narvi@haldjas.folklore.ee said: :- 3) the EABI has the advantage of an existing toolchain with no changes required at all. Also, all you Apple people, please try to keep the needs of embedded applications in mind. I believe that embedded applications will prove to be at least as important to FreeBSD as running on apple hardware. But then I'm biased. --------------------------------------------------------------------- Robert Withrow, R.W. Withrow Associates, Swampscott MA, witr@rwwa.COM To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-ppc" in the body of the message From owner-freebsd-ppc Sat Jun 10 16:56:49 2000 Delivered-To: freebsd-ppc@freebsd.org Received: from rm-rstar.sfu.ca (rm-rstar.sfu.ca [142.58.120.21]) by hub.freebsd.org (Postfix) with ESMTP id 2CE0737BE27 for ; Sat, 10 Jun 2000 16:56:47 -0700 (PDT) (envelope-from ballanty@sfu.ca) Received: from sfu.ca (rob3.ucs.sfu.ca [142.58.28.52]) by rm-rstar.sfu.ca (8.10.1/8.10.1/SFU-5.0H) with ESMTP id e5ANuja24247 for ; Sat, 10 Jun 2000 16:56:45 -0700 (PDT) Message-ID: <3942D5BC.9F72B75@sfu.ca> Date: Sat, 10 Jun 2000 16:56:44 -0700 From: Rob X-Mailer: Mozilla 4.73 (Macintosh; U; PPC) X-Accept-Language: en,pdf MIME-Version: 1.0 To: freebsd-ppc@freebsd.org Subject: ABIs Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-freebsd-ppc@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG Hello all, In case you haven't found it yet. At the Apple Developer site there is a document entitiled "Mac OS X Runtime Architecture" (http://developer.apple.com/techpubs/macosx/System/Documentation/ Developer/DeveloperTools/MacOSXRuntime/MacOSXRuntime.pdf) that describes, although very briefly, the Mach-O data types and calling conventions. I've not found anything that discusses object file formats or executable file formats. Someone mentioned that sysv4 ABI sucked. I've been reading through the Sunsoft version of it dated Sept'95 and I don't see anything that obviously sucks. I would like to see a little more description of what was right/wrong with each choice before we move to a decision about which ABI to adobt. What is NetBSD using? Can I steal their toolchain just to get underway. I would like to write a couple of test programs that interface to OF. OF3.0 understands the sysv4 elfs directly (I think) does anyone have a tool chain that builds those? Cheers, Rob To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-ppc" in the body of the message From owner-freebsd-ppc Sat Jun 10 17:34:22 2000 Delivered-To: freebsd-ppc@freebsd.org Received: from lor.watermarkgroup.com (lor.watermarkgroup.com [207.202.73.33]) by hub.freebsd.org (Postfix) with ESMTP id 46C8B37B84F for ; Sat, 10 Jun 2000 17:34:20 -0700 (PDT) (envelope-from luoqi@watermarkgroup.com) Received: (from luoqi@localhost) by lor.watermarkgroup.com (8.10.1/8.10.1) id e5B0Y2L14033; Sat, 10 Jun 2000 20:34:02 -0400 (EDT) Date: Sat, 10 Jun 2000 20:34:02 -0400 (EDT) From: Luoqi Chen Message-Id: <200006110034.e5B0Y2L14033@lor.watermarkgroup.com> To: ballanty@sfu.ca, freebsd-ppc@FreeBSD.ORG Subject: Re: ABIs Sender: owner-freebsd-ppc@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG > Someone mentioned that sysv4 ABI sucked. I've been reading > through the Sunsoft version of it dated Sept'95 and I don't see > anything that obviously sucks. I would like to see a little more > description of what was right/wrong with each choice before we > move to a decision about which ABI to adobt. > I think the svr4 abi is better than the aix abi (MacOS X uses a variant of aix abi, I believe). Under aix abi, a function pointer is not a normal pointer, it points to some descriptor structure, the advantage is all code is pic, but at the expense of speed; minimal stack space is 8 words as opposed to 4 under svr4 abi; quad integer has to be returned on stack... > > What is NetBSD using? Can I steal their toolchain just to > get underway. I would like to write a couple of test programs > that interface to OF. OF3.0 understands the sysv4 elfs directly > (I think) does anyone have a tool chain that builds those? > Both NetBSD and Linux adopted the svr4 abi. I think we should too. > Cheers, > > Rob > -lq To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-ppc" in the body of the message