HECnet October 2012

hecnet@lists.dfupdate.se

27 participants
335 discussions

embedded libraries for PDP-11, was Re: [HECnet] DECUS RSX C compiler

by jg＠jordi.guillaumes.name

Al 23/10/12 02:19, En/na Johnny Billquist ha escrit: Nice idea. Or the libc from DECUS C? Hmm yes. Will there be compiler compatibility issues there? (like with anything done in assembler, asm/C interface etc?) You could use the DECUS C libc, but you need to rewrite part of it. I bet the calling convention for gcc isn't the same as for DECUS C. Also, DECUS C does not expect to run on bare metal, so some basic I/O functionality needs to be written as well. But this is all pretty simple if you know your assembler. Well, gas (the GNU assembler) is not source-compatible with MACRO-11, so there would be some previous work before even trying to compile the stuff... For a starter, gas uses the '$' symbol to prefix immediate arguments, while MACRO-11 uses '#'. And this one is easy, since gas can work in a sort of "compatibility mode". The macro and symbol handling is also different, and some pseudoinstructions are different... Nothing that can't be worker around, but it would be a lot of work. The 2.11 BSD libc, on the other hand, is mostly written inC, and the parts in assembler are written for the BSD assembler, which is much more compatible with gas, and the calling standard is the same (R0 and R1 get clobbered, the return value comes in R0 for words and R0:R1 for longs, the called routine saves R3-R5, R5 is the frame pointer and the caller cleans the stack after the return). Oh, and the BSD license is liberal enough to not to worry about those pesky issues; (I have not checked the DECUS-C license though).

12 years, 1 month

DECnet on 4.3BSD/TOPS-20 and other questions

by b4＠gewt.net

On 23 Oct 2012, at 11:29, Dave McGuire <mcguire at neurotica.com> wrote: On 10/22/2012 04:06 PM, Cory Smelosky wrote: Was anyone ever crazy enough to try to implement any DECnet bits for 4.3BSD? Nope, not outside of Ultrix. I wonder if the sources ever escaped DEC. Wait, I have the source code to Ultrix 4.2a. I wonder if it's in there. Hmm, I wonder as well. Also, it's probably better I mention connecting here so all the crazy solutions get mentioned. Dynamic IP, located in Ohio, US. I have several spare UNIX systems to run a bridge on, and I would prefer to get several systems connected (what area should I use? I'll get back with node names once I figure out how to change them in RSTS/E and TOPS-20) I'm not very good with using emacs, maybe I should just stick with a more basic editor. Or find a better terminal emulator I don't have any real DEC terminals or any real DEC hardware yet. Where in Ohio? I'm in western PA, just a hair northeast of Pittsburgh. Near Columbus. I can hook you up with some hardware. I have, erm, a bit of a surplus at the moment. What do you have as surplus? -Dave -- Dave McGuire, AK4HZ New Kensington, PA

12 years, 1 month

DECnet on 4.3BSD/TOPS-20 and other questions

by mcguire＠neurotica.com

On 10/22/2012 04:06 PM, Cory Smelosky wrote: Was anyone ever crazy enough to try to implement any DECnet bits for 4.3BSD? Nope, not outside of Ultrix. I wonder if the sources ever escaped DEC. Wait, I have the source code to Ultrix 4.2a. I wonder if it's in there. Also, it's probably better I mention connecting here so all the crazy solutions get mentioned. Dynamic IP, located in Ohio, US. I have several spare UNIX systems to run a bridge on, and I would prefer to get several systems connected (what area should I use? I'll get back with node names once I figure out how to change them in RSTS/E and TOPS-20) I'm not very good with using emacs, maybe I should just stick with a more basic editor. Or find a better terminal emulator I don't have any real DEC terminals or any real DEC hardware yet. Where in Ohio? I'm in western PA, just a hair northeast of Pittsburgh. I can hook you up with some hardware. I have, erm, a bit of a surplus at the moment. -Dave -- Dave McGuire, AK4HZ New Kensington, PA

12 years, 1 month

Ultrix-11 tapes are available on Minnie

by form＠pdp-11.org.ru

Gregg Levine wrote: Hello! Does anyone have any experience with the collection of tapes that contain Ultrix-11 there? I have them, and, ah, I'm still working out the details behind everything. Tape images of ULTRIX-3.0 and 3.1 (build using TUHS files): http://pdp-11.org.ru/files/unix/ultrix30.zip http://pdp-11.org.ru/files/unix/ultrix31.zip Installation and generation log (under SimH): http://pdp-11.org.ru/~form/files/setup-Ultrix-3.1.html

12 years, 1 month

Ultrix-11 tapes are available on Minnie

by slgarwood＠charter.net

IIRCV there was a bunch of discussion on http://minnie.tuhs.org/mailman/listinfo/tuhs in 2004 about the tapes and how to boot them On Oct 22, 2012, at 6:56 PM, Gregg Levine <gregg.drwho8 at gmail.com> wrote: Hello! Does anyone have any experience with the collection of tapes that contain Ultrix-11 there? I have them, and, ah, I'm still working out the details behind everything. ----- Gregg C Levine gregg.drwho8 at gmail.com "This signature fought the Time Wars, time and again."

12 years, 1 month

embedded libraries for PDP-11, was Re: [HECnet] DECUS RSX C compiler

by bqt＠softjar.se

On 2012-10-23 00:07, Dave McGuire wrote: On 10/22/2012 05:35 PM, Paul_Koning at Dell.com wrote: The main barrier to a bare metal binary, or for that matter a not so bare metal one, is the supporting libraries. I did a *really* crude hack job on newlib to give me just enough bare metal support that I could run the GCC test suite under SIMH. But that certainly isn't enough for real bare metal. A pdp11 port of newlib would be a nice project. I tried... But it is too big. I think I got a fatty library because I selected the floating point support... I tried to use sprintf() and it didn't even fit into the PDP address space :(. The printf() family is GIGANTIC. In newlib there are non-FP-enabled versions of those functions; iprintf(), siprintf(), etc. They are much, much smaller than their floating-point counterparts. I suggest you try again with that and see where things land. I'm using them on ARM7 with great results. Granted that's got a much larger address space, but my resultant binaries and stack utilization are still pretty small. The 2.11BSD lib looks leaner, and I am using it in my pet OS project (I pick the pieces I need "on the run", I have not done a complete port... yet). If the above suggestion doesn't get you anywhere, perhaps avrlibc could be hacked into a PDP-11 library. All of the hardware support will have to be ripped out of course, but there are nice tight implementations of generic library functions in there. Nice idea. Or the libc from DECUS C? Hmm yes. Will there be compiler compatibility issues there? (like with anything done in assembler, asm/C interface etc?) You could use the DECUS C libc, but you need to rewrite part of it. I bet the calling convention for gcc isn't the same as for DECUS C. Also, DECUS C does not expect to run on bare metal, so some basic I/O functionality needs to be written as well. But this is all pretty simple if you know your assembler. Johnny -- Johnny Billquist || "I'm on a bus || on a psychedelic trip email: bqt at softjar.se || Reading murder books pdp is alive! || tryin' to stay hip" - B. Idol

12 years, 1 month

Ultrix-11 tapes are available on Minnie

by gregg.drwho8＠gmail.com

Hello! Does anyone have any experience with the collection of tapes that contain Ultrix-11 there? I have them, and, ah, I'm still working out the details behind everything. ----- Gregg C Levine gregg.drwho8 at gmail.com "This signature fought the Time Wars, time and again."

12 years, 1 month

embedded libraries for PDP-11, was Re: [HECnet] DECUS RSX C compiler

by mcguire＠neurotica.com

On 10/22/2012 05:35 PM, Paul_Koning at Dell.com wrote: The main barrier to a bare metal binary, or for that matter a not so bare metal one, is the supporting libraries. I did a *really* crude hack job on newlib to give me just enough bare metal support that I could run the GCC test suite under SIMH. But that certainly isn't enough for real bare metal. A pdp11 port of newlib would be a nice project. I tried... But it is too big. I think I got a fatty library because I selected the floating point support... I tried to use sprintf() and it didn't even fit into the PDP address space :(. The printf() family is GIGANTIC. In newlib there are non-FP-enabled versions of those functions; iprintf(), siprintf(), etc. They are much, much smaller than their floating-point counterparts. I suggest you try again with that and see where things land. I'm using them on ARM7 with great results. Granted that's got a much larger address space, but my resultant binaries and stack utilization are still pretty small. The 2.11BSD lib looks leaner, and I am using it in my pet OS project (I pick the pieces I need "on the run", I have not done a complete port... yet). If the above suggestion doesn't get you anywhere, perhaps avrlibc could be hacked into a PDP-11 library. All of the hardware support will have to be ripped out of course, but there are nice tight implementations of generic library functions in there. Nice idea. Or the libc from DECUS C? Hmm yes. Will there be compiler compatibility issues there? (like with anything done in assembler, asm/C interface etc?) -Dave -- Dave McGuire, AK4HZ New Kensington, PA

12 years, 1 month

embedded libraries for PDP-11, was Re: [HECnet] DECUS RSX C compiler

by Paul_Koning＠Dell.com

On Oct 22, 2012, at 3:32 PM, Dave McGuire wrote: On 10/22/2012 02:56 PM, Jordi Guillaumes i Pons wrote: The main barrier to a bare metal binary, or for that matter a not so bare metal one, is the supporting libraries. I did a *really* crude hack job on newlib to give me just enough bare metal support that I could run the GCC test suite under SIMH. But that certainly isn't enough for real bare metal. A pdp11 port of newlib would be a nice project. I tried... But it is too big. I think I got a fatty library because I selected the floating point support... I tried to use sprintf() and it didn't even fit into the PDP address space :(. The printf() family is GIGANTIC. In newlib there are non-FP-enabled versions of those functions; iprintf(), siprintf(), etc. They are much, much smaller than their floating-point counterparts. I suggest you try again with that and see where things land. I'm using them on ARM7 with great results. Granted that's got a much larger address space, but my resultant binaries and stack utilization are still pretty small. The 2.11BSD lib looks leaner, and I am using it in my pet OS project (I pick the pieces I need "on the run", I have not done a complete port... yet). If the above suggestion doesn't get you anywhere, perhaps avrlibc could be hacked into a PDP-11 library. All of the hardware support will have to be ripped out of course, but there are nice tight implementations of generic library functions in there. Nice idea. Or the libc from DECUS C? paul

12 years, 1 month

DECnet on 4.3BSD/TOPS-20 and other questions

by b4＠gewt.net

On Oct 22, 2012, at 4:51 PM, Jason Stevens <neozeed at gmail.com> wrote: On Mon, Oct 22, 2012 at 4:06 PM, Cory Smelosky <b4 at gewt.net> wrote: Was anyone ever crazy enough to try to implement any DECnet bits for 4.3BSD? I doubt it was very highly demanded, thus the "crazy" bit. I'd consider trying to implement it myself, but I don't know C, BSD internals, or in-depth DECnet info and the craziest I've done is gotten a crippled Perl 5 built on Quasijarus. (while i'm on that topic, anyone ever manage to build autoconf/automake on 4.3BSD?) My eventual end goal is to build irssi on 4.3BSD and ignoring the fact it's pretty much impossible. I've just built the old ircII stuff... and that was involved to say the least. the University of Wisconsin version of 4.3 BSD seemed more usable as its got some SUN magic in there like NFS and various other cleanups ... I haven't looked at irssi as I'd imagine it'd be ... involved. Very involved. I need to get a newer shell to build pkg-config or pkgconf (pkg-config compatible thing, probably easier to use it) so I can get glib built. But I doubt pkg-config will ever build., so the plan stops there unfortunately It seems GNU configure scripts are..picky about shells. They hate pdksh, tcsh, proper ksh, and it seems pkgconf needs a modern bash. I've tried to build several shells as of yet no luck. Seems sys/time.h needs a bit of work for the history functions of bash3 to build. Maybe I can backport that from FreeBSD? Probably too many dependencies for that. I managed to get locale.h and sys/cdefs.h ported back to build with older GCC, i don't trust it but it built :-). But it would probably be a bit easier on a different 4.3BSD flavour.

12 years, 1 month

← Newer
1
...
24
25
26
27
28
29
30
...
34
Older →

Jump to page:

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

HECnet October 2012