Linux Port
Pages: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Tristan M. (2946) 1039 posts |
Is the “Networking” section of the source nonfunctional still or am I just doing something wrong? e:Sorry about the mistake. Anyway what I mean, now I have a keyboard, is whether there is any translation layer for networking, or just the networking modules with nothing to talk to. |
Timothy Baldwin (184) 242 posts |
The stub Internet module translates Socket_ SWI calls to the equivalent Linux system calls, it currently currently does not implement the FIOSLEEPTW and FIOASYNC ioctls (Taskwindow sleep and Internet events), SO_DEBUG, SO_TYPE, SO_RCVTIMEO, and various IP socket options. Netsurf works, and have just implemented FIONREAD which is sufficient to make WebJames work. What are you trying? |
Tristan M. (2946) 1039 posts |
*ping some.lan.ip.address Netsurf says no DNS servers are configured. In “Network” only Econet isn’t greyed out. Packman can’t resolve anything. That’s about as far as I got. |
Timothy Baldwin (184) 242 posts |
Tristan, you need to either run a DNS server on 127.0.0.1 (localhost) or configure Resolver with the IP address of your DNS server using the command line interface:
I have tried altered the validation to accept opening RAW sockets, but ping crashes. It would also require extra permissions.
In configure? How did you manage to open it? |
Tristan M. (2946) 1039 posts |
Posting from RISC OS Linux running in Armbian mainline on a Pinebook. That did the trick. Excellent! Sorry about the Configure thing. I was thinking of “Configuration” I’m back in business now, so to speak. |
Tristan M. (2946) 1039 posts |
I saw mention of a patched !GCC somewhere I think. Is that available? |
Timothy Baldwin (184) 242 posts |
I’ve done a bit of work on GCCSDK and have uploaded a patch and modified SharedLibs to: http://timbaldwin.fastmail.co.uk/GCCSDK_RO_Linux_1/ It needs work doing on the signal handling etc, at present it will crash if you press escape etc. |
Tristan M. (2946) 1039 posts |
I haven’t tried gcc yet. |
Tristan M. (2946) 1039 posts |
Does the desktop still run if the SDL or OpenGL renderer isn’t executed? Just a thought anyway. My main use case is so I can use something a little more powerful than a headless Pi Zero running RO. |
Tristan M. (2946) 1039 posts |
I had a post disappear into the great unknown earlier apparently. That’s unfortunate. Short version. I got networking and VNC set up and altered a copy of run_RISC_OS that doesn’t run the front end. It worked fine. |
Alex Farlie (3076) 16 posts |
Was this a new version of https://sourceforge.net/projects/riscose/ which I recently found again? |
Richard Walker (2090) 431 posts |
No. This port allows RISC OS itself to run on Linux. Therefore you can run RISC OS apps, because RISC OS is running. Riscose aims to implement RISC OS SWIs with a load of C code and includes an ARM emulator. Therefore, you can run RISC OS applications, because there is a RISC OS compatible runtime. I don’t think riscose got very far. There would be many Acorn and third party SWIs to implement. It is a bit like WINE, which tries to achieve the same for Windows apps. I think someone else started a similar project to riscose, but I forget its name. |
George T. Greenfield (154) 748 posts |
Are there any tests or benchmarks comparing app speed running natively to app speed under the port, on identical hardware? AIUI, this Linux port has relatively low overhead costs, but utilises hardware more fully than current native OS versions, so should in practice be faster. |
David Boddie (1934) 222 posts |
Are you thinking of this? ROLF – An alternative GUI for Linux |
Richard Walker (2090) 431 posts |
That’s the one! I remember reading that blog post about running AWViewer some time ago. I have not tried the Linux port, as I don’t have a machine set up running Linux. I guess I could create another SD card for my Pi, but RISC OS already runs on there… |
Tristan M. (2946) 1039 posts |
What’s the correct incantation for running run_RISC_OS with the OpenGL fronted? |
Timothy Baldwin (184) 242 posts |
|
riscos@weatherpost.org (5379) 3 posts |
My RiscPc is getting very long in the tooth like me and I wondered if there’s any progress towards a “snap” of this port that I could try on my Ubuntu 19.04? I dont have any software compilation tools or mind paying for the port if I can run some of my software like Artworks and Impression. |
Chris Hall (132) 3554 posts |
This may sound silly but why not get a Raspberry Pi model 2 or 3 (cost £30), a USB pen drive to transfer data, the free Aemulor and then you know that it will all just work. If your copy of Impression predates the dongle-free version then £60 to !Store will get you Impression-X (as well as 5.13, the final, dongle-free version). |
riscos@weatherpost.org (5379) 3 posts |
Not silly at all. Ive RPis (x3) but wanted to run on one box, USB external storage, keyboard and mouse.. Actually that’s wrong as I have a portable as well and they all use a NAS as well which can take a pen drive with the RISCOS software on it. for storage. Hence a snap. |
David Feugey (2125) 2709 posts |
Stupid question, but does someone as a working version for the latest Raspbian used by the Pi4? |
Timothy Baldwin (184) 242 posts |
Why does FPEmulator as provided by ROOL fail to assemble on IXFS (over Tmpfs or Btrfs) but succeed on Filecore, TBAFS, DOSFS (with only Sources.HWSupport.FPASC being a DOSFS image) and Rpcemu HostFS? It fails by trying to include a directory as a assembly source file:
The workaround is to change “GET RISCOS” into “GET Hdr:RISCOS”, but what is the actual cause? IXFS behaves similar to Filecore:
Arranging for H%=OPENUP"X" to return an error does not fix FPEmulator builds. |
Chris Gransden (337) 1207 posts |
The pre-built RISC_OS from latest source always aborts with the following. Using a working rom to build the latest version from source also aborts. The last working version I have is built from source on 31/12/18. 14 of 14 interrupts counted. Starting kernel... HAL initialised IICInit InitCMOSCache entry InitCMOSCache failed IMB_Full done InitDynamicAreas InitVectors InitIRQ1 IMB_Full VduInit ExecuteInit Machine ID duff,zero substituted KeyInit OscliInit Enabling IRQs IRQs on Debug terminal on HAL_InitDevices Error: UndefinedInstruction:Undefined instruction at &2201ABB0 (Error Number &80000000) * |
Chris Gransden (337) 1207 posts |
Current versions don’t seem to work but an older one from December runs fine on the RPi 4 using Raspbian. |
Timothy Baldwin (184) 242 posts |
I’ve tried it with a Amazon Web Services A1, and it also fails. But it worked on my Chromebook and Raspberry Pi 1. I had made a mistake in calculating the address of RISCOS_AddDevice (the offsets in table of kernel entries are relative to the base of the table, not the base of the kernel) so the HAL was jumping into the middle of a literal pool and faulting on “ Then I found that in a test case I had misaligned a label so it was calling 2 bytes before the the code. I have uploaded a fixed version. |
Pages: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20