OpenEuphoria

1. ARM and Position Independent Code

• Posted by mattlewis (admin) Mar 04, 2013
• 2696 views

Can anyone confirm or deny whether ARM requires Position Independent Code (PIC) for building shared libraries? x86-64 on Linux does, but none of the other platforms that we support do. This required an additional runtime library for translating x86-64 shared libraries.

I'd like to get rid of this for platforms that don't need it, but I don't have access to any ARM stuff.

Matt

3. Re: ARM and Position Independent Code

• Posted by rkdavis Mar 04, 2013
• 2629 views

actually this is really going to be a helpful reply :)

ust asked a guy who has written an ARM shared lib and he says he doesn't think so :)

so the defintive answer is

yes & no

5. Re: ARM and Position Independent Code

• Posted by jimcbrown (admin) Mar 04, 2013
• 2605 views

I wouldn't be suprised if, for example, the answer was different on ARMv5 and ARMv6.

7. Re: ARM and Position Independent Code

• Posted by EUWX Mar 04, 2013
• 2569 views

.so files by their very nature have Position Independent Code. You might consider using .so use temporarily during the development. Some languages like QB64 for windows platform use several .dll files and never a final .exe till the user writes his own code and compiles it.

9. Re: ARM and Position Independent Code

• Posted by Jerome Mar 04, 2013
• 2576 views

My original plan was to recompile libraries I often use on my ARM devices but without -fPIC. I did this for the Pi (Armv6) using wiringPi and on my Nokia (Armv7) with wxEuphoria. Both shared libraries still work as expected. After reading this http://stackoverflow.com/questions/5311515/gcc-fpic-option, I might be oversimplifying things though. I guess there could still be instances where not using -fPIC might cause problems. Is there something in particular I can test?

Thanks,
Ira

11. Re: ARM and Position Independent Code

• Posted by jimcbrown (admin) Mar 05, 2013
• 2472 views

Why? To eliminate the need for the extra shared object?

13. Re: ARM and Position Independent Code

• Posted by jimcbrown (admin) Mar 05, 2013
• 2503 views

Thinking it over, we might want to keep it as an optional feature. We don't really target embedded systems yet, but using fPIC on the right platform can save us both memory and speed. (I've read that, unlike x86, fPIC actually results in faster code on ARM.)

The benefits of fPIC show up best if we have thousands of processes loading and unloading the same several thousand shared objects. These scenarios are probably rare, but (as long as its optional), why not?

15. Re: ARM and Position Independent Code

• Posted by mattlewis (admin) Mar 05, 2013
• 2456 views

Two ARMs and a Xeon walk into a bar...

Matt

17. Re: ARM and Position Independent Code

• Posted by jimcbrown (admin) Mar 05, 2013
• 2424 views

FreeDOS only works on x86 cpus, so that's simply not possible until someone does a port or makes a clone. (It's possible to emulate it, of course, but that's an obvious non-starter.)

(Even today, most kernels do not approach the 100-megabyte range...)

19. Re: ARM and Position Independent Code

• Posted by EUWX Mar 05, 2013
• 2419 views

Once upon a time there was x86 processor, needing help for video. So developers (e.g. Nokia) created a separate video processor taking the UNREASONABLE load off the x86 processor. Then one day some smart guys discovered that the video processor was actually more powerful than the x86 processor and its mate the Arithmatic processor unit, and they actually went about developing software to work on the video processor - Have you heard of CUDA?
Then came the phone and tablet revolution and the smart guys modified the essentials of a video processor to create the enormously powerful ARM processor, extremely strong on video, all AV, etc. And grandfather Google gave Adroid OS to it too.
Now people are again discovering the power of the Arm processors and are trying to get a proper Linux on it, and Granddad Linus has given Kernel version 3.7 so there would be less problem getting Linux onto these ENORMOUSLY POWERFUL ARM PROCESSORS, shaking the very foundations of Microsoft.

Where does the useless C-64 and olde-dos come into the picture? The users of Arm processor on the various phones and tablets have never heard of the Olde dos or the new dos or the modern Dos and shrink away from the console mode of Linux. They only want graphic interface. so let us get this useless talk out of this thread and put it into a cage called "commodore 64" and BURY it, and then proceed with the Linux version of Euphoria as powerful or more powerful than the x86 version of Euphoria with the enormous graphics power harnessed to a great user experience.

21. Re: ARM and Position Independent Code

• Posted by jaygade Mar 05, 2013
• 2417 views

Umm, you know that ARM has been around since the 1980s, right?

Although I wouldn't compare it to a C64 either, since it's a 32-bit processor.

While I don't have one, I've never heard of people having problems running Linux on ARM or Raspberry Pi. Of course, ARM itself encompasses an entire series of CPUs and MCUs in a large number of configurations.

23. Re: ARM and Position Independent Code

• Posted by useless_ Mar 05, 2013
• 2393 views

The chip on the Pi is a graphics processor and an ARM on one chip.

Sounds very anti-windoze to me.

useless

25. Re: ARM and Position Independent Code

• Posted by jaygade Mar 05, 2013
• 2353 views
• Last edited Mar 06, 2013

Well, that means "improved" to me. I mean, I guess you could call a Ferrari a "vastly improved" Model T if you wanted to. Heck, you could even call a Prius or a Smart Car a "vastly improved" Model T and still be just as relevant.

Yeah, I'm aware that some people have had some success in getting Euphoria to run on the Raspberry Pi and other ARM systems, but that there are still some bugs to work out.

What does that have to do with my statement about running Linux on these devices?

27. Re: ARM and Position Independent Code

• Posted by jaygade Mar 06, 2013
• 2360 views

Okay. As usual, I have no idea what you mean here.

Umm, actually I think that it kinda is. Which is why it has been around for forty plus years.

There are certain "recommended" versions of *nix to run on almost *any* device. Not just the Pi.

Sure, if you want. I'm very good at nitpicking and slinging urls and quoting. Calling this "Euphorum" is just stupid, however.

29. Re: ARM and Position Independent Code

• Posted by jimcbrown (admin) Mar 06, 2013
• 2362 views

I'd like to rephrase that as follows:

Euphoria, by and large, runs on the Pi, although there have been a number of loose ends to clean up, which people have spent a lot of time on.

31. Re: ARM and Position Independent Code

• Posted by useless_ Mar 06, 2013
• 2288 views

That IS what i said, no?

useless

33. Re: ARM and Position Independent Code

• Posted by jimcbrown (admin) Mar 06, 2013
• 2238 views

I believe so, yes. I'm worried that some simpleton might misread your original wording as "Eu isn't running on the the Pi", however, and thought the clarification might help.

35. Re: ARM and Position Independent Code

• Posted by mattlewis (admin) Mar 06, 2013
• 2309 views

I don't see that happening. Most of my time is spent on Linux, but I still do Windows stuff. And I know that most people run euphoria on Windows.

There's always someone around who will have a beef with something that anyone says. Especially if you lead with attacks on something else. I don't think anyone is against additional ports, but that's a lot harder if the target doesn't include an OS that we already support or that is at least very similar to one that we do.

Matt

37. Re: ARM and Position Independent Code

• Posted by useless_ Mar 06, 2013
• 2287 views

It wasn't an attack on the Pi to say the ARM was a vastly improved 6502. [edit]Neither is it a slam to say the Broadcom video chip with the ARM in it is a 700Mhz C64, as it incorporates almost all the features of the C64 in the one chip. If i wanted to slam the Pi, i could make a list.[/edit]

The ARM1 was developed with the intention to use the 6502 design philosopy, and a visit to Western Design Center, the IP holder, and is still home to Bill Mensch, one of the original designers and promoters of the 6502, and someone i have emailed with a bit. The ARM team (before there was an ARM) had been using 6502 in their systems for years, and had considered a 16bit cpu, found all had issues, etc etc etc, and after visiting WDC went home and designed the ARM1 on a 6502 system. And like i have suggested for Eu before, they coded the BASIC interpreter for the ARM1 and a kernal as a package, much like the C64 (it has bios/kernal and Basic) and later the IBM PC, was done. But the ARM was most definitely as much an outgrowth and upgrade of the 6502 as the 6502 was a clone of the 6800, but without the lawsuits. The ARM1 was designed by 6502 KISS principles, and was simply wider and had more registers. I consider the low end AVR to be clones of the early ARMs. This stuff is online, people.

useless

39. Re: ARM and Position Independent Code

• Posted by useless_ Mar 06, 2013
• 2276 views

Ah. Ok. Let me clarify, maybe. . .

I was going back to the development of the Pi, when it was two chips, an ARM (any ARM, pick an ARM, maybe a new one?) and a video chip. Back then one had full access to the ARM and to the video processing, this isn't so with the current Pi (have you read the data sheet on that chip?). It was possible to add more external memory, after a fashion, now it isn't. To me, it's like looking at the existing Pi (vs the original Pi) and the C64 (vs the VIC20) and asking myself why they shot themselves in the foot while doing the upgrade. Anyhow, all that work was thrown out. Pi v2 was initially programmed with Fedora, then Debian was backtracked from ARM7 to ARM6 to make Raspbian, and other *nixes were also modified to various results. There's likely an army of people working alone or small groups to make the Pi useable on the version of *nix they prefer, that won't be able to run this, that, or the other thing from one Pi to the next. AFAIK, only Raspbian uses the Broadcom chip's floating point math hardware.

While designing software to a specific piece of hardware is easier (in my experience) than designing software for multiple OSs on various hardware (like Euphoria is), there was still 1000's of man-hours put into just backtracking Debian to be Raspbian, and that alone is more than all the effort put into the ARM1, near as i can tell, and that is not including the work to bring about Debian or Fedora in the first place! It's just an astonishing amount of work, when compared to the same goal of making an entry level computer 30 years ago.

It's highly unlikely Broadcom will put a ARM8 or 9 into the same package the Pi uses now, and they are so closely coupled to suit Broadcom's needs that even if Pi gets a free license to make a new chip with new ARM in it, Pi will haveto shoulder the costs ($millions) of layout, debugging, and foundry. I can't see that happening. Therefore any improvements in whatever *nix is used, those will need to be backtracked, even degraded, spending time undoing upgrades in future releases, ironically to keep the OS in the Pi up to date.

Considering how Pi was developed from the start, i'd think the original 2-chip Pi was a better hardware design, and more future-proofed, and easier to code for, than the existing 1-chip design is. The VIC-20 to C64 trip was also about chip integration and cost, even while adding a few bells and whistles, and it also caused issues. After a couple more ARM versions, or a couple more OS versions, it's real likely the Pi will be an orphan. At least with the 2-chip Pi, and the C64, you could plug in different cpus. Much like there were clones muddying the waters in the early 1980's, there's Pi clones now too, it's deja vu all over again. And i am saddened to see it happening, it is like a cruel joke.

If new reports are to be believed, it is very much easier now, a year after the Pi was introduced, to boot and run the Pi using Raspbian (not necessarily so with other OSs), but there's still no OS or apps included, you need another computer and ram chip to start up the Pi.

useless

41. Re: ARM and Position Independent Code

• Posted by useless_ Mar 06, 2013
• 2288 views
• Last edited Mar 07, 2013

You are saying the ATmega644, all by it's lonesome, generated the video and ran an OS with apps too? Didn't that perfboard have at least 2 other chips on it? I am sure it did, ask Eben. I am picturing the small perfbd with three DIP chips side by side across it.

I'd like to point out that older data online is being covered over (updated, replaced, etc as time marches by), and even
http://en.wikipedia.org/wiki/FileFile:Raspberry_Pi_board_at_TransferSummit_2011_cropped.jpg
is saying that pic is of a Model A, when the Model A had no ethernet, and that plainly does. All i am finding online in 2-chips now is the first and the current version with ethernet, which by then had the RAM-GPU chip. But pre-release, i remember reading everyone there was quite pleased they found the integrated Broadcom chip, and could reduce chip count.

Initial sales commenced 29 February 2012[50] at 06:00 UTC;. At the same 
time, it was announced that the Model A, originally to have had 128 MB of RAM 

I don't consider going from 128MB to 512MB minor, given they want to run a full *nix on it, with apps. Even 512MB limit is annoying, considering that the virtual address space is 4GB using the MMU in the Broadcom chip. Sadly, the 32bit ARM chip has what.. only 27 address lines? Wasn't there a line in the datasheet about executeable code needing to be in the bottom 67MB (2^26) of ram?

There's 30,000 Debian versions? This from Wikipedia:

The ARM11 is based on version 6 of the ARM architecture (ARMv6), which due 
to its age is no longer supported by several popular versions of Linux, 
including Ubuntu which dropped support for processors below ARMv7 in 2009.[91] 

I am not counting RiscOS, mostly because it was developed in the 1980's specifically for the ARM.

That would be nice. Then they can also fix that audio pin swap issue. I heard someone recently got Android installed painlessly. I am certainly not an expert on the Pi, i wish i could use it, but i'm not into *nix, C, C+, or Python and have other things to do than climb another learning curve. Ditto all the other ARM-based SBC out there. I wasn't going to say anything about the Pi other than my first post.

useless

43. Re: ARM and Position Independent Code

• Posted by rkdavis Mar 06, 2013
• 2274 views

ye gods and little fishes talk diliberately twisting facts and selective cutting and pasting.

so shall we do this in a civilized manner or do i just call you a troll and have done with it?

ok lets start

To quote Eben, the designer http://www.raspberrypi.org/archives/264

These boards use an Atmel ATmega644 microcontroller clocked at 22.1MHz, and a 512K SRAM for data and framebuffer storage. 19 of the Atmel’s 32 GPIO lines are used to drive the SRAM address bus. To generate a 320×240 component video signal, the Atmel rapidly increments the address, and the data lines are fed via 74HC-series buffers to a trio of simple summing-point DACs; during horizontal and vertical blanking, it is free to perform other operations.

Nothing about separate video chip there is there?

Well for a start that is a beta board and not a Model A. You do know that the Model A wasn't released until a month ago don't you? so no photo from 2011 could possibly be a Model A no matter what wikipedia or anyone else claims.

do you think you could paste the whole quote?

I didn't say versions I said packages

see: http://www.raspbian.org/FrontPage

oh dear this is getting boring so for my final trick

no you didn't you got shown to be a troll of the highest order. it's been fun and i could go and refute every claim you've made but it's beginning to get boring now as it's too easy to show you up so i'll leave it at that and let others decide if you are worth reading at least when ARM or the Raspberry Pi is being discussed.