1. ARM Updates
- Posted by mattlewis (admin) Apr 02, 2013
- 2875 views
Hey, euphoria for ARM early adopters! I've made some changes and have pushed them up into the default branch. I recently acquired a raspberry pi, and I've been using the raspbian distro, which uses the hard float (i.e., hardware floating point) capabilities. I've been using the raspbian specific cross compiler here: git://github.com/raspberrypi/tools.git (specifically, the stuff in arm-bcm2708/gcc-linaro-arm-linux-gnueabihf-raspbian).
I've updated it so that stuff like t_callc.e now passes. The hardfloat calling convention is similar to the AMD64 convention. Anyways, I'd be really interested to hear from someone running a soft float system, if there's anyone out there.
It should build correctly, assuming the compiler defaults are correct for your system, or you set the flags correctly to tell it to use softfloat. Basically, some online research showed that when the predefined macro __ARM_PCS_VFP equals 1, we're compiling for hardfloat systems.
Matt
2. Re: ARM Updates
- Posted by Jerome Apr 02, 2013
- 2817 views
I've updated it so that stuff like t_callc.e now passes. The hardfloat calling convention is similar to the AMD64 convention. Anyways, I'd be really interested to hear from someone running a soft float system, if there's anyone out there.
Thanks for the updates! I just cross-compiled the new changes for my soft-float system and things are definately better. I did have an issue with one of my older cross-compilers (based on gcc 4.2.1). It looks like lines 350 and 351 are duplicated definitions for cddfuncF in be_callc.c. Also, I had an undefined referece to 'WEXITSTATUS' which was fixed with '#include <sys/wait.h>' in be_runtime.c.
But now t_callc.e only has one failed test which looks like a simple rounding issue!
Nokia-N900:/home/user/Code/Euphoria/tests# eui t_callc.e -all | grep failed failed: Can call and things are passed correctly for ten argument functions, expected: 1.12589993095027e+15 but got: 1.12589991858202e+15 96 tests run, 95 passed, 1 failed, 99% success
Thanks,
Ira
3. Re: ARM Updates
- Posted by EUWX Apr 03, 2013
- 2764 views
From "Build Error" thread.
I don't fully understand the ARM ecosystem, but there seem to be a lot of different architectures out there, where something built for one chip doesn't work on another, similar to something like i686 vs i386. Then there are the soft vs hard float issues, which are part hardware, part software.
It's probably not as bad as my (poor, incomplete) understanding is, but doing the translation on a more maturely supported (by/for euphoria) platform seems less risky. Of course, it doesn't hurt to try, and if it doesn't work, I don't think it will be subtle, but quick and catastrophic.
Another issue I had was that I couldn't use the cross compiler from my distro, because it linked to GLIBC 2.15, but raspbian uses GLIBC 2.13. I tried several things to hack around this, but couldn't get it to work. The provided cross compilers, OTOH, worked right away with no extra hacks or worrying about setting the right options to target the specific chip, since that was already baked into everything.
For all of these reasons, I suspect that we probably won't be able to offer a single binary ARM distribution of euphoria that we can expect to work everywhere.
Matt
As I said before, Rasbian is still lagging behind officially.
This guy is worth following, but he does not publish his methodology.
http://www.element14.com/community/blogs/mirandasoft/2013/02/19/raspberry-pi-now-running-linux-kernel-380-freshly-compiled-in-the-philippines
You are on the right track with Linaro. They are still going for separate builds, because of the Rapberry Pi guys, but eventually there will be a merged and more generic distro. Eventually you will be able to offer a single binary ARM distro.
4. Re: ARM Updates
- Posted by mattlewis (admin) Apr 05, 2013
- 2653 views
I've just pushed up some more updates. All of the tests pass interpreted here. I haven't run the translated tests...maybe overnight, if I remember to kick it off.
Matt
5. Re: ARM Updates
- Posted by coconut Apr 06, 2013
- 2673 views
I've just pushed up some more updates. All of the tests pass interpreted here. I haven't run the translated tests...maybe overnight, if I remember to kick it off.
Matt
Hi Matt,
Did you compile binaries for your RPi? My efikaMx smartbook is repaired. If you have binaries for your RPi I would try them on my Efika. You could place a link on the wiki page about euphoria and Raspberry pi. There is 2 links on that page pointing to binaries. The soft float version doesn't work at all. The hard float version fail sanity.ex at 4%.
If I had a procedure to cross-compile on my desktop running ubuntu 12.04LTS I would try to compile the binaries myself. Out of question to compile them on Efika, even if it succeed, it would take hours.
Jacques
6. Re: ARM Updates
- Posted by mattlewis (admin) Apr 07, 2013
- 2555 views
Did you compile binaries for your RPi? My efikaMx smartbook is repaired. If you have binaries for your RPi I would try them on my Efika. You could place a link on the wiki page about euphoria and Raspberry pi. There is 2 links on that page pointing to binaries. The soft float version doesn't work at all. The hard float version fail sanity.ex at 4%.
I've uploaded my latest:
If I had a procedure to cross-compile on my desktop running ubuntu 12.04LTS I would try to compile the binaries myself. Out of question to compile them on Efika, even if it succeed, it would take hours.
Yes, before launch, I'm going to have to update some of the packaging to allow use of cross compiled binaries. I'm currently running kubuntu 12.04, so I'm sure you could use the same stuff I'm using. Here's what I did:
- clone this repository: git://github.com/raspberrypi/tools.git
- add to your PATH: path/to/git/repo/arm-bcm2708/gcc-linaro-arm-linux-gnueabihf-raspbian/bin
- configure with --cc-prefix arm-linux-gnueabihf-
Additionally, I've used sshfs to mount my development environment on my raspberry pi, so I can edit and build as normal, and then test via an ssh session in my pi.
Matt
7. Re: ARM Updates
- Posted by coconut Apr 07, 2013
- 2587 views
Did you compile binaries for your RPi? My efikaMx smartbook is repaired. If you have binaries for your RPi I would try them on my Efika. You could place a link on the wiki page about euphoria and Raspberry pi. There is 2 links on that page pointing to binaries. The soft float version doesn't work at all. The hard float version fail sanity.ex at 4%.
I've uploaded my latest:
If I had a procedure to cross-compile on my desktop running ubuntu 12.04LTS I would try to compile the binaries myself. Out of question to compile them on Efika, even if it succeed, it would take hours.
Yes, before launch, I'm going to have to update some of the packaging to allow use of cross compiled binaries. I'm currently running kubuntu 12.04, so I'm sure you could use the same stuff I'm using. Here's what I did:
- clone this repository: git://github.com/raspberrypi/tools.git
- add to your PATH: path/to/git/repo/arm-bcm2708/gcc-linaro-arm-linux-gnueabihf-raspbian/bin
- configure with --cc-prefix arm-linux-gnueabihf-
Additionally, I've used sshfs to mount my development environment on my raspberry pi, so I can edit and build as normal, and then test via an ssh session in my pi.
Matt
Matt, No luck with your build neither with mine on EfikaMX (I used linaro building tools). When trying to run eui the loader complain the file doesn't exist although "file eui" report
jacques@efikamx:~/Téléchargements/euphoria-4.1.0-Linux-ARM/bin$ file eui eui: ELF 32-bit LSB executable, ARM, version 1 (SYSV), dynamically linked (uses shared libs), for GNU/Linux 2.6.26, not stripped
Don't know if it is difference between freeScale imx515 and BMC2708 or something related to OS. Is this a problem with different version of kernel? or libc? I don't know much about it.
Jacques
8. Re: ARM Updates
- Posted by jimcbrown (admin) Apr 07, 2013
- 2565 views
Did you compile binaries for your RPi? My efikaMx smartbook is repaired. If you have binaries for your RPi I would try them on my Efika. You could place a link on the wiki page about euphoria and Raspberry pi. There is 2 links on that page pointing to binaries. The soft float version doesn't work at all. The hard float version fail sanity.ex at 4%.
I've uploaded my latest:
If I had a procedure to cross-compile on my desktop running ubuntu 12.04LTS I would try to compile the binaries myself. Out of question to compile them on Efika, even if it succeed, it would take hours.
Yes, before launch, I'm going to have to update some of the packaging to allow use of cross compiled binaries. I'm currently running kubuntu 12.04, so I'm sure you could use the same stuff I'm using. Here's what I did:
- clone this repository: git://github.com/raspberrypi/tools.git
- add to your PATH: path/to/git/repo/arm-bcm2708/gcc-linaro-arm-linux-gnueabihf-raspbian/bin
- configure with --cc-prefix arm-linux-gnueabihf-
Additionally, I've used sshfs to mount my development environment on my raspberry pi, so I can edit and build as normal, and then test via an ssh session in my pi.
Matt
Matt, No luck with your build neither with mine on EfikaMX (I used linaro building tools). When trying to run eui the loader complain the file doesn't exist although "file eui" report
jacques@efikamx:~/Téléchargements/euphoria-4.1.0-Linux-ARM/bin$ file eui eui: ELF 32-bit LSB executable, ARM, version 1 (SYSV), dynamically linked (uses shared libs), for GNU/Linux 2.6.26, not stripped
Don't know if it is difference between freeScale imx515 and BMC2708 or something related to OS. Is this a problem with different version of kernel? or libc? I don't know much about it.
Jacques
What happens when you run ldd on the file?
9. Re: ARM Updates
- Posted by mattlewis (admin) Apr 07, 2013
- 2603 views
Matt, No luck with your build neither with mine on EfikaMX (I used linaro building tools). When trying to run eui the loader complain the file doesn't exist although "file eui" report
jacques@efikamx:~/Téléchargements/euphoria-4.1.0-Linux-ARM/bin$ file eui eui: ELF 32-bit LSB executable, ARM, version 1 (SYSV), dynamically linked (uses shared libs), for GNU/Linux 2.6.26, not stripped
Don't know if it is difference between freeScale imx515 and BMC2708 or something related to OS. Is this a problem with different version of kernel? or libc? I don't know much about it.
Difficult to say. Try something like:
$ /lib/ld-linux.so --list ./eui
That should show you what might be missing.
Matt
10. Re: ARM Updates
- Posted by coconut Apr 07, 2013
- 2490 views
Matt, No luck with your build neither with mine on EfikaMX (I used linaro building tools). When trying to run eui the loader complain the file doesn't exist although "file eui" report
jacques@efikamx:~/Téléchargements/euphoria-4.1.0-Linux-ARM/bin$ file eui eui: ELF 32-bit LSB executable, ARM, version 1 (SYSV), dynamically linked (uses shared libs), for GNU/Linux 2.6.26, not stripped
Don't know if it is difference between freeScale imx515 and BMC2708 or something related to OS. Is this a problem with different version of kernel? or libc? I don't know much about it.
Difficult to say. Try something like:
$ /lib/ld-linux.so --list ./eui
That should show you what might be missing.
Matt
jacques@efikamx:~/Téléchargements/euphoria-4.1.0-Linux-ARM/bin$ /lib/ld-linux.so.3 --list ./eui libm.so.6 => /lib/libm.so.6 (0x2aab4000) libdl.so.2 => /lib/libdl.so.2 (0x2ab5c000) libresolv.so.2 => /lib/libresolv.so.2 (0x2ab68000) libnsl.so.1 => /lib/libnsl.so.1 (0x2ab7c000) libpthread.so.0 => /lib/libpthread.so.0 (0x2ab99000) libc.so.6 => /lib/libc.so.6 (0x2abb9000) /lib/ld-linux-armhf.so.3 => /lib/ld-linux.so.3 (0x54aaa000)this the same information as 'ldd eui' but doesn't report any missing library. Without a better understanding of building linux application I can't do much more, hopefully python works well on efika so I don't really need euphoria.
Thanks Matt for all your efforts.
regards,
Jacques
11. Re: ARM Updates
- Posted by jimcbrown (admin) Apr 08, 2013
- 2520 views
this the same information as 'ldd eui' but doesn't report any missing library. Without a better understanding of building linux application I can't do much more
So both /lib/ld-linux-armhf.so.3 and /lib/ld-linux.so.3 exist on your system? Is one of them a symlink to the other?
What does something like 'ldd /bin/sh' say?
hopefully python works well on efika so I don't really need euphoria.
Wait a moment. I thought you already had eubins that worked! http://openeuphoria.org/forum/m/121251.wc
There's still http://openeuphoria.org/forum/m/121279.wc and http://openeuphoria.org/forum/m/121268.wc
12. Re: ARM Updates
- Posted by jimcbrown (admin) Apr 08, 2013
- 2544 views
If I had a procedure to cross-compile on my desktop running ubuntu 12.04LTS I would try to compile the binaries myself. Out of question to compile them on Efika, even if it succeed, it would take hours.
Jacques
Maybe take a look at http://www.raspberrypi.org/phpBB3/viewtopic.php?t=11315&p=129461
13. Re: ARM Updates
- Posted by coconut Apr 08, 2013
- 2532 views
this the same information as 'ldd eui' but doesn't report any missing library. Without a better understanding of building linux application I can't do much more
So both /lib/ld-linux-armhf.so.3 and /lib/ld-linux.so.3 exist on your system? Is one of them a symlink to the other?
What does something like 'ldd /bin/sh' say?
hopefully python works well on efika so I don't really need euphoria.
Wait a moment. I thought you already had eubins that worked! http://openeuphoria.org/forum/m/121251.wc
There's still http://openeuphoria.org/forum/m/121279.wc and http://openeuphoria.org/forum/m/121268.wc
Yes there is a version of eubins that run up to 4% of sanity.ex that's the best I got up to now. It fail at the poke, peek test.
Jacques
14. Re: ARM Updates
- Posted by jimcbrown (admin) Apr 08, 2013
- 2537 views
Yes there is a version of eubins that run up to 4% of sanity.ex that's the best I got up to now. It fail at the poke, peek test.
But Python can't do poke/peek at all - http://mail.python.org/pipermail/cplusplus-sig/2010-October/015792.html
If you actually need that functionality, I can write up a small C lib that'll do it for you.
15. Re: ARM Updates
- Posted by coconut Apr 08, 2013
- 2537 views
If I had a procedure to cross-compile on my desktop running ubuntu 12.04LTS I would try to compile the binaries myself. Out of question to compile them on Efika, even if it succeed, it would take hours.
Jacques
Maybe take a look at http://www.raspberrypi.org/phpBB3/viewtopic.php?t=11315&p=129461
Thanks Jim for the link, but I don't know if I will take all the time needed to learn about that new stuff. It seem a lot to learn and understand. And the time a I take for it is less time for my main project.
Jacques
16. Re: ARM Updates
- Posted by mattlewis (admin) Apr 08, 2013
- 2547 views
Yes there is a version of eubins that run up to 4% of sanity.ex that's the best I got up to now. It fail at the poke, peek test.
But Python can't do poke/peek at all - http://mail.python.org/pipermail/cplusplus-sig/2010-October/015792.html
If you actually need that functionality, I can write up a small C lib that'll do it for you.
The poke / peek issue goes deeper than that, though, since it's a critical part of how the front end of the interpreter communicates with the back end.
Matt
17. Re: ARM Updates
- Posted by jimcbrown (admin) Apr 08, 2013
- 2533 views
Yes there is a version of eubins that run up to 4% of sanity.ex that's the best I got up to now. It fail at the poke, peek test.
But Python can't do poke/peek at all - http://mail.python.org/pipermail/cplusplus-sig/2010-October/015792.html
If you actually need that functionality, I can write up a small C lib that'll do it for you.
The poke / peek issue goes deeper than that, though, since it's a critical part of how the front end of the interpreter communicates with the back end.
Matt
Based on the fact that he was able to get to the point of failing with std/map.eissues during a translation attempt, I suspect that this part is working fine. That is, peek()/poke() doesn't work from Euphoria code run by the eubin, but the front-backend communication is functional.
18. Re: ARM Updates
- Posted by mattlewis (admin) Apr 08, 2013
- 2511 views
The poke / peek issue goes deeper than that, though, since it's a critical part of how the front end of the interpreter communicates with the back end.
Based on the fact that he was able to get to the point of failing with std/map.eissues during a translation attempt, I suspect that this part is working fine. That is, peek()/poke() doesn't work from Euphoria code run by the eubin, but the front-backend communication is functional.
It only fails for some values. For instance, if you have a pointer that's too big for a signed integer of the right size, you'll get a bad value. So if you have a big literal value in your code, it will show up as zero. This was why the tests for text:format() were failing.
I don't think this was the cause of the map issues, but it's hard to tell. Anyways, it's a serious problem, especially since it's not obvious why some bits of code fail.
Matt
19. Re: ARM Updates
- Posted by jimcbrown (admin) Apr 08, 2013
- 2466 views
I don't think this was the cause of the map issues, but it's hard to tell.
I didn't either.
It only fails for some values. For instance, if you have a pointer that's too big for a signed integer of the right size, you'll get a bad value. So if you have a big literal value in your code, it will show up as zero. This was why the tests for text:format() were failing. Anyways, it's a serious problem, especially since it's not obvious why some bits of code fail.
Matt
You seem to be saying that there are cases where front-backend communication will fail because of this bug, but in subtle ways instead of blowing up big.
Would this cause issues during translation? If not, I don't see why this is a serious issue here (though I agree that it's a pretty big deal in general).
20. Re: ARM Updates
- Posted by mattlewis (admin) Apr 08, 2013
- 2456 views
It only fails for some values. For instance, if you have a pointer that's too big for a signed integer of the right size, you'll get a bad value. So if you have a big literal value in your code, it will show up as zero. This was why the tests for text:format() were failing. Anyways, it's a serious problem, especially since it's not obvious why some bits of code fail.
You seem to be saying that there are cases where front-backend communication will fail because of this bug, but in subtle ways instead of blowing up big.
Would this cause issues during translation? If not, I don't see why this is a serious issue here (though I agree that it's a pretty big deal in general).
The same bug hits the translator and the interpreter. Consider the interpreter. It reads in literal values, and pokes them into memory for the back end to read. So instead of #4000_0000 being poked, actually #0 goes into memory. That's obviously a translator issue manifesting itself in the interpreter.
But if you create a large value somehow...say by allocating memory and poking the pointer value somewhere, you get the same result.
Matt
21. Re: ARM Updates
- Posted by jimcbrown (admin) Apr 08, 2013
- 2452 views
The same bug hits the translator and the interpreter. Consider the interpreter. It reads in literal values, and pokes them into memory for the back end to read. So instead of #4000_0000 being poked, actually #0 goes into memory. That's obviously a translator issue manifesting itself in the interpreter.
But if you create a large value somehow...say by allocating memory and poking the pointer value somewhere, you get the same result.
Matt
I see. So it might cause translation to blow up sometimes, but not everytime (depending on which address you get).
I guess in Jacques's case, he's better off cross-translating or cross-compiling then.
22. Re: ARM Updates
- Posted by jimcbrown (admin) Apr 08, 2013
- 2483 views
Matt, No luck with your build neither with mine on EfikaMX (I used linaro building tools).
I missed this the first time I read it. You're just trying to run Matt's eubins right? Where does linaro come into play?
When trying to run eui the loader complain the file doesn't exist although "file eui" report
jacques@efikamx:~/Téléchargements/euphoria-4.1.0-Linux-ARM/bin$ file eui eui: ELF 32-bit LSB executable, ARM, version 1 (SYSV), dynamically linked (uses shared libs), for GNU/Linux 2.6.26, not stripped
Don't know if it is difference between freeScale imx515 and BMC2708 or something related to OS. Is this a problem with different version of kernel? or libc? I don't know much about it.
Actually, I find it strange that the softfloat version on the wiki failed. Looking it up, it looks like it has the same type of processor (ARM Cortex-A8) that my Nook Color has. http://wiki.debian.org/EfikaMX
What was the exact error with that version?
Difficult to say. Try something like:
$ /lib/ld-linux.so --list ./eui
That should show you what might be missing.
Matt
Actually, try
LD_DEBUG=all /lib/ld.so ./eui
That should produce a lot more output.
Also, we can use the readelf command, which will go down to the individual symbol level.
readelf -d ./eui
It'd be nice to be able to compare that output to something like this:
readelf -d /bin/sh
23. Re: ARM Updates
- Posted by coconut Apr 09, 2013
- 2404 views
Matt, No luck with your build neither with mine on EfikaMX (I used linaro building tools).
I missed this the first time I read it. You're just trying to run Matt's eubins right? Where does linaro come into play?
When trying to run eui the loader complain the file doesn't exist although "file eui" report
jacques@efikamx:~/Téléchargements/euphoria-4.1.0-Linux-ARM/bin$ file eui eui: ELF 32-bit LSB executable, ARM, version 1 (SYSV), dynamically linked (uses shared libs), for GNU/Linux 2.6.26, not stripped
Don't know if it is difference between freeScale imx515 and BMC2708 or something related to OS. Is this a problem with different version of kernel? or libc? I don't know much about it.
Actually, I find it strange that the softfloat version on the wiki failed. Looking it up, it looks like it has the same type of processor (ARM Cortex-A8) that my Nook Color has. http://wiki.debian.org/EfikaMX
What was the exact error with that version?
Difficult to say. Try something like:
$ /lib/ld-linux.so --list ./eui
That should show you what might be missing.
Matt
Actually, try
LD_DEBUG=all /lib/ld.so ./eui
That should produce a lot more output.
Also, we can use the readelf command, which will go down to the individual symbol level.
readelf -d ./eui
It'd be nice to be able to compare that output to something like this:
readelf -d /bin/sh
Here's what I have done:
I written a simple hello.c application I compile the file on the EfikaMX and crosscompile it on my ubuntu 12.04 desktop. The file compiled on the Efika works no problem. For the version cross compiled I get different result depending on the method used to load it. here some screen capture. the hello-xc is the cross compiled one (trying to run on Efika):
jacques@efikamx:~/Téléchargements$ ./hello-xc -bash: ./hello-xc: No such file or directory jacques@efikamx:~/Téléchargements$ /lib/ld.so ./hello-xc hello world! from xcompile on ubuntu 12.04 # ************that way it works**************** jacques@efikamx:~/Téléchargements$ /lib/ld.so hello-xc hello-xc: error while loading shared libraries: hello-xc: cannot open shared object file: No such file or directory jacques@efikamx:~/Téléchargements$The problem is with linux configuration or maybe libraries version. The Efika as older kernel and libc dans RaspberryPi linaro project.
jacques@efikamx:~/Téléchargements$ file /lib/ld-linux.so.3 /lib/ld-linux.so.3: symbolic link to `ld-2.11.3.so' jacques@efikamx:~/Téléchargements$ file /lib/ld-2.11.3.so /lib/ld-2.11.3.so: ELF 32-bit LSB shared object, ARM, version 1 (SYSV), dynamically linked, stripped jacques@efikamx:~/Téléchargements$ uname -a Linux efikamx 2.6.31.14.27-efikamx #1 Tue Feb 21 15:43:17 UTC 2012 armv7l GNU/Linux
readelf -d hello and readelf -d hello-xc gives same result except for entry point values.
jacques@efikamx:~/Téléchargements$ readelf -d hello-xc Dynamic section at offset 0x4f0 contains 24 entries: Tag Type Name/Value 0x00000001 (NEEDED) Shared library: [libc.so.6] 0x0000000c (INIT) 0x82a0 0x0000000d (FINI) 0x84a0 0x00000019 (INIT_ARRAY) 0x104e4 0x0000001b (INIT_ARRAYSZ) 4 (bytes) 0x0000001a (FINI_ARRAY) 0x104e8 0x0000001c (FINI_ARRAYSZ) 4 (bytes) 0x00000004 (HASH) 0x8194 0x00000005 (STRTAB) 0x820c 0x00000006 (SYMTAB) 0x81bc 0x0000000a (STRSZ) 65 (bytes) 0x0000000b (SYMENT) 16 (bytes) 0x00000015 (DEBUG) 0x0 0x00000003 (PLTGOT) 0x105d8 0x00000002 (PLTRELSZ) 32 (bytes) 0x00000014 (PLTREL) REL 0x00000017 (JMPREL) 0x8280 0x00000011 (REL) 0x8278 0x00000012 (RELSZ) 8 (bytes) 0x00000013 (RELENT) 8 (bytes) 0x6ffffffe (VERNEED) 0x8258 0x6fffffff (VERNEEDNUM) 1 0x6ffffff0 (VERSYM) 0x824e 0x00000000 (NULL) 0x0 jacques@efikamx:~/Téléchargements$ readelf -d hello Dynamic section at offset 0x4b0 contains 25 entries: Tag Type Name/Value 0x00000001 (NEEDED) Shared library: [libc.so.6] 0x0000000c (INIT) 0x82c4 0x0000000d (FINI) 0x8464 0x00000019 (INIT_ARRAY) 0x104a4 0x0000001b (INIT_ARRAYSZ) 4 (bytes) 0x0000001a (FINI_ARRAY) 0x104a8 0x0000001c (FINI_ARRAYSZ) 4 (bytes) 0x00000004 (HASH) 0x818c 0x6ffffef5 (GNU_HASH) 0x81b4 0x00000005 (STRTAB) 0x8230 0x00000006 (SYMTAB) 0x81e0 0x0000000a (STRSZ) 65 (bytes) 0x0000000b (SYMENT) 16 (bytes) 0x00000015 (DEBUG) 0x0 0x00000003 (PLTGOT) 0x105a0 0x00000002 (PLTRELSZ) 32 (bytes) 0x00000014 (PLTREL) REL 0x00000017 (JMPREL) 0x82a4 0x00000011 (REL) 0x829c 0x00000012 (RELSZ) 8 (bytes) 0x00000013 (RELENT) 8 (bytes) 0x6ffffffe (VERNEED) 0x827c 0x6fffffff (VERNEEDNUM) 1 0x6ffffff0 (VERSYM) 0x8272 0x00000000 (NULL) 0x0
Jacques
24. Re: ARM Updates
- Posted by jimcbrown (admin) Apr 10, 2013
- 2236 views
Here's what I have done:
I written a simple hello.c application I compile the file on the EfikaMX and crosscompile it on my ubuntu 12.04 desktop. The file compiled on the Efika works no problem.
Thanks for helping to get to the bottom of this.
For the version cross compiled I get different result depending on the method used to load it.
That is weird.
The problem is with linux configuration or maybe libraries version. The Efika as older kernel and libc dans RaspberryPi linaro project.
I'm inclined to agree here.
jacques@efikamx:~/Téléchargements$ /lib/ld.so hello-xc hello-xc: error while loading shared libraries: hello-xc: cannot open shared object file: No such file or directory jacques@efikamx:~/Téléchargements$
That's normal. ld.so / ld-linux.so won't look in the current directory by default - it only searches the PATH.
here some screen capture. the hello-xc is the cross compiled one (trying to run on Efika):
jacques@efikamx:~/Téléchargements$ ./hello-xc -bash: ./hello-xc: No such file or directory jacques@efikamx:~/Téléchargements$ /lib/ld.so ./hello-xc hello world! from xcompile on ubuntu 12.04 # ************that way it works****************
I wonder why that worked. Makes me think that the cross-compiled version uses something different (maybe a.out vs ELF? or ld.so supports the older ARMEL instead of ARMHF as ld-linux-arm-hf.so does?) It's like /lib/ld.so points to a completely different loader altogether.
jacques@efikamx:~/Téléchargements$ file /lib/ld-linux.so.3 /lib/ld-linux.so.3: symbolic link to `ld-2.11.3.so' jacques@efikamx:~/Téléchargements$ file /lib/ld-2.11.3.so /lib/ld-2.11.3.so: ELF 32-bit LSB shared object, ARM, version 1 (SYSV), dynamically linked, stripped jacques@efikamx:~/Téléchargements$ uname -a Linux efikamx 2.6.31.14.27-efikamx #1 Tue Feb 21 15:43:17 UTC 2012 armv7l GNU/Linux
What about file /lib/ld.so ?
readelf -d hello and readelf -d hello-xc gives same result except for entry point values.
jacques@efikamx:~/Téléchargements$ readelf -d hello-xc Dynamic section at offset 0x4f0 contains 24 entries: Tag Type Name/Value 0x00000001 (NEEDED) Shared library: [libc.so.6] 0x0000000c (INIT) 0x82a0 0x0000000d (FINI) 0x84a0 0x00000019 (INIT_ARRAY) 0x104e4 0x0000001b (INIT_ARRAYSZ) 4 (bytes) 0x0000001a (FINI_ARRAY) 0x104e8 0x0000001c (FINI_ARRAYSZ) 4 (bytes) 0x00000004 (HASH) 0x8194 0x00000005 (STRTAB) 0x820c 0x00000006 (SYMTAB) 0x81bc 0x0000000a (STRSZ) 65 (bytes) 0x0000000b (SYMENT) 16 (bytes) 0x00000015 (DEBUG) 0x0 0x00000003 (PLTGOT) 0x105d8 0x00000002 (PLTRELSZ) 32 (bytes) 0x00000014 (PLTREL) REL 0x00000017 (JMPREL) 0x8280 0x00000011 (REL) 0x8278 0x00000012 (RELSZ) 8 (bytes) 0x00000013 (RELENT) 8 (bytes) 0x6ffffffe (VERNEED) 0x8258 0x6fffffff (VERNEEDNUM) 1 0x6ffffff0 (VERSYM) 0x824e 0x00000000 (NULL) 0x0 jacques@efikamx:~/Téléchargements$ readelf -d hello Dynamic section at offset 0x4b0 contains 25 entries: Tag Type Name/Value 0x00000001 (NEEDED) Shared library: [libc.so.6] 0x0000000c (INIT) 0x82c4 0x0000000d (FINI) 0x8464 0x00000019 (INIT_ARRAY) 0x104a4 0x0000001b (INIT_ARRAYSZ) 4 (bytes) 0x0000001a (FINI_ARRAY) 0x104a8 0x0000001c (FINI_ARRAYSZ) 4 (bytes) 0x00000004 (HASH) 0x818c 0x6ffffef5 (GNU_HASH) 0x81b4 0x00000005 (STRTAB) 0x8230 0x00000006 (SYMTAB) 0x81e0 0x0000000a (STRSZ) 65 (bytes) 0x0000000b (SYMENT) 16 (bytes) 0x00000015 (DEBUG) 0x0 0x00000003 (PLTGOT) 0x105a0 0x00000002 (PLTRELSZ) 32 (bytes) 0x00000014 (PLTREL) REL 0x00000017 (JMPREL) 0x82a4 0x00000011 (REL) 0x829c 0x00000012 (RELSZ) 8 (bytes) 0x00000013 (RELENT) 8 (bytes) 0x6ffffffe (VERNEED) 0x827c 0x6fffffff (VERNEEDNUM) 1 0x6ffffff0 (VERSYM) 0x8272 0x00000000 (NULL) 0x0
Jacques
That looks normal.
25. Re: ARM Updates
- Posted by coconut Apr 10, 2013
- 2328 views
jacques@efikamx:~/Téléchargements$ file /lib/ld-linux.so.3 /lib/ld-linux.so.3: symbolic link to `ld-2.11.3.so' jacques@efikamx:~/Téléchargements$ file /lib/ld-2.11.3.so /lib/ld-2.11.3.so: ELF 32-bit LSB shared object, ARM, version 1 (SYSV), dynamically linked, stripped jacques@efikamx:~/Téléchargements$ uname -a Linux efikamx 2.6.31.14.27-efikamx #1 Tue Feb 21 15:43:17 UTC 2012 armv7l GNU/Linux
What about file /lib/ld.so ?
[/quote]
That looks normal.
[/quote]
ld.so is a soft link I created to save keystrokes. It point to /lib/ld-linux-2.11.3.so
Jacques
26. Re: ARM Updates
- Posted by jimcbrown (admin) Apr 10, 2013
- 2472 views
jacques@efikamx:~/Téléchargements$ file /lib/ld-linux.so.3 /lib/ld-linux.so.3: symbolic link to `ld-2.11.3.so' jacques@efikamx:~/Téléchargements$ file /lib/ld-2.11.3.so /lib/ld-2.11.3.so: ELF 32-bit LSB shared object, ARM, version 1 (SYSV), dynamically linked, stripped jacques@efikamx:~/Téléchargements$ uname -a Linux efikamx 2.6.31.14.27-efikamx #1 Tue Feb 21 15:43:17 UTC 2012 armv7l GNU/Linux
What about file /lib/ld.so ?
That looks normal.
[/quote]
ld.so is a soft link I created to save keystrokes. It point to /lib/ld-linux-2.11.3.so
Jacques
[/quote]
Hmm. What's the relationship between /lib/ld-2.11.3.so and /lib/ld-linux-2.11.3.so ?
27. Re: ARM Updates
- Posted by coconut Apr 10, 2013
- 2304 views
jacques@efikamx:~/Téléchargements$ file /lib/ld-linux.so.3 /lib/ld-linux.so.3: symbolic link to `ld-2.11.3.so' jacques@efikamx:~/Téléchargements$ file /lib/ld-2.11.3.so /lib/ld-2.11.3.so: ELF 32-bit LSB shared object, ARM, version 1 (SYSV), dynamically linked, stripped jacques@efikamx:~/Téléchargements$ uname -a Linux efikamx 2.6.31.14.27-efikamx #1 Tue Feb 21 15:43:17 UTC 2012 armv7l GNU/Linux
What about file /lib/ld.so ?
That looks normal.
ld.so is a soft link I created to save keystrokes. It point to /lib/ld-linux-2.11.3.so
Jacques
[/quote]
Hmm. What's the relationship between /lib/ld-2.11.3.so and /lib/ld-linux-2.11.3.so ?
[/quote]
/lib/ld-linux-so.3 is hard link to /lib/ld-2.11.3.so
there is no /lib/ld-linux-2.11.3.so
Jacques
28. Re: ARM Updates
- Posted by coconut Apr 10, 2013
- 2316 views
Who has built this one? http://openeuphoria.org/eubins/linux/4.1.0/arm-32-bit/eubin-arm-cortex-a8-2012-01-13-ee49c6b8a340.tar.gz This is the only one that load and run on my EfikaMX. It run somes of the demos but crash sanity.ex at 4% (poke(), peek() test).
What was the build environment?
All others build generate segmentation fault when loaded with /lib/ld-linux.3.so ./eui
Jacques
29. Re: ARM Updates
- Posted by Jerome Apr 10, 2013
- 2327 views
Yes there is a version of eubins that run up to 4% of sanity.ex that's the best I got up to now. It fail at the poke, peek test.
Who has built this one? http://openeuphoria.org/eubins/linux/4.1.0/arm-32-bit/eubin-arm-cortex-a8-2012-01-13-ee49c6b8a340.tar.gz This is the only one that load and run on my EfikaMX.
Jacques, the two eubins you referred to are two different abi's: hard and soft. We should really know what your system is and then work from there. With the 'hello.c' program you wrote, can you do the following on your EfikaMX:
gcc -v gcc hello.c -o hello readelf -A hello
This is just a quick test to check the abi settings of your native compiler.
Thanks,
Ira
30. Re: ARM Updates
- Posted by coconut Apr 10, 2013
- 2313 views
Yes there is a version of eubins that run up to 4% of sanity.ex that's the best I got up to now. It fail at the poke, peek test.
Who has built this one? http://openeuphoria.org/eubins/linux/4.1.0/arm-32-bit/eubin-arm-cortex-a8-2012-01-13-ee49c6b8a340.tar.gz This is the only one that load and run on my EfikaMX.
Jacques, the two eubins you referred to are two different abi's: hard and soft. We should really know what your system is and then work from there. With the 'hello.c' program you wrote, can you do the following on your EfikaMX:
gcc -v gcc hello.c -o hello readelf -A hello
This is just a quick test to check the abi settings of your native compiler.
Thanks,
Ira
done:
jacques@efikamx:~/Téléchargements$ gcc -v Using built-in specs. Target: arm-linux-gnueabi Configured with: ../src/configure -v --with-pkgversion='Debian 4.4.5-8' --with-bugurl=file:///usr/share/doc/gcc-4.4/README.Bugs --enable-languages=c,c++,fortran,objc,obj-c++ --prefix=/usr --program-suffix=-4.4 --enable-shared --enable-multiarch --enable-linker-build-id --with-system-zlib --libexecdir=/usr/lib --without-included-gettext --enable-threads=posix --with-gxx-include-dir=/usr/include/c++/4.4 --libdir=/usr/lib --enable-nls --enable-clocale=gnu --enable-libstdcxx-debug --enable-objc-gc --disable-sjlj-exceptions --enable-checking=release --build=arm-linux-gnueabi --host=arm-linux-gnueabi --target=arm-linux-gnueabi Thread model: posix gcc version 4.4.5 (Debian 4.4.5-8) jacques@efikamx:~/Téléchargements$ gcc hello.c -ohello jacques@efikamx:~/Téléchargements$ readelf -A hello Attribute Section: aeabi File Attributes Tag_CPU_name: "ARM9TDMI" Tag_CPU_arch: v4T Tag_ARM_ISA_use: Yes Tag_THUMB_ISA_use: Thumb-1 Tag_ABI_PCS_wchar_t: 4 Tag_ABI_FP_denormal: Needed Tag_ABI_FP_exceptions: Needed Tag_ABI_FP_number_model: IEEE 754 Tag_ABI_align8_needed: Yes Tag_ABI_align8_preserved: Yes, except leaf SP Tag_ABI_enum_size: intno hardfload! but the i.MX515 has a FPU
what it give with eui for eui from eubins-armv6hf
jacques@efikamx:~/Téléchargements/eubin-armv6hf$ readelf -A eui Attribute Section: aeabi File Attributes Tag_CPU_name: "6" Tag_CPU_arch: v6 Tag_ARM_ISA_use: Yes Tag_THUMB_ISA_use: Thumb-1 Tag_VFP_arch: VFPv2 Tag_ABI_PCS_wchar_t: 4 Tag_ABI_FP_denormal: Needed Tag_ABI_FP_exceptions: Needed Tag_ABI_FP_number_model: IEEE 754 Tag_ABI_align8_needed: Yes Tag_ABI_align8_preserved: Yes, except leaf SP Tag_ABI_enum_size: int Tag_ABI_HardFP_use: SP and DP Tag_ABI_VFP_args: VFP registers Tag_CPU_unaligned_access: v6 Tag_DIV_use: Not allowed
What can I do from here?
Jacques
P.S. no need to answer I found the softfloat compiler in rpi tools directory and launched a cross-compile with it.
Thanks Ira for pointing me in the right direction.
31. Re: ARM Updates
- Posted by jimcbrown (admin) Apr 10, 2013
- 2261 views
Who has built this one? http://openeuphoria.org/eubins/linux/4.1.0/arm-32-bit/eubin-arm-cortex-a8-2012-01-13-ee49c6b8a340.tar.gz This is the only one that load and run on my EfikaMX.
I built it.
I've built a newer version from the same system here: http://openeuphoria.org/eubins/linux/4.1.0/arm-32-bit/eubin-arm-cortex-a8-2013-04-10-e767d3c187d1.tar.gz
This is using a revision from today, and should have all the latest updates and fixes that were committed to trunk.
32. Re: ARM Updates
- Posted by jimcbrown (admin) Apr 10, 2013
- 2407 views
done:
jacques@efikamx:~/Téléchargements$ gcc -v Using built-in specs. Target: arm-linux-gnueabi Configured with: ../src/configure -v --with-pkgversion='Debian 4.4.5-8' --with-bugurl=file:///usr/share/doc/gcc-4.4/README.Bugs --enable-languages=c,c++,fortran,objc,obj-c++ --prefix=/usr --program-suffix=-4.4 --enable-shared --enable-multiarch --enable-linker-build-id --with-system-zlib --libexecdir=/usr/lib --without-included-gettext --enable-threads=posix --with-gxx-include-dir=/usr/include/c++/4.4 --libdir=/usr/lib --enable-nls --enable-clocale=gnu --enable-libstdcxx-debug --enable-objc-gc --disable-sjlj-exceptions --enable-checking=release --build=arm-linux-gnueabi --host=arm-linux-gnueabi --target=arm-linux-gnueabi Thread model: posix gcc version 4.4.5 (Debian 4.4.5-8) jacques@efikamx:~/Téléchargements$ gcc hello.c -ohello jacques@efikamx:~/Téléchargements$ readelf -A hello Attribute Section: aeabi File Attributes Tag_CPU_name: "ARM9TDMI" Tag_CPU_arch: v4T Tag_ARM_ISA_use: Yes Tag_THUMB_ISA_use: Thumb-1 Tag_ABI_PCS_wchar_t: 4 Tag_ABI_FP_denormal: Needed Tag_ABI_FP_exceptions: Needed Tag_ABI_FP_number_model: IEEE 754 Tag_ABI_align8_needed: Yes Tag_ABI_align8_preserved: Yes, except leaf SP Tag_ABI_enum_size: intno hardfload! but the i.MX515 has a FPU
It's a software issue as well - your libraries (libc and so on, if not the kernel itself) need to be built with hardfloat support on. You can't mix and match hardfloat and softfloat object files together.
(In addition to hardfloat and 100% software float emulation, there's also something called VFP. Urgh.)
So, if your distro is using all softfloat libraries, then the binaries that you build need to be the same.
what it give with eui for eui from eubins-armv6hf
jacques@efikamx:~/Téléchargements/eubin-armv6hf$ readelf -A eui Attribute Section: aeabi File Attributes Tag_CPU_name: "6" Tag_CPU_arch: v6 Tag_ARM_ISA_use: Yes Tag_THUMB_ISA_use: Thumb-1 Tag_VFP_arch: VFPv2 Tag_ABI_PCS_wchar_t: 4 Tag_ABI_FP_denormal: Needed Tag_ABI_FP_exceptions: Needed Tag_ABI_FP_number_model: IEEE 754 Tag_ABI_align8_needed: Yes Tag_ABI_align8_preserved: Yes, except leaf SP Tag_ABI_enum_size: int Tag_ABI_HardFP_use: SP and DP Tag_ABI_VFP_args: VFP registers Tag_CPU_unaligned_access: v6 Tag_DIV_use: Not allowed
What can I do from here?
Jacques
P.S. no need to answer I found the softfloat compiler in rpi tools directory and launched a cross-compile with it.
Thanks Ira for pointing me in the right direction.
Brilliant! That's good to know. Someone should probably add a EfikaMX Debian 4.4.5-8 specific note to the wiki about this.
33. Re: ARM Updates
- Posted by coconut Apr 10, 2013
- 2230 views
Who has built this one? http://openeuphoria.org/eubins/linux/4.1.0/arm-32-bit/eubin-arm-cortex-a8-2012-01-13-ee49c6b8a340.tar.gz This is the only one that load and run on my EfikaMX.
I built it.
I've built a newer version from the same system here: http://openeuphoria.org/eubins/linux/4.1.0/arm-32-bit/eubin-arm-cortex-a8-2013-04-10-e767d3c187d1.tar.gz
This is using a revision from today, and should have all the latest updates and fixes that were committed to trunk.
Load and run but still crash sanity.ex at same point poke() peek() test.
Jacques
34. Re: ARM Updates
- Posted by jimcbrown (admin) Apr 10, 2013
- 2256 views
Who has built this one? http://openeuphoria.org/eubins/linux/4.1.0/arm-32-bit/eubin-arm-cortex-a8-2012-01-13-ee49c6b8a340.tar.gz This is the only one that load and run on my EfikaMX. It run somes of the demos but crash sanity.ex at 4% (poke(), peek() test).
What was the build environment?
All others build generate segmentation fault when loaded with /lib/ld-linux.3.so ./eui
Jacques
I built it. The build environment was a Ubuntu 9.10 image wth kernel 2.6.32.9 and PREEMPT on armv7l running via chroot under an Android 2.2 system (so 2.6.32.9 is actually the Android kernel) - a Nook Color tablet, to be precise.
35. Re: ARM Updates
- Posted by jimcbrown (admin) Apr 10, 2013
- 2235 views
Load and run but still crash sanity.ex at same point poke() peek() test.
Jacques
Um, wow. It's happening for me too. This is so embarrassing.
36. Re: ARM Updates
- Posted by mattlewis (admin) Apr 10, 2013
- 2250 views
Load and run but still crash sanity.ex at same point poke() peek() test.
Jacques
Um, wow. It's happening for me too. This is so embarrassing.
I'll have to get my pi fired up in the morning. I haven't thought about sanity.ex in a long time. What about unit tests? Specifically, t_machine.e, t_convert.e, t_text.e and t_literal.e?
Matt
37. Re: ARM Updates
- Posted by Jerome Apr 10, 2013
- 2260 views
P.S. no need to answer I found the softfloat compiler in rpi tools directory and launched a cross-compile with it. Thanks Ira for pointing me in the right direction.
Jacques, I'm glad this helped!
Someone should probably add a EfikaMX Debian 4.4.5-8 specific note to the wiki about this.
I really like the way Archlinux handles their ARM developement http://archlinuxarm.org/developers/distcc-cross-compiling, we could have eubins for ARMv5, ARMv6, and ARMv7 with hard and soft float clearly specified. The different eubins are easy to generate with cross-compiling scripts.
Thanks,
Ira
38. Re: ARM Updates
- Posted by mattlewis (admin) Apr 11, 2013
- 2195 views
Load and run but still crash sanity.ex at same point poke() peek() test.
Jacques
Um, wow. It's happening for me too. This is so embarrassing.
I'll have to get my pi fired up in the morning. I haven't thought about sanity.ex in a long time. What about unit tests? Specifically, t_machine.e, t_convert.e, t_text.e and t_literal.e?
OK, I get that sanity.ex failure, too. Lots going on in that bit. It's definitely going into our unit tests.
poke(addr, {77, -1, 5.1, -1.1}) if compare(peek({addr, 4}), {77, 255, 5, 255}) then crash(generic_msg)
Matt
39. Re: ARM Updates
- Posted by mattlewis (admin) Apr 11, 2013
- 2146 views
Load and run but still crash sanity.ex at same point poke() peek() test.
I just pushed up a fix. It's another difference in how ARM and x86 cast doubles to ints. We'll need to do the same fix for other poke sizes. I added a test to t_machine.e, and it now passes interpreted and translated.
I also ifdef'd pit the call() test in sanity.ex. I'm not sure what the equivalent ARM version should be to return immediately. Now sanity.ex passes on my Pi.
Matt
40. Re: ARM Updates
- Posted by jimcbrown (admin) Apr 11, 2013
- 2200 views
I also ifdef'd pit the call() test in sanity.ex. I'm not sure what the equivalent ARM version should be to return immediately.
Just guessing here, but I think it's #CB00 - http://infocenter.arm.com/help/topic/com.arm.doc.faqs/ka9568.html
Edit: I am assuming that this should be a near function call, though - http://infocenter.arm.com/help/topic/com.arm.doc.faqs/ka9225.html
I think that the bugfix confirms my initial suspicions after all - this was a minor peek/poke bug that didn't affect the big picture (i.e. the ability to boostrap the interpreter). Though hindsight is 20/20 I guess.
41. Re: ARM Updates
- Posted by mattlewis (admin) Apr 11, 2013
- 2222 views
I also ifdef'd pit the call() test in sanity.ex. I'm not sure what the equivalent ARM version should be to return immediately.
Just guessing here, but I think it's #CB00 - http://infocenter.arm.com/help/topic/com.arm.doc.faqs/ka9568.html
Edit: I am assuming that this should be a near function call, though - http://infocenter.arm.com/help/topic/com.arm.doc.faqs/ka9225.html
I think that the bugfix confirms my initial suspicions after all - this was a minor peek/poke bug that didn't affect the big picture (i.e. the ability to boostrap the interpreter). Though hindsight is 20/20 I guess.
Nope. After an objdump on some test code, I found the following works:
addr = allocate_code({0xe52db004, 0xe8bd0800, 0xe12fff1e}, 4)
...which objdump tells me is:
e52db004 push {fp} ; (str fp, [sp, #-4]!)
e8bd0800 pop {fp}
e12fff1e bx lr
Matt
42. Re: ARM Updates
- Posted by coconut Apr 11, 2013
- 2136 views
Bingo!
added in config.gnu
MSIZE= -mcpu=cortex-a8
the build result in a hardfloat but it works.
jacques@efikamx:~$ readelf -A $EUDIR/bin/eui
Attribute Section: aeabi
File Attributes
Tag_CPU_name: "Cortex-A8"
Tag_CPU_arch: v7
Tag_CPU_arch_profile: Application
Tag_ARM_ISA_use: Yes
Tag_THUMB_ISA_use: Thumb-2
Tag_VFP_arch: VFPv2
Tag_ABI_PCS_wchar_t: 4
Tag_ABI_FP_denormal: Needed
Tag_ABI_FP_exceptions: Needed
Tag_ABI_FP_number_model: IEEE 754
Tag_ABI_align8_needed: Yes
Tag_ABI_align8_preserved: Yes, except leaf SP
Tag_ABI_enum_size: int
Tag_ABI_HardFP_use: SP and DP
Tag_CPU_unaligned_access: v6
Tag_DIV_use: Not allowed
Tag_Virtualization_use: TrustZone
jacques@efikamx:~$ eui --version
Euphoria Interpreter v4.1.0 development
32-bit Linux, Using System Memory
Revision Date: 2013-04-11 11:32:27, Id: 6100:1973ffe5a389
jacques@efikamx:~$ eui $EUDIR/demo/sanity.ex
Euphoria SANITY TEST ...
100%
PASSED (100%)
Thanks guys you have been very helpfull.
Jacques
43. Re: ARM Updates
- Posted by jimcbrown (admin) Apr 11, 2013
- 2123 views
Bingo!
Thanks guys you have been very helpfull.
Jacques
I know I'm being greedy here, but is there any way you can send the eubins to us? Or share them with the rest of the world?
44. Re: ARM Updates
- Posted by coconut Apr 11, 2013
- 2145 views
Bingo!
Thanks guys you have been very helpfull.
Jacques
I know I'm being greedy here, but is there any way you can send the eubins to us? Or share them with the rest of the world?
No problem,
eubins-cortex-a8-2013-04-11.tar.gz
Jacques
