![]() |
#226 |
Going Viral
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() Posts: 17,212
Karma: 18210809
Join Date: Feb 2012
Location: Central Texas
Device: No K1, PW2, KV, KOA
|
The build process should have left the (hidden) .config file around for uClibc -
see if the options where set to build the software floating point routines (they should be provided by uClibc, not by gcc). Those options might have been disabled (since everything else in the world has floating point hardware - except your Marvell chip (isn't that a comic book publisher?) ) Last edited by knc1; 05-17-2016 at 06:09 PM. |
![]() |
![]() |
![]() |
#227 | |
Carpe diem, c'est la vie.
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() Posts: 6,433
Karma: 10773670
Join Date: Nov 2011
Location: Multiverse 6627A
Device: K1 to PW3
|
Quote:
PHP Code:
I looked up the missing "__udivsi3", and it says (for this and other related functions): "The integer arithmetic routines are used on platforms that don't provide hardware support for arithmetic operations on some modes." But that description was for libgcc, and we should have that (because we have gcc), I think. No build errors making tcc anyway... We know that float is missing, but I also recall reading that integer division is missing too (and likewise on many newer arm processors as well). EDIT: More googling for those missing symbools PLUS uclibc finds complaints of "stripped down" libraries, but a common solution is to all "-lm" to link in the math library. Two problems there though -- that is the FIRST thing I tried (unsuccessfully), and what math functions does a "Hello World" app need anyway? Well, apparently it needs ALL those missing functions, but why? And google chrome still locks me out of large parts of uclibc.org (bad certificate), so I need to use a different browser to visit uclibc.org. Last edited by geekmaster; 05-17-2016 at 06:46 PM. |
|
![]() |
![]() |
Advert | |
|
![]() |
#228 |
Carpe diem, c'est la vie.
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() Posts: 6,433
Karma: 10773670
Join Date: Nov 2011
Location: Multiverse 6627A
Device: K1 to PW3
|
Java 1.0 was released in 1996. The PXA255 came into existence in 2003. The ARM "Jazelle" native Java bytecode execution began with the ARMv5TEJ, but the PXA255 is missing the 'J'. That means a Java bytecode interpreter for the K1 would need to be run in software, and such things are available even on the lowliest of processors (like Arduino).
|
![]() |
![]() |
![]() |
#229 |
Going Viral
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() Posts: 17,212
Karma: 18210809
Join Date: Feb 2012
Location: Central Texas
Device: No K1, PW2, KV, KOA
|
Try including libm in the link ( -lm ) - those names look like libm names (IIRC - I didn't look before I posted).
|
![]() |
![]() |
![]() |
#230 | |
Carpe diem, c'est la vie.
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() Posts: 6,433
Karma: 10773670
Join Date: Nov 2011
Location: Multiverse 6627A
Device: K1 to PW3
|
Quote:
PHP Code:
Last edited by geekmaster; 05-17-2016 at 07:04 PM. |
|
![]() |
![]() |
Advert | |
|
![]() |
#231 |
Going Viral
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() Posts: 17,212
Karma: 18210809
Join Date: Feb 2012
Location: Central Texas
Device: No K1, PW2, KV, KOA
|
Your posts are so verbose, I have trouble extracting the information.
The 'key' to reading those symbols (usually is, but Eric might think differently) one '_' library internal, two '__' library internal, access by one (of possibly multiple) "#define" or "#typedef" statements. = = = = If you attached a copy of the .config Aboriginal used to build your copy of the uClibc library - I missed it. |
![]() |
![]() |
![]() |
#232 | |
Carpe diem, c'est la vie.
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() Posts: 6,433
Karma: 10773670
Join Date: Nov 2011
Location: Multiverse 6627A
Device: K1 to PW3
|
Quote:
https://github.com/landley/aborigina.../master/config And the uClibc baseconfig: https://github.com/landley/aborigina...econfig-uClibc However, recent builds of aboriginal linux are musl-based rather than uclibc-based, and the armv4l I am using is an unconverted uclibc version. There is an archive of old aboriginal versions at landley.net. I do NOT know what config was used to build my working arm4vl version. Last edited by geekmaster; 05-17-2016 at 10:40 PM. |
|
![]() |
![]() |
![]() |
#233 |
Carpe diem, c'est la vie.
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() Posts: 6,433
Karma: 10773670
Join Date: Nov 2011
Location: Multiverse 6627A
Device: K1 to PW3
|
I loaded up my arm4vl build environment (which boots its own kernel), and did "find /|grep config" inside it, which found these:
PHP Code:
PHP Code:
"UCLIBC_HAS_FLOATS=y" Is that correct for a K1? I suppose I could change that line in the "DO NOT EDIT" file to "UCLIBC_HAS_FLOATS=n" and recompile, eh? Or instead perhaps "# UCLIBC_HAS_FLOATS is not set" like the others? Probably none of the above, perhaps run configure with cmd line switches or env vars instead? I am a configure newbie... I wonder how to rebuild it, and what errors I will encounter along the way... I think it was cmd line switches when I configured tinycc, so probably the same here. "configure -h" should say what is available. Probably build from that github source code, and cross my fingers... Last edited by geekmaster; 05-17-2016 at 11:17 PM. |
![]() |
![]() |
![]() |
#234 |
Going Viral
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() Posts: 17,212
Karma: 18210809
Join Date: Feb 2012
Location: Central Texas
Device: No K1, PW2, KV, KOA
|
None of the above - floats look properly supported
But notice the ...si... in the missing symbol names? and the # DO_XSI_MATH is not set change that one (with the uClibc menuconfig utility) and re-build the library. |
![]() |
![]() |
![]() |
#235 |
Carpe diem, c'est la vie.
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() Posts: 6,433
Karma: 10773670
Join Date: Nov 2011
Location: Multiverse 6627A
Device: K1 to PW3
|
Just spent hours at landley.net. Got sidetracked from uClibc config into reams of his posts describing political drama in his projects. He seems to hate uClibc with a passion, probably why his recent switch to musl. Though quite fascinating, such historical reading detracts from "getting sh*t done". Now back to configging the errors out of my build process...
|
![]() |
![]() |
![]() |
#236 |
Going Viral
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() Posts: 17,212
Karma: 18210809
Join Date: Feb 2012
Location: Central Texas
Device: No K1, PW2, KV, KOA
|
Check out his opinion on GPLv3.
|
![]() |
![]() |
![]() |
#237 | ||
Carpe diem, c'est la vie.
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() Posts: 6,433
Karma: 10773670
Join Date: Nov 2011
Location: Multiverse 6627A
Device: K1 to PW3
|
As you wish...
![]() Quote:
Quote:
![]() Last edited by geekmaster; 05-18-2016 at 04:50 PM. |
||
![]() |
![]() |
![]() |
#238 |
Carpe diem, c'est la vie.
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() Posts: 6,433
Karma: 10773670
Join Date: Nov 2011
Location: Multiverse 6627A
Device: K1 to PW3
|
I am building aboriginal from current github source tree, for armv4l target. I see that it was just downloading musl, so I am curious (though doubtful) that it will work on my K1, considering that NiLuJe's musl test builds failed on my K1. I guess we shall see.
I suppose I need to go find OLD source that was still uClibc-based, from the landley.net download archives... |
![]() |
![]() |
![]() |
#239 | |
Connoisseur
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() Posts: 86
Karma: 186294
Join Date: Jun 2011
Device: Kindle k3G 3.4.2; DXG 2.5.8; DXG 3.1; Iriver Story HD
|
Quote:
DId you get the K1 rootfs image unsquashed? Did you find the remaining files in /usr/default someplace and get those added to your K1 copy for a complete system? That should allow you to chroot into the virtual K1 and run your actual K1 programs except with the modern kernel imitating a K1 kernel. That is as far as I got. I was going to install the header files from Amazon sources for the K1 for the 2.6.10 kernel and the headers for the uClibc into the K1 copy. Those are needed to make programs compatible with the K1. I quit at that point to do other things. The only big problem would be to install the correct gcc into the K1 filesystem copy. Programs compatible with the K1 could then be built with the virtual K1. The gcc could then use the original K1 uclibc.so to build a dynamic executable as long as gcc is told to link with the uclibc.so and the programs include the correct headers. That should work unless there are compatibility problems between gcc versions. That potential problem could be solved by using the virtual K1 gcc to build a correct K1 gcc able to run on the K1 not just the emulator. The resulting programs should then run perfectly on the K1. I never got that far. Having a correct build setup should make those other problems go away. I am too busy now and for the summer to ever finish so so long. |
|
![]() |
![]() |
![]() |
#240 |
Going Viral
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() Posts: 17,212
Karma: 18210809
Join Date: Feb 2012
Location: Central Texas
Device: No K1, PW2, KV, KOA
|
|
![]() |
![]() |
![]() |
|
![]() |
||||
Thread | Thread Starter | Forum | Replies | Last Post |
geekmaster vacation | geekmaster | Kindle Developer's Corner | 2 | 03-19-2012 09:18 PM |