Small linux distro for my embedded box

Announce/Discuss projects you are working on for MiniMyth

Moderator: Pablo

laga
Member
Posts: 11
Joined: Sat Jan 20, 2007 4:48 pm

Post by laga »

Hey there,

I'm the guy who made Juski buy his S100 ;)

I've started to tinker with gar-minimyth a little bit and i must say that i'm impressed. The build system is clean and easy to modify - i think, I have just finished a normal build a few minutes.

Anyways, I plan on making minimyth work on my S100. To accomplish this, I'll need Xorg 7.0 or older because it's required by the IEGD toolkit *sigh*. Do you guys still have those old Makefiles lying around or will I have to do it from scratch? i've also looked in the subversion repository but I haven't been able to find anything but xorg-7.2.

I also found some bugs/problems when building gar-minimyth on Ubuntu Edgy, but I was able to fix everything. I'm going to post my fixes later.

Are you still interested in having S100 support in minimyth? In the Changelog for 0.19-14, it says there is "initial support for the T-Online S100". Is it still there in recent versions or did you have to drop it?

Regards,

Michael

edit: Oops! I found it! It was in the subversion repository where I'd looked before. Stupid eyes.
Pablo
Site Admin
Posts: 4182
Joined: Tue Dec 14, 2004 2:13 am
Location: La Jolla
Contact:

Post by Pablo »

laga wrote:Anyways, I plan on making minimyth work on my S100. To accomplish this, I'll need Xorg 7.0 or older because it's required by the IEGD toolkit *sigh*. Do you guys still have those old Makefiles lying around or will I have to do it from scratch? i've also looked in the subversion repository but I haven't been able to find anything but xorg-7.2.
xorg-7.0 was last in perforce changeset 1056 which should correspond to svn changeset 1011. However, I believe that the easiest thing would be for my to check xorg-7.0 into svn head, so I have done that.

In addition, I added back IEGD (package 'X11/IEGD').

It sure would be nice to have IEGD work on Xorg 7.2 since Xorg 7.2 is needed for Intel's new open source video drivers. These drivers are the reason that I have been move MiniMyth to Xorg 7.2 and Linux kernel 2.6.19.
I also found some bugs/problems when building gar-minimyth on Ubuntu Edgy, but I was able to fix everything. I'm going to post my fixes later.
That would be great. Thank you.
Are you still interested in having S100 support in minimyth? In the Changelog for 0.19-14, it says there is "initial support for the T-Online S100". Is it still there in recent versions or did you have to drop it?
That initial S100 support was dropped since it was not be actively used/developed. I would be fine with including it in MiniMyth. One problem that came up last time was memory. I am not sure that it will be possible to get MiniMyth running in 128MB of memory.
MiniMyth running on an Acer ApireRevo 3610 and a Zotac ZBOX-ID80-U. Find out more at my MythTV page.
laga
Member
Posts: 11
Joined: Sat Jan 20, 2007 4:48 pm

Post by laga »

Pablo wrote: However, I believe that the easiest thing would be for my to check xorg-7.0 into svn head, so I have done that.
In addition, I added back IEGD (package 'X11/IEGD').
Thanks! That'll save me a lot of work.

It sure would be nice to have IEGD work on Xorg 7.2 since Xorg 7.2 is needed for Intel's new open source video drivers. These drivers are the reason that I have been move MiniMyth to Xorg 7.2 and Linux kernel 2.6.19.
Intel's release notes say that there only are prebuilt drivers for X.org up to release 7.0. Is there a chance of getting them to work with 7.2?


laga wrote:I also found some bugs/problems when building gar-minimyth on Ubuntu Edgy
My 'fixes' are just ugly workarounds that might actually break stuff, but at least i managed to compile gar-minimyth. If you want me to, I should able to reproduce these errors:

* *snicker* This is a good one: "git" was listed as a dependency. I just type sudo aptitude install git to install it which turned out to be some file manager. The real packages are called cogito and git-core. In my case, I had to run sudo update-alternatives --config git to enable the real thing, probably because I still had that file manager installed.


* "rxvt" didn't compile: "config.cache not found" right after or during configure. I fixed it by removing the --cache-file=config.cache \ from script/X11/rxvt/Makefile.

* "splashutils" failed to compile due to undefined references to some freetype stuff, if my memory serves me right. I just changed /system/splashutils/Makefile to read

Code: Select all

       --without-ttf \
       --without-ttfkern
instead of

Code: Select all

       --with-ttf \
       --with-ttfkern
I'm afraid this will break $stuff, though ;).

* "stella": "stella-2.3-src.tar.gz failed checksum test! "
Even after downloading the file again, the checksum test was still failing. I had to update the checksums file with the corresponding md5sum and all went well.

* There were some problems with bash-specific code in minimyth.mk.

Code: Select all

laga@prometheus:~/dev/gar-minimyth/gar-minimyth-0.20-21/script/meta/minimyth$ diff -u minimyth.mk.orig minimyth.mk
--- minimyth.mk.orig    2007-01-20 16:19:43.397695250 +0100
+++ minimyth.mk 2007-01-20 16:20:13.703589250 +0100
@@ -364,8 +364,8 @@
        @mkdir -p $(mm_ROOTFSDIR)/home/minimyth/.mythtv
        @ln -sf $(sysconfdir)/lircrc                     $(mm_ROOTFSDIR)/home/minimyth/.mythtv/lircrc
        @mkdir -p $(mm_ROOTFSDIR)$(sysconfdir)/rc.d/rc.d
-       @index=10 ; $(foreach file, $(MM_INIT_START), index=$$((index+2)) ; ln -sf ../init.d/$(file) $(mm_ROOTFSDIR)$(sysconfdir)/rc.d/rc.d/S$${index}$(file) ; )
-       @index=10 ; $(foreach file, $(MM_INIT_KILL), index=$$((index+2)) ; ln -sf ../init.d/$(file) $(mm_ROOTFSDIR)$(sysconfdir)/rc.d/rc.d/K$${index}$(file) ; )
+       @index=10 ; $(foreach file, $(MM_INIT_START), index=$(($index+2)) ; ln -sf ../init.d/$(file) $(mm_ROOTFSDIR)$(sysconfdir)/rc.d/rc.d/S$${index}$(file) ; )
+       @index=10 ; $(foreach file, $(MM_INIT_KILL), index=$(($index+2)) ; ln -sf ../init.d/$(file) $(mm_ROOTFSDIR)$(sysconfdir)/rc.d/rc.d/K$${index}$(file) ; )

 mm-remove-pre:
        @# Remove unwanted binaries, etc, shares and libraries.
@@ -392,7 +392,7 @@
                | sed -e 's%:.*%%'`" ; \
        bin_list=`echo $${bin_list}` ; \
        while test $${lib_count} -gt 0 ; do \
-               pass=$$((pass+1)) ; \
+               pass=$(($pass+1)) ; \
                lib_list="`find $(mm_ROOTFSDIR) -exec file '{}' \; \
                        | grep -i 'ELF ..-bit LSB shared object' \
                        | sed -e 's%:.*%%'`" ; \
That initial S100 support was dropped since it was not be actively used/developed. I would be fine with including it in MiniMyth. One problem that came up last time was memory. I am not sure that it will be possible to get MiniMyth running in 128MB of memory.
Memory will be a problem. We cannot afford to hold MiniMyth in a ramdisk :(. The most sensible approach would be "Network Boot with an NFS Root File System" which is just what I might end up doing for testing purposes. However, nfsroot over WLAN is not exactly something you would want to do for various reasons: it'll be slow and AFAIK there are no PXE stacks (read: etherboot ;)) that can handle encrypted WLAN.

In my opinion, the S100 will need the ability to boot from hard disks, be it USB or plain old IDE. Setting this up won't be a problem for *me*, since I've already found the place where the kernel configs live in gar-minimyth :). However, I assume there must be a reason why it's not already possible to boot from local disks.

I'm reluctant to propose the inclusion of S100 support in MiniMyth for a few reasons:

* IEGD: well, it's not open source and it requires X.org 7.0 and older. In order to enable DRI, the kernel needs to be patched. I don't know if the OSS intel drivers will still work after that.
* MythTV needs to be patched to be used with IEGD - at least that was the case the old IEGD, I haven't tried the new 6.1 one yet.
* MiniMyth would have to boot from a local hard disk.

I cannot ask you to put restrictions to the development of MiniMyth by including IEGD and friends. The only sensible way would be building different flavours of MiniMyth, eg Xorg 7.2+Openchrome+whatever and Xorg 7.0+IEGD+patched Mythtv. Another solution would be forking MiniMyth which I'd like to avoid - it's too much work and I'm lazy :).

What do you think?

-- Michael
Pablo
Site Admin
Posts: 4182
Joined: Tue Dec 14, 2004 2:13 am
Location: La Jolla
Contact:

Post by Pablo »

laga wrote:
Pablo wrote: It sure would be nice to have IEGD work on Xorg 7.2 since Xorg 7.2 is needed for Intel's new open source video drivers. These drivers are the reason that I have been move MiniMyth to Xorg 7.2 and Linux kernel 2.6.19.
Intel's release notes say that there only are prebuilt drivers for X.org up to release 7.0. Is there a chance of getting them to work with 7.2?
It is possible that it will work. However, X may need to be run with the -ignoreABI option.
laga wrote:I also found some bugs/problems when building gar-minimyth on Ubuntu Edgy
My 'fixes' are just ugly workarounds that might actually break stuff, but at least i managed to compile gar-minimyth. If you want me to, I should able to reproduce these errors:

* *snicker* This is a good one: "git" was listed as a dependency. I just type sudo aptitude install git to install it which turned out to be some file manager. The real packages are called cogito and git-core. In my case, I had to run sudo update-alternatives --config git to enable the real thing, probably because I still had that file manager installed.
I will add a comment to the documentation about the different gits.
* "rxvt" didn't compile: "config.cache not found" right after or during configure. I fixed it by removing the --cache-file=config.cache \ from script/X11/rxvt/Makefile.
Odd, config.cache should have been copied from the files directory. If you run 'make clean ; make patch' in the 'X11/rxvt' directory, does the WORKSRC directory (work/main.d/xrvt-2.6.4' contain 'config.cache'?
* "splashutils" failed to compile due to undefined references to some freetype stuff, if my memory serves me right. I just changed /system/splashutils/Makefile to read

Code: Select all

       --without-ttf \
       --without-ttfkern
instead of

Code: Select all

       --with-ttf \
       --with-ttfkern
I'm afraid this will break $stuff, though ;).
Yes, that is likely to break the splash screen. I suspect that the problem is with freetype rather than splashutils. Did freetype build a static library (images/main/usr/lib/libfreetype.a)?
* "stella": "stella-2.3-src.tar.gz failed checksum test! "
Even after downloading the file again, the checksum test was still failing. I had to update the checksums file with the corresponding md5sum and all went well.
I suspect that stella was changed without changing the version number. I hate it when software developers change software without changing version numbers.
* There were some problems with bash-specific code in minimyth.mk.

Code: Select all

laga@prometheus:~/dev/gar-minimyth/gar-minimyth-0.20-21/script/meta/minimyth$ diff -u minimyth.mk.orig minimyth.mk
--- minimyth.mk.orig    2007-01-20 16:19:43.397695250 +0100
+++ minimyth.mk 2007-01-20 16:20:13.703589250 +0100
@@ -364,8 +364,8 @@
        @mkdir -p $(mm_ROOTFSDIR)/home/minimyth/.mythtv
        @ln -sf $(sysconfdir)/lircrc                     $(mm_ROOTFSDIR)/home/minimyth/.mythtv/lircrc
        @mkdir -p $(mm_ROOTFSDIR)$(sysconfdir)/rc.d/rc.d
-       @index=10 ; $(foreach file, $(MM_INIT_START), index=$$((index+2)) ; ln -sf ../init.d/$(file) $(mm_ROOTFSDIR)$(sysconfdir)/rc.d/rc.d/S$${index}$(file) ; )
-       @index=10 ; $(foreach file, $(MM_INIT_KILL), index=$$((index+2)) ; ln -sf ../init.d/$(file) $(mm_ROOTFSDIR)$(sysconfdir)/rc.d/rc.d/K$${index}$(file) ; )
+       @index=10 ; $(foreach file, $(MM_INIT_START), index=$(($index+2)) ; ln -sf ../init.d/$(file) $(mm_ROOTFSDIR)$(sysconfdir)/rc.d/rc.d/S$${index}$(file) ; )
+       @index=10 ; $(foreach file, $(MM_INIT_KILL), index=$(($index+2)) ; ln -sf ../init.d/$(file) $(mm_ROOTFSDIR)$(sysconfdir)/rc.d/rc.d/K$${index}$(file) ; )

 mm-remove-pre:
        @# Remove unwanted binaries, etc, shares and libraries.
@@ -392,7 +392,7 @@
                | sed -e 's%:.*%%'`" ; \
        bin_list=`echo $${bin_list}` ; \
        while test $${lib_count} -gt 0 ; do \
-               pass=$$((pass+1)) ; \
+               pass=$(($pass+1)) ; \
                lib_list="`find $(mm_ROOTFSDIR) -exec file '{}' \; \
                        | grep -i 'ELF ..-bit LSB shared object' \
                        | sed -e 's%:.*%%'`" ; \
Definitely a bug. I wonder why it worked for me. However, I would think that they should be index=$$(($${index}+2)) and pass=$$(($${pass}+1)). If you change them so that they have two $ signs, does it still work for you?
MiniMyth running on an Acer ApireRevo 3610 and a Zotac ZBOX-ID80-U. Find out more at my MythTV page.
laga
Member
Posts: 11
Joined: Sat Jan 20, 2007 4:48 pm

Just a quick update:

Post by laga »

Hi,

thanks for your feedback. I'm currently rebuilding MiniMyth from the svn repository to see if my problems still persist. I'll keep you posted. I should also try to install IEGD on xorg 7.2, but it'll take some time till I'll get to that point.

Regards,

Michael
laga
Member
Posts: 11
Joined: Sat Jan 20, 2007 4:48 pm

Post by laga »

With regard to rxvt:

After running make clean; make patch, I do have a config.cache file:

Code: Select all

-rw-r--r-- 1 laga laga 126 2007-01-21 18:43 work/main.d/rxvt-2.6.4/config.cache
However, it still doesn't work. I'll include the output of make patch and make build below:

Code: Select all

laga@prometheus:~/dev/gar-minimyth/minimyth/gar-minimyth/script/X11/rxvt$ make patch
[===== NOW BUILDING:    rxvt-2.6.4      =====]
install -d cookies/main.d
install -d download
install -d download/partial
 ==> Grabbing download/rxvt-2.6.4.tar.gz
        ==> Trying file//files/rxvt-2.6.4.tar.gz
make[1]: *** [file//files/rxvt-2.6.4.tar.gz] Error 1
        ==> Trying file///home/laga/dev/gar-minimyth/minimyth/gar-minimyth/source/rxvt-2.6.4/rxvt-2.6.4.tar.gz
make[1]: *** [file///home/laga/dev/gar-minimyth/minimyth/gar-minimyth/source/rxvt-2.6.4/rxvt-2.6.4.tar.gz] Error 1
        ==> Trying http//easynews.dl.sourceforge.net/sourceforge/rxvt/rxvt-2.6.4.tar.gz
--18:43:01--  http://easynews.dl.sourceforge.net/sourceforge/rxvt/rxvt-2.6.4.tar.gz
           => `download/partial/rxvt-2.6.4.tar.gz'
Resolving easynews.dl.sourceforge.net... 69.16.168.245
Connecting to easynews.dl.sourceforge.net|69.16.168.245|:80... connected.
HTTP request sent, awaiting response... 200 OK
Length: 307.334 (300K) [application/x-tar]

100%[=======================================================>] 307.334      165.27K/s

18:43:03 (164.81 KB/s) - `download/partial/rxvt-2.6.4.tar.gz' saved [307334/307334]

 ==> Grabbing download/config.cache.i386
        ==> Trying file//files/config.cache.i386
        [fetch] complete for rxvt.
 ==> Running checksum on rxvt-2.6.4.tar.gz
6b33ae7724e5e669ab9b2e79ae9a7758  download/rxvt-2.6.4.tar.gz
file rxvt-2.6.4.tar.gz passes checksum test!
 ==> Running checksum on config.cache.i386
f9eeb72cbc2d1326f6561cc12737c28d  download/config.cache.i386
file config.cache.i386 passes checksum test!
        [checksum] complete for rxvt.
install -d work/main.d
 ==> Extracting download/rxvt-2.6.4.tar.gz
        [extract] complete for rxvt.
        [patch] complete for rxvt.

Code: Select all

laga@prometheus:~/dev/gar-minimyth/minimyth/gar-minimyth/script/X11/rxvt$ make build
[===== NOW BUILDING:    rxvt-2.6.4      =====]
        [fetch] complete for rxvt.
        [checksum] complete for rxvt.
        [extract] complete for rxvt.
        [patch] complete for rxvt.
make[1]: Entering directory `/home/laga/dev/gar-minimyth/minimyth/gar-minimyth/script/X11/rxvt'
make[1]: `../../lang/c/cookies/main.d/install' is up to date.
make[1]: `../../xorg/xorg/cookies/main.d/install' is up to date.
make[1]: Leaving directory `/home/laga/dev/gar-minimyth/minimyth/gar-minimyth/script/X11/rxvt'
 ==> Running configure in work/main.d/rxvt-2.6.4
loading cache config.cache
.: 567: config.cache: not found
make: *** [configure-work/main.d/rxvt-2.6.4/configure] Fehler 2
laga@prometheus:~/dev/gar-minimyth/minimyth/gar-minimyth/script/X11/rxvt$ ls -al work/main.d/rxvt-2.6.4/config.cache
-rw-r--r-- 1 laga laga 126 2007-01-21 18:43 work/main.d/rxvt-2.6.4/config.cache
laga@prometheus:~/dev/gar-minimyth/minimyth/gar-minimyth/script/X11/rxvt$           
HTH,

Michael


edit:Update:

Splashutils fails with the following error message:

Code: Select all

laga@prometheus:~/dev/gar-minimyth/minimyth/gar-minimyth/script/system/splashutils$ make build
[===== NOW BUILDING:    splashutils-1.3 =====]
        [fetch] complete for splashutils.
        [checksum] complete for splashutils.
        [extract] complete for splashutils.
        [patch] complete for splashutils.
make[1]: Entering directory `/home/laga/dev/gar-minimyth/minimyth/gar-minimyth/script/system/splashutils'
make[1]: `../../lang/c/cookies/main.d/install' is up to date.
make[1]: `../../kernel/kernel/cookies/main.d/install' is up to date.
make[1]: `../../lib/libjpeg/cookies/main.d/install' is up to date.
make[1]: `../../lib/zlib/cookies/main.d/install' is up to date.
make[1]: `../../lib/freetype/cookies/main.d/install' is up to date.
make[1]: Leaving directory `/home/laga/dev/gar-minimyth/minimyth/gar-minimyth/script/system/splashutils'
 ==> Running configure in work/main.d/splashutils-1.3
config.h successfully created.
You can run `make` now.
        [configure] complete for splashutils.
 ==> Running make in work/main.d/splashutils-1.3
/bin/sh: [[: not found
/bin/sh: [[: not found
/bin/sh: [[: not found
/bin/sh: [[: not found
make[1]: Entering directory `/home/laga/dev/gar-minimyth/minimyth/gar-minimyth/script/system/splashutils/work/main.d/splashutils-1.3'
[: 1: ==: unexpected operator
[: 1: ==: unexpected operator
[: 1: ==: unexpected operator
[: 1: ==: unexpected operator
[: 1: ==: unexpected operator
[: 1: ==: unexpected operator
[: 1: ==: unexpected operator
[: 1: ==: unexpected operator
[: 1: ==: unexpected operator
[: 1: ==: unexpected operator
objs/kernel/kernel.o: In function `handle_init':
kernel.c:(.text.handle_init+0x174): undefined reference to `boot_message'
objs/kernel/kernel.o: In function `main':
kernel.c:(.text.main+0x299): undefined reference to `TTF_Init'
kernel.c:(.text.main+0x2c9): undefined reference to `boot_message'
objs/kernel/render.o: In function `prep_bgnds':
render.c:(.text.prep_bgnds+0x237): undefined reference to `text_get_xy'
render.c:(.text.prep_bgnds+0x282): undefined reference to `global_font'
render.c:(.text.prep_bgnds+0x28b): undefined reference to `boot_msg_width'
objs/kernel/render.o: In function `render_objs':
render.c:(.text.render_objs+0x320): undefined reference to `TTF_Render'
render.c:(.text.render_objs+0x377): undefined reference to `boot_message'
render.c:(.text.render_objs+0x39a): undefined reference to `global_font'
render.c:(.text.render_objs+0x3a2): undefined reference to `boot_msg_width'
render.c:(.text.render_objs+0x3bc): undefined reference to `TTF_Render'
render.c:(.text.render_objs+0x3c6): undefined reference to `boot_message'
render.c:(.text.render_objs+0x3f4): undefined reference to `global_font'
render.c:(.text.render_objs+0x3fc): undefined reference to `boot_msg_width'
render.c:(.text.render_objs+0x414): undefined reference to `TTF_Render'
objs/kernel/common.o: In function `do_getpic':
common.c:(.text.do_getpic+0x36): undefined reference to `load_fonts'
collect2: ld returned 1 exit status
make[1]: *** [objs/splash_helper] Error 1
make[1]: Leaving directory `/home/laga/dev/gar-minimyth/minimyth/gar-minimyth/script/system/splashutils/work/main.d/splashutils-1.3'
make: *** [build-work/main.d/splashutils-1.3/Makefile] Fehler 2
laga@prometheus:~/dev/gar-minimyth/minimyth/gar-minimyth/script/system/splashutils$     
To answer your question: Yes, freetype has built a static library:

Code: Select all

-rw-r--r-- 1 laga laga 462130 2007-01-21 15:29 images/main/usr/lib/libfreetype.a
Pablo
Site Admin
Posts: 4182
Joined: Tue Dec 14, 2004 2:13 am
Location: La Jolla
Contact:

Post by Pablo »

For rxvt, if you change

Code: Select all

--cache-file=config.cache
to

Code: Select all

--cache-file=./config.cache
in 'Makefile', then does it work?
Unlike other configure scripts, it appears that rxvt's configure script does not add the './'.

For splashutils, after running 'make clean; make patch', if you change

Code: Select all

config_opt = $(shell if [[ -e config.h && -n "`egrep '^\#define[[:space:]]+$(1)([[:space:]]+|$$)' config.h`" ]]; 
to

Code: Select all

config_opt = $(shell if [ -e config.h ] && [ -n "`egrep '^\#define[[:space:]]+$(1)([[:space:]]+|$$)' config.h`" ] ; 
in 'work/main.d/splashutils-1.3/Makefile', then does it work?
MiniMyth running on an Acer ApireRevo 3610 and a Zotac ZBOX-ID80-U. Find out more at my MythTV page.
juski
Contributor
Posts: 127
Joined: Sun Jun 26, 2005 10:14 pm
Location: Manchester, UK

Post by juski »

Just a little chip-in from me....

I think there's not much point even including bootsplash just yet because there's no way to initialise the fs454 tv encoder until the IEGD loads. I remember reading something about this on the zenega-user forums so maybe somebody found a workaround (a hack to write to the i2c ports to initialise the fs454 chip).
The words 'digital' and 'quality' are not synonymous
laga
Member
Posts: 11
Joined: Sat Jan 20, 2007 4:48 pm

Post by laga »

Pablo wrote:For rxvt, if you change

Code: Select all

--cache-file=config.cache
to

Code: Select all

--cache-file=./config.cache
in 'Makefile', then does it work?
Unlike other configure scripts, it appears that rxvt's configure script does not add the './'.
Yep! It's working:

Code: Select all

make[1]: Leaving directory `/home/laga/dev/gar-minimyth/minimyth/gar-minimyth/script/X11/rxvt'
 ==> Running configure in work/main.d/rxvt-2.6.4
loading cache ./config.cache
For splashutils, after running 'make clean; make patch', if you change

Code: Select all

config_opt = $(shell if [[ -e config.h && -n "`egrep '^\#define[[:space:]]+$(1)([[:space:]]+|$$)' config.h`" ]]; 
to

Code: Select all

config_opt = $(shell if [ -e config.h ] && [ -n "`egrep '^\#define[[:space:]]+$(1)([[:space:]]+|$$)' config.h`" ] ; 
in 'work/main.d/splashutils-1.3/Makefile', then does it work?
Yes! It compiles just fine :) I still get those weird errors, though. Is this another incompatibility between bash and sh? I still don't get it why I'm seeing this while you're (obviously) not. Isn't the build system supposed to provide a consistent environment?

Code: Select all

==> Running configure in work/main.d/splashutils-1.3
config.h successfully created.
You can run `make` now.
        [configure] complete for splashutils.
 ==> Running make in work/main.d/splashutils-1.3
make[1]: Entering directory `/home/laga/dev/gar-minimyth/minimyth/gar-minimyth/script/system/splashutils/work/main.d/splashutils-1.3'
[: 1: ==: unexpected operator
[: 1: ==: unexpected operator
[: 1: ==: unexpected operator
[: 1: ==: unexpected operator
[: 1: ==: unexpected operator
[: 1: ==: unexpected operator
[: 1: ==: unexpected operator
[: 1: ==: unexpected operator
[: 1: ==: unexpected operator
[: 1: ==: unexpected operator
[: 1: ==: unexpected operator
[: 1: ==: unexpected operator
[: 1: ==: unexpected operator
[: 1: ==: unexpected operator
[: 1: ==: unexpected operator
[: 1: ==: unexpected operator
[: 1: ==: unexpected operator
[: 1: ==: unexpected operator
[: 1: ==: unexpected operator
[: 1: ==: unexpected operator
[: 1: ==: unexpected operator
[: 1: ==: unexpected operator
[: 1: ==: unexpected operator
make[1]: Leaving directory `/home/laga/dev/gar-minimyth/minimyth/gar-minimyth/script/system/splashutils/work/main.d/splashutils-1.3'
        [build] complete for splashutils.
laga@prometheus:~/dev/gar-minimyth/minimyth/gar-minimyth/script/system/splashutils$     
Thanks for helping me out with those problems, It's much appreciated! I'll keep you posted, but I won't be able to get anything done until next weekend.

Have a nice week,

Michael
laga
Member
Posts: 11
Joined: Sat Jan 20, 2007 4:48 pm

Post by laga »

juski wrote:Just a little chip-in from me....

I think there's not much point even including bootsplash just yet because there's no way to initialise the fs454 tv encoder until the IEGD loads. I remember reading something about this on the zenega-user forums so maybe somebody found a workaround (a hack to write to the i2c ports to initialise the fs454 chip).
Yep. I just thought it'd be smart to stay as close to the real thing, eg MiniMyth, as possible. I can see myself trimming down the FS by leaving out packages for my own purposes, but that's just because because I can. :)
Pablo
Site Admin
Posts: 4182
Joined: Tue Dec 14, 2004 2:13 am
Location: La Jolla
Contact:

Post by Pablo »

laga wrote:Yes! It compiles just fine :) I still get those weird errors, though. Is this another incompatibility between bash and sh?
I have made some changes that I hope will fix this. They are in SVN and in 0.20-22b6 in the test download directory.
I still don't get it why I'm seeing this while you're (obviously) not. Isn't the build system supposed to provide a consistent environment?
Unfortunately, I have not figured out a way to isolate the running build system from the shell in which it is running without making rather significant changes to the GAR build system. In general, this has not been a problem except for with splashutils (there have been problems before). I suspect splashutils has problems because it was developed for a very specific distribution (Gentoo).
MiniMyth running on an Acer ApireRevo 3610 and a Zotac ZBOX-ID80-U. Find out more at my MythTV page.
laga
Member
Posts: 11
Joined: Sat Jan 20, 2007 4:48 pm

Post by laga »

Hello Pablo,

rsdvd made my tinker with minimyth again :)

We've made good progress this weekend. IEGD is working now and we can watch TV on the S100. There's still trouble with the colors being off but we'll get the sorted out eventually.

I'm still keen on merging our efforts into the main minimyth branch. I've just modified our patches to MythTV and to the kernel so they'll only be applied if mm_CHIPSETS contains the string "iegd"

Code: Select all

PATCHFILES += $(if $(filter iegd, $(mm_CHIPSETS)),  $(DISTNAME)-iegd-intelagp.patch)
Would this be acceptable?

Our next tasks:

* make the remote work. It's presented to the system as a normal keyboard with non-standard keycodes. Is LIRC the standard on minimyth or would it be acceptable to use a .xmodmap file? Personally, I'd use inputlirc, a nice & simple app.

* disable the blinking LED located in the front panel. Would you accept an additional init script that'd be coupled to a new minimyth.conf entry? Or would you want to put it into the example minimyth.script because it won't affect many users?

* integrate the IEGD config options into minimyth.conf. This won't be easy, I'll give it some more thought and propose some possible solutions tomorrow.

What do you think?


Have a nice Sunday!

-- Michael
Pablo
Site Admin
Posts: 4182
Joined: Tue Dec 14, 2004 2:13 am
Location: La Jolla
Contact:

Post by Pablo »

laga wrote:We've made good progress this weekend. IEGD is working now and we can watch TV on the S100. There's still trouble with the colors being off but we'll get the sorted out eventually.
My understanding is that some versions of the Xbox use the same TV chip. Do you know whether or not they have made any progress with getting it to work without the IEGD?
I'm still keen on merging our efforts into the main minimyth branch. I've just modified our patches to MythTV and to the kernel so they'll only be applied if mm_CHIPSETS contains the string "iegd"

Code: Select all

PATCHFILES += $(if $(filter iegd, $(mm_CHIPSETS)),  $(DISTNAME)-iegd-intelagp.patch)
Would this be acceptable?
Are these patches incompatible with anything? To date, the mm_CHIPSETS only impacts the assembly of the root file system image. It does not impact the compilation of any packets. I would prefer to keep it that way.
Our next tasks:

* make the remote work. It's presented to the system as a normal keyboard with non-standard keycodes. Is LIRC the standard on minimyth or would it be acceptable to use a .xmodmap file? Personally, I'd use inputlirc, a nice & simple app.
Yes, LIRC is the standard. My understanding is the LIRC can read event devices. Is there some reason not to use LIRC?
* disable the blinking LED located in the front panel. Would you accept an additional init script that'd be coupled to a new minimyth.conf entry? Or would you want to put it into the example minimyth.script because it won't affect many users?
That would be fine.
* integrate the IEGD config options into minimyth.conf. This won't be easy, I'll give it some more thought and propose some possible solutions tomorrow.
That would be fine. The init already does special things for certain hardware (e.g. NVIDIA graphics). The special behavior is keyed off of the output of 'lspci -mn', 'lsusb', 'dmidecode -t baseboard' and 'dmidecode -t processor'. Could you provide this information for the S100?
MiniMyth running on an Acer ApireRevo 3610 and a Zotac ZBOX-ID80-U. Find out more at my MythTV page.
laga
Member
Posts: 11
Joined: Sat Jan 20, 2007 4:48 pm

Post by laga »

Hello Pablo,

sorry for not answering earlier. Real life has kept me busy 8)
Pablo wrote: My understanding is that some versions of the Xbox use the same TV chip. Do you know whether or not they have made any progress with getting it to work without the IEGD?
Someone managed to talk to the fs454 over the i2c bus. They did enable the tv-out, but IEGD provides way more options. I think they just enabled the most basic things, i.e. composite output. You won't be able to easily adjust X and Y offsets, for example. A free solution would be nice, but it's not really usable yet and probably will never be. I don't think people are still working on it.
Are these patches incompatible with anything?
To be honest, I don't know. The patch for MythTV just changes which colorspace is preferred, IMHO. It shouldn't break anything.
However, the Intel agp kernel module needs to be patched as well. That patch seems like an addition, but I'd have to read it again to understand what it really does. For now, I'll give it the benefit of the doubt and state that it doesn't break anything :)
To date, the mm_CHIPSETS only impacts the assembly of the root file system image. It does not impact the compilation of any packets. I would prefer to keep it that way.
Yes, that's sensible. Shall we introduce an mm_ENABLE_IEGD_HACKS variable in minimyth.conf.mk? It's an annoyingly messy situation :(
Yes, LIRC is the standard. My understanding is the LIRC can read event devices. Is there some reason not to use LIRC?
No, LIRC will be fine. I just like the simplicity of inputlircd, but minimyth doesn't really need two lircds I suppose :)
laga wrote:* disable the blinking LED located in the front panel. Would you accept an additional init script that'd be coupled to a new minimyth.conf entry? Or would you want to put it into the example minimyth.script because it won't affect many users?
That would be fine.
Which one of the two choices would be fine?
laga wrote:* integrate the IEGD config options into minimyth.conf. This won't be easy, I'll give it some more thought and propose some possible solutions tomorrow.
I'll have to give it some more thought, my week was quite busy.
That would be fine. The init already does special things for certain hardware (e.g. NVIDIA graphics). The special behavior is keyed off of the output of 'lspci -mn', 'lsusb', 'dmidecode -t baseboard' and 'dmidecode -t processor'. Could you provide this information for the S100?
Yes. I'll give it to you tomorrow, the S100 gets to rest today :)

Thanks for your reply!

-- Michael
Pablo
Site Admin
Posts: 4182
Joined: Tue Dec 14, 2004 2:13 am
Location: La Jolla
Contact:

Post by Pablo »

laga wrote:
Pablo wrote: Are these patches incompatible with anything?
To be honest, I don't know. The patch for MythTV just changes which colorspace is preferred, IMHO. It shouldn't break anything.
However, the Intel agp kernel module needs to be patched as well. That patch seems like an addition, but I'd have to read it again to understand what it really does. For now, I'll give it the benefit of the doubt and state that it doesn't break anything :)
In that case, I think that it would be easier to apply the patches all the time. If you point me to the patches, then I will add them.

By the way, can the preferred colorspace be changed without patching MythTV?
To date, the mm_CHIPSETS only impacts the assembly of the root file system image. It does not impact the compilation of any packets. I would prefer to keep it that way.
Yes, that's sensible. Shall we introduce an mm_ENABLE_IEGD_HACKS variable in minimyth.conf.mk? It's an annoyingly messy situation :(
I would rather make the patches part of the mainstream (if possible).
laga wrote:* disable the blinking LED located in the front panel. Would you accept an additional init script that'd be coupled to a new minimyth.conf entry? Or would you want to put it into the example minimyth.script because it won't affect many users?
That would be fine.
Which one of the two choices would be fine?
Sorry, that was very unclear. It is fine to make it part of the normal init scripts. Whether or not it runs can be based on the identified hardware.
* integrate the IEGD config options into minimyth.conf. This won't be easy, I'll give it some more thought and propose some possible solutions tomorrow.
In general, I have tried to configure things automatically based on the detected hardware whenever possible. It would be nice to be able to use the existing configuration options along with the detected hardware to configure the binary drivers.
MiniMyth running on an Acer ApireRevo 3610 and a Zotac ZBOX-ID80-U. Find out more at my MythTV page.
Post Reply