| |
| |
| |
|
Page: 1 2 3 4
Comments:
<0> I'm building the kernel in a separate directory from the source, the build scripts ***uming this is not the case
<0> include/asm-generic does not exist in the build directory, only the source
<0> I suppose I can override kernel_do_stage, what do you think, does this make sense?
<0> currently all it's references appear to be ***uming the copy will happen from the build directory, if I'm reading this right
<1> cu
<2> good morning all
<3> hi koen
<2> hey Jin^eLD
<3> trying to get a MIPS based board to work in oe.dev.. turned out to be not an easy task :>
<2> mipsel, mipseb?
<3> mipsel
<3> I talked to tmbinc and merged the glibc that is used in the dreambox branch into my dev tree
<3> there's still some weirdness though, gcc 3.4.5 is needed co compile glibc-intermediate, but then for the rest gcc 4.1.1 is being used
<3> anyway, lets see how far I get
<4> Koen : I will be in your neighboorhoud in 2 weeks
<2> gerwinin: cool
<2> you have my phonenumber :)
<5> Jin^eLD : I have build images for the rb5xx boards
<4> Koen: I have you seen likewise recently ?
<2> last tuesday
<4> Koen: I fixed my problem , some rotten memory module
<5> koen : hi !
<2> hey Ifaistos
<6> koen org.oe.dev * rd3de105f... / (1 packages/dates/dates_0.3.bb): dates: add 0.3
<3> Ifaistos: using oe.dev branch?
<3> Ifaistos: what distro did you take as a starting point, which glibc/gcc versions did you use?
<3> just setting the machine appropriately did not work for me, I could only cleanly build the dreambox branch for mipsel
<7> http://bugs.openembedded.org/show_bug.cgi?id=1842 <--- anyone grab a look for me?
<5> Jin^eLD : I have build my own distro, although it's using the 'old' OE stuff. check http://owmnr.digital-opsis.com
<3> Ifaistos: thanks, I'll take a look
<5> Jin^eLD : hopefully will get it up to speed withing the next millenium :)
<3> I'd like to keep in touch with oe.dev though, thats why I am trying to merge stuff from dreambox
<3> :)
<5> Jin^eLD : my routerboard turned into a door-stopper had to return it to mtik
<0> cbrake:fixed my problem by copying over include files from ${S} to ${B} before staging, thx for help
<3> basically I got it to compile in .dev now, but I messed something up with threads, my system is panic'ing with "cannot set up thread-local storage", I guess it might be related to nptl stuff which dreambox glibc is configured for
<8> natezg: dunno. OE is kind of like Ruby on Rails. If you do things the OE way, it is really easy. If not, well ...
<3> Ifaistos: heh...
<5> Jin^eLD : the mips cpu they are using has some 'locking' mechanism in the bootloader
<8> natezg: no problem -- glad its working :-)
<3> Ifaistos: I am trying to OE the DeCypher 8100 board
<0> cbrake:well....built the new kernel now for testing :-O
<5> Jin^eLD : are the new db boxes using this ? dm7020 or something ?
<0> koen:FYI edb9302a board support is in the mainstream kernel as of the latest patch, thx for pointing me that way
<3> Ifaistos: no idea.. I do not think so.. I only took the dreambox branch because zecke suggested to look at it - since dreambox is also some sort of mipsel based stuff
<5> Jin^eLD : basicaly is ppc, mipsel is only one machine i think
<3> ah.. I see
<5> Jin^eLD : some ibm 405 variant with mpeg decoding
<3> aha.. the 8100 has hidef decoders and stuff like that
<5> Jin^eLD : it should :) These kind of cpu's would not be able to handle these kind of stuff....
<3> yep :)
<5> Jin^eLD : even for voip,echo cancelation i am trying to use them anything below 600Mhz is a waste of time
<3> heh.. well, my target app is video playback using the hw decoders, so that's ok I guess :>
<3> my problem was that it just did not work in oe.dev, most people seem to work on ARM based stuff which is really well supported
<5> Jin^eLD : do you have access to the specs of the decoder or do they give you a binary blob just as it was the case with dreambox
<5> Jin^eLD : Yeah here almost everyone is ARMed to the teeth :)
<3> there is some binary firmware image, I am not yet sure about controlling playback but there are some sources of demo apps provided
<3> http://www.micronas.com/products/by_function/decypher_dhm8100a/product_information/index.html
<5> Jin^eLD : mipsel should work without much trouble i think
<3> the supplied SDK is difficult to work with though, based on RedHat 8 / RedHat 9 for building, uses old kernel, actually a release candidate that only compiles with gcc 3.3.4 and so on
<3> Ifaistos: well I tried, mipsel did not work out of the box in current dev :) that was my first attempt
<3> dreambox branch has some mipsel patches for glibc and uses a different glibc version
<3> and for some reason they need gcc-cross-initial = "3.4.5" (I think else glibc is not building) and then use gcc-cross = "4.1.1"
<5> Jin^eLD : meshcube is mipsel
<5> Jin^eLD : mtx-1
<3> well, has anyone tried building it recently? :)
<3> which distro does the mtx-1 use?
<5> Jin^eLD : i think there were some updates pushed lately...
<5> Jin^eLD : nylon
<3> I tried building for a mipsel machine yesterday or wednesday or so, so pretty recently
<3> nylon was my starting point, but it does not work
<5> Jin^eLD : where is the problem ? glibc ?
<3> I would have to retry to give a definite answer, also the nylon stuff uses quite old gcc versions and so on; after it did not work I was pointed to the dreambox branch which uses more recent stuff and builds
<3> probably good idea to try again with original oe.dev things and see what exactly the problem was
<3> I just wanted to have something working and tested, that's why I took dreambox
<3> I don't feel confident hacking around the glibc :)
<5> Jin^eLD : sometimes you need to hunt around for patches on certain arch's...
<5> Jin^eLD : btw are you sure the cpu is mipsel ? it says mips32 24KE....
<3> the supplied toolchain is: mipsel-linux-gnu/gcc-3.3.4-glibc-2.3.2
<3> moment phone
<3> re
<3> Ifaistos: hmm.. well how can I find that out for sure?
<3> allthough I am pretty sure its mipsel..
<5> Jin^eLD : not sure... did they supply a pre-build toolchain ?
<3> yes, the prebuild toolchain says mipsel-linux-gcc-3.3.4 (GCC) 3.3.4
<3> so I think its mipsel after lal
<3> all
<3> well. let's see if I can bring it up with glibc compiled from the dreambox branch... if I do I can investigate a little why oe.dev defaults did not work
<5> Jin^eLD : have a look at the versions they are using i.e gcc,libc and try to build them using OE
<5> Jin^eLD : also check what patches they are applying
<3> Ifaistos: I tried to build gcc 3.3.4 in OE but it somehow failed, glibc 2.3.2 is quite old too..
<3> hmm, i only have the prebuilt toolchain, no idea about any patches they might have applied to glibc/gcc
<3> I can probably ask for it
<5> Jin^eLD : You can also use the pre-build toolchain with OE to build the rest
<3> I know.. but I wanted to avoid that
<5> Jin^eLD : not much fun though :)
<3> not really :) well, I have to build the kernel using their gcc, it's the only one that works
<3> but I really wanted to have the rest from OE
<3> since it gets packaged nicely and so on
<3> also the glibc
<5> Jin^eLD : copying is the best form of flattery :) try to build the same stuff to gain experience on the architecture.... then you can try to get newer versions
<3> well, I could not compile the same versions in OE, I did not investigate it further, it just did not build "out of the box"
<3> but indeed, if my current attempt fails I will have to start over
<5> Jin^eLD : mastering OE + a new arch is a big task... been there done that ;0
<3> :>
<3> hmm crap, still the thread problem
<3> I think I'll disable nptl :)
<9> anyone thought about further breaking down the site files beyond arm-linux and endian and stuff? there are obviously a lot of tests which are just os speciifc, not architecture, adn vise versa
<7> http://bugs.openembedded.org/show_bug.cgi?id=1842 <--- anyone grab a look for me?
<10> I made a new espgs package
<10> I'll post it in the bugtracker
<10> I'm mostly sure it works
<10> And an Exult package, but that needs some more testing
<10> But I built an ipk allright
<3> any ideas what "cannot set up thread-local storage: unknown error" could be?
<3> is my glibc bad or is it just some misconfiguration of boot options?
<3> ...still fighting with mipsel..
<2> kergoth: there should be a 'common' sitefile somewhere in OE
<2> B_Lizzard: isn't 'S = "${WORKDIR}/espgs-${PV}"" a bit redundant?
<10> Nope
<10> Um, yes
<10> Now that I think of it :)
<10> It was in the last .bb file, I DON'T KNOW :D
<2> B_Lizzard: have a look at the abiword 2.5 recipes on how to get rid of hardcoding 8.15
<10> Will do
<10> By the way, exult requires the host CPU to run some binary it makes, how do I get past that?
<2> with some goats, hemlock and a silver bullet
<10> Aha
<10> Damn, stores are closed
<2> you're in luck, it's full moon tonight
<10> My local satanist group are on a road trip
<10> I'll ask Stan directly
<10> I HATE IT WHEN PEOPLE DO THAT
<10> And Abiword has something like that
<10> Having to run some binaries
<2> only a perl script
<10> At least someone made a perl script
<10> Normally, you'd have to run a binary
<2> the abiword people changed it after I they attended my fosdem presentation that mention abiword being broken :)
<2> s/I they/they/
<10> I've wasted weeks trying to build Abiword 1.0.7
<10> :)
<10> Only to get cyclic "undefined reference" errors between wv and abiword
<10> Bastards
<10> Couldn't it just be cut down with "cut"?
<10> Or does it HAVE to be in python?
<11> B_Lizzard: libx11_X11R7.1-1.0.1.bb builds a binary that runs on the build host (manually, before executing oe_runmake)
<10> Aha
<10> I'll look into that
<10> Why isn't FVWM in OE?
Return to
#oe
or
Go to some related
logs:
+hula +fedora +x86_64
#qemu
python24 lowercase
pbot.pl script
#linux
dapper + module nvidia not found
backuppc Xlib: connection to
wifi-radar connected ubuntu
#debian
#ubuntu
|
|