Re: Working bcm2035 (bluetooth)
just an update on this... successfully sent/received a file via obex push to/from a palm zire 72, and successfully established a connection via rfcomm to my notebook. opie locks up when trying to send a file via filemanager, but everything works fine when using obextool push. also, opie never asked me for a pin, so i configured hcid to use security 'auto' instead of 'user'. On 4/2/07, Michael Kramer [EMAIL PROTECTED] wrote: Hi, Ive managed to proper initialize the bcm2035 on my a780. No kernel hacking was necessary, i just removed ezx-bt.patch from my series file. The chip works fine without ever touching the 2 gpio lines(for now). The bcm2035 initialization in bluez hciattach is buggy, and i only managed to attach at 115200 and bdaddr of , so i wrote my own tool, based on a trace of motorola's dload application. For what its worth i have a similar tool (still in rough form) to initialize the bcm2045 on the a1200 (also via stracing the results of the mbtdload app on the a1200) as well as the bcm2035 on the e680/a780 (i'm working on one binary that can detect/handle either) For the a1200 version i can get bluez to start (e.g. hciconfig looks good and 'hcitool scan' works fine) but as soon as i make an interesting connection (rfcomm, bnep, etc) the phone reboots so i'm not sure if its my hardware or something funny going on w/ the se kernel on the a1200. i've been trying to build my own openezx image and rootfs; the image seems to build fine but i cannot seem to get 'bitbake gpe-image' or 'opie-image' to build successfully; can anyone tell me what rootfs image they are building that works w/ the resulting zimage? thanks, mike -- EOF Daniel Ribeiro
Re: Working bcm2035 (bluetooth)
2007/4/1, Stefan Schmidt [EMAIL PROTECTED]: Hmm, ok. Somehow I doubt we don't need them. But it is nice to see it working without. Perhaps we need to use them later anyway. I am sure we will need at least one of them when working on power management.. Of course the end goal is to fix it in hciattach if it is not EZX specific or to ugly. Anyway good work. :) I dont think its EZX specific... And it will be really easy to fix hciattach. From your README I see that it also works with standard bcm firmware. I would suggest to work with this one instead of do unknown patching to it. The motorola patches are not even included on the .tgz.. :) I am trying to determine what it does differently than the original FW, any news on this and i will report here.. Marcel, any comments about the hciattach problem? I think Daniel would be glad to describe the problem further if you have questions about it. The goal is to have BT working ou-of-the-box with 2.6 kernel and hopefully standard bluez-utils. Well.. i can submit a patch to hciattach if requested. Its just a wrong argument to the 'change baud' command and lacking command to set bdaddr on the bcm2035 init sequence. -- EOF Daniel Ribeiro
Re: Working bcm2035 (bluetooth)
2007/4/2, Michael Kramer [EMAIL PROTECTED]: For what its worth i have a similar tool (still in rough form) to initialize the bcm2045 on the a1200 (also via stracing the results of the mbtdload app on the a1200) as well as the bcm2035 on the e680/a780 (i'm working on one binary that can detect/handle either) Any differences on the FW upload protocol??? The BDADDR is also set with a command on the init sequence?? Please tell me everything! And share your code asap! :) For the a1200 version i can get bluez to start (e.g. hciconfig looks good and 'hcitool scan' works fine) but as soon as i make an interesting connection (rfcomm, bnep, etc) the phone reboots so i'm not sure if its my hardware or something funny going on w/ the se kernel on the a1200. Humm... i have not tried anything else than hcitool scan. But i think that in your case its the a1200 kernel.. I will try a connection and report on the list. i've been trying to build my own openezx image and rootfs; the image seems to build fine but i cannot seem to get 'bitbake gpe-image' or 'opie-image' to build successfully; can anyone tell me what rootfs image they are building that works w/ the resulting zimage? OE stops on which package?? Please give us more info so we can help you. For me it works with both, gpe and opie. -- EOF Daniel Ribeiro
Re: Working bcm2035 (bluetooth)
Hello. [cc'ed Marcel Holtmann as he is bluez maintainer and also have an A780] On Sun, 2007-04-01 at 14:33, Daniel Ribeiro wrote: Ive managed to proper initialize the bcm2035 on my a780. Wow, great. :) No kernel hacking was necessary, i just removed ezx-bt.patch from my series file. The chip works fine without ever touching the 2 gpio lines(for now). Hmm, ok. Somehow I doubt we don't need them. But it is nice to see it working without. Perhaps we need to use them later anyway. The bcm2035 initialization in bluez hciattach is buggy, and i only managed to attach at 115200 and bdaddr of , so i wrote my own tool, based on a trace of motorola's dload application. Of course the end goal is to fix it in hciattach if it is not EZX specific or to ugly. Anyway good work. :) bcm2035-tool can properly set the bdaddr, change baudrare and patch the firmware of the bcm2035 connected to BTUART on EZX phones. From your README I see that it also works with standard bcm firmware. I would suggest to work with this one instead of do unknown patching to it. You can download it from http://people.openezx.org/wyrm/bcm2035-tool.tgz Marcel, any comments about the hciattach problem? I think Daniel would be glad to describe the problem further if you have questions about it. The goal is to have BT working ou-of-the-box with 2.6 kernel and hopefully standard bluez-utils. regards Stefan Schmidt signature.asc Description: Digital signature
