IRC log of #cubox of Sun 30 Sep 2012. All times are in CEST < Back to index

01:40 PrickelPit rabeeh, it was 12v. cubox just flashed the internal led for 2 seconds, since then its dead.
01:40 rabeeh PrickelPit: it's probably dead. sorry
01:40 PrickelPit no problem. things happen.
01:41 rabeeh PrickelPit: probably the DC-DC converters blown away
01:41 rabeeh their max voltage is 6-7V
01:41 Punkley_Chillin ouch :(
01:43 Punkley_Chillin todays task new kernel
01:44 rabeeh we had few developers that burned their cubox with wrong cabling.
01:49 cbxbiker61 i have fixed my xbmc patches to build against current xbmc master...it seems to work well on my x86_64 machines...problem is that someone added a java dependancy (seems kinda silly on arm based boxes) so i haven't finished building on cubox yet
01:56 cbxbiker61 i also added an onkyo tx-nr717 receiver to my arsenal, it does network streaming of pandora and spotify as well as others, i have to disassemble my tx-sr608 to fix the remote control receiver circuit so i wanted a replacement in case that repair doesn't go well
02:17 dbsx rabeeh: I have tested your new u-boot and it solves my tftp/nfs hassles
02:17 rabeeh dbsx: great. i'v seen your post.
02:17 rabeeh it's been bugging me long time
02:18 dbsx a good find
02:18 rabeeh well. actually with 1gbps the issue is not there; only with fast ethernet
02:20 dbsx It was a cause of confusion (works for one not for another). Good to see the end of it
02:34 dbsx biker: I used your 3.5.2 cubox patch and have built a headless kernel that I am happy with. I still dont seem to have MV_CESA operating. Do you plan a new cubox kernel anytime soon?
02:35 shesselba_ dbsx: getting cesa into dove is easy
02:35 dbsx tshesselba: tell me
02:35 shesselba_ 3.6 should have support for it already.. lemme look for the patch
02:36 cbxbiker61 i have been using the latest kernel, just haven't put it up, it has the graphics patches that require the newer marvell china userspace drivers
02:37 shesselba_ http://lkml.indiana.edu/hypermail/linux/kernel/1208.0/02052.html
02:39 dbsx biker: I have most of the hdmi nobbled. So I get max memory, the kernel I am using is for "puppet" clients and it replaces sheevaplugs. If you put you kernel up I will test without graphics.
02:39 dbsx thanks shessel
02:40 shesselba_ https://patchwork.kernel.org/patch/1308081/
02:40 shesselba_ this allows you to download the actual patch
02:40 shesselba_ to enable it you ll need to add a call to orion_crypto_init() to dove_init() in arch/arm/mach-dove/common.c
02:43 cbxbiker61 dbsx, 3.5.4 is online
02:43 dbsx thanks biker
02:43 cbxbiker61 yep
02:53 Punkley_Chillin u telling me i've been building 3.5.3 and now 3.5.4 is there :)
02:54 Punkley_Chillin does it have all the patches :)
14:33 jaapio hi
14:34 jaapio I need some help with geexbox 3.0 on my cubox
14:34 jaapio is there someone who can help me with this?
14:40 jaapio for some reason the boot won't work
16:08 shesselba rabeeh: ping
19:36 rabeeh shesselba: pong
19:37 rabeeh jaapio: yes. how can i help?
19:39 shesselba rabeeh: I am wondering how you prefer to store ethaddr in mainline u-boot? The current env maybe will not be compatible with new u-boot
19:39 rabeeh it will get first 3 bytes of a sector
19:39 rabeeh currently u-boot resides in 0-0x00080000 (512KB)
19:40 rabeeh there would be 64KB sector, mostly empty except 6 bytes that will hold the MAC address.
19:40 shesselba ok
19:40 rabeeh notice that the SPI flash being used also supports OTP; but i don't think we should go that far.
19:40 rabeeh what do you think?
19:40 rabeeh ok?
19:41 shesselba not necessary as you can always set ethaddr in u-boot yourself
19:41 shesselba any write protected area is useless then, except for somebody who wipes his eeprom
19:42 shesselba but you could just put a sticker with the mac address on cubox somewhere ;)
19:43 rabeeh that's the plan. mac address on the bottom and a separate sector for the mac address.
19:43 rabeeh that's only since upgrading u-boot (sometimes) wipes out env variables, so we don't really want to loose the MAC address that easily
19:44 rabeeh if the user wishes to change the MAC, he can always do that either via that sector, or runtime in Linux for example (ifconfig)
19:50 shesselba yep
20:09 shesselba rabeeh: in eth phy init, 1000Mbps gets disabled.. correct?
21:21 yawniek_ is the archlinux image hardfloat?
21:39 rabeeh shesselba:
21:39 rabeeh /* Resume AN avdertisement */
21:39 rabeeh mvEthPhyRegWrite(ethphyHalData.phyAddr[ethPortNum],4,0x1e1);
21:39 rabeeh mvEthPhyRegRead(ethphyHalData.phyAddr[ethPortNum],9,®);
21:39 rabeeh // reg |= (BIT8|BIT9); // Enable 1000Mbps
21:39 rabeeh reg &= ~(BIT8|BIT9); // Disable 1000Mbps
21:39 rabeeh mvEthPhyRegWrite(ethphyHalData.phyAddr[ethPortNum],9,reg);
21:39 rabeeh ./board/marvell/mv_hal/eth-phy/mvEthPhy.c
21:39 rabeeh look for // Disable 1000Mbps up
21:39 shesselba yeah, I was asking if you know about that
21:40 rabeeh yawniek: arch is hardfp
21:40 shesselba I have issues when using mvgbe on latest u-boot..
21:40 shesselba it doesn't transmit
21:41 shesselba is 1Gbps broken somehow on dove?
22:01 jaapio rabeeh: someone told me that I had to flash the cubox with a new u-boot
22:01 jaapio but I am not able to do that. For some reason nfs fails
23:12 jaapio finally after some trial and error I found a way to get the new u-boot in the system. And it's up and running again :-D
23:13 jaapio the only way it worked for me was using usb
23:45 rabeeh shesselba: it's exactly the same MAC as in Kirkwood.
23:46 rabeeh I think it is even better by support TCP checksum offload on jumbo frames.
23:46 rabeeh shesselba: can u github your code?