| 05:41 | Juggie | dxtr, which version of xbmc 12 or 13? |
| 08:37 | jnettlet | rabeeh, when you get a chance can you look at the memory init in my patch. I am getting random hangs and it is very all over the place so my guess is the DDR is not being initialized properly. |
| 10:59 | jnettlet | hmm the emi slow clock is not running at 22Mhz under the new u-boot. |
| 11:28 | jnettlet | Oh that is for NOR or NAND. I guess it just needs to be gated off. |
| 13:59 | jnettlet | aha. excellent. gcc 4.8 and u-boot is bad. You have to set -mno-unaligned-access to get things to not blow up amazingly. |
| 14:02 | jnettlet | and the carrier one is refusing to let go of my cat-5 cable |
| 14:05 | jnettlet | rabeeh, the rj-45 connector is really stuck in the ethernet adapter. Seen this before? |
| 14:06 | jnettlet | phew got it out |
| 14:34 | dv_ | jnettlet: is this related to the gcc 4.8 kernel problems? |
| 14:37 | jnettlet | dv_, it could be possible. Theoretically the kernel is supposed to handle unaligned accesses since 2.6.28 |
| 14:37 | jnettlet | I am going to test with gcc 4.7 and see where it gets me. |
| 14:40 | dv_ | jnettlet: https://lists.yoctoproject.org/pipermail/meta-freescale/2013-July/003534.html |
| 14:40 | dv_ | that was biting me in the arse while trying to get the OE-built kernel to work |
| 14:42 | dxtr | Juggie: the one in geexbox-hardfp-devel :) |
| 14:48 | _rmk_ | jnettlet: erm... ARMv7 is supposed to handle unaligned accesses in hardware except for some very few corner cases |
| 14:48 | _rmk_ | jnettlet: you need the U bit set and A bit clear in the control register |
| 14:49 | dv_ | _rmk_: do you share rabeeh's preference for hardfp? |
| 14:50 | _rmk_ | dv_: well, as I understand it, most stuff is moving towards hardfp. I think ubuntu will be dropping softfp at some point |
| 14:51 | jnettlet | hardfp is definitely the way to go |
| 14:52 | jnettlet | _rmk_, yeah I know but there appears to be a problem somewhere. |
| 14:52 | dv_ | do you have some figures to show that hardfp does have significant gains? |
| 14:54 | jnettlet | dv_, we did early testing but I would have to dig for the numbers. Depending what you were doing we saw between 10 and 20% improvements. |
| 14:55 | jnettlet | I think Linaro did a big study as well which pretty much matched our findings. |
| 14:59 | jnettlet | and I am back |
| 14:59 | jnettlet | https://wiki.linaro.org/OfficeofCTO/HardFloat/Benchmarks201205 |
| 15:05 | jnettlet | and you would think we would be safe but. |
| 15:05 | jnettlet | RAM Configuration: |
| 15:05 | jnettlet | Bank #0: 10000000 512 MiB |
| 15:05 | jnettlet | relocation Offset is: 18783000 |
| 15:05 | jnettlet | data abort |
| 15:05 | jnettlet | MAYBE you should read doc/README.arm-unaligned-accesses |
| 15:05 | jnettlet | pc : [<178013e4>] lr : [<46dc62b4>] |
| 15:05 | jnettlet | sp : 18783000 ip : 00000000 fp : e59ff014 |
| 15:05 | jnettlet | r10: ea000013 r9 : 2f580f38 r8 : 00000000 |
| 15:05 | jnettlet | r7 : 00000094 r6 : 020d8000 r5 : 00901ddc r4 : e8891002 |
| 15:05 | jnettlet | r3 : 178165b0 r2 : 17836b30 r1 : 17800008 r0 : 00091002 |
| 15:05 | jnettlet | Flags: Nzcv IRQs off FIQs off Mode SVC_32 |
| 15:05 | jnettlet | Resetting CPU ... |
| 15:07 | _rmk_ | and the faulting instruction was... |
| 15:09 | jnettlet | in this case a memset I believe. It was bouncing around. |
| 15:10 | jnettlet | This happens when it is trying to load the environment in memory |
| 15:10 | dv_ | jnettlet: just like in the link I posted? |
| 15:11 | jnettlet | dv_, uboot already has that patch |
| 15:22 | jnettlet | _rmk_, did you hear back about the MMC hotplug gpio issue? |
| 15:26 | _rmk_ | not yet |
| 15:27 | _rmk_ | I get the impression that the people who know aren't around at the moment - they haven't been posting to the lists recently |
| 15:34 | jnettlet | hmmm gcc 4.7 doesn't cause the data abort but hangs |
| 15:36 | jnettle | 15:36 * jnettlet wishes for JTAG |
| 15:42 | bencoh | :) |
| 17:10 | jnettle | 17:10 * jnettlet has decided what to do for his first Cubox-i mod. Replace the top of the cover with a capactive touchpad and make the two front feet buttons. Turn the entire cubox into a touchpad interface :-P |
| 20:41 | _rmk_ | jnettlet: I thought you might want it to have a gesture interface too |
| 21:46 | purch | which usb-to-uart converter should I get for C1? |
| 21:46 | purch | all work? |
| 21:50 | cbxbiker61 | http://www.amazon.com/gp/product/B008AGDTA4/ref=oh_details_o03_s00_i00?ie=UTF8&psc=1&tag=thelinactsho-20 |
| 21:50 | cbxbiker61 | works perfectly |
| 21:51 | purch | looks nice and w/o any board |
| 21:52 | cbxbiker61 | pulls power from usb side, so don't connect the +v wire |
| 22:01 | purch | aaargh, amazon and 25 $ limit |
| 22:06 | ojn | purch: http://www.adafruit.com/products/954 |
| 22:07 | ojn | same product, higher price, no limit. |
| 22:15 | purch | http://www.miniinthebox.com/pl2303hx-converter-usb-to-ttl-usb-to-com-cable-module-black-1m_p421817.html |
| 22:16 | purch | 3,99$ |
| 22:34 | dv_ | purch: make sure it uses TTL levels |
| 22:34 | dv_ | the one cbxbiker61 links to works fine |
| 22:40 | purch | and how to make sure? |
| 22:40 | jnettlet | _rmk_, now I am wondering if a leap motion controller on the top would be better. Although I think that would stick out the sides. |
| 22:40 | jnettlet | but then you could controll everything through the eair |
| 22:41 | cbxbiker61 | purch, if it works with 3.3v and 5v, it's fine |
| 22:46 | purch | well, Other Function cables: Red cable:VCC; Black cable: GND; Green cable: TXD; White cable: RXD |
| 22:47 | purch | so I connect black, green and white ( and cross fingers) |