| 13:05 | rabeeh | vpeter: about the eMMC errors - do you have a 4GByte eMMC? |
| 13:05 | rabeeh | we saw this with the 4GB eMMC and once we moved to 8GB B041 from Samsung those errors disappeared. |
| 13:06 | rabeeh | vpeter: refer to this - https://wiki.solid-run.com/doku.php?id=products:ibx:microsom:emmc |
| 16:07 | vpeter | rabeeh: Yes, it is 4 GB. |
| 16:09 | vpeter | What is mean with storage only? If I would boot from sd card then messages would disappear? |
| 16:10 | vpeter | I'm running libreelec from it. With kernel 4.4.13 and 4.10.14 - no difference. |
| 16:15 | vpeter | And what those messages mean to data on emmc? Are they safe or not? |
| 17:49 | vpeter | Seems booting from SD card doesn't help with those issues for me :( |
| 18:27 | rabeeh | vpeter: yeah; sorry. those are early drops of the hardware |
| 18:28 | rabeeh | i can't recall exactly what the problem was; Malte will probably |
| 18:28 | rabeeh | power management related stuff |
| 18:56 | vpeter | I'm testing image with newer kernel. Maybe will be better. If not I will just use SD card. |
| 18:56 | vpeter | But when I was adding CEC stuff to libreelec I think I didn't noticed this issues. But maybe Ijust missed them. |
| 18:56 | rabeeh | vpeter: what do you think about the CEC stuff? |
| 18:57 | rabeeh | vpeter: once malte is online; i'll ask him to update the hardware |
| 18:57 | rabeeh | really; the 4GB was a piece of crap (TM) |
| 18:57 | rabeeh | the 8GB B041 rev can do > 110MB/sec read |
| 18:57 | vpeter | When Gerald was making firmware for handling cec I made images for testing libreelec. |
| 18:58 | rabeeh | and we had very limited build of it; so the majority are 8GB |
| 18:58 | rabeeh | vpeter: i tried that image |
| 18:58 | vpeter | I didn't use it much till now. |
| 18:58 | rabeeh | which TV? |
| 18:58 | vpeter | Samsung |
| 19:02 | rabeeh | hope you noticed the wake-on-usb |
| 19:03 | rabeeh | i haven't tried it yet; but you should be able to wake the machine from the MCU |
| 19:03 | rabeeh | via wake-on-usb |
| 19:03 | vpeter | This issue is still open. |
| 19:03 | vpeter | It worked for me but not 100% (I think). |
| 19:03 | vpeter | Forgot a litle - there were only 3 users who were using this :-) |
| 19:04 | vpeter | And one use fried his solidpc :( |
| 19:06 | vpeter | rabeeh: Can solidpc work safely without fan? |
| 19:06 | rabeeh | define safely |
| 19:07 | rabeeh | phyiscal safely? |
| 19:07 | vpeter | yes. Can it overheat? |
| 19:08 | rabeeh | any modern processor from Intel for the last 5 years or more has built-in speedstep tech |
| 19:08 | vpeter | I had cputemp little over 60 degrees. |
| 19:08 | rabeeh | so when the die junction reaches 90c in Braswell; it will slow down |
| 19:08 | rabeeh | 60 is very cool |
| 19:09 | vpeter | I'm sure I would notice 90 degrees. With smell :-) |
| 19:09 | rabeeh | those measurements are from the die junction; so my guess it would be 45c on the aluminium |
| 19:09 | rabeeh | no; 90c can't be noticed too |
| 19:09 | vpeter | Just like with some other system 2 weeks back when white smoke come from it :D |
| 19:09 | rabeeh | really; this is processor junction; not even package temperatures |
| 19:10 | rabeeh | come on; the white smoke was something else you are burning dear :) |
| 19:10 | bencoh | :] |
| 19:10 | vpeter | Nope, it was one capacitor in short. |
| 19:10 | rabeeh | you had a short? |
| 19:10 | rabeeh | really? |
| 19:10 | rabeeh | have a picture? |
| 19:11 | rabeeh | i can run 45Mbps h.264 when running a single channel DDR without even a heatsink |
| 19:11 | vpeter | Great. |
| 19:11 | rabeeh | that was your img drop with the CEC support |
| 19:12 | vpeter | Now I'm not worried anymore :) |
| 19:12 | rabeeh | if you look at the "allowed" temperatures; Intel Braswell can go up to 90c |
| 19:13 | rabeeh | i.MX6 can go 105 |
| 19:13 | rabeeh | Marvell Armada 105 and there are skus for even higher |
| 19:13 | vpeter | Really? This is dangerous to touch. |
| 19:13 | rabeeh | those temperatures becomes normal in industrial kind of deployments when the test being done is with 60c ambient |
| 19:13 | rabeeh | vpeter: depends on which package |
| 19:14 | rabeeh | ceramic package will feel really hot |
| 19:14 | rabeeh | bare die on 105c and beyond will get you white skin |
| 19:14 | vpeter | true |
| 19:14 | rabeeh | nothing really dangerous; just annoying white skin |
| 19:14 | vpeter | Will not try. |
| 19:15 | rabeeh | ofcourse for the long term you can't really run a processor on max speed for multiple years |
| 19:15 | vpeter | Once in the life was enough (whole hand). As a child. |
| 19:17 | vpeter | rabeeh: Can I ask something off topic? Do you know any free disassembler which would show function names and other strings on a shared library? |
| 19:17 | rabeeh | objdump -lhad ? |
| 19:18 | rabeeh | or if it's cross then - 'arm-linux-gnueabi-objdump -lhad ' |
| 19:18 | rabeeh | needless to say 'arm-linux-gnueabi-' is an example of cross toolchain prefix |
| 19:19 | vpeter | Yes, cross. And this commans shows something. I'm trying to remove few system calls from it. |
| 19:21 | rabeeh | oh; so you want to actually edit the binary? |
| 19:21 | vpeter | Yes. |
| 19:22 | rabeeh | ok |
| 19:22 | vpeter | Long time back I was using IDA Pro. |
| 19:22 | rabeeh | there are two ways to do it - |
| 19:22 | rabeeh | 1. objdump + locate binary in ELF file + hexedit and modify it |
| 19:23 | rabeeh | 2. some old project that i once used that can edit ELF binaries |
| 19:23 | rabeeh | i really can't even locate the second one; i'll search more |
| 19:23 | rabeeh | but for the first it's really straightforward. for example - |
| 19:23 | vpeter | It is proprietary library. I doubt I can do it. |
| 19:24 | bencoh | hte has support for elf edition iirc (but not for arm) |
| 19:25 | rabeeh | hte? |
| 19:26 | bencoh | an opensource hexeditor with elf support |
| 19:26 | rabeeh | just brute force it |
| 19:26 | rabeeh | objdump - locate a string of hex that you can locate the string inside the actual binary file |
| 19:27 | rabeeh | and then replace it with whatever you need |
| 19:27 | rabeeh | for example - following is disassembly of some arm binary - |
| 19:27 | rabeeh | 00008114 <_init>: |
| 19:27 | rabeeh | _init(): |
| 19:27 | rabeeh | 8114: e92d4008 push {r3, lr} |
| 19:27 | rabeeh | 8118: eb000265 bl 8ab4 |
| 19:27 | rabeeh | 811c: e8bd8008 pop {r3, pc} |
| 19:28 | rabeeh | hexedit the binary and locate 'e92d4008eb000265'; once found this is where the 0x8114 is located inside the elf binary; you can replace it with 'nop' for instance |
| 19:29 | vpeter | Seems easy ... for someone :-) |
| 19:32 | vpeter | The library in question is package imx-gpu-viv 5.0.11.p7.4-hfp. Seems it sets signal handler for ALL signal numbers (SIGTERM, SIGUSR1, ...). Which means main app can't handle them :( |