| 07:28 | sanjeevs | Hello All |
| 07:28 | sanjeevs | Do we have iomux mapping of imx6Som i4pro |
| 07:43 | sanjeevs | Any help would be appreciated |
| 07:46 | rabeeh_ | sanjeevs: no we don't have iomux |
| 07:46 | sanjeevs | hmmm |
| 07:46 | rabeeh_ | it was all developed directly on the code (u-boot and device tree) |
| 07:46 | sanjeevs | Ok but can you clarify one doubt |
| 07:46 | rabeeh | yeah sure |
| 07:48 | sanjeevs | I am using quad core imx6som i4pro version of hardware and currently trying to figure it out which code will be executed for usdhc2 in uboot https://github.com/SolidRun/u-boot-imx6/blob/imx6/board/solidrun/mx6_cubox-i/mx6_cubox-i.c#L225 |
| 07:49 | sanjeevs | This is #define CONFIG_MX6QDL which means this code is executed for dual and quad core |
| 07:49 | sanjeevs | is that correct ? |
| 07:49 | sanjeevs | or this could will be executed # defined(CONFIG_MX6Q) || defined(CONFIG_MX6DL) |
| 07:50 | sanjeevs | for both #define usdhc2_pad cantrol is different |
| 07:55 | sanjeevs | I would like to know what are the correct PAD control for the SD2 port |
| 07:55 | rabeeh | if you are using SPL u-boot (which you should) then look only at #if defined(CONFIG_MX6QDL) |
| 07:55 | rabeeh | https://github.com/SolidRun/u-boot-imx6/blob/imx6/board/solidrun/mx6_cubox-i/mx6_cubox-i.c#L53 |
| 07:56 | rabeeh | the definition would be USDHC_PAD_CTRL |
| 07:56 | sanjeevs | and what about usdhc2_pads ? |
| 07:57 | rabeeh | this is an old code before the SPL code was added |
| 07:58 | rabeeh | but it would be the same thing; pull up for 47kohm internally \ |
| 07:58 | rabeeh | notice that this is over riden in the kernel dts again |
| 08:00 | sanjeevs | in usdhc2_pads[]:There is different setting for USDHC_PAD_CLK_CTRL and USDHC_PAD_GPIO_CTRL |
| 08:10 | rabeeh | ok |
| 08:10 | rabeeh | and..? |
| 08:10 | rabeeh | are you asking about the difference? why there is a difference? |
| 08:11 | rabeeh | if so then i would assume it's related to that one is a clock and the others are data |
| 08:11 | rabeeh | but regardless; the tricky part is really in the kernel; in u-boot things are pretty simple since the clk is 50mhz max |
| 08:12 | rabeeh | in hummingboard there is support for uhs-1; where the clock goes way beyond and then you should be looking for fine details of impedence of the driver etc... |
| 08:33 | sanjeevs | ok |
| 08:34 | sanjeevs | So what you would recommend to use |
| 08:34 | heap_ | rabeeh: hi, does have cubox peoper implementation of esata port? |
| 08:51 | heap_ | this is really issue now: http://pastie.org/10574787 |
| 09:19 | rabeeh | heap_: which volume is the esata port? |
| 09:19 | rabeeh | is it /dev/sdd? |
| 09:21 | heap_ | rabeeh: sda and sdd |
| 09:21 | rabeeh | port multiplier? |
| 09:21 | heap_ | cuboxi-(Esata)-> enclosure (esata) -: drive sda and sdd |
| 09:21 | heap_ | rabeeh: yes that enclosure has port multiplier |
| 09:22 | rabeeh | what about sdb and sdc? |
| 09:22 | rabeeh | they also look 2.7TB drives |
| 09:22 | rabeeh | my guess is that it's a btrfs issue |
| 09:22 | rabeeh | the volumes are there; and they are not disconnected (lsblk) |
| 09:23 | heap_ | rabeeh: sdb and sdc are inside enclosure but they are part of LVM |
| 09:23 | rabeeh | what about the kernel? does it spit any thing? |
| 09:23 | rabeeh | ahci related? |
| 09:23 | heap_ | rabeeh: another problem is while writing data onto drives im getting so many errors as link reset, drives resets etc |
| 09:24 | rabeeh | have a dump? |
| 09:24 | rabeeh | which kernel btw? |
| 09:24 | heap_ | Linux universe 4.1.13+ #1151 |
| 09:25 | heap_ | im so confused ;-( |
| 09:26 | rabeeh | do you have kernel messages? |
| 09:26 | heap_ | even when i do write on lvm or btrfs errors are still there |
| 09:26 | rabeeh | for the link reset and drive resets |
| 09:26 | heap_ | yes |
| 09:26 | heap_ | rabeeh: http://pastie.org/10574598 |
| 09:26 | heap_ | rabeeh: http://pastie.org/10574608 |
| 09:27 | rabeeh | how long is the esata to the enclosure cable? |
| 09:28 | rabeeh | SErr 0xc00501 |
| 09:29 | heap_ | rabeeh: i think 1m i tried 3 cables |
| 09:29 | rabeeh | ^^^ this is the main error in the sata connection; let me try to decode this |
| 09:29 | heap_ | rabeeh: sure, thanks |
| 09:32 | rabeeh | it says it's 10b to 8b decoding error |
| 09:32 | rabeeh | meaning that the SATA receiver saw bit errors when receiving frames. |
| 09:33 | rabeeh | OR - SATA receiver overflow |
| 09:33 | Exaga | morning rabeeh ;) |
| 09:33 | rabeeh | hi Exaga |
| 09:33 | rabeeh | :) |
| 09:33 | rabeeh | heap_: what this means is either one of the two - |
| 09:33 | rabeeh | 1. the port multiplier is sending corrupted 8b encoded to 10b |
| 09:33 | heap_ | rabeeh: means? is is faulty cable, esata port or multiplier or enclosure or chipset inside enclosure? |
| 09:34 | rabeeh | 2. the port multiplier is not obeying the rx flow control commands |
| 09:34 | heap_ | rabeeh: basically problem is in the multiplier right? |
| 09:34 | rabeeh | i'v seen this before with Silicon Image port multiplier (can you please confirm if this is a silicon image one?) |
| 09:34 | heap_ | rabeeh: its this crap ICYCube MB561U3S-4S |
| 09:34 | rabeeh | heap_: i can't really tell; but this is what the decoded errors means. It's on the receive side of the SATA |
| 09:35 | rabeeh | what's the chip inside? |
| 09:35 | heap_ | rabeeh: ;-( no idea how to find out it? |
| 09:35 | heap_ | rabeeh: or is there any way how to find out? |
| 09:36 | rabeeh | open the box :) |
| 09:36 | rabeeh | i need to disappear for 1hr |
| 09:36 | heap_ | rabeeh: i can do it |
| 09:36 | heap_ | rabeeh: or do you have tip for good working multiplier? |
| 09:36 | Exaga | see you later rabeeh. take care o/ |
| 09:43 | schribl | i want to use the fantec QB-35US3-6G (should be the same as mediasonic ProBox HF2-SU3S2) with my cubox i4-pro. does somebody know if this port multiplier will work? |
| 09:58 | heap_ | rabeeh: pls ping me when you are back, thanks |
| 10:33 | heap_ | anyone has experience with that product? http://www.addonics.com/products/ad5sarm6g.php if it works properly, if its uspported by kernel etc |
| 11:00 | rabeeh | heap_: can you please read the register 0x02200170? |
| 11:00 | rabeeh | you can do this by running 'devmem 0x02200170' as a root |
| 11:01 | rabeeh | or 'devmem2 0x02200170' |
| 12:21 | kasvain | rabeeh: any binary kernels for cubox lite/dual lite? |
| 12:22 | kasvain | artox compiled 4.x for me because i can't get any official work but the old one in ubuntu 12 from the installer |
| 12:24 | kasvain | artox's binary is made without hardware its meant for, almost made it boot but faced undefined instuction |
| 12:47 | vpeter | For ClearFog I'm looking some very small UPS or even better using some external adapter 230V/xxV and some DC/DC converter with battery which would provide 5V and 12V. Is there something off-the-shelf available for such purposes? |
| 12:54 | heap_ | rabeeh: give me sec |
| 12:56 | heap_ | rabeeh: there is no devmem bin |
| 12:56 | rabeeh | install / build one then |
| 12:56 | rabeeh | if you are in Debian i think there is a package called devmem2 |
| 12:56 | rabeeh | http://free-electrons.com/pub/mirror/devmem2.c |
| 12:57 | heap_ | its debian |
| 12:57 | heap_ | but there is no such package ;/ |
| 12:58 | heap_ | rabeeh: |
| 12:58 | heap_ | root@universe:/home/debian# ./a.out 0x02200170 |
| 12:58 | heap_ | /dev/mem opened. |
| 12:58 | heap_ | Memory mapped at address 0x76eff000. |
| 12:58 | heap_ | Value at address 0x2200170 (0x76eff170): 0x88 |
| 12:58 | heap_ | cc devmem2.c ; a.out 0x02200170 |
| 12:58 | rabeeh | lets play with the 0x88 value then |
| 12:58 | rabeeh | i suggest the following - |
| 12:58 | rabeeh | ./aout 0x02200170 w 0x28 |
| 12:59 | rabeeh | this will lower the RX fifo threshold |
| 12:59 | rabeeh | please try it and see if you still get sata errors |
| 13:00 | heap_ | what a hack : |
| 13:01 | rabeeh | clearly a reboot will reset the register to it's default value |
| 13:02 | heap_ | rabeeh: i dont think it works |
| 13:02 | heap_ | ;/ |
| 13:02 | rabeeh | you still see sata errors? |
| 13:02 | heap_ | root@universe:/home/debian# ./a.out 0x02200170 w 0x28 |
| 13:02 | heap_ | /dev/mem opened. |
| 13:02 | heap_ | Memory mapped at address 0x76fd7000. |
| 13:02 | heap_ | Value at address 0x2200170 (0x76fd7170): 0x88 |
| 13:02 | heap_ | Written 0x28; readback 0x88 |
| 13:02 | rabeeh | i'm not interesting in btrfs stuff; just kernel messages with ahci messages |
| 13:02 | heap_ | root@universe:/home/debian# ./a.out 0x02200170 |
| 13:02 | heap_ | /dev/mem opened. |
| 13:02 | heap_ | Memory mapped at address 0x76fd6000. |
| 13:02 | heap_ | Value at address 0x2200170 (0x76fd6170): 0x88 |
| 13:02 | heap_ | its still 0x88 ? |
| 13:02 | rabeeh | oh |
| 13:04 | rabeeh | what does ./aout 0x02200118 provide? |
| 13:08 | heap_ | rabeeh: |
| 13:08 | heap_ | root@universe:/home/debian# ./a.out 0x02200118 |
| 13:08 | heap_ | /dev/mem opened. |
| 13:08 | heap_ | Memory mapped at address 0x76f47000. |
| 13:08 | heap_ | Value at address 0x2200118 (0x76f47118): 0x2C817 |
| 13:08 | rabeeh | ok; lets do the following thing and lets see if it survives - |
| 13:09 | rabeeh | ./a.out 0x02200118 0x2c816; ./a.out 0x02200170 w 0x28; ./a.out 0x02200118 0x2c817 |
| 13:09 | rabeeh | three commands in a row; please cut and paste to pastebin |
| 13:10 | heap_ | http://pastie.org/10575820 |
| 13:11 | rabeeh | ./a.out 0x02200118 w 0x2c816; ./a.out 0x02200170 w 0x28; ./a.out 0x02200118 w 0x2c817 |
| 13:11 | rabeeh | :) |
| 13:11 | rabeeh | i'v missed the 'w' in two commands |
| 13:12 | heap_ | rabeeh: http://pastie.org/10575823 |
| 13:12 | rabeeh | ok; valus 0x28 is written now |
| 13:12 | rabeeh | please 'dmesg -c' and see if you get more sata errors |
| 13:15 | kasvain | rabeeh: can you make updated version of the installer? to support your elder products. only arch linux and ubuntu 12 headless installation works out of the box, all other available in the menu are not found anymore or kernel image is not compatible with the cubox lite/dual lite |
| 13:16 | rabeeh | kasvain: you mean the first cubox? |
| 13:16 | kasvain | yeah! :) |
| 13:16 | kasvain | i have one, cant get anything work |
| 13:16 | rabeeh | :) |
| 13:16 | kasvain | but the two i mentioned |
| 13:16 | rabeeh | how is it behaving? |
| 13:17 | rabeeh | (when you can install things on it) |
| 13:17 | kasvain | wait, i'll look at the old pastes i provided to artox who have helped me with this |
| 13:17 | kasvain | ubuntu 12 installs and boots for first time, but the second time it can't find the image anymore |
| 13:17 | kasvain | i managed to reflash the spi image to make sure its not about uboot |
| 13:18 | heap_ | rabeeh: ill try it |
| 13:19 | heap_ | rabeeh: but do you mean its good way with that software hack? |
| 13:19 | rabeeh | heap_: lets see if it even works |
| 13:19 | heap_ | what do you mean works? no errors? |
| 13:19 | rabeeh | this hack lowers the threshold when the AHCI controller tells the port multiplier to backoff |
| 13:19 | rabeeh | works === no errors; btrfs is totally different story |
| 13:20 | rabeeh | if you are running on 4.1 then most probably btrfs should be functional. i never played with btrfs on imx6 |
| 13:20 | kasvain | can't find those pastes anymore and can't reproduce the error, but artox found out the uboot im using does not have support for dtb+zImage booting |
| 13:20 | heap_ | rabeeh: that errors corrupted btrfs ;/ |
| 13:20 | rabeeh | heap_: if that doesn't work then the next thing would be trying a 0.5m eSata cable |
| 13:20 | heap_ | rabeeh: and if it wont work with 0.5m? |
| 13:21 | heap_ | rabeeh: wrong chipset or enclosure or port multiplier? |
| 13:21 | rabeeh | heap_: it's really hard to tell. i'm seeing receiver errors |
| 13:21 | rabeeh | from the log that you have sent |
| 13:21 | heap_ | receiver errors mean? |
| 13:22 | rabeeh | AHCI receiver errors is that imx6 AHCI controller (the SATA controller) is getting bad frames from the port multiplier |
| 13:22 | rabeeh | this can be because of tons of reasons... tons.. |
| 13:22 | rabeeh | one of them is rx overflow, other can be signal integrity (i.e. smaller cable typically helps) |
| 13:25 | heap_ | rabeeh: or bad multiplier? |
| 13:25 | rabeeh | rx overflow is a bad multiplier implementation |
| 13:25 | rabeeh | signal integrity can be cable, or bad box implementation |
| 13:27 | heap_ | rabeeh: yeah so what i said if i order new multiplier it might work |
| 13:27 | heap_ | rabeeh: i found http://www.addonics.com/products/ad5sarm6g.php |
| 13:28 | rabeeh | if it's using the same chipset then most probably it will fail too (if it's rx overflow issue) |
| 13:28 | heap_ | no idea about chipset |
| 13:29 | heap_ | or which one is with good/verified chipset |
| 13:29 | kasvain | btw is there newer u-boot for 1st gen cubox, mine is 2009.08, couldn't find any newer version as binary :o also can't compile own, i have no devices nor a working environment |
| 13:29 | kasvain | that might solve lots of problems in my case if the zImage booting support is in it |
| 13:32 | heap_ | rabeeh: do u have any experience with good esata multipliers that works with cuboxi? |
| 13:32 | rabeeh | heap_: nop |
| 13:32 | rabeeh | sorry... |
| 13:32 | heap_ | ok |
| 13:34 | rabeeh | but first please try the register modifications above |
| 13:42 | heap_ | rabeeh: sure |
| 14:20 | heap_ | rabeeh: are there many ppl who occuring those esata problems? |
| 15:20 | kasvain | malte_: do you have any quite new uImage for 1st gen cubox |
| 15:20 | kasvain | (and unitrd) |
| 15:20 | kasvain | i have been battling with this cubox for a year now and no-one but artox have helped this far |
| 15:22 | kasvain | also another solution would be to have enough new u-boot, solidruns ftp provides only old which support uImage but not zImage |
| 15:25 | kasvain | rabeeh: pls help with this problem..would like to use this cubox, i haven't managed to get this work after i bought this |
| 15:54 | heap_ | rabeeh: but those erorrs corrupted data on f |
| 15:55 | heap_ | rabeeh: fs, is it normal or expected that i u get those multiplier restarts it can corrupt data? |
| 15:56 | kasvain | Artox: pong |
| 15:56 | Artox | kasvain: pong |
| 16:14 | bencoh | 44 |
| 16:14 | bencoh | woops |
| 18:55 | heap_ | rabeeh: http://pastie.org/10576422 |