**** BEGIN LOGGING AT Thu Jul 28 02:59:58 2016 Jul 28 09:11:17 Hi, has anyone managed to run the Beaglebone black rev. c SPI at high speed? (e.g. 50 MHz SPI clock) Jul 28 09:12:52 if you need to run SPI at 50MHz, then you need reevalute your life values Jul 28 09:17:32 spi runs that quick?! Jul 28 09:23:25 it is possible to run spi up to 100MHz... i've seen it done Jul 28 09:23:29 but it ain't pretty Jul 28 09:23:52 I would imagine you gotta be quite careful about your traces Jul 28 09:24:09 the traces are not that much of a problem Jul 28 09:24:20 but rise/fall times are Jul 28 09:24:38 normal cmos chips have rise/fall times in the 2-30ns range Jul 28 09:24:41 traces/signal integrity :P Jul 28 09:25:26 veremit: 100MHZ = 3m wave length Jul 28 09:25:35 veremit: any trace will be electrically short Jul 28 09:25:47 oh I know that .. 3ghz = 1cm iirc Jul 28 09:25:54 300/lambda Jul 28 09:26:04 or .. that formula :P Jul 28 09:26:15 its early/pre-coffee still .. XD Jul 28 09:26:46 I probably need another coffee.. Jul 28 09:27:10 I need a couple nights decent sleep -sigh- Jul 28 11:50:07 hi, I tried to clean my bbb and I don't want to mess up things. bbb linux kernel is 4.6.2-bone3. I would like to purge some linux-headers, Jul 28 11:50:55 when I do dpkg --list | grep linux-header* I don't see linux-header-4.6.2-bone3 Jul 28 11:51:51 is it normal to have linux-image and linux-header different ? Jul 28 11:58:58 2 linux-headers and 3 linux-images. There are all "ii" Jul 28 11:59:13 welcome to debian Jul 28 11:59:43 there is a reason why i reinstall my systems from scratch every 10 years Jul 28 12:00:43 dirkk: did you install that kernel manually? they are just packages, you can mix and match them (or not, if e.g. you don't need headers for your business) Jul 28 12:06:21 samael: no I did with apt-get install linux-image-X and I had error messages which mentionned some headers, so I installed it. I don't really know what's a linux header... just header file for kernel function ? Jul 28 12:09:37 samael: How to know if one header is being used ? Should I keep one or could I delete them all ? Jul 28 12:19:50 dirkk: headers are used when compiling kernel/modules/other software which explicitly need them. if you don't know, then you most likely don't need them installed Jul 28 12:20:58 you can safely remove them. if you prefer, you can leave installed the headers for the running kernel, but this is not mandatory Jul 28 12:21:14 you can always reinstall them if/when needed Jul 28 12:22:10 am I missing something? have pru code to read r31 gpio pins, it just fails (not entirely sure). change "r31" to "r3" and it all works fine... Jul 28 12:22:20 (reading some random values, anyway) Jul 28 12:23:38 samael: ok, are they needed for pruss ? Jul 28 12:23:57 msh: whats the exact command ? Jul 28 12:24:16 msh: using pasm or clpru ? Jul 28 12:26:48 dirkk: If you are just compiling PRU code, no headers are not needed. If a kernel module, yes headers are needed Jul 28 12:32:36 ZeekHuge: clpru, "and r28, r31.b0, 0xff" Jul 28 12:32:40 though have tried various variants Jul 28 12:34:35 guess I should try qbbs, though I don't see why straight reading wouldn't work Jul 28 12:36:44 its working on my device . Jul 28 12:36:49 msh: ^ Jul 28 12:36:54 ZeekHuge: *nod* thanks Jul 28 12:37:26 msh: you got it working ? Jul 28 12:37:27 perhaps I need to fiddle with resource tables or things. though I woudld've thought it would at least read _something_ Jul 28 12:37:30 nope Jul 28 12:37:34 thanks for testing Jul 28 12:37:44 is your code up somewhere ? Jul 28 12:38:00 msh: ^ Jul 28 12:38:15 just a minute Jul 28 12:38:37 don't know hoe resource table can cause a problem. Jul 28 12:38:49 you sure clpru is the latest version ? Jul 28 12:40:16 ZeekHuge: ok, (stupid question) sysfs are part of userspace or kernel space ? Jul 28 12:40:39 http://evil.ucc.asn.au/~matt/b/pru/asmpr0.asm Jul 28 12:40:40 ahh .. they connect the two .. Jul 28 12:41:11 dirkk: so, they expose some features/attributes of the kernel to the userspace Jul 28 12:41:14 that's a sampling loop for pru0, then writes to pru1 which sends rpmsg etc Jul 28 12:41:30 so that all works fine if I send a counter etc. just when I try the actual r31 read, it all stops working Jul 28 12:46:48 yeah 2.1.2 compiler Jul 28 12:48:09 msh: thats a clean compile Jul 28 12:48:23 it didn show any error Jul 28 12:49:33 yep. lookat at the program counter sys/kernel/debug/remoteproc/remoteproc1/regs it's running Jul 28 12:49:40 but it isn't doing the xout to the pru1 Jul 28 12:49:59 yet when I change "r31" to something else, it does. anyway, I'll keep poking about at it. must be something silly. Jul 28 12:52:24 msh: good luck . btw, great code. Jul 28 12:53:47 thanks. trying to read some mems microphones. if I get it working I'll probably put it up somewhere Jul 28 12:57:24 can I remove safely files in /lib/modules/image-purged ? Jul 28 13:06:23 msh: I am missing a lot of points, and I believe that you know this, but still, you may be able to use 'mov r28,, r31.b0' and .. for register to register transfer mov would only work LDI is for constant to register transfer Jul 28 13:09:05 hm Jul 28 13:09:28 I've not noticed I don't reset rindex, so maybe that's wrong Jul 28 13:12:32 huzzah, that seems to be it Jul 28 13:16:33 cool ! Jul 28 13:23:31 Hi People Jul 28 13:41:12 hello Jul 28 13:41:24 I am working on beagle bone Jul 28 13:41:37 and trying to integrate simcom 800 with it Jul 28 13:42:28 I have write api's for call and to send messages Jul 28 13:42:33 they are working fine Jul 28 13:42:52 but when i integrate GPRS with it then facing issues Jul 28 13:43:15 i am accessing gprs through pppd package of linux Jul 28 13:43:45 Now problem arise, when gprs is working on it through pppd package Jul 28 13:43:59 and my api's starting to make a call Jul 28 13:44:29 I am getting error: Resources Temporarily unavailble Jul 28 13:44:47 Can any please repsond Jul 28 13:44:48 ? Jul 28 14:04:40 I am unable to help you. Jul 28 14:05:05 but out of curiosity, are you actually using a beaglebone? The old one? Jul 28 14:05:36 oh. he/she left. Jul 28 14:47:34 what is the expected release date of beaglebone-x15? Jul 28 14:49:47 heeeey is anybody there? Jul 28 14:50:40 ok I found it Jul 28 14:50:42 First 900 boards shipping.......................8/22/2016 Jul 28 14:50:53 bye losers Jul 28 15:11:41 hmm.. lives in turkey and calles us losers? Jul 28 15:15:50 heh Jul 28 15:16:23 I guess now that X15 is shipping I should dust off mine to be able to field questions Jul 28 15:17:06 Why would living in Turkey make you a loser? Either way, typical internet hostility. Some people use 'losers' as a term of antagonistic endearment though. Jul 28 15:17:50 But I wouldn't use it with that intent with random strangers unless I was itching to be misunderstood. Jul 28 16:16:49 Hello I am trying to run some code on my debian jessie beaglebone black. In my code I import matplotlib.pyplot. When I do, it returns the error " ImportError:No module named matplotlib.pylot" any ideas on how to fix this? Jul 28 16:17:28 * jkridner wonders if there is a matlab group on freenode. Jul 28 16:17:40 gurm_: your issue sounds very matlab specific. Jul 28 16:18:26 is freenode another python chatroom? Jul 28 16:19:07 I am using matplotlib. not matlab Jul 28 16:19:16 I am using matplotlib. not matlab Jul 28 16:22:47 gurm_ you need to install matplotlib Jul 28 16:23:52 not matlab, but python specific FWIW Jul 28 16:27:28 methuzla how do I do that? Jul 28 16:28:09 http://matplotlib.org/users/installing.html Jul 28 16:30:06 sorry my connection timed out Jul 28 16:44:39 when I try "sudo apt-get install matplotlib" it returns "Reading package lists... Done Building dependency tree Reading state information... Done E: Unable to locate package matplotlib" Jul 28 16:47:43 gurm__, http://matplotlib.org/users/installing.html Jul 28 16:52:34 E: You must put some 'source' URIs in your sources.list Jul 28 17:11:09 gurm__: k. no, freenode is the server on which this IRC chat is. Jul 28 17:13:17 gurum__: https://packages.debian.org/search?searchon=contents&keywords=matplotlib&mode=path&suite=stable&arch=any Jul 28 17:13:48 "sudo apt-get install python-matplotlib" Jul 28 17:21:58 note to self: use -iquote instead of -I for telling gcc about paths to project-local include directories Jul 28 17:23:50 it's kinda weird that #include "..." also searches system include dirs though Jul 28 17:24:28 but at least iquote avoids any risk of a project header file accidently shadowing an identically named system header file Jul 28 17:25:02 ok, shopping time, maybe I can still avoid the rain Jul 28 17:48:51 is beaglebone black bugged @ handling more than one usb connected? Jul 28 17:52:03 hello ? Jul 28 17:53:42 I have two identical wifi dongles. If I connect at boot it freezes after a few seconds Jul 28 17:54:51 do i need to do something to get the driver into two instances or is there a problem with linux version or with the sitara(read it could be this) Jul 28 18:02:59 do they have unique mac addresses? Jul 28 18:16:25 disturbing that such a question needs to be asked >< Jul 28 18:17:27 you'd think that if they don't bother getting an official MAC range then *at least* generate them randomly to ensure essentially zero chance of conflict Jul 28 18:18:04 disturbing, yes, real word problem, yes. :/ Jul 28 18:19:44 doesn't linux detect and fix this though? Jul 28 18:21:42 i know it didn't on sparc as of something like 8-10 years ago, for what it's worth Jul 28 18:22:21 supposedly there's a valid use case for it ... Jul 28 18:25:39 Part of a MAC address acts as a sort of vendor id, so I think you could potentially confuse a lot of software by randomly generating them Jul 28 18:55:09 MathOnNapkins: you'd set bit 1 of course to indicate it's a "locally administered" (read: randomly generated) MAC rather than a officially assigned one Jul 28 18:55:30 Oh, I wasn't aware that was a thing Jul 28 18:55:32 thanks Jul 28 18:56:14 The company I work for was looking into purchasing a MAC address range at one point, but we didn't go through with it. I didn't recall running across that. Jul 28 18:57:14 I have a small problem Jul 28 18:57:19 can anybody help Jul 28 18:57:28 most likely if you use the bbb via usb networking at least one of the two sides will also use a random MAC Jul 28 18:57:42 hiu Jul 28 18:57:55 hi guys I have a small confusion Jul 28 18:57:59 (in theory it doesn't need to, the AM335x comes with three officially registered MAC addresses per device, but that doesn't seem well known) Jul 28 18:58:10 in uboot Jul 28 18:58:21 crap I really gotto hurry to stay ahead of the rain Jul 28 18:58:29 The AM335x has everything but I don't know about any of it :/ Jul 28 18:58:44 My problem is regarding pin multiplexing in spl Jul 28 18:58:46 Or rather it seems that it's buried deep in the manual Jul 28 18:59:08 So as we know that beaglebone uses uart0 to communicate via serial console on pc Jul 28 18:59:22 MathOnNapkins: every now and then just pick some random chapter of the TRM you're not familiar with and read a bit... read the intro, browse the functional overview Jul 28 18:59:29 i checked where the pins are actually muxed in the spl file. Jul 28 19:00:02 Is somebody have experience on this topic Jul 28 19:00:03 ? Jul 28 19:00:13 obviously spl will do so, otherwise you have no serial console Jul 28 19:00:22 in some cases it may even be done by boot ROM Jul 28 19:00:46 Yes, I was just making myself familiar how the pins start behaving that way Jul 28 19:01:14 so I read trm's control module chapter as what values must be filled in uart0 Jul 28 19:01:37 the address I used is 0x44E1_0980 Jul 28 19:01:40 board/ti/am335x/mux.c Jul 28 19:01:44 to read from this register. Jul 28 19:01:48 Yes I know this file. Jul 28 19:02:13 I am printing this value after preloader_console_init() function Jul 28 19:02:28 preloader_console_init();//it will print uboot version date and time information mfi = *(unsigned int *)(0x44E10980); Jul 28 19:02:35 this value came as Jul 28 19:02:37 37 Jul 28 19:02:57 this means I am reading wrong value, as this seems to be a gpio? Jul 28 19:04:19 0x44E10980 is the mux register for P9.26 on the expansion header Jul 28 19:04:29 (pin 96) Jul 28 19:04:58 check https://goo.gl/Jkcg0w for reference (especially the orange-marked tabs, those have BBB-specific info) Jul 28 19:05:12 the pin numbers I use are index into the pinmux array Jul 28 19:05:38 Umm control module base register is 44E1_0000, right? Jul 28 19:05:52 and offset for conf_uart0_rx is 980 Jul 28 19:06:17 So its 44E1_0980 for uart0, am I wrong? Jul 28 19:06:31 it is beaglebone black Jul 28 19:08:42 I don't know where you're getting your data from, but my list shows uart0_rx to be pin 92 Jul 28 19:08:52 and 0x44e10800 + 92 * 4 == 0x44e10970 Jul 28 19:09:19 you might be having 0-based/1-based fun Jul 28 19:09:33 since pin 96 is uart1 rxd Jul 28 19:10:12 you are right Jul 28 19:10:23 (my spreadsheet consistently uses 0-based numbering, linux and the TRM are however inconsistent with each other, between different versions, and across products) Jul 28 19:10:28 I just checked this address and booted my beaglebone black Jul 28 19:10:39 it showed 0x30 value that is correct Jul 28 19:10:50 my advice is: just say no to 1-based numbering :P Jul 28 19:11:05 rcn seems to have done the same in the DTs for the BBB kernels Jul 28 19:12:16 If your first birthday your 0th birthday or your 1st birthday? Jul 28 19:12:33 If the former that means I'll stay 33 in a month or so :) Jul 28 19:12:47 a birthday counts the number of years since you were born Jul 28 19:12:54 (If you consider a 'birthday' to be the event of a year of life passing) Jul 28 19:12:59 technically the day you were born was your 0th birthday Jul 28 19:13:08 0x44e10800 why you used this address Jul 28 19:13:12 Shhhhh I'm bending the definition to make it work Jul 28 19:13:20 Guest90155: it's the base address of the pinmux array Jul 28 19:13:44 ok I will search this base address in TRM Jul 28 19:13:53 https://github.com/dutchanddutch/jbang/blob/master/include/ti/subarctic/ctrl.h#L582 Jul 28 19:14:25 it's in the control module register summary Jul 28 19:14:39 the TRM also lists all registers by... name -.- Jul 28 19:14:56 I absolutely dispise the practice of naming pins after *one* of the possible functions of that pin Jul 28 19:15:16 since it's confusing as fuck if you use it for a different purpose, and names are not convenient for indexing into arrays Jul 28 19:16:28 AM335x_techincal_reference_manual.pdf is it the correct TRM? Jul 28 19:16:34 the pinmux array is the saviour of sanity on the am335x, basically every pin of relevance (and a few irrelevant ones) is in there, thus yielding a nice contiguous pin numbering to use Jul 28 19:16:35 I didn Jul 28 19:16:52 found this base address Jul 28 19:16:57 http://www.ti.com/product/AM3358/technicaldocuments Jul 28 19:17:22 at the top you find latest datasheet and errata, if you scroll further down you find the latest TRM under User guides Jul 28 19:17:34 now I really gotto go, back later Jul 28 19:17:52 Very big thanks Jul 28 19:17:54 to you Jul 28 19:38:03 Hi all. I was wondering if drivers exist for BeagleBone Black support of S-Video or composite video. Or would I realistically have to go straight for a BeagleBoard-xM? Jul 28 19:50:15 is the code in the rom in sitara on the beaglebone readable ? Jul 28 20:54:56 rufuss: public rom is. you can dump it at physical address 0x40020000 (length 0xC000 iirc) and there's some memory map overview in the initialization chapter of the TRM Jul 28 20:55:36 * zmatt has returned, evidently Jul 28 20:58:42 rufuss: if you're looking for anything in particular I might be able to give some pointers Jul 28 20:58:57 although I haven't examined am335x rom as much as dm814x rom (but they Jul 28 20:59:51 *they're very similar, in fact the am335x 1.0 had an erratum that it identified itself as DM814x via bootp when netbooting) Jul 28 22:20:24 its the wiki of beagle board (the original one ) out of service Jul 28 23:07:18 the TI docs for the WiLink8 module drivers is one of the WORSE in the world :( Jul 28 23:14:41 imagine that... Jul 28 23:55:41 anyone there? Jul 28 23:56:36 kali/ubuntu/mint users here? Jul 29 00:40:59 ds2: they're not terribly enlightening no Jul 29 00:41:29 that isn't the problem Jul 29 00:41:45 it is their attempt at putting out version specific stuff rather then a generic "stock" tree so anyone can build it Jul 29 00:41:54 cuz they did that, every one of their "wiki" crap is out of date Jul 29 00:42:17 i.e. there is damn reason to have a page for Angstrom and one for Debian Jul 29 00:42:22 like it took us a while to discover the pcm/i2s connection isn't all that useful: their own a2dp demo actually just sends the data over serial and has the cpu play it back Jul 29 00:42:34 ok that's pretty stupid indeed Jul 29 00:42:52 if they just put a tarball + diffs against ANY version, it would have been easier Jul 29 00:43:07 now I got to unweave their damn docs to get it to the point where it can be patched into a kernel Jul 29 00:43:23 zmatt: a2dp typically doesn't use PCM or I2S Jul 29 00:44:15 well I didn't know that :P and it might actually support "assisted a2dp", but only for that crappiest lowest-common-denominator codec Jul 29 00:45:18 I'm still periodically stupefied by the idiocy of the BBGW design Jul 29 00:45:19 the PCM channel is only for HSP/HFP stuff Jul 29 00:45:36 yes now I know that, but that part isn't clear in the docs Jul 29 00:45:50 hehe... true Jul 29 00:46:08 knew the audio stuff from other BT things Jul 29 00:46:27 they show that connection being made, no mention that if you have no use for HSP/HFP then you can save four traces and complications in interfacing to the BBB Jul 29 00:46:46 heeheh... Jul 29 00:47:08 If you had consulted with me before.... ;) Jul 29 00:47:25 I still discovered it fairly early in the schematic design phase Jul 29 00:47:29 wel that's the thing Jul 29 00:47:41 one consults if one sees a potential problem or question Jul 29 00:48:10 but all documents seem to imply they're simply part of "the connection" Jul 29 00:48:14 to your host system Jul 29 00:49:04 by fortunate accident there was something that made me go "hey, wait a minute..." Jul 29 00:49:35 helps to have worked with other modules and understand the protocol Jul 29 00:49:58 yes, but that wasn't the case, and (as usual) They are in a hurry Jul 29 00:50:36 what's the old saying...haste makes waste Jul 29 00:51:29 myeah, but not having stuff to show makes lack of customers and investors :P Jul 29 00:52:45 the way too frequent rushwork is one thing that really annoys me, but I do also get that bills need to be paid... it's a small and fairly new company, there's no line of established products ensuring a steady income while you do R&D Jul 29 00:54:20 and in some ways it's also good for me to learn to make "adequate" stuff... I do otherwise suffer from a bit of a perfectionism Jul 29 00:55:44 ds2: have you seen the BBGW schem btw? Jul 29 00:56:58 zmatt: yes, I have Jul 29 00:58:13 brilliant isn't it, leaving the former ethernet pins unused Jul 29 00:58:49 even though those just so happen to have as mux options... mmc2, serial, mcasp Jul 29 00:59:24 and they're on their own supply rail so if they had used those pins then no level shifters would have been required Jul 29 00:59:47 got to realize.... now they can seperate out the module and turn it into a software compatible cape ;) Jul 29 01:00:10 if you took over the ethernet pins like you suggests, think of the software issues if one made a cape! Jul 29 01:00:21 eh, none? just a DT difference Jul 29 01:00:47 and how do you keep it off? Jul 29 01:00:54 s/keep/key/ Jul 29 01:01:03 ? Jul 29 01:01:04 can't you see the brilliance Jul 29 01:01:18 not only that, it contributes to the level shifter economy Jul 29 01:01:21 honestly? no, I really can't Jul 29 01:01:36 :P Jul 29 01:01:52 with the level of unemployement...every cent helps Jul 29 01:05:22 it is pretty clear whoever did the design had no idea what a pinmux is Jul 29 01:07:15 ok they at least reused a few eth pins Jul 29 01:09:54 of course, the BBB started going down the path of burning accessible pins Jul 29 01:10:00 can't completely fault the BBGW Jul 29 01:10:39 the BBB had no choice Jul 29 01:11:40 the BBGW otoh chopped off ethernet, thereby making a set of IOs available that (coincidence?) happens to match beautifully to the wilink 8, AND THEN DIDN'T FUCKING USE IT Jul 29 01:11:41 why did you say that? Jul 29 01:11:43 :/ Jul 29 01:12:00 lcd pins have no alternative location Jul 29 01:12:19 using the LCD pins the way they did it, didn't really screw up much stuff Jul 29 01:12:31 what about the eMMC fiasco? Jul 29 01:13:03 no other mux option available that doesn't conflict with ethernet Jul 29 01:14:23 if they just replaced the uSD socket the eMMC and let the socket sit on the shared pins, it would become "inert" unless on really wants to use the socket Jul 29 01:14:50 there would be no mux option issue Jul 29 01:14:59 how exactly would the flasher work then if you can't copy anything between SD and eMMC ? :P Jul 29 01:15:12 how does that preclude copying? Jul 29 01:15:24 ohh Jul 29 01:15:26 I misread Jul 29 01:15:26 one can free up the pins by just not putting in a uSD card Jul 29 01:15:42 none of this silly BBW/BBB case or having to use a disable pin Jul 29 01:16:02 i find that precedence for burning as many pins as possible on the connector :D Jul 29 01:16:18 SD is only 4-bit Jul 29 01:16:26 of course on my own board, there is no ethernet or eMMC and wireless is going on the ehternet pins Jul 29 01:16:37 I don't find a noiceable performance difference Jul 29 01:17:04 that's weird, certainly the kingston should be able to max it out for reads Jul 29 01:18:02 now I'm tempted to configure a DT for 4-bit to see if you are right :P Jul 29 01:18:54 I do compiles locally Jul 29 01:19:05 and use a decent class 10 microSDHC card Jul 29 01:19:16 we don't want cards anyhow Jul 29 01:19:29 if you used the kingston uSD card from the xM days... it is of course faste Jul 29 01:19:31 r Jul 29 01:19:48 even molasses in the northpole would flow faster thenthose cards Jul 29 01:19:52 we don't really use μSD cards at all Jul 29 01:20:04 I'll take the uSD card over emmc Jul 29 01:20:22 insufficient assurance it will make proper contact and continue to do so Jul 29 01:20:37 you can solder that on ;) Jul 29 01:20:45 but in anycase, the point was 8bit vs 4bit Jul 29 01:21:19 yes, and iirc bootROM also treats mmc0 and mmc1 differently Jul 29 01:21:43 something about compat with >2GB cards Jul 29 01:26:20 why the hell can't I locate it on the network... Jul 29 01:27:50 ok, router is being weird Jul 29 01:28:00 how so? Jul 29 01:28:05 I regularly use 8G cards on the socket Jul 29 01:28:20 no idea, I just remember something about it in the TRM Jul 29 01:33:12 ok, apparently my omap5-uevm is picking a new random mac address EVERY BOOT... jesus fucking christ, just hash the unique chip id, take the first six bytes, and force the multicast bit to zero and the locally-administered bit to 1, or something Jul 29 01:41:42 ds2: anyhow I'm not really convinced by your argument, since if you want to be able to occasionally use the card slot then to transfer stuff, you'd still need to avoid reusing those pins for other purposes Jul 29 01:41:46 boot also gets complicated Jul 29 01:42:01 right now it can automatically fall back to sd boot if emmc boot fails Jul 29 01:46:02 plus it would have rendered the GPMC inaccessible, and I know there are people explicitly using that Jul 29 01:46:34 omap5-uevm? HE HE HE HE HE Jul 29 01:46:45 so your idea is interesting :) but simply would not have been an option for the BBB Jul 29 01:47:07 so I stand by: they didn't really have choice, and seemed to have done a fairly okay job Jul 29 01:47:18 while the BBGW is really just pointless stupiditu Jul 29 01:47:20 y Jul 29 01:47:32 zmatt: doing it my way would help somefolks and make the remainer no worse then it is right now Jul 29 01:48:01 not true, you can disable eMMC now and regain BBW compat including GPMC Jul 29 01:48:08 oh wait Jul 29 01:48:16 you wouldn't have swapped the interfaces of course Jul 29 01:50:20 my bad... the boot issue however is specific to not swapping the interfaces Jul 29 01:50:37 (and yes, the omap5-uevm is really embarrasing to the name "evm") Jul 29 01:51:44 "MMC0 supports booting from the MMC/SD card cage and also supports booting from eMMC/eSD/managed NAND memory devices with less than 4GB capacity" Jul 29 01:52:03 "MMC1 supports booting from eMMC/eSD/managed NAND memory devices with 4GB capacity or greater" Jul 29 01:52:40 it's not black and white however Jul 29 01:52:48 :) Jul 29 01:52:52 many eMMC might work on either, and ditto SD cards Jul 29 01:53:24 but there are cards known incompatible with the init procedure of MMC1, and eMMC known incompatible with the init procedure of MMC0 Jul 29 01:54:03 if you want to retain freedom for shop for memory suppliers, you don't want to speculate on that Jul 29 01:54:19 shop? heh heh Jul 29 01:54:48 better description is if you want to retain the ability to buy from a memory supplier.... Jul 29 01:55:03 whatever, you get my point Jul 29 01:56:24 you don't want to use MMC1 for the card cage and then get user complaints about cards not working, let alone use MMC0 for the eMMC and then find out the eMMC available isn't compatible Jul 29 02:02:25 that'sjust for booting... Jul 29 02:02:30 anyways Jul 29 02:02:56 I just did a quick test Jul 29 02:03:25 pv /dev/mmcblk1 >/dev/null showed just over 38 MiB/s Jul 29 02:03:44 iirc the clock is 48 MHz, so 4-bit would have limited it to 24 MiB/s absolute theoretical max Jul 29 02:03:57 so 8-bit actually *is* much faster Jul 29 02:04:38 that's on the kingston eMMC though Jul 29 02:04:54 the micron is shit, 4-bit would easily have sufficed Jul 29 02:07:59 my boss has tried to see if he could specifically get BBBs with the kingston flash, but unfortunately not Jul 29 02:09:04 you sort of can Jul 29 02:09:20 might have to pay a premium or you can rework boards as needed Jul 29 02:10:33 well I don't know to whom he talked but they basically couldn't tell anymore, the boards are packaged and they don't keep the batches separated or something, dunno Jul 29 02:11:39 it's annoying they still pretend those two eMMCs are interchangeable :/ Jul 29 02:11:45 you might have to talk to speciality folks... depending on your quantities.. if the need comes up again, ping me Jul 29 02:12:06 electrically, they are :D Jul 29 02:12:10 the desire never went away Jul 29 02:12:36 gotta run... **** ENDING LOGGING AT Fri Jul 29 02:59:58 2016