#qi-hardware IRC log for Sunday, 2012-06-24

Aylahow many pages on the NAND should be considered as being 2048-bytes wide? Only the first two?17:36
larsceither the first two or first four, can't remember exactly, but there is a section about this in the datasheet17:41
Aylaok, so it's not the full eraseblock17:46
larscno it's just the few pages which are loaded by the internal bootrom17:49
Aylaok, that's what I wanted to know17:49
larscthe max pagesize supported by the bootrom is 204817:51
AylaI'm trying to chainload uC/OS-II17:54
Aylafrom ubiboot17:54
wpwrakit's interesting how circuits behave if you expose them to rapidly changing acceleration forces in the order of 3-5 g. solder joint stress fractures. yummy :-)20:24
larscare you building a rocket again?20:30
wpwraknaw, a communication device for moments of public disquiet20:48
wpwrakhaving a bit of trouble with the timing, though. the effect only lasts for less than 100 ms, so you have to shake the stick very rapidly. right now, it's mounted on an IC tube, which flexes a bit. need to find some small aluminium rail ...20:56
rohwpwrak: hihi21:11
rohwpwrak: i like the project desc.21:12
qi-bot[commit] Paul Cercueil: Don't scan the NAND as it's already done elsewhere. (master) http://qi-hw.com/p/gmenu2x/64db1d721:15
qi-bot[commit] Paul Cercueil: Check if USB cable is plugged before returning the battery level. (master) http://qi-hw.com/p/gmenu2x/7484ad221:15
qi-bot[commit] Paul Cercueil: Open battery/USB file handlers only when needed. (master) http://qi-hw.com/p/gmenu2x/0d9977e21:15
qi-bot[commit] Paul Cercueil: Don't re-execute gmenu2x on error, just call main() again. (master) http://qi-hw.com/p/gmenu2x/a21650d21:15
qi-bot[commit] Paul Cercueil: Use function getSkinPath() instead of reading the dir of the binary (master) http://qi-hw.com/p/gmenu2x/cf20f3621:15
qi-bot[commit] Paul Cercueil: Removed variable/function to get the directory of the executable. (master) http://qi-hw.com/p/gmenu2x/124152321:15
larsci build something like that once, but without the accelerometer. it was rather hard to match the speed. and well people around me started to be annoyed by all the waving ;)21:18
wpwrakroh: RF and acceleration sensor work perfectly. it's actually a nice example for OTA updates with atben/atusb on the other end :)21:45
wpwraklarsc: maybe it was to big then :) i need the acceleration sensor because it's text. if i had small symmetric images, i could do without.21:46
larscwpwrak: you are a usb expert. if i do a bulk transfer on a in endpoint, is this basically just telling the hardware to listen for incomming packets or is it also sending some kind of command to the other side of the link?21:49
wpwrakerr, which side are you on ? in the case of host, you need to poll. on the side of device, you wait.21:53
wpwrakroles may look different depending on which layer you're on, too21:54
larsci'm on both side basically ;)21:54
wpwrakoh dear :)21:54
larscfirmware on one side, libusb on the other and i'm trying to understand how it works21:55
wpwrakthe host always initiates the communication. BUT the perspective changes a bit with how the device works21:56
larscthe datasheet for the chip claims that the device has a 1024 byte usb buffer, but the software always sends a 4096 byte bluk request21:56
wpwrakdo you know this introduction to usb yet ? http://www.beyondlogic.org/usbnutshell/usb1.shtml21:56
larscso i'm wondering will I see only transfers on the host side where only 1024 byte of those 4096 bytes buffers are used?21:58
wpwraki think you're likely to see only 1024 bytes at a time, yes21:59
wpwrakunless your host combines things behind your back21:59
larscok, thanks22:00
larschm, on the other hand the standard says that a bulk transfer will be 512 bytes max22:09
wpwrakare you sure ?22:13
wpwrakat what speed are you operating ?22:13
larschigh-speed i guess22:16
wpwrakthat's just the packet size. you can have more than one packet per transfer22:18
larscah ok22:20
larscso you send a request, the device respons with multiple data packets followed by an ack22:22
wpwrakat the wire level, i think you only have one DATAx per ACK. but if your api lets you use larger sizes (e.g., 4096), you have multiple IN-DATAx-ACK.22:25
Aylaguys, there's proprietary-licensed stuff on the Qi repo22:27
Aylathere's the Arial font on the source tree of gmenu2x22:28
larsceverybody hides ;)22:38
kristianpaulAyla: is just hosted or already used?22:59
qi-bot[commit] Paul Cercueil: Remove unused (proprietary-licensed for some) XCF fonts. (master) http://qi-hw.com/p/gmenu2x/16d66d022:59
Aylawhat are you talking about? :)22:59
AylaI just removed them23:03
Aylathey were hosted, but not used23:03
kristianpaulah, thanks !23:05
kristianpaulhe just saw the commit log :-|23:05
--- Mon Jun 25 201200:00

Generated by irclog2html.py 2.9.2 by Marius Gedminas - find it at mg.pov.lt!