#milkymist IRC log for Friday, 2013-09-06

GitHub124[migen] sbourdeauducq pushed 1 new commit to master: http://git.io/w80fTw14:35
GitHub124migen/master 4fb3e97 Sebastien Bourdeauducq: genlib/fsm: add entering/ongoing/leaving methods14:35
sb0ftdi_eeprom is so incredibly bad ...15:42
sb0for any practical purpose, you need to hack into its source to hardcode what you need because the normal behaviour is completely unusable15:43
wpwrakyou're overly specific. ftdi sucks. it's as simple as that :)15:44
sb0anyway, the mixxeo now enumerates correctly on USB :)15:45
sb0about to load the first bitstream, after finding out why the urjtag svn won't compile15:46
wpwrak(ftdi avoidance) doing fast jtag is a little hard, but if all you use it for is flashing and serial anyway, you may as well use an MCU. keep the fpga tri-stated, then dfu to the flash through the MCU.15:46
wpwrakso you got the boards ? you ought to post about that, with pictures. keep people interested :)15:47
sb0I have two fully assembled boards, and 8 more have gone through SMT and still need through-hole components15:48
sb0need to find a camera somewhere. also, I'm on the plane every second day atm ...15:49
sb0with, of course, maximum annoyance at airports regarding the contents and/or weight of my luggage. the mass spectrometer I had on one segment got my name called on the PA 5min after checking in the bag :)15:51
wpwrak;-)))15:51
wpwrakyou had that with you on then way into syria or out ? :)15:52
sb0checking for LIBUSB... yes15:53
sb0checking for LIBFTDI... no15:53
sb0checking for LIBFTDI... yes15:53
sb0...15:53
larscchecking for LIBFTDI... maybe ;)15:55
wpwrakchecking for LIBFTDI... maybe15:55
larscha!15:55
wpwrak(-:C15:55
wpwrakchecking for LIBFTDI... [y/n] ? _15:56
sb0abort/retry/fail?15:56
wpwrakoh, do i WISH we could abort autoconf ...15:56
sb0JTAG works :) FPGA is detected16:19
sb0and yet another s6 stepping ...16:19
sb0fjmem loaded16:22
wpwrakkewl16:23
sb0hmm, and now this: http://pastebin.com/MtAn8cZS16:24
wpwraki could kick myself for not thinking of the direct usb-to-flash approach earlier. that avoids the whole jtag and ftdi mess.16:24
wpwrakhmm, that's when you wish you had used spi flash ...16:25
wpwrakwell, let's hope it's nothing too messy16:25
sb0yes, and hopefully SPI flash would not have caused the corruption nightmare either16:34
sb0I wanted an easy route from the ML401 and that's what I got ...16:35
sb0interestingly enough, Xilinx seems to get away without reset chip. but not me.16:35
wpwrakthe chip in question is also quite weak when it comes to protection features. there are better approaches.16:36
sb0led is blinking :)16:36
wpwrakwhat i like best is "soft" protection plus a trapdoor: start with soft protection disabling all writes, then have "trapdoor" write-protect bits that can be set but are only cleared on reset.16:38
wpwrakthat way, a boot loader can set up any protection scheme one wants, then pass control to the application. for a controlled shutdown, the application could then protect everything through the trapdoor mechanism.16:39
wpwraksince all the protection is volatile, you also don't have incremental tearing down of protections, like first clearing the (persistent) "lock" bit and later doing the actual corruption16:40
wpwrakthe trap door mechanism does exist in some flash chips. i don't think any start with the "soft" protection set to read-only, though. (probably in order to be compatible with ancient chips that had none of this)16:42
wpwrak(led blinking) mission accomplished :)16:42
cdeit's a small voltage drop for leds, but a giant step for sb0!16:56
cdeor should I say: "#milkymist, this board has landed" ;)16:57
wpwrakis the "NORton, we have a problem" solved ?17:07
sb0was in a (quick) meeting. forgot that the clock pin has been changed...17:16
sb0recompiling fjmem ...17:16
sb0"highly Zeitgeist-compatible" :))17:21
wpwrakone's paranoia is another one's opportunity :)17:22
sb0with the correct clock pin, flash detection works17:26
sb0:)17:26
sb0BIOS is running, now memtest fails17:30
wpwrakwhee !17:31
sb0hmm, when bit-banging the SDRAM, it responds correctly though17:32
sb0but memtest reports 100% failed words. sounds like a relatively easy logic design bug.17:32
sb0memory reads from LM32 are all 0's17:33
sb0hmm17:34
sb0_ethernet is ok20:54
wpwrakHDMI ? :)20:59
sb0_needs to fix SDRAM first... and with a solid 1hr runtime of the ISE shitware on my travel laptop, this takes a while21:01
cdehave you tried turning it off and on again?21:01
wpwrak;-))21:03
acathla:)21:04
GitHub172[mibuild] sbourdeauducq pushed 1 new commit to master: http://git.io/pNgjaA21:10
GitHub172mibuild/master d57344b Sebastien Bourdeauducq: platforms/mixxeo: add LED21:10
GitHub15[fjmem-m1] sbourdeauducq pushed 2 new commits to master: http://git.io/UfT5bw21:13
GitHub15fjmem-m1/master 8417e28 Sebastien Bourdeauducq: Add build folder21:13
GitHub15fjmem-m1/master 8c1ea67 Sebastien Bourdeauducq: add Mixxeo support21:13
GitHub39[fjmem-m1] sbourdeauducq pushed 1 new commit to master: http://git.io/uFgYnQ21:19
GitHub39fjmem-m1/master aff0207 Sebastien Bourdeauducq: README: update21:19
sb0_found the problem with SDRAM. Florent's patch had broken it.21:36
--- Sat Sep 7 201300:00

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