#milkymist IRC log for Tuesday, 2012-05-29

wpwraklooks reasonable00:58
wpwrakas long as they don't forget to work on getting those goodies into mainline00:58
kristianpaulwolfspraul: http://wiki.xilinx.com/zynq-linux01:16
wolfspraulwhat do you think?01:17
kristianpaulbesides ". A minimal BSP for Zynq has been pushed to the mainline (kernel.org) and is available there. "01:17
kristianpaulwolfspraul: i like :)01:17
wolfspraulxilinx's strategy seems pretty clear to me, as with any good faith effort of course I wish them success!01:17
wolfspraulthe zynq chips will only be generally available next year I think01:17
wolfsprauland we don't know prices01:17
wolfsprauland we have very good problems to solve on our current m101:18
kristianpauli mean if all goes upstream and their zynq cheaps get to floss comunity is a small step01:18
wolfspraulthings that make us more agile and faster in the future01:18
kristianpaulooh sure01:18
kristianpauljust thinking about upstream01:18
wolfsprauland even if the chip is out and price becomes clear, I would still wait a year to see whether it gains tracking01:18
wolfspraulor two01:18
wolfspraulsemiconductor business is tough01:18
wolfspraulmost chips "don't make it"01:18
wolfspraulbut xilinx sees that they come under margin pressure for their 'bare bones' fpgas01:19
kristianpaulyeah i know01:19
wolfspraulso they use those chips increasingly as a cash cow and try to invest in future high-margin businesses01:19
kristianpauli'm in hurry to test lastest xilinx -7 new cheap, but just reading01:19
wolfspraulall fine by me, just that I don't care since they don't give me a monthly paycheck :-)01:19
wolfsprauloh definitely, the -7 chips and ddr3 and all that is something we should already go for01:20
wolfsprauland llhdl :-)01:20
kristianpaulmost chips "don't make it", and both xilinx and altera had been trying this for years..01:22
kristianpaullets see :)01:22
kristianpaulit looks like they got upstream in qemu as well, not bad01:25
kristianpaulanyway :)01:25
wolfsprauland brands are important in chips01:32
wolfspraulat least I wouldn't know how to navigate the semiconductor industry without brands01:33
rohhey wolfspraul01:40
rohseen this: http://www.cl.cam.ac.uk/~sps32/sec_news.html#Assurance01:41
rohi hear that its about ProASIC3 .. but its afaik not noted in the paper01:42
rohin short: backdoor via static aes key in a 'secure' fpga which enables decrypting and dumping of the design stored in there crypted01:42
wolfspraulbut what does that mean for us?01:43
rohthat we are on the right path searching for a way to opensource hardware (and have tests to see if what we do isnt building backdoors)01:44
rohand that even fpga doesnt mean one is safe against such backdoors and attacks (well.. in this case we dont need to care as long as we dont try to trust the hw to store keys for us)01:45
kristianpaulyes, remenber sebastien post about that topic?01:45
rohi like the idea of zync (if open enough)01:46
lekernelhopefully, this publication will make some noise and hackers will give a bit more of a shit about hw, and we won't see things like www.fsf.org/blogs/community/task2-openmoko anymore07:44
bkerowhat publication?08:00
lekernelahem... http://www.cl.cam.ac.uk/~sps32/Silicon_scan_draft.pdf08:36
GitHub54[llvm-lm32] sbourdeauducq pushed 1 new commit to master: http://git.io/2zokFw14:25
GitHub54[llvm-lm32/master] Add bare metal 'elf' environment - Sebastien Bourdeauducq14:25
GitHub95[clang-lm32] sbourdeauducq pushed 1 new commit to master: http://git.io/EvgBWg14:26
GitHub95[clang-lm32/master] Bare metal 'ELF' toolchain that runs as and ld directly - Sebastien Bourdeauducq14:26
GitHub31[milkymist-ng] sbourdeauducq pushed 2 new commits to master: http://git.io/vRxywQ14:41
GitHub31[milkymist-ng/master] software/common.mak: use clang + lm32-elf binutils only - Sebastien Bourdeauducq14:41
GitHub31[milkymist-ng/master] software: platform configuration file for compiler-rt - Sebastien Bourdeauducq14:41
--- Wed May 30 201200:00

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