#qi-hardware IRC log for Sunday, 2014-09-21

DocScrutinizer05wpwrak: sounds about right, yes. 03:19
DocScrutinizer05particularly when you talk about ETB/ETM with all the hw-triggers and funnel and whatnot stuff associated03:20
DocScrutinizer05but for the power it offers, this stuff asks a price in complexity03:21
DocScrutinizer05aside from my polish colleague at ST-M being the one to develop proprietary solution for ETM, I also had a look into that stuff for myself in private, since that colleague's task gave me the idea of implementing something similar like that "ETB local loopback" for linux. Finally I found out there's already some project aiming at exactly that03:26
DocScrutinizer05"post mortem" backtrace of last several 1000s of branch instructions (and obviously all the unconditional linear code in between), plus maybe even data trace for all writes, all incl timestamps, and "for free" from a performance POV. Sounds like mad useful03:29
DocScrutinizer05especially when you don't need that damn debug hw IF with its proprietary "JTAG" and high speed databus or whatever it is, but can do this on *any* ARM Cortex core (which has ETM IP)03:31
DocScrutinizer05indeed the documentation isn't really welcoming you, learning curve is pretty steep03:33
DocScrutinizer05heck, even ARM assembler code is indigestible, particularly when your compiler optimizes the hell out of source code, moving around and swaping and inverting whole sections of code03:34
DocScrutinizer05more often than you'd like to, you find some code doing something en passant that maybe is needed 150 lines of sourcecode later, or been in relation to a sourcode line 150 lines earlier and not been neeeded until now03:36
DocScrutinizer05when you think you could find a few bits of your 'if then else' in code so you understand what it's just doing... forget it, the assembler code structure often has not the faintest similarities with what you wrote in a high level language03:42
DocScrutinizer05wpwrak: only loosely on topic, but at least funny they mention ETM and *MUSB* as an example ;-) http://elinux.org/images/d/d7/Elc2011_gadiyar.pdf04:48
DocScrutinizer05wpwrak: I'm sure you already found http://lxr.free-electrons.com/source/arch/arm/kernel/etm.c04:50
wpwrakDocScrutinizer05: hah, nothing like ETB/ETM. it's much simpler: new chip arrived as "secure". if it's "secure" you can't debug or program it, but you may still be able to do a mass-erase. it says mass-erase is allowed. so i tried to do that.12:41
wpwrakbut ... it doesn't take the command. doesn't ack it. so i went through all the layers to see if any of them had complains, if i had gotten any of the many configuration and control bits wrong, etc.12:42
wpwrakbut i can't find any flaw in what i'm doing. the core is in reset. i tried the debug domain power up process in all possible variations, etc., checked acks and errors at all three layers involved, and so on. nothing.12:45
whitequarkwpwrak: have you complained to the vendor?12:45
whitequarkoccasionally, it works really well12:46
wpwraki'm not sure they can be of much help. i went through all the resources (fora, gdb source, etc.), where these problems are being discussed. and it seems that i'm doing everything right. in fact, there are several areas i address that nobody else seems to mention when struggling with this specific problem.12:52
whitequarkperhaps it's a counterfeit chip?12:56
whitequarki mean, someone i know had similar questions for ti, and eventually they asked him to send the chip back and in a week or two added an errata :D12:57
wpwrakpretty unlikely. a) it comes from digi-key, b) the chip right next to it on the cut tape works fine, c) besides not accepting the mass erase it seems to work fine12:58
wpwrak(errata) hehe :) lemme check if they have an update on these ...12:58
wpwrakof course, having to send my board around wouldn't be such a great thing for me12:59
wpwrakwhee, seems that today officially spring started ! :)12:59
whitequarkoh right southern hemisphere13:00
kyakdo you really call it "spring"?13:02
wpwrakhmm no, no new errata13:03
wpwrakwell, "primavera" in spanish :)13:03
pcercueithat means that fall started here :(13:05
kyakso it's really "spring".. i thought season names are tied to months worldwide13:07
kyakno matter what seasons it feels like13:07
pcercueino, in the southern hemisphere winter is in july/august and summer is in december/january13:09
DocScrutinizer05kyak: now that sounds odd13:11
DocScrutinizer05snow in summer, heatwaves in winter. NAH!13:11
kyakit sounds ok for winter/summer.. but for spring/autumn - dunno :)13:11
DocScrutinizer05Autumn, Summer, Spring, Winter then?13:13
DocScrutinizer05well, given the sun also runs "the wrong way round" on southern hemisphere, it might actually sound correct when seasons also run backwards ;-P13:14
whitequarkso if one cleverly organizes their position, it's possible to achieve eternal summer13:16
pcercueigo to Colombia13:18
pcercueieternal summer there13:18
wpwrakDocScrutinizer05: the sub still rises in the east and sets in the west, even here in the south ;-)13:18
DocScrutinizer05I know13:18
DocScrutinizer05you even ruined the compass ;-P13:19
wpwraksuN even. we have many miracles, but no flying submarines just yet :)13:19
DocScrutinizer05yeah, mueacles like sun from north in the noon13:20
wpwrakon more than half of the planet, that's something entirely normal to experience every year13:21
wpwrakhere, we have it at least always in the north. not sneaking behind our backs and shining from the south. something our northern comrades have to endure :)13:23
DocScrutinizer05figure poles! ;-D13:46
DocScrutinizer05sun from any arbitrary direction 13:46
DocScrutinizer05a NIGHTMARE! ;-P13:47
DocScrutinizer05only mildly mitigated by a 6 months of contiguous night, which I'd adore a lot13:47
dos1uselessd: fork of systemd which tries to make it somewhat sane - http://uselessd.darknedgy.net/ :)14:37
pcercueiI saw this earlier today. Looks interesting, definitely15:08
whitequarkwhat would look interesting is a bonfire15:09
viricburn, code burn15:10
whitequarkcode, flesh, etc15:11
DocScrutinizer05dos1: http://xkcd.com/927/15:45
dos1DocScrutinizer05: this strip is actually indirectly referenced on that page :)15:46
dos1"Wasnt there a famous and overused comic involving stick figures on the folly of standardization?... Nah, couldnt be." :D15:46
DocScrutinizer05nice chess opening though. Finally somebody not only criticises but actually shows that stuff could have done better15:47
DocScrutinizer05dos1: that's how I found it ;-D15:47
DocScrutinizer05(done better) IOW: expose incompetence of systemd cabal15:49
DocScrutinizer05redhat should get nuked15:49
DocScrutinizer05maybe drop canonical on it, from 10000 km height15:51
whitequarkcanonical as a kinetic weapon15:53
DocScrutinizer05ooh, it will have enough kinetic energy when falling down from 10'000km altitude, I'd hope16:19
whitequarknot much more than from 1000km, say16:21
whitequarkinverse square law and all that16:21
DocScrutinizer05yeah, probably true16:53
DocScrutinizer05maybe we need to shoot canonical around the moon for a nice lil slingshot maneuver to gain some 20km/s speed16:54
DocScrutinizer05might take a year til impact at redhat, when we accelerate it contrary to the earth's vector around the sun16:56
DocScrutinizer05actually prolly only 6 months16:56
--- Mon Sep 22 201400:00

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