#qi-hardware IRC log for Wednesday, 2014-08-27

apeleteHi larsc 07:11
apeletelarsc: is it possible to do early debugging on mips ?07:11
apeletetrying to debug the dma-jz4740.c driver (modified to work with jz4770), and my kernel seems to be going into panic *before* kgdb driver is loaded07:14
apeletehttp://paste.debian.net/117757/07:14
apeletewhich could mean dma driver is loaded before the serial_8250 driver07:16
apeletelarsc: is it possible to load the serial driver earlier in order to have early debugging ?07:19
larscapelete: yes, I think we have support for early_printk on jz474008:16
larscset CONFIG_EARLY_PRINTK=y08:17
larscsee arch/mips/jz4740/prom.c and prom_putchar()08:18
apeletelarsc: I set CONFIG_EARLY_PRINTK=y on jz4770 too, but it didn't help08:19
larscif there are multiple UARTs you may have to select the correct one08:21
larsccheck the jz4770 prom_putchar function08:22
apeletelarsc: looks like prom_putchar is writing to a UART2_BASE register address08:26
apeletedoes that mean it might be the wrong UART to write to ?08:26
larscI don't know, which uart are you connected to?08:27
apeletettyS2 from the cmdline point of view08:27
larscsounds like uart 208:28
apeletelarsc: compared to an old jz4740 boot log, serial driver is loaded late too -> http://paste.debian.net/117756/08:30
apeleteloads at line 8208:30
larscearly printk bypasses the serial driver09:06
apeletelarsc: so it's not possible to launch the serial driver earlier than what is done on jz4740 ?09:43
DocScrutinizer05make sure your UART doesn't load DMA driver ;-)12:40
DocScrutinizer05aaah >><larsc> early printk bypasses the serial driver<< makes sense12:41
DocScrutinizer05wpwrak: (browsers) 12:43
DocScrutinizer05TECHNISCHE WARNUNG TW-T14/007912:43
DocScrutinizer05    Titel: Sicherheitsupdate für den Google Chrome Browser12:43
DocScrutinizer05>> Der Google Chrome Browser vor Version 37.0.2062.94 für Windows, Mac und     Linux enthält mehrere kritische Sicherheitslücken. Diese können von einem     entfernten Angreifer ausgenutzt werden, um beliebiegen Programmcode     auszuführen, .....<<12:45
DocScrutinizer05oops ECHAN12:46
DocScrutinizer05well, not that bad anyway12:46
nicksydneyFREE! http://www.engadget.com/2014/08/27/imagination-tech-mips-dev-board/?ncid=rss_truncated13:01
ysionneauwould be cool if Ingenic would liberate as open source HDL its Xburst core14:28
ysionneaulike it would be cool if I win the lottery also14:29
larscotror space travel!14:40
larscfor everybody!14:40
wpwrakDocScrutinizer05:  hmm,sounds nasty. thanks ! let's hope chromium on ubuntu gets fixed quickly14:45
wpwrakwhitequark: speaking of quantum conscience and such ... http://www.smbc-comics.com/index.php?id=2657#comic15:24
arhuacowpwrak: Old but good! (new to me).15:47
wpwrak(old) yeah, found smbc only recently. and i can't spend all day to catch up :)15:49
arhuacowpwrak: Happens to me with XKCD. I do not know if it is (just) me, but now they are way too many and I can no longer keep up with them. Even if they are in my feeds.15:52
wpwrakyes, the list keeps growing. slowly, but steadily. so far, i can still manage but the comicalypse seems inevitable in the long run15:55
pcercueiwej, hi, I have a request for GMU: make the file selector point to $HOME when you start GMU, instead of the data directory (which is generally in /usr/share in Linux)15:56
Action: kristianpaul removed its xkcd rss time ago15:59
wejpcercuei: hi, actually, that directory is configurable (DefaultFileBrowserPath), which by default is set to the working directory (which might be /usr/share as you said). The reason for that "current directory" default simply was that on the handhelds were Gmu originated, it was useful to assume that directory. On PCs and similar devices the user's home directory probably is more useful.16:00
wejmaybe i'll change the default to the user's home16:00
pcercueihmmok16:00
pcercueiI could do DefaultFileBrowserPath=/usr/local/home but then it may break if we decide to move the home directory16:01
pcercueiother than that, I made an OPK for it16:02
wejmaybe i could add support for ~ in the path, so one could specify that and it picks the right home directory of the user, whatever that is16:02
pcercueiah, that would be great, yes16:02
weji'll add it to my todo list ;)16:02
wejso i don't forget :D16:02
pcercueithanks!16:02
pcercueiyou said that GMU is resolution-independent now?16:04
wejyes, it is16:04
wejyou can even run it windowed and resize its window16:04
wejon platforms that support windows that is ;)16:07
pcercueiand does it use the highest resolution by default?16:09
pcercueior the current screen's resolution?16:09
wejusually you specify the desired resolution it Gmu's config file16:09
wejSDL_frontend.Width=64016:09
wejSDL_frontend.Height=48016:09
wejlike so16:09
wejand then there is SDL_frontend.Fullscreen=yes/no16:09
wejfor devices that support windowed and fullscreen operation16:10
pcercueiso if I set SDL_frontend.Width=640 SDL_frontend.Height=480 but I only support 320x240 maximum, will it crash?16:10
wejit won't crash, but it won't run properly either. the sdl frontend will try to use that resolution and if that fails it does not continue to start16:10
pcercueihmmok16:10
apeletelarsc: testing kgdb with a working kernel -> http://paste.debian.net/117889/19:40
apeleteit loads after the serial driver as expected, which itself loads late in the boot process, after the lcd driver...19:41
apelete...looks like I'm screwed :-(19:42
larscI still don[t understand the problem19:43
mthwhat do you want to debug with it?19:44
larscor is the problem that kgdb is not working?19:45
apeleteI want to debug the dma driver19:45
mthdo you need to catch it before it initializes?19:46
apeleteproblem is the kernel is crashing in the dma driver before kgdb driver is loaded19:46
larscyou can try to build the dma driver as a module19:46
apeletedidn't think about that, silly me19:46
mthalthough that won't work if the module is on an SD card and the mmc driver depends on the DMA module...19:47
apeletemth: by the time I start working on the mmc driver I expect the dma one to be stable enough19:47
apeleteat least I hope so :)19:48
pcercueikgdb requires you to hit some keyboard keys, no?20:02
wh1tequarkfun thing: http://hackaday.com/2014/08/27/thp-semifinalst-laser-solder-paste/20:19
wh1tequarkif pointless a bit20:19
apeletepcercuei: the kernel can be configured to automatically pause during boot and wait for gdb to be attached to it20:33
DocScrutinizer05that's cool21:06
whitequarkDocScrutinizer05: the laser thing?21:15
DocScrutinizer05laser thing? nah, gdb sync21:29
whitequarkoh, right21:29
DocScrutinizer05laser thing the 3W lasers intrigue me. Would like to use those for proper soldering tasks - though the absolute failure to work on covered solder points like BGA is obvious21:34
whitequarkindeed, it's sorta pointless except as a curiosity21:35
whitequarkI wonder what is the absorption spectrum of solder21:36
whitequarklike, maybe IR won't be reflected? I don't know21:36
DocScrutinizer05WOW!  https://sites.google.com/site/dtrlpf/home/diodes/ndb7a75-3-5w-445nm-diodes21:47
whitequarkhuh21:49
DocScrutinizer05with *that* laser power you actually can start tooling21:51
whitequarkI dunno, you can get a tube with hundreds or thousands of W pretty easily21:52
DocScrutinizer05I don't think you can get it (operating) for that pricetag21:55
DocScrutinizer05nor is it in a format where you could integrate it into e.g. a CNC head21:55
DocScrutinizer05nevertheless I might be interested *where* to get such tube easily. Since I'm musing about building a nice show laser since long21:57
DocScrutinizer05it's simply I'm adicted to light emitting objects, and particularly monochromatic ones. And when they are even coherent and focused, all the better21:58
DocScrutinizer05doesn't mean the tritium light keyring in my pocket lost its fascination ;-)21:59
whitequarkDocScrutinizer05: ebay22:02
whitequarksome CO2 laser tube is pretty cheap22:03
whitequarkwell22:06
whitequark$60022:06
whitequarkmaybe half that if you DIY22:06
DocScrutinizer05CO2 is fine, but has one obvious flaw for an addict to light emitting objects - that addiction is pretty much a perception driven thing22:16
DocScrutinizer05yeah, I checked for 50W range CO2 lasers nly 6 months ago. Wasn't too easy to spot something that sounded like it might actually pan out and work in the end22:17
DocScrutinizer05the 70$ blue one is awesome  https://sites.google.com/site/dtrlpf/home/diodes/9mm-445nm22:29
--- Thu Aug 28 201400:00

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