| 00:15.50 | *** join/#edev kergoth (n=kergoth@ip68-3-163-239.ph.ph.cox.net) |
| 01:26.22 | MonMotha | where do all of you manage to find racks? |
| 01:26.26 | MonMotha | they always seem so expensive |
| 01:27.45 | agaffney | I found an old-ass crappy one at a local corporate off-lease reseller |
| 01:34.01 | MonMotha | even the old, beat up ones I find people seem to be asking several hundred for |
| 01:43.10 | solar | Mirell: yeah 1U server is overkill for a soho router. |
| 01:45.33 | Mirell | solar: Howso? |
| 01:45.43 | Mirell | I can get a Dell 1U for $650 |
| 01:46.08 | solar | well. Do you have gigabit into the home? |
| 01:46.18 | Mirell | Nope |
| 01:46.32 | solar | then for like $65 you can fill the same role |
| 01:47.01 | Mirell | Eh. |
| 01:47.02 | ds2 | hey solar, did you loose power yesterday |
| 01:47.40 | solar | ds2: I don't think I did |
| 01:48.08 | solar | Mirell: ever checked out routerboards? |
| 01:48.14 | Mirell | I have not. |
| 01:48.16 | solar | http://routerboard.com/prices.html |
| 01:48.25 | ds2 | guess that was not that important of a power line... saw a power line dancing on 101 shoulder near Mathilda |
| 01:49.05 | solar | I picked up one several years ago when they were mipsel based. It was the only firmware in the my history of dealing with embedded/sbc devices that I did not want to gut the default firmware |
| 01:49.20 | Mirell | Well, their website crashed Safari... |
| 01:49.38 | Mirell | Oh, there we go.. |
| 01:49.54 | solar | Safari does not make a linux client. So you are doing something wrong from the start :p |
| 01:50.11 | Mirell | I gave up on Linux on the Desktop for right now. |
| 01:50.37 | solar | lots of developers seem to be doing that. |
| 01:50.58 | solar | OSCON,LWE,CELF last year. seemed like 90% of the ppl were on macbook pros etc.. |
| 01:51.04 | ds2 | Linux desktop stuff is getting less and less usable |
| 01:51.06 | Mirell | 'tis what I have |
| 01:51.15 | ds2 | it is quickly nipping on MS's heels for lack of usability :( |
| 01:52.12 | Mirell | is trying to figure out exactly what RouterBoard is... |
| 01:52.17 | solar | dunno. I open files, read email, listen to music, code in my editor |
| 01:52.30 | ds2 | what mailer do you use? |
| 01:52.57 | solar | Mirell: just sbc's with lots of eth ports or a decent number of minipci slots |
| 01:53.06 | solar | evolution |
| 01:53.15 | solar | it's outlookish |
| 01:53.44 | ds2 | evolution was unuseable (at least on the EeePC) |
| 01:54.02 | solar | how so? like slow? |
| 01:54.21 | solar | eepc is about ~1ghz ? |
| 01:54.27 | ds2 | no, the UI... too many clicks to do things and it won't give me control of what it shows |
| 01:54.42 | ds2 | 1.6GHz but it is more of equiv to a 600-800MHz Celery |
| 01:56.10 | solar | so feels like a pent-2 ? (the celery was never known for it's clock rate to prefromance ratio) |
| 01:56.31 | ds2 | yep |
| 01:56.52 | ds2 | older Eee's was an underclocked celery and it is compareable to the 1.6GHz Atom |
| 01:57.40 | Mirell | Only reason I used Evolution was because of Exchange... |
| 01:57.50 | solar | does it handle stuff like youtube or mplayer ? |
| 01:58.10 | ds2 | yep, it ships with a legal white washed mplayer and has stock adobe flash |
| 01:58.38 | solar | got a kill-a-watt handy? |
| 01:58.40 | ds2 | it is sort of usable after sticking normal KDE in there; the 900 does have a annoying touch pad button tough |
| 01:58.55 | ds2 | yeah, I do but I am not home ATM |
| 01:59.35 | ds2 | the other irritating factor is the power brick is 12V regulated in |
| 01:59.47 | ds2 | makes it a pain to find cheap car power sources w/o using an inverter |
| 02:00.12 | solar | if you don't mind. could you take a few readings off it? I've been toying around with the idea of getting one to lower my power needs at home. current laptop ranges from 45W idle 95W peak |
| 02:00.27 | ds2 | sure, I'd be happy to |
| 02:00.51 | ds2 | think the cube is rated for 36W max; but I'll get some real numbers |
| 02:00.58 | ds2 | you could always go ARM ;) |
| 02:01.33 | solar | while I like arm. I don't have a fast enough build host to justify trying to put it on a desktop |
| 02:01.54 | solar | and I don't feel like fixing that many bugs quite yet |
| 02:02.55 | ds2 | it is NOT THAT buggy ;) |
| 02:03.08 | ds2 | the whining isn't too loud with the Beagle folks |
| 02:03.30 | solar | I mean bugs with cross-compiles and all the pkgs in foss |
| 02:03.44 | ds2 | Oh that, thought you meant the alignment issues on ARM |
| 02:04.42 | ds2 | dist-cc with a farm of ARM's powered up on demand? |
| 02:05.01 | ds2 | PM on top of PM |
| 02:06.18 | solar | thats even more buggy really. distcc's gains are debateable these days. the linker itself is where the toolchain spends a good deal of time. |
| 02:06.29 | solar | and ./configure scripts etc that can't be offloaded |
| 02:10.13 | ds2 | <PROTECTED> |
| 02:10.41 | ds2 | too much time spent fighting them; they just aren't worth it |
| 02:11.28 | solar | I'll take autotools any day of the week over perl's ./Configure :) |
| 02:11.57 | ds2 | I'll take editing config.h's over either of those 2 |
| 02:12.05 | solar | Mirell: I use one of these for my router http://soekris.com/pictures/net4521/net4521_BO_front_big.jpg |
| 02:14.26 | *** join/#edev brolin (n=brolin@190.157.13.75) |
| 02:21.13 | Mirell | solar: Hmm... |
| 02:21.30 | Mirell | I just wonder about their ability to be a router/IDS/vpn concentrator |
| 02:24.24 | Mirell | I mean, it appears quite cool |
| 02:27.12 | ds2 | solar: what's that board draw power wise? |
| 02:30.05 | MonMotha | oddly, Beagleboard has some quite nasty errata |
| 02:30.37 | ds2 | snickers |
| 02:30.44 | ds2 | 'some' |
| 02:31.54 | ds2 | MonMotha: which ones caught your interest? |
| 02:56.48 | MonMotha | ds2: I remember something about the USB host not working |
| 02:56.58 | MonMotha | that seems to be a pretty major errata |
| 02:57.26 | ds2 | Oh that... there are 2 USB ports on there; the USB host only one is not working on RevB but the OTG port is (sort of) |
| 02:58.09 | MonMotha | I see |
| 02:58.32 | MonMotha | yeah, I do kinda recall that |
| 02:58.43 | ds2 | at ESC in Boston, we had it working with a DVI monitor + USB keyboard + USB mice |
| 02:58.54 | ds2 | but there are other gotchas |
| 02:59.03 | MonMotha | the impression I got was that it is usable, but you have to be careful |
| 02:59.37 | MonMotha | seems to have been developed on a budget, so perhaps not as much "step back and think about what we're doing here" as should have occured |
| 02:59.41 | ds2 | MUSB is not exactly a robust implementation; I have personally seen 2 of them and... |
| 03:00.08 | ds2 | actually the beagle bugs are present in all the OMAP3 boards |
| 03:00.25 | MonMotha | really? |
| 03:00.28 | ds2 | apparently routing the USB Host only (EHCI) block through certain pins don't work |
| 03:00.41 | MonMotha | they just fudged up the reference design or something? |
| 03:01.01 | ds2 | and it was discovered that routing it through yet another set does work (solely by coincidence on another board) so revC fixes it by rerouting the USBport |
| 03:01.04 | MonMotha | oh, that's a muxable function? I guess with an external PHY it could be. |
| 03:01.23 | ds2 | yes, it is muxable...that port can be routed/used at least 5 different ways |
| 03:01.54 | MonMotha | I guess I'm used to onboard USB PHYs, which are generally not used as muxable I/O lines due to the somewhat unique physical layer requirements of USB (especially high speed) |
| 03:01.55 | ds2 | I have worked with the more expensive, "official" evalboards and they all do the same. The EHCI port does not work right |
| 03:02.12 | ds2 | they have an onboard PHY mode too, donno if that works or not |
| 03:02.32 | MonMotha | wait, if there's an onboard PHY, why wouldn't you use it? |
| 03:03.04 | ds2 | donno, maybe there is another errata for it ;) |
| 03:03.09 | MonMotha | heh |
| 03:03.20 | MonMotha | and I thought the Atmel parts had bad errata :) |
| 03:03.43 | MonMotha | "Major advertised function X of the chip doesn't work. Workaround: Do not use X." |
| 03:04.05 | ds2 | the OMAP3 processors have several eval boards (from TI) - Beagle, Zoom, EVM, SDP (in asscending price order) along with other SBC's based on it |
| 03:04.32 | MonMotha | Sounds like they may all be based off of one reference design that has issues |
| 03:04.42 | MonMotha | (well, and the part has issues, too) |
| 03:05.45 | ds2 | SDP is the mother of all boards |
| 03:06.31 | MonMotha | I hate broken reference designs |
| 03:06.39 | solar | ds2: had to break my internet connection to find out. But looks like 6W. I'm using 14.3M on the fs http://rafb.net/p/2xVOa276.html |
| 03:06.44 | MonMotha | kinda defeats the point |
| 03:06.47 | ds2 | they are only on their 3rd major rev of the silicon ;) |
| 03:06.55 | MonMotha | jeesh |
| 03:07.02 | MonMotha | sounds like they majorly messed it up |
| 03:07.19 | ds2 | solar: not bad... same ball park as the WRT's and that is an x86 board |
| 03:07.55 | solar | looks like I have the wifi off |
| 03:08.02 | MonMotha | I've been meaning to build myself a router |
| 03:08.03 | solar | let up up that and see |
| 03:08.12 | MonMotha | I think I can come in at about 3W sans wifi |
| 03:08.12 | ds2 | there is so much crap on there, it is hard not to mess up... 2 processors + at least 3 different blocks they do not own plus other glue |
| 03:08.39 | MonMotha | ds2: true, but I'd like them to work the bugs out before offering the silicon up for sale |
| 03:09.06 | ds2 | MonMotha: to be fair, only 2 major revs of the silicon was officially offered for sale |
| 03:09.12 | MonMotha | dev cycles seem to have gotten so short in the embedded/mobile industry that it's almost impossible to buy bug-free parts, and that makes bug-free end designs hard |
| 03:09.26 | ds2 | Note that I say 'OMAP3' not OMAP35xx or another name |
| 03:10.05 | ds2 | I really would like to get a better view of how the Freescale i.MX stuff is going... but they don't seem to be as easy to get close to |
| 03:10.32 | MonMotha | ds2: oddly, I had the opposite experience |
| 03:10.51 | MonMotha | the TI people seemed uninterested, but we had Freescale people who seemed trying very hard to get a design win |
| 03:11.11 | ds2 | *shrug* Motorola and now Freescale always seem to be hard to get close to |
| 03:11.22 | MonMotha | weird |
| 03:11.41 | MonMotha | anything "non-standard" from TI seemed impossible to buy, and that included things like OMAP |
| 03:11.52 | MonMotha | they just didn't seem to care unless you were talking 100ku+ |
| 03:11.59 | ds2 | OMAP being purchaseable is a new thing; like since last June |
| 03:12.16 | MonMotha | that's the timeframe I'm talking in |
| 03:12.26 | MonMotha | perhaps our TI area rep just wasn't "with the times" |
| 03:12.41 | ds2 | there are actually 2 OMAP3's being sold by 2 groups... one you have to guarantee in stone that you will buy 394324801234890234901234890234890238492384902340 chips before you can see the specs |
| 03:12.54 | ds2 | the other group's version is available through digikey |
| 03:12.58 | MonMotha | I see |
| 03:13.07 | ds2 | Catalog vs Wireless |
| 03:13.09 | MonMotha | I wonder why the difference |
| 03:13.38 | ds2 | different groups |
| 03:13.49 | MonMotha | well, clearly |
| 03:13.58 | MonMotha | but I wonder why the two groups take such a different stance |
| 03:14.24 | ds2 | I can guess but it is probally not appropriate to say in a public forum ;) |
| 03:15.42 | ds2 | time to fight traffic |
| 03:15.45 | ds2 | bbl |
| 03:26.47 | Mirell | solar: Do you use RouterOS on it, or do your own? |
| 03:27.49 | solar | Mirell: the routerboard is what I had a few of years ago. I'm using the http://soekris.com/net4521.htm now. |
| 03:29.00 | solar | it could take a vpn mini-pci card. But my existing ssh traffic does not really seem to warrant the need for one for me. |
| 03:30.05 | Mirell | Ahh |
| 03:30.50 | Mirell | 'cause the RouterBoard looks pretty cool as a general hacker-friendly router thingie |
| 03:31.11 | solar | yeah |
| 03:31.30 | solar | the routeros. yeah that the was one that did not make me puke |
| 03:32.06 | solar | had forgot the name of that one |
| 03:33.47 | Mirell | Well, I figure I'd try for my own variant, seems a standard MIPSEL based board... |
| 03:34.49 | solar | the routerboard? I think they are real ports vs that hacked up setup that linksys used |
| 03:39.28 | *** join/#edev davidc__ (n=davidc__@S01060002553240a8.vc.shawcable.net) |
| 03:56.41 | Mirell | Yah, so I'm reading. |
| 03:56.43 | Mirell | How odd... |
| 03:56.57 | Mirell | Half my apartment building lost power.. |
| 03:59.54 | Mirell | Wonder if some kid flipped the breaker for this building... |
| 04:01.03 | MonMotha | Mirell: utility company might have dropped a phase |
| 04:01.09 | *** join/#edev JoeLlama (n=snork@unaffiliated/joellama) |
| 04:01.17 | MonMotha | that could result in chunks of the building losing power while the rest is fine |
| 04:02.34 | MonMotha | though normally that'd result in either 1/3 or 2/3 going out, not 1/2 |
| 04:18.49 | *** join/#edev mbuf (n=shakthim@61.16.248.242) |
| 04:30.14 | Mirell | MonMotha: Well, 1/2 was a random number, more than my place, less than the whole building |
| 04:30.29 | Mirell | But huh, hadn't thought of that. |
| 04:31.45 | *** join/#edev redblue (i=star@ppp171.108-253-207.mtl.mt.videotron.ca) |
| 04:32.17 | *** join/#edev DrZimmerman (n=theo@77-56-240-146.dclient.hispeed.ch) |
| 04:36.00 | *** join/#edev harpal (n=Harpal@122.169.108.195) |
| 04:41.19 | *** join/#edev harpal_ (n=Harpal@59.162.116.80) |
| 04:42.42 | *** join/#edev onnoval (n=sam@123.252.146.52) |
| 05:51.02 | *** join/#edev davidc__ (n=davidc__@S01060002553240a8.vc.shawcable.net) |
| 05:59.44 | *** join/#edev redblue (i=star@ppp013.108-253-207.mtl.mt.videotron.ca) |
| 06:03.03 | *** join/#edev onnoval (n=sam@123.252.146.52) |
| 06:05.01 | *** join/#edev JoeLlama (n=snork@unaffiliated/joellama) |
| 06:17.12 | *** join/#edev mparnaudeau (n=parnaude@little-black-box.vmware.com) |
| 06:24.24 | *** join/#edev sleipnir (n=Sleipnir@p4FDD5ED6.dip.t-dialin.net) |
| 07:08.10 | *** join/#edev flameman (n=toor@net-93-65-1-167.cust.dsl.vodafone.it) |
| 07:11.36 | *** join/#edev lyakh (n=lyakh@p57BD2C42.dip0.t-ipconnect.de) |
| 07:30.07 | methril|work | yo |
| 07:30.17 | MonMotha | hello |
| 07:35.31 | flameman | hello |
| 07:36.00 | flameman | it seem the Angstrom has added the "dht-walnut" in their supported stuff |
| 07:55.38 | *** join/#edev KazaLite (n=mohsin@116.58.8.130) |
| 08:08.22 | *** join/#edev boris_OmegA (n=boris@194.51.141.172) |
| 08:17.22 | *** join/#edev hw (n=hw@p578b3905.dip0.t-ipconnect.de) |
| 08:19.10 | *** join/#edev davidc__ (n=davidc__@S01060002b360aacd.vc.shawcable.net) |
| 09:10.36 | KazaLite | hi all |
| 09:11.20 | KazaLite | on arm architecture, if my function takes four or less than four parameters, these parameters are copied in registers instead of stack? |
| 09:12.45 | MonMotha | I do believe that the calling convention goes something like that, yes |
| 09:15.21 | mparnaudeau | KazaLite: yes, but maybe not the case with some exotic ABI |
| 09:17.04 | mparnaudeau | There are (were ?) many ABIs on ARM |
| 09:21.46 | KazaLite | is there not just one ABI specified by ARM? |
| 09:22.23 | MonMotha | most architectures do not have a single specified ABI |
| 09:22.30 | MonMotha | in fact, Linux has used two ABIs on ARM over time |
| 09:24.50 | KazaLite | i see |
| 09:26.43 | KazaLite | another thing is i have experience of writing device drivers for nucleus plus rtos...i tried to figure out how should i write device driver for linx, but it seemed very strange.....its not very clear that how to write one for linux |
| 09:26.44 | MonMotha | there are actually like 3 or 4 common ABIs on x86, for example |
| 09:27.06 | MonMotha | device drivers in Linux depend very much on what subsystems you need to hook into |
| 09:27.10 | MonMotha | they all have their own nuances |
| 09:27.28 | KazaLite | looked at couple of books like linux device drivers etc but nothing offers some practical approach |
| 09:27.39 | MonMotha | best thing to do is usually comb through the source for a similar device (i.e. one that uses the same busses, userspace APIs, etc.) and start looking up things you don't know |
| 09:28.26 | KazaLite | do you have some document about any written low level driver's code breakup? |
| 09:28.39 | MonMotha | no, not really |
| 09:28.50 | MonMotha | it really does depend on what you're trying to talk to |
| 09:29.02 | MonMotha | a driver for a PCI device looks very different from a driver for a USB device |
| 09:29.18 | MonMotha | and a driver that shows up as a network interface looks different from a driver that shows up as a character device |
| 09:29.30 | KazaLite | i seriously feel there should be some reference document for every subsystem |
| 09:29.41 | KazaLite | well written document |
| 09:29.53 | MonMotha | some systems do have one |
| 09:30.02 | KazaLite | does usb have? |
| 09:30.09 | MonMotha | other subsystems have "example" drivers that attempt to exercise most functionality and are well commented |
| 09:30.11 | KazaLite | for usb function controller drivers? |
| 09:30.12 | MonMotha | I think so |
| 09:30.45 | KazaLite | cool |
| 09:32.01 | MonMotha | http://www.lrr.in.tum.de/Par/arch/usb/usbdoc/ |
| 09:32.12 | MonMotha | also read various device drivers. You'll notice some patterns |
| 09:32.32 | MonMotha | the "Linux LXR" is very handy for probing your way through the source. It'll hotlink all procedure names, struct names, etc. |
| 09:33.14 | KazaLite | isn't this document outdated? |
| 09:33.24 | KazaLite | or still practical? |
| 09:34.16 | MonMotha | I don't think the APIs have changed a whole lot since then |
| 09:34.35 | MonMotha | also check out drivers/usb/usb-skeleton.c |
| 09:34.49 | KazaLite | cool |
| 09:34.56 | MonMotha | that's a working USB driver that doesn't do anything |
| 09:35.55 | KazaLite | and how class drivers and controller drivers are separated in linux? if i layer things it will be [hardware][controller driver][usb stack...in linux i say usb sub-system][class driver][application] |
| 09:36.11 | KazaLite | is this hierarchy true for linux as well? |
| 09:40.19 | MonMotha | no |
| 09:40.30 | MonMotha | the USB heirarchy is 100% based on Windows driver model |
| 09:40.34 | MonMotha | Linux does not fully adopt that |
| 09:40.56 | MonMotha | however, it's not totally off-base, either |
| 09:53.28 | KazaLite | cool |
| 10:01.47 | Mirell | Hmm... |
| 10:36.25 | *** join/#edev VK4MSL (n=vk4msl@203.94.56.252) |
| 10:43.24 | *** join/#edev likewise (n=likewise@82-171-51-231.ip.telfort.nl) |
| 11:18.54 | *** join/#edev gustavoz (n=gustavoz@imhotep.toptech.com.ar) |
| 11:30.19 | *** join/#edev mbuf (n=shakthim@61.16.248.242) |
| 11:37.50 | *** join/#edev pirho (i=pirho@gateway/gpg-tor/key-0x2CEEC9CB) |
| 12:12.05 | *** join/#edev g1powermac (n=g1powerm@unaffiliated/g1powermac) |
| 12:16.46 | *** join/#edev linac (n=lin@123.249.167.20) |
| 12:20.32 | *** join/#edev linac (n=lin@123.249.167.20) |
| 13:07.13 | KazaLite | is __pure keyword valid for all compilers for ARM? |
| 13:39.49 | *** join/#edev likewise (n=likewise@82-171-51-231.ip.telfort.nl) |
| 13:52.50 | *** join/#edev samrobb (n=samrobb@eng.arriad.com) |
| 13:54.57 | wmat | KazaLite: it is in RealView and GCC, but it's usage is slightly different I believe |
| 13:55.35 | wmat | and it's not available in really old versions of GCC, earlier than 2.96 |
| 13:55.52 | KazaLite | cool |
| 13:57.35 | wmat | here's a GCC example: int square (int) __attribute__ ((pure)); |
| 13:58.42 | wmat | here's a RealView example: int factr(int n) __pure |
| 14:17.41 | *** join/#edev GPSFan (n=kenm@64.92.145.112) |
| 14:17.41 | *** mode/#edev [+o GPSFan] by ChanServ |
| 14:18.26 | *** join/#edev harpal_ (n=Harpal@59.162.116.80) |
| 14:41.54 | *** join/#edev harpal__ (n=Harpal@122.169.108.195) |
| 14:54.24 | *** join/#edev jayakumar (n=jayakuma@118.100.11.6) |
| 15:00.20 | *** join/#edev harpal_ (n=Harpal@59.162.116.80) |
| 15:04.57 | *** join/#edev chouimat|work (n=dieu@kde/developer/chouinard) |
| 15:06.00 | *** join/#edev kergoth (n=kergoth@65.200.49.156) |
| 15:06.37 | kergoth | morning |
| 15:07.59 | chouimat|work | morning |
| 15:13.09 | methril|work | morning kergoth |
| 15:30.02 | KazaLite | why don't we need return type for ISR? |
| 15:32.45 | *** join/#edev kergoth (n=kergoth@65.200.49.156) |
| 15:48.20 | *** join/#edev pocek_ (i=pock@87-205-79-201.adsl.inetia.pl) |
| 15:55.49 | *** join/#edev prpplague (n=dave@mail.americanmicrosystems.com) |
| 15:56.15 | *** part/#edev prpplague (n=dave@mail.americanmicrosystems.com) |
| 15:57.43 | solar | ds2: speaking of arm laptops. Just ran across this one on DX for dirt cheap. http://www.dealextreme.com/details.dx/sku.18355 |
| 15:59.24 | *** join/#edev CosmicPenguin (n=nobody@66-162-99-237.static.twtelecom.net) |
| 16:08.54 | *** join/#edev mbuf (n=shakthim@61.16.248.242) |
| 16:24.36 | *** join/#edev adyer1 (n=adyer@adsl-69-223-13-78.dsl.chcgil.ameritech.net) |
| 16:37.07 | *** join/#edev kergoth (n=kergoth@65.200.49.156) |
| 17:18.43 | CosmicPenguin | http://i.gizmodo.com/5163636/asus-eee-keyboard-pc-should-arrive-in-may-or-june-for-400+600 |
| 17:18.47 | CosmicPenguin | That will be a fun toy |
| 17:21.58 | *** join/#edev dijenerate (n=dijenera@65.48.155.36) |
| 17:36.32 | *** join/#edev hzhzhzRAmiz (n=A_G@41.178.199.66) |
| 17:48.41 | *** join/#edev Crofton (n=balister@65.14.229.26) |
| 18:03.22 | *** join/#edev adyer (n=adyer@adsl-69-223-13-78.dsl.chcgil.ameritech.net) |
| 18:05.00 | *** join/#edev kantor (i=bird@79.114.86.231) |
| 18:05.10 | kantor | hi |
| 18:05.51 | kantor | embedded |
| 18:06.02 | kantor | sorry |
| 18:06.21 | kantor | on embedded system a char type variable can be less or more then 8 bytes ? |
| 18:07.09 | CosmicPenguin | by universal convention, a char is always 8 bits, at least for any operating system that this group would be interested in |
| 18:07.16 | kantor | 8 bit, sorry |
| 18:08.02 | mgsnr | well, in C, a char is always CHAR_BITS, and it's always the smallest addressable unit. |
| 18:08.16 | mgsnr | some DSPs don't have CHAR_BITS == 8; might be 32. |
| 18:08.36 | CosmicPenguin | well, we have to draw a line in the sand |
| 18:08.59 | mgsnr | i agree; but he mentioned "embedded system" and DSP's certainly fall in that group. |
| 18:09.05 | CosmicPenguin | if you have a specific device in mind, then you should go ask the specific experts for that platform |
| 18:09.30 | CosmicPenguin | For 99% of us, if you said "I have a char" we would think 8 bits |
| 18:09.37 | CosmicPenguin | but we don't actually do embedded here |
| 18:09.37 | mgsnr | no doubt. |
| 18:09.39 | CosmicPenguin | we just pretend |
| 18:09.49 | mgsnr | reads the /topic... |
| 18:12.54 | kantor | so if I write program for embedded systems it is not necessary to use the int8_t type from the stdint.h, I can use char for 8 bits, no ? |
| 18:12.58 | *** join/#edev sleipnir (n=Sleipnir@p4FDD5ED6.dip.t-dialin.net) |
| 18:13.33 | mgsnr | kantor: what kind of target do you have? |
| 18:16.01 | kantor | I'm writing the program for PC actually, but a friend of mine has a router with linux as the routers OS, and I would like to integrate some parts of the program in that linux OS (only for curiosity) |
| 18:18.12 | mgsnr | i think you'd have to know something of how the router's controller works |
| 18:18.47 | mgsnr | but on the PC sure sizeof (int8_t) would be sizeof (char) (which is 1) |
| 18:19.07 | mgsnr | the router is quite likely the same |
| 18:21.23 | chouimat|work | mgsnr: the might be the same size but some cpu uses signed char |
| 18:22.30 | mgsnr | chouimat|work: yeah, sometimes char == signed char, and sometimes it's unsigned char |
| 18:28.07 | *** join/#edev harpal (n=Harpal@59.162.116.80) |
| 18:32.46 | ds2 | morning |
| 18:34.34 | CosmicPenguin | mgsnr: indeed, thats a famous early ARM bug - applications using getopt() would assume char == signed and be very sad |
| 18:34.57 | chouimat|work | CosmicPenguin: and an easy one to forget :) |
| 18:35.50 | mnemoc | when you care about signess, better cast to int |
| 18:36.02 | CosmicPenguin | ever since those days, I've always used signed char for getopt(), just one of those things, I guess |
| 18:36.08 | CosmicPenguin | its funny the little foibles we pick up |
| 18:36.31 | chouimat|work | CosmicPenguin: hehe |
| 18:40.19 | *** join/#edev g1powermac_PB (n=g1powerm@unaffiliated/g1powermac) |
| 18:40.26 | ds2 | isn't there a -f flag to force chars to default to sign on gcc? |
| 18:43.15 | Madsy | yes |
| 18:45.09 | mgsnr | CosmicPenguin: are you talking about the return value of getopt? |
| 18:45.28 | CosmicPenguin | it doesn't matter - it was a side comment |
| 18:45.46 | mgsnr | ok... i always use int. |
| 18:46.04 | CosmicPenguin | bully for you then |
| 18:46.06 | mgsnr | heh |
| 18:49.34 | *** join/#edev flavioribeiro (n=avaty@189.71.66.144) |
| 18:51.58 | *** join/#edev davidc__ (n=davidc__@S01060002b360aacd.vc.shawcable.net) |
| 19:06.32 | *** join/#edev JoeLlama (n=snork@unaffiliated/joellama) |
| 19:11.54 | *** join/#edev DrZimmerman (n=theo@77-56-240-146.dclient.hispeed.ch) |
| 19:15.10 | *** part/#edev adyer (n=adyer@adsl-69-223-13-78.dsl.chcgil.ameritech.net) |
| 20:14.09 | *** join/#edev kergoth (n=kergoth@65.200.49.156) |
| 20:17.03 | kergoth | CosmicPenguin: ooh, that eee keyboard pc looks nifty |
| 20:27.28 | *** join/#edev pirho (i=pirho@gateway/gpg-tor/key-0x2CEEC9CB) |
| 20:30.09 | CosmicPenguin | kergoth: yeah, seems like some hacking fun |
| 20:30.19 | CosmicPenguin | but its x86, so not too much fun without the graphics eyecandy |
| 20:34.53 | *** join/#edev likewise (n=likewise@82-171-51-231.ip.telfort.nl) |
| 20:37.37 | *** join/#edev frikker_ (n=blaine@107.135.68.216.DED-DSL.fuse.net) |
| 20:52.20 | *** join/#edev brolin (n=brolin@200.24.16.55) |
| 20:53.49 | *** join/#edev harpal_ (n=Harpal@59.162.116.80) |
| 21:13.44 | *** join/#edev harpal (n=Harpal@59.162.116.80) |
| 21:22.02 | *** join/#edev harpal_ (n=Harpal@59.162.116.80) |
| 21:53.06 | *** join/#edev octet (n=septet@116.71.158.87) |
| 21:56.00 | octet | in realtime systems, we do not need to program for virtual memory...as it may affect determinism of RTOS. right? |
| 22:08.02 | *** join/#edev pirho (i=pirho@gateway/gpg-tor/key-0x2CEEC9CB) |
| 22:19.02 | *** join/#edev kergoth (n=kergoth@ip68-3-163-239.ph.ph.cox.net) |
| 22:25.36 | MonMotha | octet: virtual memory need not affect determinism |
| 22:25.57 | MonMotha | you just have to be sure that there will be no page faults or that the time required to service the page fault is "OK" |
| 22:26.47 | MonMotha | now, swap (i.e. virtual memory backed by disk) is probably not going to work out so well |
| 22:29.57 | roxfan | virtual memory does not necessarily mean paging |
| 22:31.06 | MonMotha | atypical virtual memory implementations that don't rely on traditional paging would be one way to assure there are no page faults |
| 22:31.38 | MonMotha | basically, the OS has to make sure the entire process is in memory whenever it is scheduled to run |
| 22:32.12 | MonMotha | you could do that by either ensuring all pages are visible or by simply treating each process as a "hunk" rather than multiple pages |
| 22:32.30 | MonMotha | some MMUs may not allow the latter approach, though |
| 22:36.46 | octet | cool |
| 22:37.32 | MonMotha | virtual memory does also allow you to make better use of things like malloc, but if you're concerned about determinism, you may want to avoid that for other reasons |
| 22:37.59 | MonMotha | biggest thing virtual memory gets you is that generally with it comes memory protection |
| 22:38.39 | MonMotha | on systems where failure is a "big deal" that can allow more graceful recovery from out-of-bounds access since you can detect it and do something about it rather than corrupt random memory from some other process |
| 22:39.40 | MonMotha | of course, avoiding such access in the first place is great, but "bad stuff" happens. If anything, it makes debugging easier |
| 22:39.48 | *** join/#edev ZeZu (n=null@c-98-227-56-17.hsd1.in.comcast.net) |
| 22:54.45 | *** join/#edev pirho (i=pirho@gateway/gpg-tor/key-0x2CEEC9CB) |
| 22:59.18 | *** join/#edev pirho (i=pirho@gateway/gpg-tor/key-0x2CEEC9CB) |
| 23:03.02 | MonMotha | purl: seen joe_bleau |
| 23:03.08 | purl | joe_bleau <n=joe_blea@99.175.79.53> was last seen on IRC in channel #edev, 4d 21h 58m 41s ago, saying: 'What's this rumor about Xilinx Spartan 6 including some sort of ARM core?'. |
| 23:05.17 | *** join/#edev Madsy (n=madsy@fu/coder/madsy) |
| 23:55.28 | *** join/#edev hzhzhzRAmiz (n=A_G@41.178.163.43) |
