| 00:50.21 | *** join/#edev melchior_ (~melchior@AStrasbourg-552-1-55-169.w83-203.abo.wanadoo.fr) |
| 01:51.45 | *** join/#edev davidc__ (~davidc__@S01060002b360aacd.vc.shawcable.net) |
| 02:37.19 | *** join/#edev JoeLlama (~snork@unaffiliated/joellama) |
| 02:45.02 | *** join/#edev SpeedEvil (1000@tor/regular/SpeedEvil) |
| 03:18.36 | *** join/#edev mbuf (~user@118.102.130.6) |
| 03:48.14 | *** join/#edev miltonm (~miltonm@71.22.98.68) |
| 04:35.50 | *** join/#edev JoeLlama (~snork@unaffiliated/joellama) |
| 05:34.41 | *** join/#edev davidc__ (~davidc__@S01060002b360aacd.vc.shawcable.net) |
| 05:44.14 | *** join/#edev kari (~mike@d206-75-24-126.abhsia.telus.net) |
| 05:44.29 | kari | hello |
| 05:44.57 | *** join/#edev rickly (who@unaffiliated/rickly) |
| 06:19.05 | *** join/#edev tsjsieb (~tsjsieb@80.101.185.65) |
| 06:42.30 | *** join/#edev hw_ (~hw@p578b3905.dip0.t-ipconnect.de) |
| 06:51.04 | *** part/#edev rickly (who@unaffiliated/rickly) |
| 07:12.02 | *** join/#edev mbuf (~user@118.102.130.6) |
| 07:31.06 | *** join/#edev flo_lap (~fuchs@Maemo/community/contributor/florian) |
| 08:22.34 | *** join/#edev purl (ibot@rikers.org) |
| 08:22.34 | *** topic/#edev is Discussion of embedded systems (Hardware/Software/Linux/WinCE/Commercial Device Repurposing) : Aboriginal Linux (formerly FWL) : Busybox and Buildroot : http://www.elinux.org |
| 08:36.08 | *** join/#edev MuscleNerd (eric@adsl-75-16-57-151.dsl.irvnca.sbcglobal.net) |
| 08:41.08 | *** join/#edev ohsix (ohsix@66.220.111.99) |
| 08:43.39 | *** join/#edev dijenerate (~dijenerat@64.210.44.37) |
| 08:49.51 | *** join/#edev likewise (~likewise@188.201.28.66) |
| 09:01.52 | *** join/#edev mbuf (~mbuf@118.102.130.6) |
| 09:15.38 | *** join/#edev florian_kc (~fuchs@Maemo/community/contributor/florian) |
| 10:14.11 | *** join/#edev pirho (debian-tor@gateway/tor-sasl/pirho) |
| 10:18.51 | *** join/#edev MuscleNerd (eric@adsl-75-16-51-17.dsl.irvnca.sbcglobal.net) |
| 10:20.39 | *** join/#edev mne (~mne@unaffiliated/mne) |
| 10:25.43 | mne | Hi. On my AT91SAM9260 based system I would like to use strong encryption. As trusted platform modules (TPM) are widely available for PCs now, I was wondering if I could use one of these modules. Usually these modules have at least a hardware random number generator (TRNG), RSA and other nice things in them. What do you think ? |
| 10:35.02 | mne | It seems that a "33MHz LPC interface" is used to talk to these modules. |
| 10:36.36 | *** join/#edev [sk] (sk@devio.us) |
| 10:44.10 | *** join/#edev linac (~lin@125.77.83.184) |
| 10:48.48 | SpeedEvil | LPC is an ISA bus over lots fewer wires |
| 10:49.38 | SpeedEvil | It's commonly used on motherboards - well - was - I'm not sure of its current usage - to connect IO chips and stuff |
| 10:49.45 | mne | I also found out that TPM modules are available with SMBus connectivity which seems to be pretty much the same as I2C. Hence it would be easy to hook one of these modules up to a AT91 controller |
| 10:49.49 | SpeedEvil | but they got eaten by the chipset recently |
| 10:50.09 | SpeedEvil | Also - do you need high speed? |
| 10:50.22 | *** join/#edev pirho (debian-tor@gateway/tor-sasl/pirho) |
| 10:50.34 | SpeedEvil | many encryption standards are not hard to implement on even quite small micros. |
| 10:50.57 | mne | Well, basically what I'm looking for is a way to get high quality random numbers that can be used for crypto operations. The application is rather low speed so software encryption should work as well. |
| 10:51.26 | SpeedEvil | AT9 is what? |
| 10:51.28 | SpeedEvil | I forget |
| 10:51.32 | mne | AT91SAM9260 |
| 10:51.42 | mne | it's a controller from atmel |
| 10:51.51 | mne | with an ARM9 core |
| 10:51.52 | SpeedEvil | 8 bit/16/... |
| 10:51.53 | SpeedEvil | ah |
| 10:51.55 | mne | 32bit |
| 10:52.06 | SpeedEvil | you could use a hardware external. |
| 10:52.09 | mne | I'm running linux on it |
| 10:52.13 | SpeedEvil | For example - white noise diode. |
| 10:52.27 | SpeedEvil | you know about how /dev/random works? |
| 10:53.02 | mne | I know. But as soon as I build something on my own it might be easy to bias |
| 10:53.15 | SpeedEvil | Why not use /dev/random? |
| 10:53.24 | SpeedEvil | And add to the entropy pool from external hw |
| 10:53.30 | mne | AFAIk /dev/random uses a pool of entropy to generate random numbers. This pool uses for example interrupts as random source |
| 10:53.35 | SpeedEvil | yes |
| 10:54.26 | mne | Well, but if my super secure homebrew random generator starts to produce biased random numbers, /dev/random will also start to be biased |
| 10:54.40 | SpeedEvil | Nope. That's the nice bit. |
| 10:54.58 | SpeedEvil | You set the estimate of entropy on the incoming bits really low. |
| 10:55.38 | SpeedEvil | So you design your external generator as best you can to give - say - 0.5 bits of entropy per sample, and tell the kernel that its randomness is .01 bits |
| 10:55.59 | mne | hmm, ok |
| 10:56.18 | mne | and how does it achieve high quality random numbers then ? |
| 10:56.57 | mne | Let's assume that my hardware RNG is really bad and produces 1s all the time ... |
| 10:57.04 | mne | for some reason |
| 10:58.58 | SpeedEvil | It maintains an entropy pool. |
| 10:59.06 | SpeedEvil | New randomness is added to this entropy pool. |
| 10:59.22 | SpeedEvil | <PROTECTED> |
| 10:59.38 | SpeedEvil | It will not output more data unless it has sufficient randomness. |
| 10:59.49 | mne | yes. So besides my hardware RNG it also uses interrupts for example. However, if the system is idle all the time, entropy will soon get low and no more random numbers are available anymore |
| 10:59.52 | SpeedEvil | Every input bit is hashed with the pool, to disturb its state. |
| 10:59.59 | mne | ok |
| 11:00.34 | SpeedEvil | However - the count of how much randomness remains in the pool - and if the next read will block - is not incremented by 1 bit for each bit fed in. |
| 11:00.47 | SpeedEvil | It's incremented by some number representing the quality of that bit. |
| 11:00.54 | SpeedEvil | the quality of randomness. |
| 11:00.57 | mne | ok |
| 11:01.30 | mne | and the quality is measured by statistical tests in the kernel |
| 11:01.33 | SpeedEvil | So if you set a very conservative estimate of the entropy from your randomness source, then it will always be properly random. |
| 11:01.34 | SpeedEvil | No |
| 11:01.52 | SpeedEvil | it's configurable - randomness tests can never measure entropy. |
| 11:02.01 | SpeedEvil | 111111111111111111111111111111111111111 can be completely random. |
| 11:02.10 | mne | rights, it's just evry unlikely |
| 11:02.23 | SpeedEvil | But it will cause randomness tests to fail. |
| 11:02.58 | mne | Do you have a link on how to add random sources to the linux kernel random pool ? |
| 11:02.59 | SpeedEvil | randomness tests are simply attempts to find weaknesses in random sources. Failing or succeeding does not generally imply anything with certainty. |
| 11:03.27 | SpeedEvil | No - this is based on reading I did some years ago. I would recommend reading around in /usr/src/linux/Documentation/ |
| 11:03.40 | mne | I guess it would be easy to use some MCU and a random source (e.g. a diode). I could then hook it up to my system over SPI/I2C |
| 11:03.40 | SpeedEvil | IIRC there is a good document explaining the random generator |
| 11:03.55 | SpeedEvil | Or even a diode and an opamp to an ADC |
| 11:04.03 | mne | cool, thanks. |
| 11:04.14 | SpeedEvil | Or work out missing sources of randomness in your device. |
| 11:04.20 | mne | What about just reading the LSB of the ADC input with nothing connected to it ? |
| 11:04.21 | SpeedEvil | For example - does it have a radio. |
| 11:04.42 | mne | no |
| 11:04.49 | SpeedEvil | That works - but is problematic in that in principle that value can be influenced by stray RF or something. |
| 11:05.03 | SpeedEvil | However - again - if you set the value very conservatively, that could work. |
| 11:05.21 | *** join/#edev ercani (~ercani@78.185.34.242) |
| 11:05.26 | mne | And the diode noise is not that easily influenced ? I guess I would need shielding as well |
| 11:05.41 | *** join/#edev Zygo (startkeylo@startkeylogger.hungrycats.org) |
| 11:05.57 | SpeedEvil | In many ways the entropy pool means you don't care. |
| 11:06.05 | mne | hehe, ok |
| 11:06.23 | SpeedEvil | It means that they not only need to be able to add signals - but they also need to be able to mask the noise signal from the diode. |
| 11:06.56 | SpeedEvil | If you've set the 'quality' of the randomness low enough - then even 1 bit in 20 if it comes from the diode, not the attacker - is plenty to keep it random. |
| 11:07.45 | mne | And it would still ensure that new random numbers are available, right ? |
| 11:08.10 | mne | If the quality is very low it would just need a long time |
| 11:16.03 | SpeedEvil | yes |
| 11:16.22 | SpeedEvil | If your quality is low, then the output bit rate is much lower than the input bit rate |
| 11:16.31 | SpeedEvil | there are 'stupid' solutions that may work too. |
| 11:16.42 | SpeedEvil | for example - a gigabyte of random numbers on SD. |
| 11:16.43 | mne | like what ? |
| 11:17.01 | mne | but they would be always the same ? |
| 11:17.03 | SpeedEvil | or random numbers over an encrypted connection you're making anyway to a trusted server. |
| 11:17.10 | SpeedEvil | you only use them once though |
| 11:17.16 | mne | i see |
| 11:18.16 | mne | I'll have a look at the kernel documentation now. Maybe it turns out that I don't need hw random numbers at all ;) |
| 11:18.46 | mne | which depends on what data is used in the pool and whether such data is continuously available on my system |
| 11:19.00 | SpeedEvil | yes. |
| 11:28.12 | *** part/#edev ercani (~ercani@78.185.34.242) |
| 11:55.39 | *** join/#edev florian (~fuchs@Maemo/community/contributor/florian) |
| 12:04.58 | *** join/#edev gustavoz (~gustavoz@201.253.130.60) |
| 12:08.19 | *** join/#edev kazaa_lite (~eddie@212.183.140.19) |
| 12:09.57 | mne | SpeedEvil, in the random generator implementation of the linux kernel I found out that noise is gathered by add_input_randomness() and add_interrupt_randomness(). I'm now especially interested which interrupts are used. So far I found out that add_interrupt_randomness() is called in the linux kernel interrupt handler if the flag IRQF_SAMPLE_RANDOM is set. However, there are only very few drivers that can be used as random source. In my case the only driver that |
| 12:09.57 | mne | is used is the macb ethernet driver. But as my device is not connected to an ethernet, that random source is useless. Are there really no other interrupts used (e.g. uart) ? |
| 12:11.04 | SpeedEvil | you can hook other sources to it |
| 12:11.18 | SpeedEvil | you need to carefully consider how much entropy they will add |
| 12:16.04 | mne | You said that the entropy of a random source can be defined. In the source code I cannot see that such a setting is actually made for add_input_randomness() ? For example in the macb driver the interrupt is just set up with the IRQF_SAMPLE_RANDOM flag and that's all |
| 12:17.24 | mne | At the moment I'm also trying to figure out which input sources are used as random source. It seems that each driver calling input_event() is used as random source |
| 12:18.25 | SpeedEvil | Unsure - it's been a while since I read the source - sorry. |
| 12:18.39 | SpeedEvil | There may be useful stuff too on the related mailing list |
| 12:18.43 | SpeedEvil | IIRC crypto-linux |
| 12:19.02 | mne | i'll have a look at it later |
| 12:20.34 | mne | As my embedded device hardly has any user interaction, there's not too much randomness either. It looks like I really need to add some hardware random source, possibly somehow with add_input_randomness() as hammering the CPU with random interrupts does not seem to be a good idea ;) |
| 12:21.07 | SpeedEvil | well - are there any external inputs? |
| 12:21.27 | SpeedEvil | If it's got ADCs - then LSBs of those may be usable. |
| 12:21.39 | mne | Yes, there will be a small keypad, but that's all |
| 12:23.45 | mne | Hmm, maybe I could just use a small MCU that constantly reads more or less good random data, e.g. an ADC with the LSB or even some zener diode noise. Then in the kernel I could set up a timer that reads some values from the MCU and calls add_input_randomness(). |
| 12:24.38 | mne | Of course I could also use an interrupt that the MCU generates each time it has new data. But then I would also need to ensure that there is no interrupt hammering |
| 12:25.45 | mne | However, my device has a "soundcard". And the atmel SSC is used to transfer samples between the CPU and the soundcard over I2S. AFAIK there are also SSC interrupts |
| 12:26.07 | mne | But they are possibly too periodic |
| 12:26.55 | *** join/#edev calculus (~calculus@gentoo/user/calculus) |
| 12:35.33 | SpeedEvil | There are also hw rng chips |
| 12:35.46 | SpeedEvil | a lot simpler than rolling your own. |
| 12:35.49 | SpeedEvil | the cheap ones at least |
| 12:36.12 | mne | I looked for hours but could not find one that it also easy to get |
| 12:37.08 | mne | The best thing I found is the RPG100 from FDK, but it doesn't seem to be easy to get either |
| 12:37.52 | mne | I already asked for samples: The minimum quantity is 5pcs with EUR 25,- per unit. IMHO that's pretty expensive. |
| 12:39.04 | mne | So if you know some chip that can be easily obtained I would be glad ;) |
| 12:39.44 | SpeedEvil | hmm |
| 12:39.55 | SpeedEvil | not off-hand - it was some years ago I was looking. |
| 12:40.44 | SpeedEvil | http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=AQ2A2-C2-T5VDC-ND - I guess on my first search - is not going to be helpful :) |
| 12:41.33 | mne | this seems to be a solid state relay ?? |
| 12:42.03 | SpeedEvil | yes - it has the word 'random' in the description |
| 12:42.11 | mne | lol, ok |
| 12:43.10 | SpeedEvil | Hmm. |
| 12:43.23 | mne | There are some chips for smartcard interfacing. But they can only be purchased in a bulk of like 500 pieces. The cost aside, I'm not sure what to do with 500 chips :) |
| 12:43.25 | SpeedEvil | A quick search reveals none of the obvious candidates. |
| 12:43.56 | SpeedEvil | I guess they never went to volume, due to the fact that anyone wanting them generally picks a CPU with it rolled in. |
| 12:44.39 | mne | Which is why I was thinking of purchasing a cheap TPM module that maybe even supports SMBus (which is more or less I2C). |
| 12:45.20 | mne | Unfortunately my controller is has not one if those neat things built in |
| 12:47.05 | SpeedEvil | http://search.digikey.com/scripts/DkSearch/dksus.dll?Cat=2556771&k=tpm&stock=1 - yeah - they don't seem to be easily available. |
| 12:47.59 | SpeedEvil | http://en.wikipedia.org/wiki/Hardware_random_number_generator |
| 12:48.03 | SpeedEvil | seems to be mostly sane |
| 12:50.05 | mne | I was thinking of something like this: http://www.google.com/products/catalog?hl=en&resnum=0&q=asus+tpm+module&um=1&ie=UTF-8&cid=1340098577167668524&ei=cYzyS9ylHYWimwPUn5GrDQ&sa=X&oi=product_catalog_result&ct=result&resnum=3&ved=0CCwQ8wIwAg# |
| 12:50.24 | mne | But I guess it is not SMBus, but LPC instead |
| 12:50.54 | mne | also I would possibly need to implement all the protocol stuff :( |
| 12:51.36 | mne | This is also interesting: http://pdfserv.maxim-ic.com/en/an/AN3469.pdf |
| 12:51.49 | mne | My controller has 4 ADC inputs |
| 12:52.07 | mne | So I could connect one of these white noise sources to the ADC |
| 13:25.19 | *** join/#edev linac (~lin@121.15.253.237) |
| 13:39.51 | SpeedEvil | In that app, there is little benefit of those chips over an inexpensive dual op-amp - say |
| 14:02.04 | *** join/#edev miltonm (~miltonm@129.33.192.72) |
| 14:17.52 | *** join/#edev kergoth_ (~clarson@ip24-251-170-95.ph.ph.cox.net) |
| 14:37.32 | *** join/#edev armin76 (~armin@gentoo/developer/armin76) |
| 14:40.40 | *** join/#edev prpplague (~danders@nat/ti/x-exiikzynotmvgvaa) |
| 14:45.47 | *** mode/#edev [+o prpplague] by ChanServ |
| 15:02.04 | *** join/#edev Crofton (~balister@pool-96-240-168-177.ronkva.east.verizon.net) |
| 15:29.45 | *** join/#edev kazaa_lite (~eddie@212.183.140.38) |
| 15:34.35 | *** join/#edev masked (masked@bewifi.org.au) |
| 15:37.13 | *** join/#edev file (~file@asterisk/developer-and-muffin-lover/file) |
| 15:39.31 | *** join/#edev linac (~lin@220.181.106.187) |
| 15:43.24 | *** join/#edev cbrake (~cbrake@oh-69-34-21-229.sta.embarqhsd.net) |
| 15:45.05 | *** join/#edev Bartron (~Bartron@83.119.172.124) |
| 15:45.33 | *** join/#edev gustavoz (~gustavoz@201.253.130.60) |
| 15:53.45 | *** join/#edev Zygo (startkeylo@startkeylogger.hungrycats.org) |
| 16:12.46 | *** join/#edev Zygo (startkeylo@startkeylogger.hungrycats.org) |
| 16:28.14 | *** join/#edev Iam3r (~Iam3r@e180073118.adsl.alicedsl.de) |
| 16:30.19 | *** join/#edev Zygo (startkeylo@startkeylogger.hungrycats.org) |
| 16:54.25 | *** join/#edev Crofton (~balister@pool-96-240-168-177.ronkva.east.verizon.net) |
| 16:58.12 | *** join/#edev SpeedEvil (1000@tor/regular/SpeedEvil) |
| 17:21.00 | kazaa_lite | hi all |
| 17:21.08 | kazaa_lite | i hae made some notes about ARM |
| 17:21.15 | kazaa_lite | can someone review them? |
| 17:21.49 | kazaa_lite | its a short highlight of capabilities of arm 9 and how things are handled in arm 9 |
| 17:28.17 | SpeedEvil | where? |
| 17:28.36 | SpeedEvil | My knowledge is limited, but I can look for typos. |
| 17:42.21 | *** join/#edev gandhijee (akp@ip67-152-15-148.z15-152-67.customer.algx.net) |
| 17:43.41 | *** join/#edev Zygo (startkeylo@startkeylogger.hungrycats.org) |
| 17:51.10 | kazaa_lite | SpeedEvil: http://pastebin.org/248990 |
| 17:51.21 | kazaa_lite | SpeedEvil: got it? |
| 18:01.16 | *** join/#edev SpeedEvil (1000@tor/regular/SpeedEvil) |
| 18:03.49 | *** join/#edev jnd (~jenda@unaffiliated/jnd) |
| 18:04.10 | *** join/#edev Zygo (startkeylo@startkeylogger.hungrycats.org) |
| 18:11.41 | *** join/#edev dijenerate (~dijenerat@64.210.44.37) |
| 18:16.04 | SpeedEvil | Might be worth mentioning what modes these are - for example - different to thumb/... mode |
| 18:17.19 | kazaa_lite | SpeedEvil: oppps.... yes.... this is all in 32bit mode.... not addressing anything for thumb mode |
| 18:18.27 | SpeedEvil | remaining pipeline is flush ^ed |
| 18:18.56 | SpeedEvil | If pipeline is - add the in middle |
| 18:19.29 | SpeedEvil | there will not be much parallelism - could maybe do with rewording |
| 18:21.50 | SpeedEvil | The entire para is a bit wordy, and reiterates stuff that could be more simply written. |
| 18:26.04 | SpeedEvil | Sorry - I'm not up to close reading of this at the moment - have been baling out pond - which was fun. |
| 18:32.16 | Iam3r | "If pipeline is too small, there will not be much parallelism with in the system and throughput might be not be great" |
| 18:32.24 | Iam3r | i think i lerned something different ;) |
| 18:32.59 | Iam3r | a short pipeline is the best, but it can't be clocked very high, so longer pipelines are needed |
| 18:35.17 | SpeedEvil | I think something along the lines of 'A pipeline is a compromise to improve performance. On the one hand dividing instructions into a dozen steps, and performing each of these very simple steps very fast for each instruction may mean that you can process some code very fast indeed. However - the problem with deep pipelines is that if there is a branch instruction, this can cause all of the partially completed results to be invalid. |
| 18:36.27 | SpeedEvil | So it's a balance between simple fast pipeline stages, and complex slower ones. Where the exact optimum tradeoff depends on the compiler and code, as well as 'raw' performance that can be achieved. |
| 19:13.44 | *** join/#edev flo_lap (~fuchs@Maemo/community/contributor/florian) |
| 19:14.16 | kazaa_lite | SpeedEvil: yes, you can correct my english.... I am from Pakistan and english is not my mother tongue, so I am all open for improvements:D |
| 19:17.56 | kazaa_lite | SpeedEvil: do you think technically I have correctly put down information about ARM9? |
| 19:18.42 | kazaa_lite | Iam3r: and could you find any mistakes in the document? |
| 19:47.23 | SpeedEvil | kazaa_lite: Sorry - I'm not knowledgable enough. There don't seem to be any glaring inconsistencies that I have noticed. |
| 19:47.49 | kazaa_lite | cool.... thanks a lot:) |
| 19:51.32 | SpeedEvil | I was initially going to comment that your english is a lot better than my Urdu - but then realised on checking http://en.wikipedia.org/wiki/Languages_of_Pakistan that it's a minority language. |
| 19:51.57 | SpeedEvil | I find it odd that the above page doesn't list how widely spoken english is - given that it lists it as an official language. |
| 19:58.12 | SpeedEvil | What was that reverse engineered wifi firmware project? |
| 20:06.43 | roxfan | well, german is an official language of belgium... there's like 1% of german speakers |
| 20:11.35 | SpeedEvil | I know. I just found it odd that there diddn't seem to be a number, yet all of the other languages were broken down. |
| 20:17.46 | *** join/#edev dijenerate (~dijenerat@64.210.44.37) |
| 20:31.39 | *** join/#edev skowroniasty (skowronias@devilroot.pl) |
| 20:32.48 | skowroniasty | hi, i want create my own linux distro, working on routerboard ( atheros cpu), |
| 20:32.48 | skowroniasty | 22:28 < skowroniasty> how start with embedded linux ? |
| 20:32.48 | skowroniasty | 22:29 < skowroniasty> for now i printing ebook (embedded linux from oreily), |
| 20:32.48 | skowroniasty | 22:30 < skowroniasty> but where on Internet, i can find some use ful tutorial, or discusion groups, or forums ? |
| 20:32.51 | skowroniasty | 22:30 < skowroniasty> any know ? |
| 20:36.34 | SpeedEvil | what sort of embedded? |
| 20:36.39 | SpeedEvil | ah |
| 20:36.40 | *** join/#edev zaz1R (~zaz@adsl-074-244-082-076.sip.asm.bellsouth.net) |
| 20:36.57 | SpeedEvil | looking at the ddwrt and openwrt and related forums may be a good start |
| 20:37.08 | skowroniasty | i want create my distro for routerboard ( with Atheros cpu ) |
| 20:37.24 | roxfan | the best would be to find a working distro for your router and start poking at it |
| 20:37.33 | SpeedEvil | you have the documentation for this routerboard/ |
| 20:37.44 | skowroniasty | not now, |
| 20:37.45 | SpeedEvil | Or do you man 'I bought a router, and want to put linux on it'? |
| 20:37.49 | SpeedEvil | mean |
| 20:37.52 | skowroniasty | I want buy some board, |
| 20:37.59 | skowroniasty | <PROTECTED> |
| 20:38.24 | SpeedEvil | A development board, or a customer device with no documents? |
| 20:38.34 | skowroniasty | <PROTECTED> |
| 20:39.13 | skowroniasty | development board - i think ( some mikrotik stuff) |
| 20:39.30 | skowroniasty | http://www.mikrotik.org.pl/urzadz.php?gr=5&kat=35 |
| 20:39.38 | roxfan | if it's a dev board, it should have linux with it |
| 20:39.43 | skowroniasty | I think about this brand, or |
| 20:39.47 | *** join/#edev mne (~mne@unaffiliated/mne) |
| 20:40.05 | SpeedEvil | Are there _DETAILED_ docs on it? |
| 20:40.18 | skowroniasty | yes, |
| 20:40.18 | SpeedEvil | Register breakdowns and datasheets for all chips - programming info, ... |
| 20:40.25 | roxfan | hm, it says "RouterOS Level3" whatever that is |
| 20:40.46 | skowroniasty | routers OS is MikroTik linux distro, |
| 20:40.47 | SpeedEvil | Also - why do you want to port linux to a board, when there are many boards that run linux already. |
| 20:40.51 | roxfan | ah |
| 20:40.54 | skowroniasty | creating for this board |
| 20:41.12 | skowroniasty | but you can buy this board without this OS |
| 20:41.13 | roxfan | well, if it's linux then you're all set |
| 20:41.30 | roxfan | just learn to compile it and start changing it |
| 20:42.00 | skowroniasty | uh... |
| 20:42.15 | SpeedEvil | If there is not an existing linux port, then it would take a skilled person, with full documents on it - several days to weeks to get it working, depending on stuff. |
| 20:42.47 | skowroniasty | i don't know what is better for start with embedde linux, |
| 20:43.08 | skowroniasty | <PROTECTED> |
| 20:43.24 | SpeedEvil | Beagleboard already has linux on it. |
| 20:43.34 | SpeedEvil | You can simply start programming your userspace app. |
| 20:44.00 | SpeedEvil | With the routerboard - unless there is a linux port - you are going to spend weeks or months getting linux running smoothly on it. |
| 20:44.10 | skowroniasty | yes, I know, but i want create my own distro for One board comp. |
| 20:44.26 | roxfan | skowroniasty: it's not a simple process |
| 20:44.33 | SpeedEvil | skowroniasty: Weeks or months then. |
| 20:44.39 | skowroniasty | and I dont know, what is better (easy for beginer) , routerboard or beagelboard |
| 20:44.43 | roxfan | as i said, better take an existing distro and start changing it |
| 20:44.47 | SpeedEvil | skowroniasty: beagleboard |
| 20:44.57 | roxfan | SpeedEvil: he said that routeros = linux |
| 20:45.04 | SpeedEvil | ah - I missed that. |
| 20:45.14 | SpeedEvil | Beagleboard has a large community of people to help. |
| 20:45.32 | SpeedEvil | Wander over to #beagle for example. |
| 20:45.54 | SpeedEvil | There are dozens of people there at any time. Is there a comparable community for the other board? |
| 20:46.03 | *** join/#edev zazenrasta (~zaz@adsl-074-244-082-076.sip.asm.bellsouth.net) |
| 20:46.36 | skowroniasty | yes, i know, createing own distro is not simple task, but i want try :) |
| 20:47.23 | skowroniasty | brb. |
| 20:47.53 | roxfan | well, you got your advice, take it or ignore it as you wish |
| 20:47.55 | SpeedEvil | What do you mean by distro? |
| 20:49.50 | Zygo | start with building your own x86 distro, then build a distro for some hardware that already has one, like beagleboard. Then you'll know enough about the process to see if you're still interested in porting to new hardware |
| 20:49.50 | Zygo | or do what sane people do, which is start with someone else's and modify ;) |
| 20:57.55 | *** join/#edev zaz1R (~zaz@adsl-074-244-082-076.sip.asm.bellsouth.net) |
| 20:59.51 | *** join/#edev likewise (~likewise@dsl-087-195-096-146.solcon.nl) |
| 21:03.05 | kazaa_lite | SpeedEvil: well..... culture in pakistan is becoming more and more westernize though this afghan war has affected it a lot by adding different problems to the country..... but after moving to uk i have realized that my english is not that good as i had expected it to be:P |
| 21:03.34 | kazaa_lite | i would say... in big cities i think 30% of the population would be able to speak english |
| 21:03.38 | *** join/#edev zazenrasta (~zaz@adsl-074-244-082-076.sip.asm.bellsouth.net) |
| 21:04.40 | kazaa_lite | however, since urdu takes many words from english and english is also a 'status symbol' so even if you dont know any urdu, people will still be able to help you out by using some broken sentences or words in english:P |
| 21:12.18 | *** join/#edev skowroni1sty (skowronias@devilroot.pl) |
| 21:12.32 | skowroni1sty | OK, Im |
| 21:12.35 | skowroni1sty | so |
| 21:13.17 | SpeedEvil | kazaa_lite: interesting. |
| 21:13.40 | SpeedEvil | kazaa_lite: I don't think you ever get a good picture of how well you speak a language till you go to the country it's spoken. |
| 21:16.32 | skowroni1sty | <PROTECTED> |
| 21:17.06 | skowroni1sty | So I need some SD card (1 or 2 GB) |
| 21:19.19 | skowroni1sty | for now i lokking some board, It's about the kind of tiles routerboard. Ultimately, it comes to such a distribution that has the basic drivers that support hardware on the board, the functionality of a switch - if it is more than an RJ, and can easily start their own utilities. |
| 21:20.29 | skowroni1sty | before, i never create my own distro, so I think Zygo, has right, when he sugest me, to create for start some x86 distro. |
| 21:21.33 | skowroni1sty | I thnik i should start with x86, |
| 21:22.07 | skowroni1sty | try this way, and next porting on other arch. |
| 21:22.21 | *** join/#edev ercani (~ercani@78.185.34.242) |
| 21:22.28 | ercani | hi |
| 21:22.47 | skowroni1sty | hi, |
| 21:22.58 | ercani | just want to ask that embedded linux kernel is as same as x86 ? |
| 21:23.21 | roxfan | if it's an x86 processor, it's mostly the same |
| 21:23.34 | kazaa_lite | SpeedEvil: exactly..... and then the worst thing is your accient:P you really have to work hard to improve |
| 21:23.35 | roxfan | if it's something else, then it's a bit more different |
| 21:23.35 | kazaa_lite | :P |
| 21:24.31 | ercani | c library is same ? |
| 21:24.42 | SpeedEvil | Maybe. |
| 21:24.53 | skowroni1sty | ercani, i think no, becouse on embedded os, You add only this stuff, whoch you nedd |
| 21:24.56 | SpeedEvil | ercani: Embedded differs _lots_. |
| 21:24.58 | skowroni1sty | need* |
| 21:25.02 | ercani | have sys/stat.h ? |
| 21:25.05 | SpeedEvil | ercani: I have 2 routers. |
| 21:25.20 | SpeedEvil | ercani: They are both running ulibc - which is a really, really stripped down libc. |
| 21:25.29 | SpeedEvil | ercani: My mobile phone uses a full libc |
| 21:25.39 | ercani | u mean uclinux ? |
| 21:25.43 | ercani | ulib |
| 21:25.54 | skowroni1sty | SpeedEvil , which mobiel You use ? with Symbian ? |
| 21:26.08 | ercani | processor is arm |
| 21:26.11 | ercani | pnd device |
| 21:26.11 | skowroni1sty | which * |
| 21:26.15 | ercani | arm11 |
| 21:26.16 | roxfan | symbian doesn't use libc :( |
| 21:26.17 | roxfan | :) |
| 21:26.30 | *** join/#edev N900evil (~Speedevil@tor/regular/SpeedEvil) |
| 21:26.32 | skowroni1sty | so, which brand ? |
| 21:26.39 | ercani | s3c2450 |
| 21:26.41 | ercani | samsung |
| 21:26.51 | N900evil | n900 :) |
| 21:26.57 | ercani | not a mobile |
| 21:27.01 | ercani | pnd device |
| 21:27.46 | roxfan | and what is "pnd"? |
| 21:27.53 | skowroni1sty | now i askig google about this mobile, nice toy :) |
| 21:28.09 | solar | would of guesses portable network device |
| 21:28.10 | ercani | it is mobile inernet device |
| 21:28.16 | ercani | mid |
| 21:28.45 | roxfan | anyway, the libc depends on the actual distro that's installed on your device |
| 21:29.00 | ercani | just wonder about sys/stat.h available in embedded linux |
| 21:29.05 | roxfan | and they usually don't have headers or gcc, so you'll need to use a cross compiler |
| 21:29.19 | ercani | actually I work on wince |
| 21:29.37 | ercani | there is no sys/stat or errno |
| 21:29.38 | *** join/#edev zaz1R (~zaz@adsl-074-244-082-076.sip.asm.bellsouth.net) |
| 21:30.02 | ercani | I wonder if embedded linux have them |
| 21:30.13 | *** join/#edev SpeedEvil (1000@tor/regular/SpeedEvil) |
| 21:30.39 | ercani | by the way cross compiler have floating point emulation setting ? |
| 21:30.53 | ercani | I am not familiar with linux |
| 21:32.44 | roxfan | there's no single cross compiler, you can build one with emulation or without |
| 21:32.59 | ercani | I see |
| 21:33.05 | roxfan | just need to match it with what's actually on the device |
| 21:33.16 | ercani | you mean bsp of the device |
| 21:33.37 | roxfan | check the manufacturer's site, sometimes they provide downloads of the compiler in addition to linux sources |
| 21:33.59 | ercani | ok thnks |
| 21:35.00 | ercani | I port some linux software to wince and some headers are not same |
| 21:35.54 | *** join/#edev SpeedEvil (~user@tor/regular/SpeedEvil) |
| 21:36.57 | ercani | what is the preprocessor definition of embedded linux ? |
| 21:37.18 | ercani | for ex wince we use WIN32_WINCE |
| 21:37.29 | SpeedEvil | There isn't one. |
| 21:37.47 | SpeedEvil | There are many different toolchains used. |
| 21:38.15 | SpeedEvil | Some linux - my phone - forex - is basically 'normal desktop' linux - only differing by being on a somewhat odd architecture. |
| 21:39.18 | ercani | I see |
| 21:39.31 | ercani | toolchain means bsp ? |
| 21:40.23 | ercani | is it board specific package of device manufacturer ? |
| 21:40.46 | SpeedEvil | Toolchain means the tools used to produce executable code. |
| 21:41.10 | SpeedEvil | It can be as simple as a 'normal' installation of GCC - or something much more elaborate |
| 21:41.43 | *** join/#edev zazenrasta (~zaz@adsl-074-244-082-076.sip.asm.bellsouth.net) |
| 21:42.17 | solar | toolchain = gcc,binutils,libc |
| 21:42.28 | solar | most of the time |
| 21:42.57 | ercani | ok I got it |
| 21:43.26 | ercani | libc is same for all ARM devices ? |
| 21:43.29 | *** join/#edev davidc__ (~davidc__@S01060002b360aacd.vc.shawcable.net) |
| 21:43.50 | ercani | I mean arm11 |
| 21:43.58 | roxfan | no |
| 21:44.18 | roxfan | but most libc implementations provide the same functionality |
| 21:44.45 | ercani | ok |
| 21:44.59 | SpeedEvil | It regrettably gets _considerably_ more involved if you want to compile software from the wild. |
| 21:45.13 | SpeedEvil | As you can't simply crosscompile stuff that relies on intermediate build steps. |
| 21:45.29 | SpeedEvil | And you have to use qemu or similar to emulate the native architecture. |
| 21:46.28 | ercani | the code will be compiled in the pc then will be installed executable to device ? |
| 21:46.45 | roxfan | that's a common scenario |
| 21:48.07 | *** part/#edev wmat (~btraynor@gromit.mixdown.ca) |
| 21:50.52 | SpeedEvil | Generally embedded devices were a lot slower than PCs. |
| 21:51.36 | ercani | yes |
| 21:52.11 | ercani | it is c code |
| 21:52.32 | ercani | fft application |
| 21:53.49 | ercani | how is debugging for embd linux ? |
| 21:58.54 | SpeedEvil | Same ways are often available as desktop. |
| 21:59.01 | SpeedEvil | gdb, printf. |
| 21:59.32 | SpeedEvil | qemu too to emulate your targer platform |
| 22:03.34 | ercani | ok thanks |
| 22:15.38 | skowroni1sty | <PROTECTED> |
| 22:15.50 | solar | never heard of it. |
| 22:16.06 | skowroni1sty | rocklinux.net |
| 22:16.07 | solar | your 3 best free choices are openembedded, buildroot and gentoo |
| 22:17.30 | skowroni1sty | I dont know gentoo, OK. I check other two sugestion .. |
| 22:21.09 | skowroni1sty | solar, please tell me, If i try use for ex. buildroot, can I try build simple distro for ARM, and run it on some virtual (emulate) hardware ? or i must have some device with this cpu ? |
| 22:21.40 | solar | yes |
| 22:21.45 | skowroni1sty | :) |
| 22:22.20 | solar | additional buildroot questions should go to the proper channel. I think #uclibc for that. |
| 22:22.47 | solar | OE should go to #openembedded and Gentoo ones to #gentoo-embedded |
| 22:23.00 | *** part/#edev flavioribeiro (~flavio@li20-198.members.linode.com) |
| 22:23.03 | *** join/#edev flavioribeiro (~flavio@li20-198.members.linode.com) |
| 22:26.54 | roxfan | http://www.karo-electronics.com/tx25.html not bad price |
| 22:26.58 | *** join/#edev prpplague (~Dave@adsl-99-110-22-28.dsl.rcsntx.sbcglobal.net) |
| 22:33.47 | *** join/#edev calculus (~calculus@gentoo/user/calculus) |
| 22:36.09 | *** join/#edev bgamari (~ben@pool-96-240-192-139.spfdma.east.verizon.net) |
| 22:38.38 | skowroni1sty | OK, many thanks for all reply for my question, now I want try use openembedded kit, |
| 22:40.18 | skowroni1sty | Thank you once again, an appeal for some time to show off progress. |
| 22:40.21 | skowroni1sty | bye |
| 22:40.33 | *** part/#edev skowroni1sty (skowronias@devilroot.pl) |
| 22:41.39 | f1yback | prpplague, hahaha dude |
| 22:41.51 | prpplague | f1yback: ?? |
| 22:41.52 | f1yback | I setup a ubuntu 10.04LTS server vm on my esxi box |
| 22:42.00 | f1yback | to compile crosstool on |
| 22:42.07 | f1yback | the hd starts to act up so had to kill it |
| 22:42.35 | f1yback | so then I setup the same type vm on a unused server at work, run into a few problems with crosstool then laney told me about crosstool-ng |
| 22:42.49 | f1yback | went thru menuconfig, etc it failed to compile but I found out why |
| 22:43.08 | f1yback | well that server is unused because of a raid controller bug that I thought I fixed, apprentely not |
| 22:43.14 | f1yback | you see why it takes me yrs to do stuff? |
| 22:43.15 | f1yback | :P |
| 22:43.52 | f1yback | fortunitely my mini-itx is done torture testing so going to setup crosstool-ng on it, it will be slow as hell to compile but it will be stable and run run :) |
| 22:44.05 | f1yback | I just thought it's funny that's all |
| 22:44.10 | f1yback | how the impossible always happens to me :P |
| 23:44.31 | *** join/#edev kergoth (~kergoth@ip24-251-170-95.ph.ph.cox.net) |
| 23:49.39 | f1yback | hey kergoth ltns |
| 23:59.50 | *** join/#edev calculus (~calculus@gentoo/user/calculus) |