*** riazantsev has quit IRC | 00:21 | |
*** fernandod has quit IRC | 00:46 | |
*** fernandod has joined #automotive | 00:46 | |
*** Luming has quit IRC | 01:24 | |
*** Luming has joined #automotive | 01:24 | |
*** riazantsev has joined #automotive | 01:25 | |
*** persia has quit IRC | 03:29 | |
*** persia has joined #automotive | 03:30 | |
*** persia has joined #automotive | 03:30 | |
*** persia has quit IRC | 03:41 | |
*** persia has joined #automotive | 03:42 | |
*** fernandod has quit IRC | 03:46 | |
*** fernandod has joined #automotive | 03:47 | |
*** apinheiro has joined #automotive | 06:13 | |
*** KlausUhl has joined #automotive | 07:00 | |
*** jeremiah has joined #automotive | 07:15 | |
jeremiah | ohai! | 07:55 |
---|---|---|
jeremiah | Is Walt Miner around? | 07:56 |
CTtpollard | hi jeremiah | 07:56 |
jeremiah | :-) | 07:56 |
jeremiah | waltminer: Are you here? I've heard you're on vacation, nice! | 07:56 |
CTtpollard | he's not in the channel list | 07:57 |
*** toscalix__ has joined #automotive | 08:03 | |
*** jacobo has joined #automotive | 08:19 | |
*** jacobo has quit IRC | 08:19 | |
dzen | hi there. I'm looking for an RFC about the can dbc file. Does anyone have some good lectures ? I'm not found of vector.com documentation | 08:28 |
*** jacobo has joined #automotive | 08:51 | |
*** wschaller has joined #automotive | 09:20 | |
*** riazantsev has quit IRC | 10:49 | |
Figure | paulsherwood: from boot time perspective busybox would be the fastest. | 10:50 |
Figure | also from the memory footprint of busybox is smallest | 10:50 |
Figure | but I guess for most use cases systemd makes sense | 10:51 |
*** apinheiro has quit IRC | 11:05 | |
paulsherwood | Figure: i'm not sure about the boot speed argument, actually, for full ivi systems at least | 11:08 |
paulsherwood | lots of stuff needs to be started, with lots of interdepends... systemd is very good at that | 11:08 |
KlausUhl | paulsherwood: the problem is the early stuff that needs to be up quickly (if you don't want to do that in kernel space). It is not clear - at least for us - if systemd is fast enough for that. | 11:10 |
KlausUhl | systemd is definitely a very good solution for the rest | 11:11 |
paulsherwood | KlausUhl: ack. we've adopted some trickery on projects for the early stuff :) | 11:11 |
KlausUhl | the same for us :) | 11:12 |
KlausUhl | but we also would like to get rid of it | 11:12 |
KlausUhl | one reason being that it is non-trivial to integrate into yocto | 11:12 |
*** wschaller has quit IRC | 11:17 | |
paulsherwood | KlausUhl: get rid of systemd? | 11:19 |
radiofree | -10 | 11:29 |
radiofree | -1million actually | 11:29 |
KlausUhl | paulsherwood: sorry, my statement was a bit imprecise. With "it" I have been referring to the trickery, not systemd. | 11:37 |
KlausUhl | So no, we do not want to get rid of systemd. | 11:37 |
KlausUhl | We want to get rid of the trickery. | 11:37 |
*** riazantsev has joined #automotive | 11:40 | |
paulsherwood | ack :) | 11:41 |
jeremiah | Figure: Do you have any data to back up the claim that busybox is faster in boot time than systemd? Boot charts for example? | 11:43 |
*** fredcadete has joined #automotive | 11:44 | |
jeremiah | KlausUhl: I think there ought to be a way to discuss this kind of 'fastboot' trickery and (hopefully) standardization no? | 11:47 |
jeremiah | KlausUhl: Or do you think things are tied to specific hardware? | 11:48 |
*** wschaller has joined #automotive | 11:51 | |
KlausUhl | jeremiah: I do not see any hardware specifics here so there should be a way to standardize fastboot (and maybe integrate the recommendation into the GENIVI compliance specification, NSM, etc.) | 11:54 |
KlausUhl | jeremiah: there are, however, some hardware speicific things that influence the timing during startup | 11:55 |
KlausUhl | jeremiah: but those should only change the amount of time that remains between the start of Linux user space (e.g. systemd) and the fastboot time target | 11:56 |
*** wschaller has quit IRC | 12:37 | |
*** jacobo has quit IRC | 12:37 | |
*** wschaller has joined #automotive | 12:41 | |
Figure | jeremiah: less code means it's faster | 12:41 |
Figure | jeremiah: I've seen 700ms to opengl application from reset it was using busybox | 12:42 |
Figure | if I recall correctly systemd requires udev busybox has much simpler implementation from dev | 12:42 |
Figure | paulsherwood: yes true it depends how you define the boot time. any how boot time is something that should be considered when designing the system. E.g. bring up the ui before the services and load the services in background | 12:43 |
*** toscalix__ is now known as toscalix | 13:16 | |
jeremiah | Figure: Less code does not mean faster always. Parellization means faster. | 13:27 |
jeremiah | 700 ms to OpenGL sounds great. | 13:29 |
jeremiah | But I know that there are implementations out there for rear view camera that are < 250 ms done with a framebuffer. | 13:29 |
jeremiah | So I think its use case that matters. | 13:30 |
jeremiah | I've seen systemd bring up everything after the kernel in 1 second | 13:30 |
*** wschaller has quit IRC | 13:31 | |
*** wschaller has joined #automotive | 13:45 | |
fredcadete | in my opinion I expecto to save more time by using systemd when setting up complex unit dependencies, watchdogs, or even sandboxing, than the time needed to implement edge cases like rear view camera | 13:56 |
fredcadete | developer time, I mean | 13:57 |
fredcadete | then with that developer time I buy more optimizations | 13:57 |
fredcadete | and get boot time | 13:57 |
fredcadete | it's a perfect plan, can't fail | 13:57 |
jeremiah | :-) | 13:57 |
CTtpollard | Hi jeremiah, who has access to http://git.projects.genivi.org/?p=browser-poc.git;a=summary ? | 14:01 |
*** KlausUhl has quit IRC | 14:02 | |
*** CTtpollard has quit IRC | 14:18 | |
*** wschaller has quit IRC | 14:18 | |
*** CTtpollard has joined #automotive | 14:19 | |
*** wschaller has joined #automotive | 14:20 | |
*** CTtpollard has quit IRC | 14:26 | |
*** CTtpollard has joined #automotive | 14:27 | |
CTtpollard | sorry for the disconnects | 14:28 |
*** CTtpollard has quit IRC | 14:44 | |
*** wschaller has quit IRC | 14:44 | |
*** CTtpollard has joined #automotive | 14:51 | |
*** wschaller has joined #automotive | 14:52 | |
jeremiah | CTtpollard: I'll check, I can easily add you if needed. :-) | 15:00 |
jeremiah | Sorry for the late reply | 15:00 |
*** ToxicGumbo-work has joined #automotive | 15:02 | |
jeremiah | CTtpollard: The Network EG in GENIVI is the access group for the browser PoC | 15:05 |
jeremiah | If you would like to patch the browser poc I'm sure that granting you access would be a good approach, let me know if that works for you. | 15:06 |
*** apinheiro has joined #automotive | 15:07 | |
CTtpollard | jeremiah: sorry I've been having network woes, the patch is only changing one line, if others think an append in genivi-demo is good enough then I'm ok with either option | 15:10 |
*** rZr has left #automotive | 15:20 | |
*** toscalix has quit IRC | 15:24 | |
*** bbranch has joined #automotive | 15:43 | |
*** Raccoon1400 has joined #automotive | 15:57 | |
*** Raccoon1400 has left #automotive | 15:57 | |
*** jeremiah has quit IRC | 16:07 | |
*** wschaller has quit IRC | 16:55 | |
*** riazantsev has quit IRC | 17:55 | |
*** riazantsev has joined #automotive | 18:11 | |
*** fernandod has quit IRC | 18:47 | |
*** fernandod has joined #automotive | 18:47 | |
*** UserTest|62480 has joined #automotive | 19:26 | |
*** UserTest|62480 has left #automotive | 19:27 | |
*** apinheiro has quit IRC | 19:30 | |
*** ToxicGumbo-work has quit IRC | 21:08 | |
*** Egy has joined #automotive | 23:10 | |
*** Egy has quit IRC | 23:11 | |
*** bbranch has quit IRC | 23:51 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!