Re: identifying a 1.75
martin wrote: On Tue, Aug 9, 2011 at 6:31 PM, Paul Fox p...@laptop.org wrote: thoughts/comments? better approaches? obvious additions? Hi Paul timely - I was just hacking on olpc-utils, bitfrost and sugar on exactly the same thing (while on the plane, no internet). Something along the lines of what you have is needed, I'll probably merge it into my hacking. And we need it as part of a mini bash function library as well, machine identification and other tasks reading from ofw are spread across olpc-utils at random. So I'll prolly hack olpc-hwinfo into a shell of what you posted (oh! the pun!) -- calling into shared function calls. And will refactor other scripts to match. something else i found this morning, while looking at #11126 -- udev uses dmi/id/product_name to decide to apply our keyboard map. this won't work on 1.75, so i guess we'll need to choose a/the canonical method of distinguishing a 1.75 from sysfs. paul cheers, m -- martin.langh...@gmail.com mar...@laptop.org -- Software Architect - OLPC - ask interesting questions - don't get distracted with shiny stuff - working code first - http://wiki.laptop.org/go/User:Martinlanghoff =- paul fox, p...@laptop.org ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
identifying a 1.75
on XO-1 and XO-1.5, we were able to discover the model of the laptop from the information under /sys/class/dmi/id. the DMI schema comes from the PC world, and we can't expect it to exist on ARM. there was also information to be found in /ofw on those machines, like serial number, and uuid. the hex model designator found there was used as a fallback if the dmi tree wasn't there (on older XO-1 firmware). on 1.75, there's no dmi tree, and /ofw has moved to /proc/device-tree, so we need to modify a lot of places that try and dig up platform info. (see #6) so i'm floating the attached script, tentatively named olpc-hwinfo, as a strawman. i think it gives access to the most often needed info, and can obviously be expanded if needed. it would go in olpc-utils, which would put it in /usr/sbin (since some clients live in /usr/sbin). thoughts/comments? better approaches? obvious additions? paul =- paul fox, p...@laptop.org olpc-hwinfo Description: - ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: identifying a 1.75
On Tue, Aug 9, 2011 at 16:31, Paul Fox p...@laptop.org wrote: on XO-1 and XO-1.5, we were able to discover the model of the laptop from the information under /sys/class/dmi/id. the DMI schema comes from the PC world, and we can't expect it to exist on ARM. there was also information to be found in /ofw on those machines, like serial number, and uuid. the hex model designator found there was used as a fallback if the dmi tree wasn't there (on older XO-1 firmware). on 1.75, there's no dmi tree, and /ofw has moved to /proc/device-tree, so we need to modify a lot of places that try and dig up platform info. (see #6) so i'm floating the attached script, tentatively named olpc-hwinfo, as a strawman. i think it gives access to the most often needed info, and can obviously be expanded if needed. it would go in olpc-utils, which would put it in /usr/sbin (since some clients live in /usr/sbin). thoughts/comments? better approaches? obvious additions? Check the CPU? Shouldn't /proc/cpuinfo tell you what you have since the major change is cpu? -- Stephen J Smoogen. The core skill of innovators is error recovery, not failure avoidance. Randy Nelson, President of Pixar University. Let us be kind, one to another, for most of us are fighting a hard battle. -- Ian MacLaren ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: identifying a 1.75
stephen john smoogen wrote: On Tue, Aug 9, 2011 at 16:31, Paul Fox p...@laptop.org wrote: on XO-1 and XO-1.5, we were able to discover the model of the laptop from the information under /sys/class/dmi/id. the DMI schema comes from the PC world, and we can't expect it to exist on ARM. there was also information to be found in /ofw on those machines, like serial number, and uuid. the hex model designator found there was used as a fallback if the dmi tree wasn't there (on older XO-1 firmware). on 1.75, there's no dmi tree, and /ofw has moved to /proc/device-tree, so we need to modify a lot of places that try and dig up platform info. (see #6) so i'm floating the attached script, tentatively named olpc-hwinfo, as a strawman. i think it gives access to the most often needed info, and can obviously be expanded if needed. it would go in olpc-utils, which would put it in /usr/sbin (since some clients live in /usr/sbin). thoughts/comments? better approaches? obvious additions? Check the CPU? Shouldn't /proc/cpuinfo tell you what you have since the major change is cpu? yeah, i thought of that. it's likely the next OLPC product will use the same processor, so we'll need something else in the future anyway. it happens that /proc/cpuinfo even says: Hardware: OLPC XO-1.75 (since ARM kernels provide slightly different info than x86 kernels), which makes it very tempting to use that. but if we're lucky, the next product might share the same kernel (so that string may change). in any case, i think i'd prefer using info that sourced from the hardware or firmware rather than a compiled in string. (but maybe i'm missing something here, and that line in /proc/cpuinfo is exactly what we should be using. anyone?) paul =- paul fox, p...@laptop.org ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: identifying a 1.75
On Tue, Aug 9, 2011 at 6:31 PM, Paul Fox p...@laptop.org wrote: thoughts/comments? better approaches? obvious additions? Hi Paul timely - I was just hacking on olpc-utils, bitfrost and sugar on exactly the same thing (while on the plane, no internet). Something along the lines of what you have is needed, I'll probably merge it into my hacking. And we need it as part of a mini bash function library as well, machine identification and other tasks reading from ofw are spread across olpc-utils at random. So I'll prolly hack olpc-hwinfo into a shell of what you posted (oh! the pun!) -- calling into shared function calls. And will refactor other scripts to match. cheers, m -- martin.langh...@gmail.com mar...@laptop.org -- Software Architect - OLPC - ask interesting questions - don't get distracted with shiny stuff - working code first - http://wiki.laptop.org/go/User:Martinlanghoff ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: identifying a 1.75
martin wrote: On Tue, Aug 9, 2011 at 6:31 PM, Paul Fox p...@laptop.org wrote: thoughts/comments? better approaches? obvious additions? Hi Paul timely - I was just hacking on olpc-utils, bitfrost and sugar on exactly the same thing (while on the plane, no internet). Something along the lines of what you have is needed, I'll probably merge it into my hacking. And we need it as part of a mini bash function library as well, machine identification and other tasks reading from ofw are spread across olpc-utils at random. So I'll prolly hack olpc-hwinfo into a shell of what you posted (oh! the pun!) -- calling into shared function calls. And will refactor other scripts to match. okay. most clients don't need hw info at high rates, so i figured a self-contained script would be sufficient (and necessary, for some clients). but certainly refactoring into sourceable chunks is a fine idea. (and more to the point, i won't commit anything -- ball's in your court. :-) paul =- paul fox, p...@laptop.org ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
