Re: Proposed communication protocol between Host and EC for 1.75
On May 25, 2010, at 10:38 AM, Richard A. Smith wrote: > On 05/24/2010 11:03 PM, Mitch Bradley wrote: >> Another protocol idea is at >> http://wiki.laptop.org/go/Talk:XO_1.75_HOST_to_EC_Protocol > > The SPI interface on the CPU is shared with the SPI boot flash. So on > boot the CPU has to be master and then switch over to slave when the > boot it finished. To use this setup we will also have to negotiate a > master slave switch when the CPU wants to change mfg data or when it > wants to update the firmware. Not quite true, as there are additional SPI interfaces available through the SSP ports. wad ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: Proposed communication protocol between Host and EC for 1.75
On 05/24/2010 11:03 PM, Mitch Bradley wrote: > Another protocol idea is at > http://wiki.laptop.org/go/Talk:XO_1.75_HOST_to_EC_Protocol The SPI interface on the CPU is shared with the SPI boot flash. So on boot the CPU has to be master and then switch over to slave when the boot it finished. To use this setup we will also have to negotiate a master slave switch when the CPU wants to change mfg data or when it wants to update the firmware. Also how does this protocol sync? Meaning how do you guarantee that a channel byte is really getting read as a channel byte? The alternate protocol with CPU as master is very similar to what EnE proposed to me while over in Taiwan. While trying to accommodate the exchanges that need more than 4 bytes I moved on to something else. We have: - EC wakeup command which needs 5 bytes - Battery ID command which has a 8 byte payload - Advanced mouse data. from what pgf tells me synaptics has an advanced mode with data > 3 bytes and that we should eventually switch to this mode. There's also the possibility that the EC might need to handle the incoming touchscreen data depending on how that all shakes out. I also wonder about sync here too. If you become un-synced how do you resync? -- Richard A. Smith One Laptop per Child ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
