Em Ter, 2009-06-16 às 15:52 +0300, Ilya Petrov escreveu: > Hi! > > I`m trying to suspend a1200 with "echo mem > /sys/power/state" and getting > this: > > http://pastebin.com/fe222ff7 > > looks like happened two things: > > 1) bp awaked ap just after sleep > 2) bp paniced or lost from ap after resume > > after reading ohci-pxa27x.c and old ipc driver i see following: > > * old driver suspends port when no data transmited during some timeout
Yes, and this was a layering violation. :) > * in new driver we disabled this behaviour and leaving suspend to hci driver Correct. > * hci driver not suspends ports, it stops and resets whole hci It disconnects, suspends and resumes the port. If the resume fails, then it resets the port. > * bp requires some actions to keep working after ap sleep Not true. > * there is also ap_rdy, which is used to wake up and for handshake and > commented out gpio_set in ezx-bp driver. It is used so AP can wake BP, not the other way around. > i`l try to shut up modem with AT+CMER=0,0,0,0,0 and set ap_rdy to 0 before > sleep > and set ap_rdy to 1 at resume. is this ok? CMER may help with the auto-resume thing. ap_rdy to 0 will probably _not_ do what you want. It is not used to signal BP that the AP is suspended, BP uses GPIO_SYS_EN or GPIO_PWR_EN instead (I dont remember which one). But you don't have control over this pin using normal GPIO registers. -- Daniel Ribeiro
signature.asc
Description: Esta é uma parte de mensagem assinada digitalmente