On 12/09/2013 11:46 AM, Mark Rutland wrote:
[...]
+ /*
+ * CORE_SW_RST above may trigger power irq if previous status of PWRCTL
+ * was either BUS_ON or IO_HIGH_V. So before we enable the power irq
+ * interrupt in GIC (by registering the interrupt handler), we need to
+ * ensure that any pending power irq interrupt status is acknowledged
+ * otherwise power irq interrupt handler would be fired prematurely.
+ */
+ irq_status = readl_relaxed(msm_host->core_mem + CORE_PWRCTL_STATUS);
+ writel_relaxed(irq_status, (msm_host->core_mem + CORE_PWRCTL_CLEAR));
+ irq_ctl = readl_relaxed(msm_host->core_mem + CORE_PWRCTL_CTL);
+ if (irq_status & (CORE_PWRCTL_BUS_ON | CORE_PWRCTL_BUS_OFF))
+ irq_ctl |= CORE_PWRCTL_BUS_SUCCESS;
+ if (irq_status & (CORE_PWRCTL_IO_HIGH | CORE_PWRCTL_IO_LOW))
+ irq_ctl |= CORE_PWRCTL_IO_SUCCESS;
+ writel_relaxed(irq_ctl, (msm_host->core_mem + CORE_PWRCTL_CTL));
+ /*
+ * Ensure that above writes are propogated before interrupt enablement
+ * in GIC.
+ */
+ mb();
Does this guarantee that the device has finished responding to the write
and deasserted the interrupt line (i.e. does the device only acknowledge
the write once that is true)?
I am not sure that i understand your concern. The write to
CORE_PWRCTL_CTL should acknowledge and deassert the interrupt.
The mb() ensures that the write has reached the device, and the device's
slave interface has acknowledged the write. On some devices this
acknowledgement of the write can be asynchronous with respect to the
device changing state in response to the write (i.e. the interrupt might
get deasserted a short time after the write completes). Typically there
is a register that should be polled to see whether the state change has
completed.
Does the acknowledgement of the write only occur once the device has
changed state? Or might it change state in the background?
Thanks for clarifying this. All this is correct, but it will be
difficult to answer, because i don't have documentation and i can only
be guessing how exactly the hardware behaves internally.
I can remove this fragment from the initial version to keep it more
simple. Meanwhile i can do some tests reading/polling the status
register to see how and when exactly it changes.
Thanks,
Georgi
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html