W dniu 2014-09-26 15:11, Russell King - ARM Linux pisze:
On Fri, Sep 26, 2014 at 02:59:18PM +0200, Janusz Użycki wrote:
But serial8250_get_mctrl() in 8250_core.c calls serial8250_modem_status()
which calls eg. uart_handle_cts_change() even if enable_ms() wasn't called.
This is the difference.
The serial8250_modem_status() is also called in the interrupt
and, what I don't understand, in serial8250_console_write().
Reading the MSR register clears the interrupts. So, whenever MSR is read,
you have to deal with any state changes which _would_ have been passed
to the interrupt function.
Plus, it's not quite as you make out above.
If enable_ms() is not called, then UART_IER_MSI will not be set in up->ier.
Hence, uart_handle_cts_change() will not be called.
Oh, now I understand the condition "up->ier & UART_IER_MSI"
in serial8250_modem_status().
So in case of gpio modem lines interrupt is the only right place like in
atmel_serial.c.
The reason for the call in the console function is to account for the
state changes during console write with CTS flow control - this again
needs the MSR register to be read, and we have to account for MSR state
changes after the console write has completed.
It means that gpio modem lines does not support flow control for console
on the moment, right Richard?
best regards
Janusz
--
To unsubscribe from this list: send the line "unsubscribe linux-serial" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
