atmel_serial might lose modem status change
Subsystems (2)
PATCH 2/3 atmel_serial might lose modem status change
From: Atsushi Nemoto <[email protected]>
I found a problem of handling of modem status of atmel_serial driver.
With the commit 1ecc26 ("atmel_serial: split the interrupt handler"), handling of modem status signal was splitted into two parts. The atmel_tasklet_func() compares new status with irq_status_prev, but irq_status_prev is not correct if signal status was changed while the port is closed.
Here is a sequence to cause problem:
1. Remote side sets CTS (and DSR).
2. Local side close the port.
3. Local side clears RTS and DTR.
4. Remote side clears CTS and DSR.
5. Local side reopen the port. hw_stopped becomes 1.
6. Local side sets RTS and DTR.
7. Remote side sets CTS and DSR.
Then CTS change interrupt can be received, but since CTS bit in irq_status_prev and new status is same, uart_handle_cts_change() will not be called (so hw_stopped will not be cleared, i.e. cannot send any data).
I suppose irq_status_prev should be initialized at somewhere in open sequence.
---
drivers/serial/atmel_serial.c | 3 +++ 1 file changed, 3 insertions(+)
drivers/serial/atmel_serial.c renamed from drivers/serial/atmel_serial.c~atmel_serial-might-lose-modem-status-change +3 -0
@@ -877,6 +877,9 @@ static int atmel_startup(struct uart_por } } +/* Save current CSR for comparison in atmel_tasklet_func() */ + atmel_port->irq_status_prev =UART_GET_CSR (port); +/* * Finally, enable the serial port */
_
Quoted 24 lines · [email protected] wrote:
From: Atsushi Nemoto <[email protected]>
I found a problem of handling of modem status of
atmel_serialdriver.With the commit 1ecc26 ("
atmel_serial: split the interrupt handler"), handling of modem status signal was splitted into two parts. Theatmel_tasklet_func()compares new status withirq_status_prev, butirq_status_previs not correct if signal status was changed while the port is closed.Here is a sequence to cause problem:
1. Remote side sets CTS (and DSR).
2. Local side close the port.
3. Local side clears RTS and DTR.
4. Remote side clears CTS and DSR.
5. Local side reopen the port.hw_stoppedbecomes 1.
6. Local side sets RTS and DTR.
7. Remote side sets CTS and DSR.Then CTS change interrupt can be received, but since CTS bit in
irq_status_prevand new status is same,uart_handle_cts_change()will not be called (sohw_stoppedwill not be cleared, i.e. cannot send any data).
Quoted 8 lines
I suppose
irq_status_prevshould be initialized at somewhere in open sequence.Acked-by Remy Bohmer <[email protected]>Cc Haavard Skinnemoen <[email protected]>Cc Marc Pignat <[email protected]>Signed-off-by Atsushi Nemoto <[email protected]>Signed-off-by Andrew Morton <[email protected]>
I think I already ACKed this. However...
drivers/serial/atmel_serial.c static int atmel_startup(struct uart_por · line 882
Quoted 10 lines
diff -puN
drivers/serial/atmel_serial.c~atmel_serial-might-lose-modem-status-changedrivers/serial/atmel_serial.cdrivers/serial/atmel_serial.c renamed from drivers/serial/atmel_serial.c~atmel_serial-might-lose-modem-status-change +3
-0@@ -877,6 +877,9 @@ static int atmel_startup(struct uart_por } } +/* Save current CSR for comparison in atmel_tasklet_func() */ + atmel_port->irq_status_prev =UART_GET_CSR (port); +
...Itai Levi pointed out that we need to initialize atmel_port->irq_status as well here. His analysis is as follows:
Quoted 28 lines
drivers/serial/atmel_serial.c renamed from drivers/serial/atmel_serial.c~atmel_serial-might-lose-modem-status-change +0
-0Regarding the second part of the patch (which resets irq_status_prev), it turns out that both versions of the patch (mine and Atsushi's)still leave enough room for faulty behavior when opening the port.
This is because we are not resetting both
irq_status_prevandirq_statusinatmel_startup()to CSR, which leads faulty behavior in the following sequences:First case:
1. closing the port while CTS line = 1 (TX not allowed)
2. setting CTS line = 0 (TX allowed)
3. opening the port
4. transmitting one char
5. Cannot transmit more chars, although CTS line is 0Second case:
1. closing the port while CTS line = 0 (TX allowed)
2. setting CTS line = 1 (TX not allowed)
3. opening the port
4. receiving some chars
5. Now we can transmit, although CTS line is 1This reason for this is that the tasklet is scheduled as a result of TX or RX interrupts (not a status change!), in steps 4 above. Inside the tasklet, the
atmel_port->irq_status(which holds the value from the previous session) is compared toatmel_port->irq_status_prev. Hence, a status-change of the CTS line is faultily detected.
Quoted 7 lines
Both cases were verified on 9260 hardware.
The solution for this, is to reset both
irq_status_prevandirq_statusto the value of CSR inatmel_startup().Here is my updated patch (you can ignore the first part which you already accepted):
drivers/serial/atmel_serial.c +4 -1
@@ -579,7 +579,7 @@ static void atmel_tx_dma(struct uart_por -if (!uart_circ_empty (xmit)) { +if (!uart_circ_empty (xmit) && !uart_tx_stopped (port)) { dma_sync_single_for_device(port->dev, pdc->dma_addr, pdc->dma_size,
@@ -805,6 +805,9 @@ static int atmel_startup(struct uart_por*/ UART_PUT_IDR (port,-1 ); + atmel_port->irq_status_prev =UART_GET_CSR (port); + atmel_port->irq_status = atmel_port->irq_status_prev; +/* * Allocate the IRQ */
Quoted 19 lines · Haavard Skinnemoen
drivers/serial/atmel_serial.c static int atmel_startup(struct uart_por · line 882
Quoted 8 lines
+++
a/drivers/serial/atmel_serial.cdrivers/serial/atmel_serial.c +3
-0@@ -877,6 +877,9 @@ static int atmel_startup(struct uart_por } } +/* Save current CSR for comparison in atmel_tasklet_func() */ + atmel_port->irq_status_prev =UART_GET_CSR (port); +...Itai Levi pointed out that we need to initialize
atmel_port->irq_statusas well here. His analysis is as follows:Quoted 7 lines
(quoted diff) +0
-0Regarding the second part of the patch (which resets irq_status_prev), it turns out that both versions of the patch (mine and Atsushi's)still leave enough room for faulty behavior when opening the port.
This is because we are not resetting both
irq_status_prevandirq_statusinatmel_startup()to CSR, which leads faulty behavior in the following sequences:
...
drivers/serial/atmel_serial.c static int atmel_startup(struct uart_por · line 813
Quoted 10 lines
drivers/serial/atmel_serial.c +3
-0@@ -805,6 +805,9 @@ static int atmel_startup(struct uart_por*/ UART_PUT_IDR (port,-1 ); + atmel_port->irq_status_prev =UART_GET_CSR (port); + atmel_port->irq_status = atmel_port->irq_status_prev; +/* * Allocate the IRQ */
Looks correct for me too. Then, what is preffered final fix?
Put irq_status initialization into my patch, or replace with Itai's patch? These patch put initialization in another place. Any comments for preferred place to fix?
---
Atsushi Nemoto
Quoted 13 lines · Atsushi Nemoto
drivers/serial/atmel_serial.c static int atmel_startup(struct uart_por · line 813
Quoted 10 lines
drivers/serial/atmel_serial.c +3
-0@@ -805,6 +805,9 @@ static int atmel_startup(struct uart_por*/ UART_PUT_IDR (port,-1 ); + atmel_port->irq_status_prev =UART_GET_CSR (port); + atmel_port->irq_status = atmel_port->irq_status_prev; +/* * Allocate the IRQ */ Looks correct for me too. Then, what is preffered final fix?
Quoted 3 lines
Put
irq_statusinitialization into my patch, or replace with Itai's patch? These patch put initialization in another place. Any comments for preferred place to fix?
I don't think it matters as long as it happens before the interrupts are enabled. But it's probably enough to do it once ;-)
So the easiest solution is probably to drop your patch (atmel_serial-might-lose-modem-status-change.patch) and keep Itai's patch (atmel_serial-might-lose-modem-status-change-fix.patch).
Andrew, if it makes things easier, feel free to drop both patches and I'll send you a new one.
Haavard