[PATCH] [RFC] serial: stm32: Move stm32_usart_transmit_chars() to interrupt thread

[Sebastian Andrzej Siewior]

On 2022-12-07 02:30:12 [+0100], Marek Vasut wrote:
> Avoid locking in hard interrupt context, move the stm32_usart_transmit_chars()
> into the threaded IRQ handler. This fixes the following splat with preempt-rt:
> 
>  BUG: scheduling while atomic: (mount)/1289/0x00010001
>  Modules linked in:
>  Preemption disabled at:
>  [<c0119127>] irq_enter_rcu+0xb/0x42
>  CPU: 0 PID: 1289 Comm: (mount) Not tainted 6.1.0-rc7-rt5-stable-standard-00006-gd70aeccb9f0f #17
>  Hardware name: STM32 (Device Tree Support)
>   unwind_backtrace from show_stack+0xb/0xc
>   show_stack from dump_stack_lvl+0x2b/0x34
>   dump_stack_lvl from __schedule_bug+0x53/0x80
>   __schedule_bug from __schedule+0x47/0x404
>   __schedule from schedule_rtlock+0x15/0x34
>   schedule_rtlock from rtlock_slowlock_locked+0x1d7/0x57e
>   rtlock_slowlock_locked from rt_spin_lock+0x29/0x3c
>   rt_spin_lock from stm32_usart_interrupt+0xa9/0x110
>   stm32_usart_interrupt from __handle_irq_event_percpu+0x73/0x14e
>   __handle_irq_event_percpu from handle_irq_event_percpu+0x9/0x22
>   handle_irq_event_percpu from handle_irq_event+0x53/0x76
>   handle_irq_event from handle_fasteoi_irq+0x65/0xa8
>   handle_fasteoi_irq from handle_irq_desc+0xf/0x18
>   handle_irq_desc from gic_handle_irq+0x45/0x54
>   gic_handle_irq from generic_handle_arch_irq+0x19/0x2c
>   generic_handle_arch_irq from call_with_stack+0xd/0x10
> 
> Signed-off-by: Marek Vasut <marex at denx.de>
> ---
> Cc: Alexandre Torgue <alexandre.torgue at foss.st.com>
> Cc: Erwan Le Ray <erwan.leray at foss.st.com>
> Cc: Greg Kroah-Hartman <gregkh at linuxfoundation.org>
> Cc: Jean Philippe Romain <jean-philippe.romain at foss.st.com>
> Cc: Jiri Slaby <jirislaby at kernel.org>
> Cc: Maxime Coquelin <mcoquelin.stm32 at gmail.com>
> Cc: Sebastian Andrzej Siewior <bigeasy at linutronix.de>
> Cc: Thomas Gleixner <tglx at linutronix.de>
> Cc: Valentin Caron <valentin.caron at foss.st.com>
> Cc: linux-arm-kernel at lists.infradead.org
> Cc: linux-stm32 at st-md-mailman.stormreply.com
> To: linux-serial at vger.kernel.org
> ---
>  drivers/tty/serial/stm32-usart.c | 19 +++++++++++++------
>  1 file changed, 13 insertions(+), 6 deletions(-)
> 
> diff --git a/drivers/tty/serial/stm32-usart.c b/drivers/tty/serial/stm32-usart.c
> index a1490033aa164..56357a7962edc 100644
> --- a/drivers/tty/serial/stm32-usart.c
> +++ b/drivers/tty/serial/stm32-usart.c
> @@ -791,11 +791,8 @@ static irqreturn_t stm32_usart_interrupt(int irq, void *ptr)
>  		}
>  	}

Why is this handler running as the primary handler to begin with?
There is
 stm32_usart_rs485_rts_disable() 
 -> mctrl_gpio_set()
    -> gpiod_set_array_value()
       -> gpiod_set_array_value_complex()
         -> gpio_chip_set_multiple()
	    -> gc->set_multiple() || gc->set() ?
         -> bitmap_alloc() boom

I don't know if the underlying gpiod is always using a raw_spinlock_t
but that bitmap_alloc() (depending on FASTPATH_NGPIO()) is not going to
work.
pm_wakeup_dev_event() is also using a spinlock_t.


> -	if ((sr & USART_SR_TXE) && !(stm32_port->tx_ch)) {
> -		spin_lock(&port->lock);
> -		stm32_usart_transmit_chars(port);
> -		spin_unlock(&port->lock);
> -	}
> +	if ((sr & USART_SR_TXE) && !(stm32_port->tx_ch))
> +		return IRQ_WAKE_THREAD;

Before that, there is a "stm32_port->throttled" check using the very
same lock.

>  	if (stm32_usart_rx_dma_enabled(port))
>  		return IRQ_WAKE_THREAD;

Sebastian