Re: drivers/input/keyboard/applespi.c:1833 applespi_drain_writes() warn: mixing irqsave and irq
From: Dmitry Torokhov
Date: Mon Jul 01 2024 - 13:14:14 EST
Hi Dan,
On Mon, Jul 01, 2024 at 06:30:38PM +0200, Dan Carpenter wrote:
> tree: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git master
> head: 6d6444ba82053c716fb5ac83346202659023044e
> commit: 038b1a05eae6666b731920e46f47d8e2332e07ff Input: add Apple SPI keyboard and trackpad driver
> config: x86_64-randconfig-161-20240628 (https://download.01.org/0day-ci/archive/20240628/202406280337.1Av9qm8V-lkp@xxxxxxxxx/config)
> compiler: clang version 18.1.5 (https://github.com/llvm/llvm-project 617a15a9eac96088ae5e9134248d8236e34b91b1)
>
> If you fix the issue in a separate patch/commit (i.e. not just a new version of
> the same patch/commit), kindly add following tags
> | Reported-by: kernel test robot <lkp@xxxxxxxxx>
> | Reported-by: Dan Carpenter <dan.carpenter@xxxxxxxxxx>
> | Closes: https://lore.kernel.org/r/202406280337.1Av9qm8V-lkp@xxxxxxxxx/
>
> smatch warnings:
> drivers/input/keyboard/applespi.c:1833 applespi_drain_writes() warn: mixing irqsave and irq
> drivers/input/keyboard/applespi.c:1845 applespi_drain_reads() warn: mixing irqsave and irq
>
> vim +1833 drivers/input/keyboard/applespi.c
>
> 038b1a05eae666 Ronald Tschalär 2019-07-15 1826 static void applespi_drain_writes(struct applespi_data *applespi)
> 038b1a05eae666 Ronald Tschalär 2019-07-15 1827 {
> 038b1a05eae666 Ronald Tschalär 2019-07-15 1828 unsigned long flags;
> 038b1a05eae666 Ronald Tschalär 2019-07-15 1829
> 038b1a05eae666 Ronald Tschalär 2019-07-15 1830 spin_lock_irqsave(&applespi->cmd_msg_lock, flags);
>
> _irqsave() sort of implies that potentially the caller has disabled IRQs,
> otherwise we would use spin_lock_irq() if we knew they hadn't disabled
> IRQs.
>
> 038b1a05eae666 Ronald Tschalär 2019-07-15 1831
> 038b1a05eae666 Ronald Tschalär 2019-07-15 1832 applespi->drain = true;
> 038b1a05eae666 Ronald Tschalär 2019-07-15 @1833 wait_event_lock_irq(applespi->drain_complete, !applespi->write_active,
> 038b1a05eae666 Ronald Tschalär 2019-07-15 1834 applespi->cmd_msg_lock);
>
> This will renable IRQs briefly and then disable them again.
>
> 038b1a05eae666 Ronald Tschalär 2019-07-15 1835
> 038b1a05eae666 Ronald Tschalär 2019-07-15 1836 spin_unlock_irqrestore(&applespi->cmd_msg_lock, flags);
>
> This will set them back to how they were at the start of the function.
>
> The issue is that if the caller really wanted IRQs disabled, then it's
> not necessarilly a good idea to enable them at all in the
> wait_event_lock_irq() macro. I suspect that in real life no callers
> disable IRQs so it's not an issue. But anyway, that's what the warning
> is about.
applespi_drain_writes() and applespi_drain_reads() are only called from
the process context with interrupts enabled, so it looks like we should
simply change spin_lock_irqsave() to spin_lock_irq().
I am taking patches ;)
Thanks.
--
Dmitry