linux-kernel - Re: [PATCH] irq: Resolve that mask_irq/unmask

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <875y12p2r0.ffs@tglx>
Date:   Wed, 13 Dec 2023 15:59:31 +0100
From:   Thomas Gleixner <tglx@...utronix.de>
To:     xiongxin <xiongxin@...inos.cn>, jikos@...nel.org,
        benjamin.tissoires@...hat.com
Cc:     linux-input@...r.kernel.org, stable@...r.kernel.org,
        Riwen Lu <luriwen@...inos.cn>, hoan@...amperecomputing.com,
        fancer.lancer@...il.com, linus.walleij@...aro.org, brgl@...ev.pl,
        andy@...nel.org, linux-gpio@...r.kernel.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH] irq: Resolve that mask_irq/unmask_irq may not be called
 in pairs

On Wed, Dec 13 2023 at 10:29, xiongxin wrote:
> 在 2023/12/12 23:17, Thomas Gleixner 写道:
> Sorry, the previous reply may not have clarified the BUG process. I 
> re-debugged and confirmed it yesterday. The current BUG execution 
> sequence is described as follows:

It's the sequence how this works and it works correctly.

Just because it does not work on your machine it does not mean that this
is incorrect and a BUG.

You are trying to fix a symptom and thereby violating guarantees of the
core code.

> That is, there is a time between the 1:handle_level_irq() and 
> 3:irq_thread_fn() calls for the 2:disable_irq() call to acquire the lock 
> and then implement the irq_state_set_disabled() operation. When finally 
> call irq_thread_fn()->irq_finalize_oneshot(), it cannot enter the 
> unmask_thread_irq() process.

Correct, because the interrupt has been DISABLED in the mean time.

> In this case, the gpio irq_chip irq_mask()/irq_unmask() callback pairs 
> are not called in pairs, so I think this is a BUG, but not necessarily 
> fixed from the irq core code layer.

No. It is _NOT_ a BUG. unmask() is not allowed to be invoked when the
interrupt is DISABLED. That's the last time I'm going to tell you that.
Only enable_irq() can undo the effect of disable_irq(), period.

> Next, when the gpio controller driver calls the suspend/resume process, 
> it is as follows:
>
> suspend process:
> dwapb_gpio_suspend()
>      ctx->int_mask   = dwapb_read(gpio, GPIO_INTMASK);
>
> resume process:
> dwapb_gpio_resume()
>      dwapb_write(gpio, GPIO_INTMASK, ctx->int_mask);

Did you actually look at the sequence I gave you?

   Suspend:

	  i2c_hid_core_suspend()
	     disable_irq();       <- Marks it disabled and eventually
				     masks it.

	  gpio_irq_suspend()
	     save_registers();    <- Saves masked interrupt

   Resume:

	  gpio_irq_resume()
	     restore_registers(); <- Restores masked interrupt

	  i2c_hid_core_resume()
	     enable_irq();        <- Unmasks interrupt and removes the
				     disabled marker


Have you verified that this order of invocations is what happens on
your machine?

Thanks,

        tglx