[PATCH 1/2] arm64: smp: Fix pseudo NMI issues w/ broken Mediatek FW

[Marc Zyngier]

On Mon, 02 Oct 2023 18:24:11 +0100,
Mark Rutland <mark.rutland at arm.com> wrote:
> 
> On Mon, Oct 02, 2023 at 09:45:29AM -0700, Douglas Anderson wrote:
> > Some mediatek devices have the property
> > "mediatek,broken-save-restore-fw" in their GIC. This means that,
> > although the hardware supports pseudo-NMI, the firmware has a bug
> > that blocks enabling it. When we're in this state,
> > system_uses_irq_prio_masking() will return true but we'll fail to
> > actually enable the IRQ in the GIC.
> > 
> > Let's make the code handle this. We'll detect that we failed to
> > request an IPI as NMI and fallback to requesting it normally. Though
> > we expect that either all of our requests will fail or all will
> > succeed, it's just as cheap to keep a per-IPI bitmap and that keeps us
> > robust.
> > 
> > Fixes: 331a1b3a836c ("arm64: smp: Add arch support for backtrace using pseudo-NMI")
> > Reported-by: Chen-Yu Tsai <wenst at chromium.org>
> > Closes: https://issuetracker.google.com/issues/197061987#comment68
> > Signed-off-by: Douglas Anderson <dianders at chromium.org>
> > ---
> > 
> >  arch/arm64/kernel/smp.c | 19 ++++++++++++-------
> >  1 file changed, 12 insertions(+), 7 deletions(-)
> 
> I'm not too keen on falling back here when we have no idea why the request failed.
> 
> I'd prefer if we could check the `supports_pseudo_nmis` static key directly to
> account for the case of broken FW, e.g. as below.
> 
> Mark.
> 
> ---->8----
> From 72fdec05c64a74f21871b44c7c760bbe07cac044 Mon Sep 17 00:00:00 2001
> From: Mark Rutland <mark.rutland at arm.com>
> Date: Mon, 2 Oct 2023 18:00:36 +0100
> Subject: [PATCH] arm64: smp: avoid NMI IPIs with broken MediaTek FW
> 
> Some MediaTek devices have broken firmware which corrupts some GICR
> registers behind the back of the OS, and pseudo-NMIs cannot be used on
> these devices. For more details see commit:
> 
>   44bd78dd2b8897f5 ("irqchip/gic-v3: Disable pseudo NMIs on Mediatek devices w/ firmware issues")
> 
> We did not take this problem into account in commit:
> 
>   331a1b3a836c0f38 ("arm64: smp: Add arch support for backtrace using pseudo-NMI")
> 
> Since that commit arm64's SMP code will try to setup some IPIs as
> pseudo-NMIs, even on systems with broken FW. The GICv3 code will
> (rightly) reject attempts to request interrupts as pseudo-NMIs,
> resulting in boot-time failures.
> 
> Avoid the problem by taking the broken FW into account when deciding to
> request IPIs as pseudo-NMIs. The GICv3 driver maintains a static_key
> named "supports_pseudo_nmis" which is false on systems with broken FW,
> and we can consult this within ipi_should_be_nmi().
> 
> Fixes: 331a1b3a836c0f38 ("arm64: smp: Add arch support for backtrace using pseudo-NMI")
> Reported-by: Chen-Yu Tsai <wenst at chromium.org>
> Closes: https://issuetracker.google.com/issues/197061987#comment68
> Signed-off-by: Mark Rutland <mark.rutland at arm.com>
> Cc: Douglas Anderson <dianders at chromium.org>
> Cc: Marc Zyngier <maz at kernel.org>
> ---
>  arch/arm64/kernel/smp.c      | 5 ++++-
>  drivers/irqchip/irq-gic-v3.c | 2 +-
>  2 files changed, 5 insertions(+), 2 deletions(-)
> 
> diff --git a/arch/arm64/kernel/smp.c b/arch/arm64/kernel/smp.c
> index 814d9aa93b21b..061c69160f90f 100644
> --- a/arch/arm64/kernel/smp.c
> +++ b/arch/arm64/kernel/smp.c
> @@ -964,7 +964,10 @@ static void smp_cross_call(const struct cpumask *target, unsigned int ipinr)
>  
>  static bool ipi_should_be_nmi(enum ipi_msg_type ipi)
>  {
> -	if (!system_uses_irq_prio_masking())
> +	DECLARE_STATIC_KEY_FALSE(supports_pseudo_nmis);
> +
> +	if (!system_uses_irq_prio_masking() ||
> +	    !static_branch_likely(&supports_pseudo_nmis))
>  		return false;
>  
>  	switch (ipi) {
> diff --git a/drivers/irqchip/irq-gic-v3.c b/drivers/irqchip/irq-gic-v3.c
> index 787ccc880b22d..737da1b9aabf2 100644
> --- a/drivers/irqchip/irq-gic-v3.c
> +++ b/drivers/irqchip/irq-gic-v3.c
> @@ -106,7 +106,7 @@ static DEFINE_STATIC_KEY_TRUE(supports_deactivate_key);
>   * - Figure 4-7 Secure read of the priority field for a Non-secure Group 1
>   *   interrupt.
>   */
> -static DEFINE_STATIC_KEY_FALSE(supports_pseudo_nmis);
> +DEFINE_STATIC_KEY_FALSE(supports_pseudo_nmis);
>  
>  DEFINE_STATIC_KEY_FALSE(gic_nonsecure_priorities);
>  EXPORT_SYMBOL(gic_nonsecure_priorities);

This last hunk is going to result in more spam from the robots about
global objects without a previous declaration. Not that I care the
least, but worth mentioning.

Other than that:

Reviewed-by: Marc Zyngier <maz at kernel.org>

Please take it via the arm64 tree with patch #1

Thanks,

	M.

-- 
Without deviation from the norm, progress is not possible.