[PATCH] Fix the message in facility unavailable exception

[PATCH] Fix the message in facility unavailable exception

[Balbir Singh]

I ran into this during some testing on qemu. The current
facility_strings[] are correct when the trap address is
0xf80 (hypervisor facility unavailable). When the trap
address is 0xf60, IC (Interruption Cause) a.k.a status
in the code is undefined for values 0 and 1. This patch
adds a check to prevent printing the wrong information
and helps better direct debugging effort.

Signed-off-by: Balbir Singh <bsingharora@gmail.com>
---
 arch/powerpc/kernel/traps.c | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)

diff --git a/arch/powerpc/kernel/traps.c b/arch/powerpc/kernel/traps.c
index d26605d..da0f634 100644
--- a/arch/powerpc/kernel/traps.c
+++ b/arch/powerpc/kernel/traps.c
@@ -1520,8 +1520,14 @@ void facility_unavailable_exception(struct pt_regs *regs)
 	}
 
 	if ((status < ARRAY_SIZE(facility_strings)) &&
-	    facility_strings[status])
-		facility = facility_strings[status];
+	    facility_strings[status]) {
+		if (!hv && status < 2) {
+			pr_warn("Unexpected facility unavailable exception "
+				"interruption cause %d\n", status);
+			facility = "Unknown";
+		} else
+			facility = facility_strings[status];
+	}
 
 	/* We restore the interrupt state now */
 	if (!arch_irq_disabled_regs(regs))
-- 
2.5.5

Re: [PATCH] Fix the message in facility unavailable exception

[Michael Ellerman]

Balbir Singh <bsingharora@gmail.com> writes:

> I ran into this during some testing on qemu. The current
> facility_strings[] are correct when the trap address is
> 0xf80 (hypervisor facility unavailable). When the trap
> address is 0xf60, IC (Interruption Cause) a.k.a status
> in the code is undefined for values 0 and 1.

OK. But how did you generate an exception with an undefined status code?

> This patch
> adds a check to prevent printing the wrong information
> and helps better direct debugging effort.
>
> diff --git a/arch/powerpc/kernel/traps.c b/arch/powerpc/kernel/traps.c
> index d26605d..da0f634 100644
> --- a/arch/powerpc/kernel/traps.c
> +++ b/arch/powerpc/kernel/traps.c
> @@ -1520,8 +1520,14 @@ void facility_unavailable_exception(struct pt_regs *regs)
>  	}
>  
>  	if ((status < ARRAY_SIZE(facility_strings)) &&
> -	    facility_strings[status])
> -		facility = facility_strings[status];
> +	    facility_strings[status]) {
> +		if (!hv && status < 2) {
> +			pr_warn("Unexpected facility unavailable exception "
> +				"interruption cause %d\n", status);

Please don't add un-ratelimited printks() in this function, otherwise if
they're user triggerable (which some are) it gives the user a way to
scrub the kernel log.

> +			facility = "Unknown";
> +		} else
> +			facility = facility_strings[status];
> +	}

I think we should instead tighten the condition on that top-level if, and
have an else clause for all cases that uses "Unknown". eg.

	if ((hv || status >= 2) &&
            (status < ARRAY_SIZE(facility_strings)) &&
	    facility_strings[status])
        {
		facility = facility_strings[status];
        } else {
		facility = "Unknown";
        }

And then if you want to we can also print the hex status value in the
existing printk().

cheers