2020-02-07 17:22:33

by Christophe Leroy

[permalink] [raw]
Subject: [PATCH] powerpc: Fix CONFIG_TRACE_IRQFLAGS with CONFIG_VMAP_STACK

When CONFIG_PROVE_LOCKING is selected together with (now default)
CONFIG_VMAP_STACK, kernel enter deadlock during boot.

At the point of checking whether interrupts are enabled or not, the
value of MSR saved on stack is read using the physical address of the
stack. But at this point, when using VMAP stack the DATA MMU
translation has already been re-enabled, leading to deadlock.

Don't use the physical address of the stack when
CONFIG_VMAP_STACK is set.

Signed-off-by: Christophe Leroy <[email protected]>
Reported-by: Guenter Roeck <[email protected]>
Fixes: 028474876f47 ("powerpc/32: prepare for CONFIG_VMAP_STACK")
---
arch/powerpc/kernel/entry_32.S | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/powerpc/kernel/entry_32.S b/arch/powerpc/kernel/entry_32.S
index 77abbc34bbe0..0713daa651d9 100644
--- a/arch/powerpc/kernel/entry_32.S
+++ b/arch/powerpc/kernel/entry_32.S
@@ -214,7 +214,7 @@ transfer_to_handler_cont:
* To speed up the syscall path where interrupts stay on, let's check
* first if we are changing the MSR value at all.
*/
- tophys(r12, r1)
+ tophys_novmstack r12, r1
lwz r12,_MSR(r12)
andi. r12,r12,MSR_EE
bne 1f
--
2.25.0


2020-02-07 17:40:22

by Guenter Roeck

[permalink] [raw]
Subject: Re: [PATCH] powerpc: Fix CONFIG_TRACE_IRQFLAGS with CONFIG_VMAP_STACK

On Fri, Feb 07, 2020 at 05:20:57PM +0000, Christophe Leroy wrote:
> When CONFIG_PROVE_LOCKING is selected together with (now default)
> CONFIG_VMAP_STACK, kernel enter deadlock during boot.
>
> At the point of checking whether interrupts are enabled or not, the
> value of MSR saved on stack is read using the physical address of the
> stack. But at this point, when using VMAP stack the DATA MMU
> translation has already been re-enabled, leading to deadlock.
>
> Don't use the physical address of the stack when
> CONFIG_VMAP_STACK is set.
>
> Signed-off-by: Christophe Leroy <[email protected]>
> Reported-by: Guenter Roeck <[email protected]>
> Fixes: 028474876f47 ("powerpc/32: prepare for CONFIG_VMAP_STACK")

Tested-by: Guenter Roeck <[email protected]>

> ---
> arch/powerpc/kernel/entry_32.S | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/arch/powerpc/kernel/entry_32.S b/arch/powerpc/kernel/entry_32.S
> index 77abbc34bbe0..0713daa651d9 100644
> --- a/arch/powerpc/kernel/entry_32.S
> +++ b/arch/powerpc/kernel/entry_32.S
> @@ -214,7 +214,7 @@ transfer_to_handler_cont:
> * To speed up the syscall path where interrupts stay on, let's check
> * first if we are changing the MSR value at all.
> */
> - tophys(r12, r1)
> + tophys_novmstack r12, r1
> lwz r12,_MSR(r12)
> andi. r12,r12,MSR_EE
> bne 1f
> --
> 2.25.0
>

2020-02-08 12:50:45

by Michael Ellerman

[permalink] [raw]
Subject: Re: [PATCH] powerpc: Fix CONFIG_TRACE_IRQFLAGS with CONFIG_VMAP_STACK

On Fri, 2020-02-07 at 17:20:57 UTC, Christophe Leroy wrote:
> When CONFIG_PROVE_LOCKING is selected together with (now default)
> CONFIG_VMAP_STACK, kernel enter deadlock during boot.
>
> At the point of checking whether interrupts are enabled or not, the
> value of MSR saved on stack is read using the physical address of the
> stack. But at this point, when using VMAP stack the DATA MMU
> translation has already been re-enabled, leading to deadlock.
>
> Don't use the physical address of the stack when
> CONFIG_VMAP_STACK is set.
>
> Signed-off-by: Christophe Leroy <[email protected]>
> Reported-by: Guenter Roeck <[email protected]>
> Fixes: 028474876f47 ("powerpc/32: prepare for CONFIG_VMAP_STACK")

Applied to powerpc fixes, thanks.

https://git.kernel.org/powerpc/c/d4bf905307a1c90a27714ff7a9fd29b0a2ceed98

cheers