[PATCH] powerpc/32e: Ignore ESR in instruction storage interrupt handler

[Christophe Leroy]

Le 28/10/2021 à 15:30, Nicholas Piggin a écrit :
> A e5500 machine running a 32-bit kernel sometimes hangs at boot,
> seemingly going into an infinite loop of instruction storage interrupts.
> The ESR SPR has a value of 0x800000 (store) when this happens, which is
> likely set by a previous store. An instruction TLB miss interrupt would
> then leave ESR unchanged, and if no PTE exists it calls directly to the
> instruction storage interrupt handler without changing ESR.
> 
> access_error() does not cause a segfault due to a store to a read-only
> vma because is_exec is true. Most subsequent fault handling does not
> check for a write fault on a read-only vma, and might do strange things
> like create a writeable PTE or call page_mkwrite on a read only vma or
> file. It's not clear what happens here to cause the infinite faulting in
> this case, a fault handler failure or low level PTE or TLB handling.
> 
> In any case this can be fixed by having the instruction storage
> interrupt zero regs->dsisr rather than storing the ESR value to it.
> 
> Link: https://lore.kernel.org/linuxppc-dev/1635306738.0z8wt7619v.astroid@bobo.none/
> Fixes: a01a3f2ddbcd ("powerpc: remove arguments from fault handler functions")

Should it go to stable as well ?

> Reported-by: Jacques de Laval <jacques.delaval at protonmail.com>
> Tested-by: Jacques de Laval <jacques.delaval at protonmail.com>
> Signed-off-by: Nicholas Piggin <npiggin at gmail.com>

Reviewed-by: Christophe Leroy <christophe.leroy at csgroup.eu>

> ---
>   arch/powerpc/kernel/head_booke.h | 15 ++++++++++++---
>   1 file changed, 12 insertions(+), 3 deletions(-)
> 
> diff --git a/arch/powerpc/kernel/head_booke.h b/arch/powerpc/kernel/head_booke.h
> index e5503420b6c6..ef8d1b1c234e 100644
> --- a/arch/powerpc/kernel/head_booke.h
> +++ b/arch/powerpc/kernel/head_booke.h
> @@ -465,12 +465,21 @@ ALT_FTR_SECTION_END_IFSET(CPU_FTR_EMB_HV)
>   	bl	do_page_fault;						      \
>   	b	interrupt_return
>   
> +/*
> + * Instruction TLB Error interrupt handlers may call InstructionStorage
> + * directly without clearing ESR, so the ESR at this point may be left over
> + * from a prior interrupt.
> + *
> + * In any case, do_page_fault for BOOK3E does not use ESR and always expects
> + * dsisr to be 0. ESR_DST from a prior store in particular would confuse fault
> + * handling.
> + */
>   #define INSTRUCTION_STORAGE_EXCEPTION					      \
>   	START_EXCEPTION(InstructionStorage)				      \
> -	NORMAL_EXCEPTION_PROLOG(0x400, INST_STORAGE);		      \
> -	mfspr	r5,SPRN_ESR;		/* Grab the ESR and save it */	      \
> +	NORMAL_EXCEPTION_PROLOG(0x400, INST_STORAGE);			      \
> +	li	r5,0;			/* Store 0 in regs->esr (dsisr) */    \
>   	stw	r5,_ESR(r11);						      \
> -	stw	r12, _DEAR(r11);	/* Pass SRR0 as arg2 */		      \
> +	stw	r12, _DEAR(r11);	/* Set regs->dear (dar) to SRR0 */    \
>   	prepare_transfer_to_handler;					      \
>   	bl	do_page_fault;						      \
>   	b	interrupt_return
>