[PATCH v3 2/2] arm64: Fix single stepping in kernel traps

[Alex Bennée]

Julien Thierry <julien.thierry at arm.com> writes:

> Hi Will,
>
> On 24/10/17 15:27, Will Deacon wrote:
>> Hi Julien,
>>
>> On Thu, Oct 12, 2017 at 02:43:03PM +0100, Julien Thierry wrote:
>>> Software Step exception is missing after stepping a trapped instruction.
>>>
>>> Ensure SPSR.SS gets set to 0 after emulating/skipping a trapped instruction
>>> before doing ERET.
>>
>> Curious, but how did you spot this?
>>
>
> It was discovered after investigating the issue of single stepping on
> GIC distributor registers accesses from KVM guests. Once the cause of
> the issue was identified as being traps/faults from KVM, it was
> speculated we might have similar issues on EL1 traps which turned out
> to be the case.
>
>>> Signed-off-by: Julien Thierry <julien.thierry at arm.com>
>>> Reviewed-by: Alex Bennée <alex.bennee at linaro.org>
>>> Cc: Catalin Marinas <catalin.marinas at arm.com>
>>> Cc: Will Deacon <will.deacon at arm.com>
>>> Cc: Mark Rutland <mark.rutland at arm.com>
>>> ---
>>>   arch/arm64/include/asm/insn.h        |  5 +++++
>>>   arch/arm64/include/asm/traps.h       |  6 ++++++
>>>   arch/arm64/kernel/armv8_deprecated.c |  8 ++++----
>>>   arch/arm64/kernel/cpufeature.c       |  2 +-
>>>   arch/arm64/kernel/traps.c            | 21 ++++++++++++++++-----
>>>   5 files changed, 32 insertions(+), 10 deletions(-)
>>>
>>> diff --git a/arch/arm64/include/asm/insn.h b/arch/arm64/include/asm/insn.h
>>> index 4214c38..de5e31a 100644
>>> --- a/arch/arm64/include/asm/insn.h
>>> +++ b/arch/arm64/include/asm/insn.h
>>> @@ -22,6 +22,11 @@
>>>
>>>   /* A64 instructions are always 32 bits. */
>>>   #define	AARCH64_INSN_SIZE		4
>>> +#define	AARCH32_INSN_SIZE		4
>>> +
>>> +/* Thumb/Thumb2 instruction sizes */
>>> +#define	AARCH32_T32_INSN_SIZE		4
>>> +#define	AARCH32_T16_INSN_SIZE		2
>>
>> The naming here is a bit misleading, since T32 is an instruction set
>> consisting of both 32-bit and 16-bit instructions. For now, you might just
>> be better off having the caller pass in an immediate directly rather than
>> add these confusing #defines.
>>
>
> I see. I guess I'll remove them.
>
>>>   #ifndef __ASSEMBLY__
>>>   /*
>>> diff --git a/arch/arm64/include/asm/traps.h b/arch/arm64/include/asm/traps.h
>>> index d131501..67b971e 100644
>>> --- a/arch/arm64/include/asm/traps.h
>>> +++ b/arch/arm64/include/asm/traps.h
>>> @@ -37,6 +37,12 @@ struct undef_hook {
>>>
>>>   void arm64_notify_segfault(struct pt_regs *regs, unsigned long addr);
>>>
>>> +/*
>>> + * Move regs->pc to next instruction and do necessary setup before it
>>> + * is executed.
>>> + */
>>> +void arm64_setup_next_instr(struct pt_regs *regs, unsigned long size);
>>> +
>>>   static inline int __in_irqentry_text(unsigned long ptr)
>>>   {
>>>   	return ptr >= (unsigned long)&__irqentry_text_start &&
>>> diff --git a/arch/arm64/kernel/armv8_deprecated.c b/arch/arm64/kernel/armv8_deprecated.c
>>> index d06fbe4..2808e56 100644
>>> --- a/arch/arm64/kernel/armv8_deprecated.c
>>> +++ b/arch/arm64/kernel/armv8_deprecated.c
>>> @@ -431,7 +431,7 @@ static int swp_handler(struct pt_regs *regs, u32 instr)
>>>   	pr_warn_ratelimited("\"%s\" (%ld) uses obsolete SWP{B} instruction at 0x%llx\n",
>>>   			current->comm, (unsigned long)current->pid, regs->pc);
>>>
>>> -	regs->pc += 4;
>>> +	arm64_setup_next_instr(regs, AARCH32_INSN_SIZE);
>>>   	return 0;
>>>
>>>   fault:
>>> @@ -512,7 +512,7 @@ static int cp15barrier_handler(struct pt_regs *regs, u32 instr)
>>>   	pr_warn_ratelimited("\"%s\" (%ld) uses deprecated CP15 Barrier instruction at 0x%llx\n",
>>>   			current->comm, (unsigned long)current->pid, regs->pc);
>>>
>>> -	regs->pc += 4;
>>> +	arm64_setup_next_instr(regs, AARCH32_INSN_SIZE);
>>>   	return 0;
>>>   }
>>>
>>> @@ -586,14 +586,14 @@ static int compat_setend_handler(struct pt_regs *regs, u32 big_endian)
>>>   static int a32_setend_handler(struct pt_regs *regs, u32 instr)
>>>   {
>>>   	int rc = compat_setend_handler(regs, (instr >> 9) & 1);
>>> -	regs->pc += 4;
>>> +	arm64_setup_next_instr(regs, AARCH32_INSN_SIZE);
>>>   	return rc;
>>>   }
>>>
>>>   static int t16_setend_handler(struct pt_regs *regs, u32 instr)
>>>   {
>>>   	int rc = compat_setend_handler(regs, (instr >> 3) & 1);
>>> -	regs->pc += 2;
>>> +	arm64_setup_next_instr(regs, AARCH32_T16_INSN_SIZE);
>>>   	return rc;
>>>   }
>>>
>>> diff --git a/arch/arm64/kernel/cpufeature.c b/arch/arm64/kernel/cpufeature.c
>>> index 21e2c95..235834e 100644
>>> --- a/arch/arm64/kernel/cpufeature.c
>>> +++ b/arch/arm64/kernel/cpufeature.c
>>> @@ -1287,7 +1287,7 @@ static int emulate_mrs(struct pt_regs *regs, u32 insn)
>>>   	if (!rc) {
>>>   		dst = aarch64_insn_decode_register(AARCH64_INSN_REGTYPE_RT, insn);
>>>   		pt_regs_write_reg(regs, dst, val);
>>> -		regs->pc += 4;
>>> +		arm64_setup_next_instr(regs, AARCH64_INSN_SIZE);
>>>   	}
>>>
>>>   	return rc;
>>> diff --git a/arch/arm64/kernel/traps.c b/arch/arm64/kernel/traps.c
>>> index 5ea4b85..f93b33f 100644
>>> --- a/arch/arm64/kernel/traps.c
>>> +++ b/arch/arm64/kernel/traps.c
>>> @@ -293,6 +293,17 @@ void arm64_notify_die(const char *str, struct pt_regs *regs,
>>>   	}
>>>   }
>>>
>>> +void arm64_setup_next_instr(struct pt_regs *regs, unsigned long size)
>>> +{
>>
>> This strikes me as a pretty broadly named function to be exposing like this.
>> How about: arm64_skip_faulting_instruction instead?
>>
>
> Hmmm, I think Alex was not fond of that kind of naming because we are
> not just skipping. Although here we specify it is for faulting
> instruction.
>
> Your suggestion suits, so if everyone agrees on it I'll change it.

I'm happy to defer to Will on this one ;-)

--
Alex Bennée