Re: [PATCH V2] arm64: tlbflush: avoid writing RES0 bits

[Philip Elcan]

On 3/27/2018 9:03 PM, Shanker Donthineni wrote:
> Hi Will,
> 
> On 03/27/2018 12:36 PM, Will Deacon wrote:
>> Hi Shanker,
>>
>> On Tue, Mar 27, 2018 at 09:53:16AM -0500, Shanker Donthineni wrote:
>>> On 03/27/2018 06:34 AM, Robin Murphy wrote:
>>>> On 27/03/18 04:21, Philip Elcan wrote:
>>>>> Several of the bits of the TLBI register operand are RES0 per the ARM
>>>>> ARM, so TLBI operations should avoid writing non-zero values to these
>>>>> bits.
>>>>>
>>>>> This patch adds a macro __TLBI_VADDR(addr, asid) that creates the
>>>>> operand register in the correct format and honors the RES0 bits.
>>>>>
>>>>> Signed-off-by: Philip Elcan <pelcan@xxxxxxxxxxxxxx>
>>>>> ---
>>>>> Â arch/arm64/include/asm/tlbflush.h | 23 ++++++++++++++++-------
>>>>> Â 1 file changed, 16 insertions(+), 7 deletions(-)
>>>>>
>>>>> diff --git a/arch/arm64/include/asm/tlbflush.h b/arch/arm64/include/asm/tlbflush.h
>>>>> index 9e82dd7..b1205e9 100644
>>>>> --- a/arch/arm64/include/asm/tlbflush.h
>>>>> +++ b/arch/arm64/include/asm/tlbflush.h
>>>>> @@ -60,6 +60,15 @@
>>>>> ÂÂÂÂÂÂÂÂÂ __tlbi(op, (arg) | USER_ASID_FLAG);ÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ \
>>>>> Â } while (0)
>>>>> Â +/* This macro creates a properly formatted VA operand for the TLBI */
>>>>> +#define __TLBI_VADDR(addr, asid)ÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ \
>>>>> +ÂÂÂ ({ÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ \
>>>>> +ÂÂÂÂÂÂÂ unsigned long __ta = (addr) >> 12;ÂÂÂÂÂÂÂ \
>>>>> +ÂÂÂÂÂÂÂ __ta &= GENMASK_ULL(43, 0);ÂÂÂÂÂÂÂÂÂÂÂ \
>>>>> +ÂÂÂÂÂÂÂ __ta |= (unsigned long)(asid) << 48;ÂÂÂÂÂÂÂ \
>>>>> +ÂÂÂÂÂÂÂ __ta;ÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ \
>>>>> +ÂÂÂ })
>>>>
>>>> I'd be inclined to make this a static inline function rather than a
>>>> macro, since it doesn't need to do any wacky type-dodging, but either
>>>> way the overall change now looks appropriate;
>>>>
>>>> Acked-by: Robin Murphy <robin.murphy@xxxxxxx>
>>>>
>>>
>>> Tested-by: Shanker Donthineni <shankerd@xxxxxxxxxxxxxx>
>>
>> [...]
>>
>>>>> @@ -154,8 +163,8 @@ static inline void __flush_tlb_range(struct vm_area_struct *vma,
>>>>> ÂÂÂÂÂÂÂÂÂ return;
>>>>> ÂÂÂÂÂ }
>>>>> Â -ÂÂÂ start = asid | (start >> 12);
>>>>> -ÂÂÂ end = asid | (end >> 12);
>>>>> +ÂÂÂ start = __TLBI_VADDR(start, asid);
>>>>> +ÂÂÂ end = __TLBI_VADDR(end, asid);
>>
>> Can you test this bit too, please? ;)
>>
> 
> I've verified the basic boot functionality on QDF2400 platform. But I can see now
> after your comments, it leads to TLB conflicts because of ASID is truncated to zero
> due to two times 48bit shift.     
> 
> Thanks for catching this one.
> 
> @@ -146,7 +155,7 @@ static inline void __flush_tlb_range(structvm_area_struct *
>                                      unsigned long start, unsigned long end,
>                                      bool last_level)
> {
> -       unsigned long asid = ASID(vma->vm_mm) << 48;
> +       unsigned long asid = ASID(vma->vm_mm);
>         unsigned long addr;
>  
>         if ((end - start) > MAX_TLB_RANGE) {
> @@ -154,8 +163,8 @@ static inline void __flush_tlb_range(struct
> vm_area_struct *
>                 return;
>         }
> 
> -       start = asid | (start >> 12);
> -       end = asid | (end >> 12);
> +       start = __TLBI_VADDR(start, asid);
> +       end = __TLBI_VADDR(end, asid);
> 
> 
>> Will
>>
>> _______________________________________________
>> linux-arm-kernel mailing list
>> linux-arm-kernel@xxxxxxxxxxxxxxxxxxx
>> http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
>>
> 

Thanks for catching that.  I'll address with a v3 patch.

Philip

-- 
Qualcomm Datacenter Technologies as an affiliate of Qualcomm
Technologies, Inc.
Qualcomm Technologies, Inc. is a member of the
Code Aurora Forum, a Linux Foundation Collaborative Project.