Re: [PATCH v2] xen/arm64: Remove READ/WRITE_SYSREG32 helper macros

[Michal Orzel <michal.orzel@xxxxxxx>]

Hi Julien,

On 09.07.2021 17:21, Julien Grall wrote:
> Hi Michal,
> 
> On 09/07/2021 13:40, Michal Orzel wrote:
>> AArch64 system registers are 64bit whereas AArch32 ones
>> are 32bit or 64bit. MSR/MRS are expecting 64bit values thus
>> we should get rid of helpers READ/WRITE_SYSREG32
>> in favour of using READ/WRITE_SYSREG.
>>
>> The last place in code making use of READ/WRITE_SYSREG32
>> on arm64 is in TVM_REG macro defining functions vreg_emulate_<register>.
>> Implement a macro WRITE_SYSREG_SZ which expands as follows:
>> -on arm64: WRITE_SYSREG
>> -on arm32: WRITE_SYSREG{32/64}
>>
>> As there are no other places in the code using these helpers
>> on arm64 - remove them.
>>
>> Signed-off-by: Michal Orzel <michal.orzel@xxxxxxx>
>> ---
>> Changes since v1:
>> -implement WRITE_SYSREG_SZ instead of duplicating the TVM_REG
>> ---
>>   xen/arch/arm/vcpreg.c               | 12 +++++++++++-
>>   xen/include/asm-arm/arm64/sysregs.h |  4 ----
>>   2 files changed, 11 insertions(+), 5 deletions(-)
>>
>> diff --git a/xen/arch/arm/vcpreg.c b/xen/arch/arm/vcpreg.c
>> index f0cdcc8a54..10c4846954 100644
>> --- a/xen/arch/arm/vcpreg.c
>> +++ b/xen/arch/arm/vcpreg.c
>> @@ -47,6 +47,16 @@
>>    *
>>    */
>>   +#ifdef CONFIG_ARM_64
>> +#define WRITE_SYSREG_SZ(sz, val, sysreg)     WRITE_SYSREG(val, sysreg)
> 
> I think you want to cast to (uint##sz##_t) to avoid any surprise. I think...
> 
But the val will always be of type uint##sz##_t because it is passed from ...
>> +#else
>> +/*
>> + * WRITE_SYSREG{32/64} on arm32 is defined as variadic macro which imposes
>> + * on the below macro to be defined like that as well.
>> + */
>> +#define WRITE_SYSREG_SZ(sz, val, sysreg...)  WRITE_SYSREG##sz(val, sysreg)
>> +#endif
>> +
>>   /* The name is passed from the upper macro to workaround macro expansion. 
>> */
>>   #define TVM_REG(sz, func, reg...)                                          
>>  \
>>   static bool func(struct cpu_user_regs *regs, uint##sz##_t *r, bool read)   
>>  \
>> @@ -55,7 +65,7 @@ static bool func(struct cpu_user_regs *regs, uint##sz##_t 
>> *r, bool read)    \
... here(*r argument).
So I do not think that such casting makes sense e.g
uint##sz##_t foo;
WRITE_SYSREG((uint##sz##_t)foo, bar);

Also whatever type we will pass to WRITE_SYSREG,  it will be casted to uint64_t 
...
>>       bool cache_enabled = vcpu_has_cache_enabled(v);                        
>>  \
>>                                                                              
>>  \
>>       GUEST_BUG_ON(read);                                                    
>>  \
>> -    WRITE_SYSREG##sz(*r, reg);                                              
>> \
>> +    WRITE_SYSREG_SZ(sz, *r, reg);                                           
>> \
>>                                                                              
>>  \
>>       p2m_toggle_cache(v, cache_enabled);                                    
>>  \
>>                                                                              
>>  \
>> diff --git a/xen/include/asm-arm/arm64/sysregs.h 
>> b/xen/include/asm-arm/arm64/sysregs.h
>> index 077fd95fb7..795901e1ba 100644
>> --- a/xen/include/asm-arm/arm64/sysregs.h
>> +++ b/xen/include/asm-arm/arm64/sysregs.h
>> @@ -87,10 +87,6 @@
>>     /* Access to system registers */
>>   -#define READ_SYSREG32(name) ((uint32_t)READ_SYSREG64(name))
>> -
>> -#define WRITE_SYSREG32(v, name) WRITE_SYSREG64((uint64_t)v, name)
> 
> ... this was intended behavior but I used the wrong cast here.
> 
>> -
>>   #define WRITE_SYSREG64(v, name) do {                    \
>>       uint64_t _r = v;                                    \
...here
>>       asm volatile("msr "__stringify(name)", %0" : : "r" (_r));       \
>>
> 

Cheers,
Michal