[v1] arm64: Refactor TLB invalidation API and implementation

[PATCH v1 01/13] arm64: mm: Re-implement the __tlbi_level macro as a C function

Posted by Ryan Roberts 1 month, 3 weeks ago

As part of efforts to reduce our reliance on complex preprocessor macros
for TLB invalidation routines, convert the __tlbi_level macro to a C
function for by-level TLB invalidation.

Each specific tlbi level op is implemented as a C function and the
appropriate function pointer is passed to __tlbi_level(). Since
everything is declared inline and is statically resolvable, the compiler
will convert the indirect function call to a direct inline execution.

Suggested-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Ryan Roberts <ryan.roberts@arm.com>
---
 arch/arm64/include/asm/tlbflush.h | 69 +++++++++++++++++++++++++------
 1 file changed, 56 insertions(+), 13 deletions(-)

diff --git a/arch/arm64/include/asm/tlbflush.h b/arch/arm64/include/asm/tlbflush.h
index a2d65d7d6aae..13a59cf28943 100644
--- a/arch/arm64/include/asm/tlbflush.h
+++ b/arch/arm64/include/asm/tlbflush.h
@@ -105,19 +105,62 @@ static inline unsigned long get_trans_granule(void)
 
 #define TLBI_TTL_UNKNOWN	INT_MAX
 
-#define __tlbi_level(op, addr, level) do {				\
-	u64 arg = addr;							\
-									\
-	if (alternative_has_cap_unlikely(ARM64_HAS_ARMv8_4_TTL) &&	\
-	    level >= 0 && level <= 3) {					\
-		u64 ttl = level & 3;					\
-		ttl |= get_trans_granule() << 2;			\
-		arg &= ~TLBI_TTL_MASK;					\
-		arg |= FIELD_PREP(TLBI_TTL_MASK, ttl);			\
-	}								\
-									\
-	__tlbi(op, arg);						\
-} while(0)
+typedef void (*tlbi_op)(u64 arg);
+
+static __always_inline void vae1is(u64 arg)
+{
+	__tlbi(vae1is, arg);
+}
+
+static __always_inline void vae2is(u64 arg)
+{
+	__tlbi(vae2is, arg);
+}
+
+static __always_inline void vale1(u64 arg)
+{
+	__tlbi(vale1, arg);
+	__tlbi_user(vale1, arg);
+}
+
+static __always_inline void vale1is(u64 arg)
+{
+	__tlbi(vale1is, arg);
+}
+
+static __always_inline void vale2is(u64 arg)
+{
+	__tlbi(vale2is, arg);
+}
+
+static __always_inline void vaale1is(u64 arg)
+{
+	__tlbi(vaale1is, arg);
+}
+
+static __always_inline void ipas2e1(u64 arg)
+{
+	__tlbi(ipas2e1, arg);
+}
+
+static __always_inline void ipas2e1is(u64 arg)
+{
+	__tlbi(ipas2e1is, arg);
+}
+
+static __always_inline void __tlbi_level(tlbi_op op, u64 addr, u32 level)
+{
+	u64 arg = addr;
+
+	if (alternative_has_cap_unlikely(ARM64_HAS_ARMv8_4_TTL) && level <= 3) {
+		u64 ttl = level | (get_trans_granule() << 2);
+
+		arg &= ~TLBI_TTL_MASK;
+		arg |= FIELD_PREP(TLBI_TTL_MASK, ttl);
+	}
+
+	op(arg);
+}
 
 #define __tlbi_user_level(op, arg, level) do {				\
 	if (arm64_kernel_unmapped_at_el0())				\
-- 
2.43.0

Re: [PATCH v1 01/13] arm64: mm: Re-implement the __tlbi_level macro as a C function

Posted by Linu Cherian 1 month ago

Ryan,

On Tue, Dec 16, 2025 at 02:45:46PM +0000, Ryan Roberts wrote:
> As part of efforts to reduce our reliance on complex preprocessor macros
> for TLB invalidation routines, convert the __tlbi_level macro to a C
> function for by-level TLB invalidation.
> 
> Each specific tlbi level op is implemented as a C function and the
> appropriate function pointer is passed to __tlbi_level(). Since
> everything is declared inline and is statically resolvable, the compiler
> will convert the indirect function call to a direct inline execution.
> 
> Suggested-by: Linus Torvalds <torvalds@linux-foundation.org>
> Signed-off-by: Ryan Roberts <ryan.roberts@arm.com>
> ---
>  arch/arm64/include/asm/tlbflush.h | 69 +++++++++++++++++++++++++------
>  1 file changed, 56 insertions(+), 13 deletions(-)
> 
> diff --git a/arch/arm64/include/asm/tlbflush.h b/arch/arm64/include/asm/tlbflush.h
> index a2d65d7d6aae..13a59cf28943 100644
> --- a/arch/arm64/include/asm/tlbflush.h
> +++ b/arch/arm64/include/asm/tlbflush.h
> @@ -105,19 +105,62 @@ static inline unsigned long get_trans_granule(void)
>  
>  #define TLBI_TTL_UNKNOWN	INT_MAX
>  
> -#define __tlbi_level(op, addr, level) do {				\
> -	u64 arg = addr;							\
> -									\
> -	if (alternative_has_cap_unlikely(ARM64_HAS_ARMv8_4_TTL) &&	\
> -	    level >= 0 && level <= 3) {					\
> -		u64 ttl = level & 3;					\
> -		ttl |= get_trans_granule() << 2;			\
> -		arg &= ~TLBI_TTL_MASK;					\
> -		arg |= FIELD_PREP(TLBI_TTL_MASK, ttl);			\
> -	}								\
> -									\
> -	__tlbi(op, arg);						\
> -} while(0)
> +typedef void (*tlbi_op)(u64 arg);
> +
> +static __always_inline void vae1is(u64 arg)
> +{
> +	__tlbi(vae1is, arg);
> +}
> +
> +static __always_inline void vae2is(u64 arg)
> +{
> +	__tlbi(vae2is, arg);
> +}
> +
> +static __always_inline void vale1(u64 arg)
> +{
> +	__tlbi(vale1, arg);
> +	__tlbi_user(vale1, arg);

Should the __tlbi_user addition be added as part of patch 3 ?

> +}
> +
> +static __always_inline void vale1is(u64 arg)
> +{
> +	__tlbi(vale1is, arg);
> +}
> +
> +static __always_inline void vale2is(u64 arg)
> +{
> +	__tlbi(vale2is, arg);
> +}
> +
> +static __always_inline void vaale1is(u64 arg)
> +{
> +	__tlbi(vaale1is, arg);
> +}
> +
> +static __always_inline void ipas2e1(u64 arg)
> +{
> +	__tlbi(ipas2e1, arg);
> +}
> +
> +static __always_inline void ipas2e1is(u64 arg)
> +{
> +	__tlbi(ipas2e1is, arg);
> +}
> +
> +static __always_inline void __tlbi_level(tlbi_op op, u64 addr, u32 level)
> +{
> +	u64 arg = addr;
> +
> +	if (alternative_has_cap_unlikely(ARM64_HAS_ARMv8_4_TTL) && level <= 3) {
> +		u64 ttl = level | (get_trans_granule() << 2);
> +
> +		arg &= ~TLBI_TTL_MASK;
> +		arg |= FIELD_PREP(TLBI_TTL_MASK, ttl);
> +	}
> +
> +	op(arg);
> +}
>  
>  #define __tlbi_user_level(op, arg, level) do {				\
>  	if (arm64_kernel_unmapped_at_el0())				\
> -- 
> 2.43.0
>

Re: [PATCH v1 01/13] arm64: mm: Re-implement the __tlbi_level macro as a C function

Posted by Ryan Roberts 1 month ago

On 05/01/2026 05:30, Linu Cherian wrote:
> Ryan,
> 
> On Tue, Dec 16, 2025 at 02:45:46PM +0000, Ryan Roberts wrote:
>> As part of efforts to reduce our reliance on complex preprocessor macros
>> for TLB invalidation routines, convert the __tlbi_level macro to a C
>> function for by-level TLB invalidation.
>>
>> Each specific tlbi level op is implemented as a C function and the
>> appropriate function pointer is passed to __tlbi_level(). Since
>> everything is declared inline and is statically resolvable, the compiler
>> will convert the indirect function call to a direct inline execution.
>>
>> Suggested-by: Linus Torvalds <torvalds@linux-foundation.org>
>> Signed-off-by: Ryan Roberts <ryan.roberts@arm.com>
>> ---
>>  arch/arm64/include/asm/tlbflush.h | 69 +++++++++++++++++++++++++------
>>  1 file changed, 56 insertions(+), 13 deletions(-)
>>
>> diff --git a/arch/arm64/include/asm/tlbflush.h b/arch/arm64/include/asm/tlbflush.h
>> index a2d65d7d6aae..13a59cf28943 100644
>> --- a/arch/arm64/include/asm/tlbflush.h
>> +++ b/arch/arm64/include/asm/tlbflush.h
>> @@ -105,19 +105,62 @@ static inline unsigned long get_trans_granule(void)
>>  
>>  #define TLBI_TTL_UNKNOWN	INT_MAX
>>  
>> -#define __tlbi_level(op, addr, level) do {				\
>> -	u64 arg = addr;							\
>> -									\
>> -	if (alternative_has_cap_unlikely(ARM64_HAS_ARMv8_4_TTL) &&	\
>> -	    level >= 0 && level <= 3) {					\
>> -		u64 ttl = level & 3;					\
>> -		ttl |= get_trans_granule() << 2;			\
>> -		arg &= ~TLBI_TTL_MASK;					\
>> -		arg |= FIELD_PREP(TLBI_TTL_MASK, ttl);			\
>> -	}								\
>> -									\
>> -	__tlbi(op, arg);						\
>> -} while(0)
>> +typedef void (*tlbi_op)(u64 arg);
>> +
>> +static __always_inline void vae1is(u64 arg)
>> +{
>> +	__tlbi(vae1is, arg);
>> +}
>> +
>> +static __always_inline void vae2is(u64 arg)
>> +{
>> +	__tlbi(vae2is, arg);
>> +}
>> +
>> +static __always_inline void vale1(u64 arg)
>> +{
>> +	__tlbi(vale1, arg);
>> +	__tlbi_user(vale1, arg);
> 
> Should the __tlbi_user addition be added as part of patch 3 ?

Yes, it should; good spot!

> 
>> +}
>> +
>> +static __always_inline void vale1is(u64 arg)
>> +{
>> +	__tlbi(vale1is, arg);
>> +}
>> +
>> +static __always_inline void vale2is(u64 arg)
>> +{
>> +	__tlbi(vale2is, arg);
>> +}
>> +
>> +static __always_inline void vaale1is(u64 arg)
>> +{
>> +	__tlbi(vaale1is, arg);
>> +}
>> +
>> +static __always_inline void ipas2e1(u64 arg)
>> +{
>> +	__tlbi(ipas2e1, arg);
>> +}
>> +
>> +static __always_inline void ipas2e1is(u64 arg)
>> +{
>> +	__tlbi(ipas2e1is, arg);
>> +}
>> +
>> +static __always_inline void __tlbi_level(tlbi_op op, u64 addr, u32 level)
>> +{
>> +	u64 arg = addr;
>> +
>> +	if (alternative_has_cap_unlikely(ARM64_HAS_ARMv8_4_TTL) && level <= 3) {
>> +		u64 ttl = level | (get_trans_granule() << 2);
>> +
>> +		arg &= ~TLBI_TTL_MASK;
>> +		arg |= FIELD_PREP(TLBI_TTL_MASK, ttl);
>> +	}
>> +
>> +	op(arg);
>> +}
>>  
>>  #define __tlbi_user_level(op, arg, level) do {				\
>>  	if (arm64_kernel_unmapped_at_el0())				\
>> -- 
>> 2.43.0
>>

Re: [PATCH v1 01/13] arm64: mm: Re-implement the __tlbi_level macro as a C function

Posted by Jonathan Cameron 1 month, 3 weeks ago

On Tue, 16 Dec 2025 14:45:46 +0000
Ryan Roberts <ryan.roberts@arm.com> wrote:

> As part of efforts to reduce our reliance on complex preprocessor macros
> for TLB invalidation routines, convert the __tlbi_level macro to a C
> function for by-level TLB invalidation.
> 
> Each specific tlbi level op is implemented as a C function and the
> appropriate function pointer is passed to __tlbi_level(). Since
> everything is declared inline and is statically resolvable, the compiler
> will convert the indirect function call to a direct inline execution.
> 
> Suggested-by: Linus Torvalds <torvalds@linux-foundation.org>
> Signed-off-by: Ryan Roberts <ryan.roberts@arm.com>
> ---

> +static __always_inline void __tlbi_level(tlbi_op op, u64 addr, u32 level)
> +{
> +	u64 arg = addr;
> +
> +	if (alternative_has_cap_unlikely(ARM64_HAS_ARMv8_4_TTL) && level <= 3) {
> +		u64 ttl = level | (get_trans_granule() << 2);
> +
> +		arg &= ~TLBI_TTL_MASK;
> +		arg |= FIELD_PREP(TLBI_TTL_MASK, ttl);

Probably don't care, but I think you could do
		FIELD_MODIFY(TLBI_TTL_MASK, &arg, ttl);
instead of those two lines. Code generation hopefully similar?
So depends on which macros you find more readable.

> +	}
> +
> +	op(arg);
> +}
>  
>  #define __tlbi_user_level(op, arg, level) do {				\
>  	if (arm64_kernel_unmapped_at_el0())				\

Re: [PATCH v1 01/13] arm64: mm: Re-implement the __tlbi_level macro as a C function

Posted by Ryan Roberts 1 month, 1 week ago

Happy new year!


On 16/12/2025 17:53, Jonathan Cameron wrote:
> On Tue, 16 Dec 2025 14:45:46 +0000
> Ryan Roberts <ryan.roberts@arm.com> wrote:
> 
>> As part of efforts to reduce our reliance on complex preprocessor macros
>> for TLB invalidation routines, convert the __tlbi_level macro to a C
>> function for by-level TLB invalidation.
>>
>> Each specific tlbi level op is implemented as a C function and the
>> appropriate function pointer is passed to __tlbi_level(). Since
>> everything is declared inline and is statically resolvable, the compiler
>> will convert the indirect function call to a direct inline execution.
>>
>> Suggested-by: Linus Torvalds <torvalds@linux-foundation.org>
>> Signed-off-by: Ryan Roberts <ryan.roberts@arm.com>
>> ---
> 
>> +static __always_inline void __tlbi_level(tlbi_op op, u64 addr, u32 level)
>> +{
>> +	u64 arg = addr;
>> +
>> +	if (alternative_has_cap_unlikely(ARM64_HAS_ARMv8_4_TTL) && level <= 3) {
>> +		u64 ttl = level | (get_trans_granule() << 2);
>> +
>> +		arg &= ~TLBI_TTL_MASK;
>> +		arg |= FIELD_PREP(TLBI_TTL_MASK, ttl);
> 
> Probably don't care, but I think you could do
> 		FIELD_MODIFY(TLBI_TTL_MASK, &arg, ttl);
> instead of those two lines. Code generation hopefully similar?
> So depends on which macros you find more readable.

Yeah that's probably slightly neater - I'll switch to this for the next version.

Thanks,
Ryan

> 
>> +	}
>> +
>> +	op(arg);
>> +}
>>  
>>  #define __tlbi_user_level(op, arg, level) do {				\
>>  	if (arm64_kernel_unmapped_at_el0())				\
>