From: Ryan Roberts <ryan.roberts@arm.com>
To: Anshuman Khandual <anshuman.khandual@arm.com>,
Catalin Marinas <catalin.marinas@arm.com>,
Will Deacon <will@kernel.org>,
Mark Rutland <mark.rutland@arm.com>,
James Morse <james.morse@arm.com>
Cc: linux-arm-kernel@lists.infradead.org, linux-kernel@vger.kernel.org
Subject: Re: [RFC PATCH v1 1/2] arm64: tlbflush: Move invocation of __flush_tlb_range_op() to a macro
Date: Thu, 11 Sep 2025 15:12:48 +0100 [thread overview]
Message-ID: <3e3f15a1-77d4-4391-91bb-7a5eb0e93a63@arm.com> (raw)
In-Reply-To: <4ac449f1-d5cc-42d6-bded-2db6984d55f0@arm.com>
On 11/09/2025 06:50, Anshuman Khandual wrote:
>
>
> On 29/08/25 9:05 PM, Ryan Roberts wrote:
>> __flush_tlb_range_op() is a pre-processor macro that takes the TLBI
>> operation as a string, and builds the instruction from it. This prevents
>> passing the TLBI operation around as a variable. __flush_tlb_range_op()
>> also takes 7 other arguments.
>>
>> Adding extra invocations for different TLB operations means duplicating
>> the whole thing, but those 7 extra arguments are the same each time.
>>
>> Add an enum for the TLBI operations that __flush_tlb_range() uses, and a
>> macro to pass the operation name as a string to __flush_tlb_range_op(),
>> and the rest of the arguments using __VA_ARGS_.
>>
>> The result is easier to add new TLBI operations to, and to modify any of
>> the other arguments as they only appear once.
>>
>> Suggested-by: James Morse <james.morse@arm.com>
>> Signed-off-by: Ryan Roberts <ryan.roberts@arm.com>
>> ---
>> arch/arm64/include/asm/tlbflush.h | 30 ++++++++++++++++++++++++------
>> 1 file changed, 24 insertions(+), 6 deletions(-)
>>
>> diff --git a/arch/arm64/include/asm/tlbflush.h b/arch/arm64/include/asm/tlbflush.h
>> index 18a5dc0c9a54..f66b8c4696d0 100644
>> --- a/arch/arm64/include/asm/tlbflush.h
>> +++ b/arch/arm64/include/asm/tlbflush.h
>> @@ -11,6 +11,7 @@
>> #ifndef __ASSEMBLY__
>>
>> #include <linux/bitfield.h>
>> +#include <linux/build_bug.h>
>> #include <linux/mm_types.h>
>> #include <linux/sched.h>
>> #include <linux/mmu_notifier.h>
>> @@ -433,12 +434,32 @@ static inline bool __flush_tlb_range_limit_excess(unsigned long start,
>> return false;
>> }
>>
>> +enum tlbi_op {
>> + TLBI_VALE1IS,
>> + TLBI_VAE1IS,
>> +};
>> +
>> +#define flush_tlb_range_op(op, ...) \
>> +do { \
>> + switch (op) { \
>> + case TLBI_VALE1IS: \
>> + __flush_tlb_range_op(vale1is, __VA_ARGS__); \
>> + break; \
>> + case TLBI_VAE1IS: \
>> + __flush_tlb_range_op(vae1is, __VA_ARGS__); \
>> + break; \
>> + default: \
>> + BUILD_BUG_ON_MSG(1, "Unknown TLBI op"); \
>> + } \
>> +} while (0)
>> +
>> static inline void __flush_tlb_range_nosync(struct mm_struct *mm,
>> unsigned long start, unsigned long end,
>> unsigned long stride, bool last_level,
>> int tlb_level)
>> {
>> unsigned long asid, pages;
>> + enum tlbi_op tlbi_op;
>>
>> start = round_down(start, stride);
>> end = round_up(end, stride);
>> @@ -452,12 +473,9 @@ static inline void __flush_tlb_range_nosync(struct mm_struct *mm,
>> dsb(ishst);
>> asid = ASID(mm);
>>
>> - if (last_level)
>> - __flush_tlb_range_op(vale1is, start, pages, stride, asid,
>> - tlb_level, true, lpa2_is_enabled());
>> - else
>> - __flush_tlb_range_op(vae1is, start, pages, stride, asid,
>> - tlb_level, true, lpa2_is_enabled());
>> + tlbi_op = last_level ? TLBI_VALE1IS : TLBI_VAE1IS;
>> + flush_tlb_range_op(tlbi_op, start, pages, stride, asid, tlb_level,
>> + true, lpa2_is_enabled());
>>
>> mmu_notifier_arch_invalidate_secondary_tlbs(mm, start, end);
>> }
>
> Should the remaining __flush_tlb_range_op() in flush_tlb_kernel_range()
> converted into flush_tlb_range_op() adding another similar enum variable
> i.e TLBI_VAALE1IS ? Because this will ensure that there is one variant
> helper i.e flush_tlb_range_op() that gets called.
Yeah maybe. I don't really have a strong opinion.
Will posted an RFC for converting all this to functions a while ago. That's the
better baseline to build this all on I think.
>
> diff --git a/arch/arm64/include/asm/tlbflush.h b/arch/arm64/include/asm/tlbflush.h
> index f66b8c4696d0..a23169751deb 100644
> --- a/arch/arm64/include/asm/tlbflush.h
> +++ b/arch/arm64/include/asm/tlbflush.h
> @@ -437,6 +437,7 @@ static inline bool __flush_tlb_range_limit_excess(unsigned long start,
> enum tlbi_op {
> TLBI_VALE1IS,
> TLBI_VAE1IS,
> + TLBI_VAALE1IS,
> };
>
> #define flush_tlb_range_op(op, ...) \
> @@ -448,6 +449,9 @@ do { \
> case TLBI_VAE1IS: \
> __flush_tlb_range_op(vae1is, __VA_ARGS__); \
> break; \
> + case TLBI_VAALE1IS: \
> + __flush_tlb_range_op(vaale1is, __VA_ARGS__); \
> + break; \
> default: \
> BUILD_BUG_ON_MSG(1, "Unknown TLBI op"); \
> } \
> @@ -517,7 +521,7 @@ static inline void flush_tlb_kernel_range(unsigned long start, unsigned long end
> }
>
> dsb(ishst);
> - __flush_tlb_range_op(vaale1is, start, pages, stride, 0,
> + flush_tlb_range_op(TLBI_VAALE1IS, start, pages, stride, 0,
> TLBI_TTL_UNKNOWN, false, lpa2_is_enabled());
> dsb(ish);
> isb();
next prev parent reply other threads:[~2025-09-11 14:12 UTC|newest]
Thread overview: 22+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-08-29 15:35 [RFC PATCH v1 0/2] Don't broadcast TLBI if mm was only active on local CPU Ryan Roberts
2025-08-29 15:35 ` [RFC PATCH v1 1/2] arm64: tlbflush: Move invocation of __flush_tlb_range_op() to a macro Ryan Roberts
2025-09-02 16:25 ` Catalin Marinas
2025-09-11 5:50 ` Anshuman Khandual
2025-09-11 14:12 ` Ryan Roberts [this message]
2025-08-29 15:35 ` [RFC PATCH v1 2/2] arm64: tlbflush: Don't broadcast if mm was only active on local cpu Ryan Roberts
2025-09-01 9:08 ` Alexandru Elisei
2025-09-01 9:18 ` Ryan Roberts
2025-09-02 16:23 ` Catalin Marinas
2025-09-02 16:54 ` Ryan Roberts
2025-09-10 23:58 ` Yang Shi
2025-09-11 1:20 ` Huang, Ying
2025-09-11 14:19 ` Ryan Roberts
2025-09-11 22:29 ` Yang Shi
2025-09-18 15:18 ` Catalin Marinas
2025-09-02 16:47 ` [RFC PATCH v1 0/2] Don't broadcast TLBI if mm was only active on local CPU Catalin Marinas
2025-09-02 16:56 ` Ryan Roberts
2025-09-15 16:05 ` Christoph Lameter (Ampere)
2025-09-03 2:12 ` Huang, Ying
2025-09-15 16:02 ` Christoph Lameter (Ampere)
2025-09-10 10:57 ` Huang, Ying
2025-09-10 12:42 ` Ryan Roberts
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=3e3f15a1-77d4-4391-91bb-7a5eb0e93a63@arm.com \
--to=ryan.roberts@arm.com \
--cc=anshuman.khandual@arm.com \
--cc=catalin.marinas@arm.com \
--cc=james.morse@arm.com \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=linux-kernel@vger.kernel.org \
--cc=mark.rutland@arm.com \
--cc=will@kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®