linux-kernel - [PATCH 2/2] riscv: mm: Clear cpu in mm_cpumask after local_flush_tlb_all

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-Id: <20250827131444.23893-3-luxu.kernel@bytedance.com>
Date: Wed, 27 Aug 2025 21:14:44 +0800
From: Xu Lu <luxu.kernel@...edance.com>
To: paul.walmsley@...ive.com,
	palmer@...belt.com,
	aou@...s.berkeley.edu,
	alex@...ti.fr
Cc: linux-riscv@...ts.infradead.org,
	linux-kernel@...r.kernel.org,
	Xu Lu <luxu.kernel@...edance.com>
Subject: [PATCH 2/2] riscv: mm: Clear cpu in mm_cpumask after local_flush_tlb_all_asid

Clear corresponding bit of current cpu in mm_cpumask after executing
local_flush_tlb_all_asid().

This reduces the number of IPI due to tlb flush:

* ltp - mmapstress01
Before: ~98k
After: 268

Signed-off-by: Xu Lu <luxu.kernel@...edance.com>
---
 arch/riscv/mm/tlbflush.c | 41 ++++++++++++++++++++++++----------------
 1 file changed, 25 insertions(+), 16 deletions(-)

diff --git a/arch/riscv/mm/tlbflush.c b/arch/riscv/mm/tlbflush.c
index 962db300a1665..571358f385879 100644
--- a/arch/riscv/mm/tlbflush.c
+++ b/arch/riscv/mm/tlbflush.c
@@ -17,7 +17,8 @@
  */
 unsigned long tlb_flush_all_threshold __read_mostly = 64;
 
-static void local_flush_tlb_range_threshold_asid(unsigned long start,
+static void local_flush_tlb_range_threshold_asid(struct mm_struct *mm,
+						 unsigned long start,
 						 unsigned long size,
 						 unsigned long stride,
 						 unsigned long asid)
@@ -27,6 +28,8 @@ static void local_flush_tlb_range_threshold_asid(unsigned long start,
 
 	if (nr_ptes_in_range > tlb_flush_all_threshold) {
 		local_flush_tlb_all_asid(asid);
+		if (mm && mm != current->active_mm)
+			cpumask_clear_cpu(raw_smp_processor_id(), mm_cpumask(mm));
 		return;
 	}
 
@@ -46,21 +49,28 @@ static void local_flush_tlb_range_threshold_asid(unsigned long start,
 	}
 }
 
-static inline void local_flush_tlb_range_asid(unsigned long start,
-		unsigned long size, unsigned long stride, unsigned long asid)
+static inline void local_flush_tlb_range_mm(struct mm_struct *mm,
+					    unsigned long start,
+					    unsigned long size,
+					    unsigned long stride)
 {
-	if (size <= stride)
+	unsigned long asid = get_mm_asid(mm);
+
+	if (size <= stride) {
 		local_flush_tlb_page_asid(start, asid);
-	else if (size == FLUSH_TLB_MAX_SIZE)
+	} else if (size == FLUSH_TLB_MAX_SIZE) {
 		local_flush_tlb_all_asid(asid);
-	else
-		local_flush_tlb_range_threshold_asid(start, size, stride, asid);
+		if (mm && mm != current->active_mm)
+			cpumask_clear_cpu(raw_smp_processor_id(), mm_cpumask(mm));
+	} else {
+		local_flush_tlb_range_threshold_asid(mm, start, size, stride, asid);
+	}
 }
 
 /* Flush a range of kernel pages without broadcasting */
 void local_flush_tlb_kernel_range(unsigned long start, unsigned long end)
 {
-	local_flush_tlb_range_asid(start, end - start, PAGE_SIZE, FLUSH_TLB_NO_ASID);
+	local_flush_tlb_range_mm(NULL, start, end - start, PAGE_SIZE);
 }
 
 static void __ipi_flush_tlb_all(void *info)
@@ -79,17 +89,17 @@ void flush_tlb_all(void)
 }
 
 struct flush_tlb_range_data {
-	unsigned long asid;
+	struct mm_struct *mm;
 	unsigned long start;
 	unsigned long size;
 	unsigned long stride;
 };
 
-static void __ipi_flush_tlb_range_asid(void *info)
+static void __ipi_flush_tlb_range_mm(void *info)
 {
 	struct flush_tlb_range_data *d = info;
 
-	local_flush_tlb_range_asid(d->start, d->size, d->stride, d->asid);
+	local_flush_tlb_range_mm(d->mm, d->start, d->size, d->stride);
 }
 
 static void __flush_tlb_range(struct mm_struct *mm,
@@ -97,7 +107,6 @@ static void __flush_tlb_range(struct mm_struct *mm,
 			      unsigned long start, unsigned long size,
 			      unsigned long stride)
 {
-	unsigned long asid = get_mm_asid(mm);
 	unsigned int cpu;
 
 	if (cpumask_empty(cmask))
@@ -107,17 +116,17 @@ static void __flush_tlb_range(struct mm_struct *mm,
 
 	/* Check if the TLB flush needs to be sent to other CPUs. */
 	if (cpumask_any_but(cmask, cpu) >= nr_cpu_ids) {
-		local_flush_tlb_range_asid(start, size, stride, asid);
+		local_flush_tlb_range_mm(mm, start, size, stride);
 	} else if (riscv_use_sbi_for_rfence()) {
-		sbi_remote_sfence_vma_asid(cmask, start, size, asid);
+		sbi_remote_sfence_vma_asid(cmask, start, size, get_mm_asid(mm));
 	} else {
 		struct flush_tlb_range_data ftd;
 
-		ftd.asid = asid;
+		ftd.mm = mm;
 		ftd.start = start;
 		ftd.size = size;
 		ftd.stride = stride;
-		on_each_cpu_mask(cmask, __ipi_flush_tlb_range_asid, &ftd, 1);
+		on_each_cpu_mask(cmask, __ipi_flush_tlb_range_mm, &ftd, 1);
 	}
 
 	put_cpu();
-- 
2.20.1