Merge patch series "riscv: ASID-related and UP-related TLB flush enhancements"

Samuel Holland <samuel.holland@sifive.com> says:

This series converts uniprocessor kernel builds to use the same TLB
flushing code as SMP builds, to take advantage of batching and existing
range- and ASID-based TLB flush optimizations. It optimizes out IPIs and
SBI calls based on the online CPU count, which also covers the scenario
where SMP was enabled at build time but only one CPU is present/online.
A final optimization is to use single-ASID flushes wherever possible, to
avoid unnecessary TLB misses for kernel mappings.

This series has a semantic conflict with the AIA patches that are in
linux-next due to the removal of the third parameter of
riscv_ipi_set_virq_range(), which is called from imsic_ipi_domain_init()
in drivers/irqchip/irq-riscv-imsic-early.c. The resolution is to remove
the extra argument from the call site.

Here are some numbers from D1 which show the performance impact:

v6.9-rc1:
 System Benchmarks Partial Index              BASELINE       RESULT    INDEX
 Execl Throughput                                 43.0        198.5     46.2
 File Copy 1024 bufsize 2000 maxblocks          3960.0      73934.4    186.7
 File Copy 256 bufsize 500 maxblocks            1655.0      20242.6    122.3
 File Copy 4096 bufsize 8000 maxblocks          5800.0     197706.4    340.9
 Pipe Throughput                               12440.0     176974.2    142.3
 Pipe-based Context Switching                   4000.0      23626.8     59.1
 Process Creation                                126.0        449.9     35.7
 Shell Scripts (1 concurrent)                     42.4        544.4    128.4
 Shell Scripts (16 concurrent)                     ---         35.3      ---
 Shell Scripts (8 concurrent)                      6.0         71.6    119.3
 System Call Overhead                          15000.0     248072.6    165.4
                                                                    ========
 System Benchmarks Index Score (Partial Only)                          110.6

v6.9-rc1 + this patch series:
 System Benchmarks Partial Index              BASELINE       RESULT    INDEX
 Execl Throughput                                 43.0        196.8     45.8
 File Copy 1024 bufsize 2000 maxblocks          3960.0      71782.2    181.3
 File Copy 256 bufsize 500 maxblocks            1655.0      21269.4    128.5
 File Copy 4096 bufsize 8000 maxblocks          5800.0     199424.0    343.8
 Pipe Throughput                               12440.0     196468.6    157.9
 Pipe-based Context Switching                   4000.0      24261.8     60.7
 Process Creation                                126.0        459.0     36.4
 Shell Scripts (1 concurrent)                     42.4        543.8    128.2
 Shell Scripts (16 concurrent)                     ---         35.5      ---
 Shell Scripts (8 concurrent)                      6.0         71.7    119.6
 System Call Overhead                          15000.0     259415.2    172.9
                                                                    ========
 System Benchmarks Index Score (Partial Only)                          113.0

* b4-shazam-lts:
  riscv: mm: Always use an ASID to flush mm contexts
  riscv: mm: Preserve global TLB entries when switching contexts
  riscv: mm: Make asid_bits a local variable
  riscv: mm: Use a fixed layout for the MM context ID
  riscv: mm: Introduce cntx2asid/cntx2version helper macros
  riscv: Avoid TLB flush loops when affected by SiFive CIP-1200
  riscv: Apply SiFive CIP-1200 workaround to single-ASID sfence.vma
  riscv: mm: Combine the SMP and UP TLB flush code
  riscv: Only send remote fences when some other CPU is online
  riscv: mm: Broadcast kernel TLB flushes only when needed
  riscv: Use IPIs for remote cache/TLB flushes by default
  riscv: Factor out page table TLB synchronization
  riscv: Flush the instruction cache during SMP bringup

Link: https://lore.kernel.org/r/20240327045035.368512-1-samuel.holland@sifive.com
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

4 files changed, 36 insertions, 74 deletions

diff --git a/arch/riscv/mm/Makefile b/arch/riscv/mm/Makefile
index 2c869f8026a8..cbe4d775ef56 100644
--- a/arch/riscv/mm/Makefile
+++ b/arch/riscv/mm/Makefile
@@ -13,14 +13,11 @@ endif
 KCOV_INSTRUMENT_init.o := n
 
 obj-y += init.o
-obj-$(CONFIG_MMU) += extable.o fault.o pageattr.o pgtable.o
+obj-$(CONFIG_MMU) += extable.o fault.o pageattr.o pgtable.o tlbflush.o
 obj-y += cacheflush.o
 obj-y += context.o
 obj-y += pmem.o
 
-ifeq ($(CONFIG_MMU),y)
-obj-$(CONFIG_SMP) += tlbflush.o
-endif
 obj-$(CONFIG_HUGETLB_PAGE) += hugetlbpage.o
 obj-$(CONFIG_PTDUMP_CORE) += ptdump.o
 obj-$(CONFIG_KASAN)   += kasan_init.o
diff --git a/arch/riscv/mm/cacheflush.c b/arch/riscv/mm/cacheflush.c
index 3ef666c7dfc7..a03c994eed3b 100644
--- a/arch/riscv/mm/cacheflush.c
+++ b/arch/riscv/mm/cacheflush.c
@@ -22,7 +22,9 @@ void flush_icache_all(void)
 {
 	local_flush_icache_all();
 
-	if (IS_ENABLED(CONFIG_RISCV_SBI) && !riscv_use_ipi_for_rfence())
+	if (num_online_cpus() < 2)
+		return;
+	else if (riscv_use_sbi_for_rfence())
 		sbi_remote_fence_i(NULL);
 	else
 		on_each_cpu(ipi_remote_fence_i, NULL, 1);
@@ -70,8 +72,7 @@ void flush_icache_mm(struct mm_struct *mm, bool local)
 		 * with flush_icache_deferred().
 		 */
 		smp_mb();
-	} else if (IS_ENABLED(CONFIG_RISCV_SBI) &&
-		   !riscv_use_ipi_for_rfence()) {
+	} else if (riscv_use_sbi_for_rfence()) {
 		sbi_remote_fence_i(&others);
 	} else {
 		on_each_cpu_mask(&others, ipi_remote_fence_i, NULL, 1);
diff --git a/arch/riscv/mm/context.c b/arch/riscv/mm/context.c
index 9b6db024103b..4abe3de23225 100644
--- a/arch/riscv/mm/context.c
+++ b/arch/riscv/mm/context.c
@@ -21,9 +21,7 @@
 
 DEFINE_STATIC_KEY_FALSE(use_asid_allocator);
 
-static unsigned long asid_bits;
 static unsigned long num_asids;
-unsigned long asid_mask;
 
 static atomic_long_t current_version;
 
@@ -82,7 +80,7 @@ static void __flush_context(void)
 		if (cntx == 0)
 			cntx = per_cpu(reserved_context, i);
 
-		__set_bit(cntx & asid_mask, context_asid_map);
+		__set_bit(cntx2asid(cntx), context_asid_map);
 		per_cpu(reserved_context, i) = cntx;
 	}
 
@@ -103,7 +101,7 @@ static unsigned long __new_context(struct mm_struct *mm)
 	lockdep_assert_held(&context_lock);
 
 	if (cntx != 0) {
-		unsigned long newcntx = ver | (cntx & asid_mask);
+		unsigned long newcntx = ver | cntx2asid(cntx);
 
 		/*
 		 * If our current CONTEXT was active during a rollover, we
@@ -116,7 +114,7 @@ static unsigned long __new_context(struct mm_struct *mm)
 		 * We had a valid CONTEXT in a previous life, so try to
 		 * re-use it if possible.
 		 */
-		if (!__test_and_set_bit(cntx & asid_mask, context_asid_map))
+		if (!__test_and_set_bit(cntx2asid(cntx), context_asid_map))
 			return newcntx;
 	}
 
@@ -129,7 +127,7 @@ static unsigned long __new_context(struct mm_struct *mm)
 		goto set_asid;
 
 	/* We're out of ASIDs, so increment current_version */
-	ver = atomic_long_add_return_relaxed(num_asids, &current_version);
+	ver = atomic_long_add_return_relaxed(BIT(SATP_ASID_BITS), &current_version);
 
 	/* Flush everything  */
 	__flush_context();
@@ -169,7 +167,7 @@ static void set_mm_asid(struct mm_struct *mm, unsigned int cpu)
 	 */
 	old_active_cntx = atomic_long_read(&per_cpu(active_context, cpu));
 	if (old_active_cntx &&
-	    ((cntx & ~asid_mask) == atomic_long_read(&current_version)) &&
+	    (cntx2version(cntx) == atomic_long_read(&current_version)) &&
 	    atomic_long_cmpxchg_relaxed(&per_cpu(active_context, cpu),
 					old_active_cntx, cntx))
 		goto switch_mm_fast;
@@ -178,7 +176,7 @@ static void set_mm_asid(struct mm_struct *mm, unsigned int cpu)
 
 	/* Check that our ASID belongs to the current_version. */
 	cntx = atomic_long_read(&mm->context.id);
-	if ((cntx & ~asid_mask) != atomic_long_read(&current_version)) {
+	if (cntx2version(cntx) != atomic_long_read(&current_version)) {
 		cntx = __new_context(mm);
 		atomic_long_set(&mm->context.id, cntx);
 	}
@@ -192,7 +190,7 @@ static void set_mm_asid(struct mm_struct *mm, unsigned int cpu)
 
 switch_mm_fast:
 	csr_write(CSR_SATP, virt_to_pfn(mm->pgd) |
-		  ((cntx & asid_mask) << SATP_ASID_SHIFT) |
+		  (cntx2asid(cntx) << SATP_ASID_SHIFT) |
 		  satp_mode);
 
 	if (need_flush_tlb)
@@ -203,7 +201,7 @@ static void set_mm_noasid(struct mm_struct *mm)
 {
 	/* Switch the page table and blindly nuke entire local TLB */
 	csr_write(CSR_SATP, virt_to_pfn(mm->pgd) | satp_mode);
-	local_flush_tlb_all();
+	local_flush_tlb_all_asid(0);
 }
 
 static inline void set_mm(struct mm_struct *prev,
@@ -228,7 +226,7 @@ static inline void set_mm(struct mm_struct *prev,
 
 static int __init asids_init(void)
 {
-	unsigned long old;
+	unsigned long asid_bits, old;
 
 	/* Figure-out number of ASID bits in HW */
 	old = csr_read(CSR_SATP);
@@ -248,7 +246,6 @@ static int __init asids_init(void)
 	/* Pre-compute ASID details */
 	if (asid_bits) {
 		num_asids = 1 << asid_bits;
-		asid_mask = num_asids - 1;
 	}
 
 	/*
@@ -256,7 +253,7 @@ static int __init asids_init(void)
 	 * at-least twice more than CPUs
 	 */
 	if (num_asids > (2 * num_possible_cpus())) {
-		atomic_long_set(&current_version, num_asids);
+		atomic_long_set(&current_version, BIT(SATP_ASID_BITS));
 
 		context_asid_map = bitmap_zalloc(num_asids, GFP_KERNEL);
 		if (!context_asid_map)
diff --git a/arch/riscv/mm/tlbflush.c b/arch/riscv/mm/tlbflush.c
index 893566e004b7..44e7ed4e194f 100644
--- a/arch/riscv/mm/tlbflush.c
+++ b/arch/riscv/mm/tlbflush.c
@@ -7,34 +7,11 @@
 #include <asm/sbi.h>
 #include <asm/mmu_context.h>
 
-static inline void local_flush_tlb_all_asid(unsigned long asid)
-{
-	if (asid != FLUSH_TLB_NO_ASID)
-		__asm__ __volatile__ ("sfence.vma x0, %0"
-				:
-				: "r" (asid)
-				: "memory");
-	else
-		local_flush_tlb_all();
-}
-
-static inline void local_flush_tlb_page_asid(unsigned long addr,
-		unsigned long asid)
-{
-	if (asid != FLUSH_TLB_NO_ASID)
-		__asm__ __volatile__ ("sfence.vma %0, %1"
-				:
-				: "r" (addr), "r" (asid)
-				: "memory");
-	else
-		local_flush_tlb_page(addr);
-}
-
 /*
  * Flush entire TLB if number of entries to be flushed is greater
  * than the threshold below.
  */
-static unsigned long tlb_flush_all_threshold __read_mostly = 64;
+unsigned long tlb_flush_all_threshold __read_mostly = 64;
 
 static void local_flush_tlb_range_threshold_asid(unsigned long start,
 						 unsigned long size,
@@ -79,10 +56,12 @@ static void __ipi_flush_tlb_all(void *info)
 
 void flush_tlb_all(void)
 {
-	if (riscv_use_ipi_for_rfence())
-		on_each_cpu(__ipi_flush_tlb_all, NULL, 1);
-	else
+	if (num_online_cpus() < 2)
+		local_flush_tlb_all();
+	else if (riscv_use_sbi_for_rfence())
 		sbi_remote_sfence_vma_asid(NULL, 0, FLUSH_TLB_MAX_SIZE, FLUSH_TLB_NO_ASID);
+	else
+		on_each_cpu(__ipi_flush_tlb_all, NULL, 1);
 }
 
 struct flush_tlb_range_data {
@@ -103,46 +82,34 @@ static void __flush_tlb_range(struct cpumask *cmask, unsigned long asid,
 			      unsigned long start, unsigned long size,
 			      unsigned long stride)
 {
-	struct flush_tlb_range_data ftd;
-	bool broadcast;
+	unsigned int cpu;
 
 	if (cpumask_empty(cmask))
 		return;
 
-	if (cmask != cpu_online_mask) {
-		unsigned int cpuid;
+	cpu = get_cpu();
 
-		cpuid = get_cpu();
-		/* check if the tlbflush needs to be sent to other CPUs */
-		broadcast = cpumask_any_but(cmask, cpuid) < nr_cpu_ids;
+	/* 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);
+	} else if (riscv_use_sbi_for_rfence()) {
+		sbi_remote_sfence_vma_asid(cmask, start, size, asid);
 	} else {
-		broadcast = true;
-	}
+		struct flush_tlb_range_data ftd;
 
-	if (broadcast) {
-		if (riscv_use_ipi_for_rfence()) {
-			ftd.asid = asid;
-			ftd.start = start;
-			ftd.size = size;
-			ftd.stride = stride;
-			on_each_cpu_mask(cmask,
-					 __ipi_flush_tlb_range_asid,
-					 &ftd, 1);
-		} else
-			sbi_remote_sfence_vma_asid(cmask,
-						   start, size, asid);
-	} else {
-		local_flush_tlb_range_asid(start, size, stride, asid);
+		ftd.asid = asid;
+		ftd.start = start;
+		ftd.size = size;
+		ftd.stride = stride;
+		on_each_cpu_mask(cmask, __ipi_flush_tlb_range_asid, &ftd, 1);
 	}
 
-	if (cmask != cpu_online_mask)
-		put_cpu();
+	put_cpu();
 }
 
 static inline unsigned long get_mm_asid(struct mm_struct *mm)
 {
-	return static_branch_unlikely(&use_asid_allocator) ?
-			atomic_long_read(&mm->context.id) & asid_mask : FLUSH_TLB_NO_ASID;
+	return cntx2asid(atomic_long_read(&mm->context.id));
 }
 
 void flush_tlb_mm(struct mm_struct *mm)