runtime: convert page allocator bitmap to sparse array

Currently the page allocator bitmap is implemented as a single giant
memory mapping which is reserved at init time and committed as needed.
This causes problems on systems that don't handle large uncommitted
mappings well, or institute low virtual address space defaults as a
memory limiting mechanism.

This change modifies the implementation of the page allocator bitmap
away from a directly-mapped set of bytes to a sparse array in same vein
as mheap.arenas. This will hurt performance a little but the biggest
gains are from the lockless allocation possible with the page allocator,
so the impact of this extra layer of indirection should be minimal.

In fact, this is exactly what we see:
    https://perf.golang.org/search?q=upload:20191125.5

This reduces the amount of mapped (PROT_NONE) memory needed on systems
with 48-bit address spaces to ~600 MiB down from almost 9 GiB. The bulk
of this remaining memory is used by the summaries.

Go processes with 32-bit address spaces now always commit to 128 KiB of
memory for the bitmap. Previously it would only commit the pages in the
bitmap which represented the range of addresses (lowest address to
highest address, even if there are unused regions in that range) used by
the heap.

Updates #35568.
Updates #35451.

Change-Id: I0ff10380156568642b80c366001eefd0a4e6c762
Reviewed-on: https://go-review.googlesource.com/c/go/+/207497
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
Reviewed-by: Cherry Zhang <cherryyz@google.com>

1 files changed, 94 insertions, 54 deletions

diff --git a/src/runtime/mpagealloc.go b/src/runtime/mpagealloc.go
index 21ea6a8525..f48b9faec3 100644
--- a/src/runtime/mpagealloc.go
+++ b/src/runtime/mpagealloc.go
@@ -9,9 +9,8 @@
 //
 // Pages are managed using a bitmap that is sharded into chunks.
 // In the bitmap, 1 means in-use, and 0 means free. The bitmap spans the
-// process's address space. Chunks are allocated using a SLAB allocator
-// and pointers to chunks are managed in one large array, which is mapped
-// in as needed.
+// process's address space. Chunks are managed in a sparse-array-style structure
+// similar to mheap.arenas, since the bitmap may be large on some systems.
 //
 // The bitmap is efficiently searched by using a radix tree in combination
 // with fast bit-wise intrinsics. Allocation is performed using an address-ordered
@@ -49,6 +48,7 @@
 package runtime
 
 import (
+	"runtime/internal/atomic"
 	"unsafe"
 )
 
@@ -74,6 +74,14 @@ const (
 	summaryLevelBits = 3
 	summaryL0Bits    = heapAddrBits - logPallocChunkBytes - (summaryLevels-1)*summaryLevelBits
 
+	// pallocChunksL2Bits is the number of bits of the chunk index number
+	// covered by the second level of the chunks map.
+	//
+	// See (*pageAlloc).chunks for more details. Update the documentation
+	// there should this change.
+	pallocChunksL2Bits  = heapAddrBits - logPallocChunkBytes - pallocChunksL1Bits
+	pallocChunksL1Shift = pallocChunksL2Bits
+
 	// Maximum searchAddr value, which indicates that the heap has no free space.
 	//
 	// We subtract arenaBaseOffset because we want this to represent the maximum
@@ -111,6 +119,26 @@ func chunkPageIndex(p uintptr) uint {
 	return uint(p % pallocChunkBytes / pageSize)
 }
 
+// l1 returns the index into the first level of (*pageAlloc).chunks.
+func (i chunkIdx) l1() uint {
+	if pallocChunksL1Bits == 0 {
+		// Let the compiler optimize this away if there's no
+		// L1 map.
+		return 0
+	} else {
+		return uint(i) >> pallocChunksL1Shift
+	}
+}
+
+// l2 returns the index into the second level of (*pageAlloc).chunks.
+func (i chunkIdx) l2() uint {
+	if pallocChunksL1Bits == 0 {
+		return uint(i)
+	} else {
+		return uint(i) & (1<<pallocChunksL2Bits - 1)
+	}
+}
+
 // addrsToSummaryRange converts base and limit pointers into a range
 // of entries for the given summary level.
 //
@@ -160,11 +188,29 @@ type pageAlloc struct {
 
 	// chunks is a slice of bitmap chunks.
 	//
-	// The backing store for chunks is reserved in init and committed
-	// by grow.
+	// The total size of chunks is quite large on most 64-bit platforms
+	// (O(GiB) or more) if flattened, so rather than making one large mapping
+	// (which has problems on some platforms, even when PROT_NONE) we use a
+	// two-level sparse array approach similar to the arena index in mheap.
 	//
 	// To find the chunk containing a memory address `a`, do:
-	//   chunks[chunkIndex(a)]
+	//   chunkOf(chunkIndex(a))
+	//
+	// Below is a table describing the configuration for chunks for various
+	// heapAddrBits supported by the runtime.
+	//
+	// heapAddrBits | L1 Bits | L2 Bits | L2 Entry Size
+	// ------------------------------------------------
+	// 32           | 0       | 10      | 128 KiB
+	// 33 (iOS)     | 0       | 11      | 256 KiB
+	// 48           | 13      | 13      | 1 MiB
+	//
+	// There's no reason to use the L1 part of chunks on 32-bit, the
+	// address space is small so the L2 is small. For platforms with a
+	// 48-bit address space, we pick the L1 such that the L2 is 1 MiB
+	// in size, which is a good balance between low granularity without
+	// making the impact on BSS too high (note the L1 is stored directly
+	// in pageAlloc).
 	//
 	// summary[len(s.summary)-1][i] should always be checked, at least
 	// for a zero max value, before accessing chunks[i]. It's possible the
@@ -176,7 +222,7 @@ type pageAlloc struct {
 	// TODO(mknyszek): Consider changing the definition of the bitmap
 	// such that 1 means free and 0 means in-use so that summaries and
 	// the bitmaps align better on zero-values.
-	chunks []pallocData
+	chunks [1 << pallocChunksL1Bits]*[1 << pallocChunksL2Bits]pallocData
 
 	// The address to start an allocation search with.
 	//
@@ -231,16 +277,6 @@ func (s *pageAlloc) init(mheapLock *mutex, sysStat *uint64) {
 	// Start with the scavAddr in a state indicating there's nothing more to do.
 	s.scavAddr = minScavAddr
 
-	// Reserve space for the bitmap and put this reservation
-	// into the chunks slice.
-	const maxChunks = (1 << heapAddrBits) / pallocChunkBytes
-	r := sysReserve(nil, maxChunks*unsafe.Sizeof(s.chunks[0]))
-	if r == nil {
-		throw("failed to reserve page bitmap memory")
-	}
-	sl := notInHeapSlice{(*notInHeap)(r), 0, maxChunks}
-	s.chunks = *(*[]pallocData)(unsafe.Pointer(&sl))
-
 	// Set the mheapLock.
 	s.mheapLock = mheapLock
 }
@@ -315,6 +351,11 @@ func (s *pageAlloc) compareSearchAddrTo(addr uintptr) int {
 	return 0
 }
 
+// chunkOf returns the chunk at the given chunk index.
+func (s *pageAlloc) chunkOf(ci chunkIdx) *pallocData {
+	return &s.chunks[ci.l1()][ci.l2()]
+}
+
 // grow sets up the metadata for the address range [base, base+size).
 // It may allocate metadata, in which case *s.sysStat will be updated.
 //
@@ -332,7 +373,6 @@ func (s *pageAlloc) grow(base, size uintptr) {
 	// Update s.start and s.end.
 	// If no growth happened yet, start == 0. This is generally
 	// safe since the zero page is unmapped.
-	oldStart, oldEnd := s.start, s.end
 	firstGrowth := s.start == 0
 	start, end := chunkIndex(base), chunkIndex(limit)
 	if firstGrowth || start < s.start {
@@ -340,23 +380,8 @@ func (s *pageAlloc) grow(base, size uintptr) {
 	}
 	if end > s.end {
 		s.end = end
-
-		// s.end corresponds directly to the length of s.chunks,
-		// so just update it here.
-		s.chunks = s.chunks[:end]
 	}
 
-	// Extend the mapped part of the chunk reservation.
-	elemSize := unsafe.Sizeof(s.chunks[0])
-	extendMappedRegion(
-		unsafe.Pointer(&s.chunks[0]),
-		uintptr(oldStart)*elemSize,
-		uintptr(oldEnd)*elemSize,
-		uintptr(s.start)*elemSize,
-		uintptr(s.end)*elemSize,
-		s.sysStat,
-	)
-
 	// A grow operation is a lot like a free operation, so if our
 	// chunk ends up below the (linearized) s.searchAddr, update
 	// s.searchAddr to the new address, just like in free.
@@ -364,11 +389,21 @@ func (s *pageAlloc) grow(base, size uintptr) {
 		s.searchAddr = base
 	}
 
-	// Newly-grown memory is always considered scavenged.
+	// Add entries into chunks, which is sparse, if needed. Then,
+	// initialize the bitmap.
 	//
+	// Newly-grown memory is always considered scavenged.
 	// Set all the bits in the scavenged bitmaps high.
 	for c := chunkIndex(base); c < chunkIndex(limit); c++ {
-		s.chunks[c].scavenged.setRange(0, pallocChunkPages)
+		if s.chunks[c.l1()] == nil {
+			// Create the necessary l2 entry.
+			//
+			// Store it atomically to avoid races with readers which
+			// don't acquire the heap lock.
+			r := sysAlloc(unsafe.Sizeof(*s.chunks[0]), s.sysStat)
+			atomic.StorepNoWB(unsafe.Pointer(&s.chunks[c.l1()]), r)
+		}
+		s.chunkOf(c).scavenged.setRange(0, pallocChunkPages)
 	}
 
 	// Update summaries accordingly. The grow acts like a free, so
@@ -395,7 +430,7 @@ func (s *pageAlloc) update(base, npages uintptr, contig, alloc bool) {
 		// Fast path: the allocation doesn't span more than one chunk,
 		// so update this one and if the summary didn't change, return.
 		x := s.summary[len(s.summary)-1][sc]
-		y := s.chunks[sc].summarize()
+		y := s.chunkOf(sc).summarize()
 		if x == y {
 			return
 		}
@@ -406,7 +441,7 @@ func (s *pageAlloc) update(base, npages uintptr, contig, alloc bool) {
 		summary := s.summary[len(s.summary)-1]
 
 		// Update the summary for chunk sc.
-		summary[sc] = s.chunks[sc].summarize()
+		summary[sc] = s.chunkOf(sc).summarize()
 
 		// Update the summaries for chunks in between, which are
 		// either totally allocated or freed.
@@ -423,7 +458,7 @@ func (s *pageAlloc) update(base, npages uintptr, contig, alloc bool) {
 		}
 
 		// Update the summary for chunk ec.
-		summary[ec] = s.chunks[ec].summarize()
+		summary[ec] = s.chunkOf(ec).summarize()
 	} else {
 		// Slow general path: the allocation spans more than one chunk
 		// and at least one summary is guaranteed to change.
@@ -432,7 +467,7 @@ func (s *pageAlloc) update(base, npages uintptr, contig, alloc bool) {
 		// every chunk in the range and manually recompute the summary.
 		summary := s.summary[len(s.summary)-1]
 		for c := sc; c <= ec; c++ {
-			summary[c] = s.chunks[c].summarize()
+			summary[c] = s.chunkOf(c).summarize()
 		}
 	}
 
@@ -479,18 +514,22 @@ func (s *pageAlloc) allocRange(base, npages uintptr) uintptr {
 	scav := uint(0)
 	if sc == ec {
 		// The range doesn't cross any chunk boundaries.
-		scav += s.chunks[sc].scavenged.popcntRange(si, ei+1-si)
-		s.chunks[sc].allocRange(si, ei+1-si)
+		chunk := s.chunkOf(sc)
+		scav += chunk.scavenged.popcntRange(si, ei+1-si)
+		chunk.allocRange(si, ei+1-si)
 	} else {
 		// The range crosses at least one chunk boundary.
-		scav += s.chunks[sc].scavenged.popcntRange(si, pallocChunkPages-si)
-		s.chunks[sc].allocRange(si, pallocChunkPages-si)
+		chunk := s.chunkOf(sc)
+		scav += chunk.scavenged.popcntRange(si, pallocChunkPages-si)
+		chunk.allocRange(si, pallocChunkPages-si)
 		for c := sc + 1; c < ec; c++ {
-			scav += s.chunks[c].scavenged.popcntRange(0, pallocChunkPages)
-			s.chunks[c].allocAll()
+			chunk := s.chunkOf(c)
+			scav += chunk.scavenged.popcntRange(0, pallocChunkPages)
+			chunk.allocAll()
 		}
-		scav += s.chunks[ec].scavenged.popcntRange(0, ei+1)
-		s.chunks[ec].allocRange(0, ei+1)
+		chunk = s.chunkOf(ec)
+		scav += chunk.scavenged.popcntRange(0, ei+1)
+		chunk.allocRange(0, ei+1)
 	}
 	s.update(base, npages, true, true)
 	return uintptr(scav) * pageSize
@@ -702,7 +741,7 @@ nextLevel:
 	// After iterating over all levels, i must contain a chunk index which
 	// is what the final level represents.
 	ci := chunkIdx(i)
-	j, searchIdx := s.chunks[ci].find(npages, 0)
+	j, searchIdx := s.chunkOf(ci).find(npages, 0)
 	if j < 0 {
 		// We couldn't find any space in this chunk despite the summaries telling
 		// us it should be there. There's likely a bug, so dump some state and throw.
@@ -744,7 +783,7 @@ func (s *pageAlloc) alloc(npages uintptr) (addr uintptr, scav uintptr) {
 		// npages is guaranteed to be no greater than pallocChunkPages here.
 		i := chunkIndex(s.searchAddr)
 		if max := s.summary[len(s.summary)-1][i].max(); max >= uint(npages) {
-			j, searchIdx := s.chunks[i].find(npages, chunkPageIndex(s.searchAddr))
+			j, searchIdx := s.chunkOf(i).find(npages, chunkPageIndex(s.searchAddr))
 			if j < 0 {
 				print("runtime: max = ", max, ", npages = ", npages, "\n")
 				print("runtime: searchIdx = ", chunkPageIndex(s.searchAddr), ", s.searchAddr = ", hex(s.searchAddr), "\n")
@@ -793,7 +832,8 @@ func (s *pageAlloc) free(base, npages uintptr) {
 	if npages == 1 {
 		// Fast path: we're clearing a single bit, and we know exactly
 		// where it is, so mark it directly.
-		s.chunks[chunkIndex(base)].free1(chunkPageIndex(base))
+		i := chunkIndex(base)
+		s.chunkOf(i).free1(chunkPageIndex(base))
 	} else {
 		// Slow path: we're clearing more bits so we may need to iterate.
 		limit := base + npages*pageSize - 1
@@ -802,14 +842,14 @@ func (s *pageAlloc) free(base, npages uintptr) {
 
 		if sc == ec {
 			// The range doesn't cross any chunk boundaries.
-			s.chunks[sc].free(si, ei+1-si)
+			s.chunkOf(sc).free(si, ei+1-si)
 		} else {
 			// The range crosses at least one chunk boundary.
-			s.chunks[sc].free(si, pallocChunkPages-si)
+			s.chunkOf(sc).free(si, pallocChunkPages-si)
 			for c := sc + 1; c < ec; c++ {
-				s.chunks[c].freeAll()
+				s.chunkOf(c).freeAll()
 			}
-			s.chunks[ec].free(0, ei+1)
+			s.chunkOf(ec).free(0, ei+1)
 		}
 	}
 	s.update(base, npages, true, false)