Skip to content

Commit

Permalink
runtime: add 24 byte allocation size class
Browse files Browse the repository at this point in the history 
This CL introduces a 24 byte allocation size class which
fits 3 pointers on 64 bit and 6 pointers on 32 bit architectures.

Notably this new size class fits a slice header on 64 bit
architectures exactly while previously a 32 byte size class
would have been used for allocating a slice header on the heap.

The main complexity added with this CL is that heapBitsSetType
needs to handle objects that aren't 16-byte aligned but contain
more than a single pointer on 64-bit architectures.

Due to having a non 16 byte aligned size class on 32 bit a
h.shift of 2 is now possible which means a heap bitmap byte might
only be partially written. Due to this already having been
possible on 64 bit before the heap bitmap code only needed
minor adjustments for 32 bit doublecheck code paths.

Note that this CL changes the slice capacity allocated by append
for slice growth to a target capacity of 17 to 24 bytes.

On 64 bit architectures the capacity of the slice returned by
append([]byte{}, make([]byte, 24)...)) is 32 bytes before and
24 bytes after this CL. Depending on allocation patterns of the
specific Go program this can increase the number of total
alloctions as subsequent appends to the slice can trigger slice
growth earlier than before. On the other side if the slice is
never appended to again above its capacity this will lower heap
usage by 8 bytes.

This CL changes the set of size classes reported in the
runtime.MemStats.BySize array due to it being limited to a
total of 61 size classes. The new 24 byte size class is now
included and the 20480 byte size class is not included anymore.

Fixes golang#8885

name                      old time/op       new time/op       delta
Template                        196ms ± 3%        194ms ± 2%    ~     (p=0.247 n=10+10)
Unicode                        85.6ms ±16%       88.1ms ± 1%    ~     (p=0.165 n=10+10)
GoTypes                         673ms ± 2%        668ms ± 2%    ~     (p=0.258 n=9+9)
Compiler                        3.14s ± 6%        3.08s ± 1%    ~     (p=0.243 n=10+9)
SSA                             6.82s ± 1%        6.76s ± 1%  -0.87%  (p=0.006 n=9+10)
Flate                           128ms ± 7%        127ms ± 3%    ~     (p=0.739 n=10+10)
GoParser                        154ms ± 3%        153ms ± 4%    ~     (p=0.730 n=9+9)
Reflect                         404ms ± 1%        412ms ± 4%  +1.99%  (p=0.022 n=9+10)
Tar                             172ms ± 4%        170ms ± 4%    ~     (p=0.065 n=10+9)
XML                             231ms ± 4%        230ms ± 3%    ~     (p=0.912 n=10+10)
LinkCompiler                    341ms ± 1%        339ms ± 1%    ~     (p=0.243 n=9+10)
ExternalLinkCompiler            1.72s ± 1%        1.72s ± 1%    ~     (p=0.661 n=9+10)
LinkWithoutDebugCompiler        221ms ± 2%        221ms ± 2%    ~     (p=0.529 n=10+10)
StdCmd                          18.4s ± 3%        18.2s ± 1%    ~     (p=0.515 n=10+8)

name                      old user-time/op  new user-time/op  delta
Template                        238ms ± 4%        243ms ± 6%    ~     (p=0.661 n=9+10)
Unicode                         116ms ± 6%        113ms ± 3%  -3.37%  (p=0.035 n=9+10)
GoTypes                         854ms ± 2%        848ms ± 2%    ~     (p=0.604 n=9+10)
Compiler                        4.10s ± 1%        4.11s ± 1%    ~     (p=0.481 n=8+9)
SSA                             9.49s ± 1%        9.41s ± 1%  -0.92%  (p=0.001 n=9+10)
Flate                           149ms ± 6%        151ms ± 7%    ~     (p=0.481 n=10+10)
GoParser                        189ms ± 2%        190ms ± 2%    ~     (p=0.497 n=9+10)
Reflect                         511ms ± 2%        508ms ± 2%    ~     (p=0.211 n=9+10)
Tar                             215ms ± 4%        212ms ± 3%    ~     (p=0.105 n=10+10)
XML                             288ms ± 2%        288ms ± 2%    ~     (p=0.971 n=10+10)
LinkCompiler                    559ms ± 4%        557ms ± 1%    ~     (p=0.968 n=9+10)
ExternalLinkCompiler            1.78s ± 1%        1.77s ± 1%    ~     (p=0.055 n=8+10)
LinkWithoutDebugCompiler        245ms ± 3%        245ms ± 2%    ~     (p=0.684 n=10+10)

name                      old alloc/op      new alloc/op      delta
Template                       34.8MB ± 0%       34.4MB ± 0%  -0.95%  (p=0.000 n=9+10)
Unicode                        28.6MB ± 0%       28.3MB ± 0%  -0.95%  (p=0.000 n=10+10)
GoTypes                         115MB ± 0%        114MB ± 0%  -1.02%  (p=0.000 n=10+9)
Compiler                        554MB ± 0%        549MB ± 0%  -0.86%  (p=0.000 n=9+10)
SSA                            1.28GB ± 0%       1.27GB ± 0%  -0.83%  (p=0.000 n=10+10)
Flate                          21.8MB ± 0%       21.6MB ± 0%  -0.87%  (p=0.000 n=8+10)
GoParser                       26.7MB ± 0%       26.4MB ± 0%  -0.97%  (p=0.000 n=10+9)
Reflect                        75.0MB ± 0%       74.1MB ± 0%  -1.18%  (p=0.000 n=10+10)
Tar                            32.6MB ± 0%       32.3MB ± 0%  -0.94%  (p=0.000 n=10+7)
XML                            41.5MB ± 0%       41.2MB ± 0%  -0.90%  (p=0.000 n=10+8)
LinkCompiler                    105MB ± 0%        104MB ± 0%  -0.94%  (p=0.000 n=10+10)
ExternalLinkCompiler            153MB ± 0%        152MB ± 0%  -0.69%  (p=0.000 n=10+10)
LinkWithoutDebugCompiler       63.7MB ± 0%       63.6MB ± 0%  -0.13%  (p=0.000 n=10+10)

name                      old allocs/op     new allocs/op     delta
Template                         336k ± 0%         336k ± 0%  +0.02%  (p=0.002 n=10+10)
Unicode                          332k ± 0%         332k ± 0%    ~     (p=0.447 n=10+10)
GoTypes                         1.16M ± 0%        1.16M ± 0%  +0.01%  (p=0.001 n=10+10)
Compiler                        4.92M ± 0%        4.92M ± 0%  +0.01%  (p=0.000 n=10+10)
SSA                             11.9M ± 0%        11.9M ± 0%  +0.02%  (p=0.000 n=9+10)
Flate                            214k ± 0%         214k ± 0%  +0.02%  (p=0.032 n=10+8)
GoParser                         270k ± 0%         270k ± 0%  +0.02%  (p=0.004 n=10+9)
Reflect                          877k ± 0%         877k ± 0%  +0.01%  (p=0.000 n=10+10)
Tar                              313k ± 0%         313k ± 0%    ~     (p=0.075 n=9+10)
XML                              387k ± 0%         387k ± 0%  +0.02%  (p=0.007 n=10+10)
LinkCompiler                     455k ± 0%         456k ± 0%  +0.08%  (p=0.000 n=10+9)
ExternalLinkCompiler             670k ± 0%         671k ± 0%  +0.06%  (p=0.000 n=10+10)
LinkWithoutDebugCompiler         113k ± 0%         113k ± 0%    ~     (p=0.149 n=10+10)

name                      old maxRSS/op     new maxRSS/op     delta
Template                        34.1M ± 1%        34.1M ± 1%    ~     (p=0.853 n=10+10)
Unicode                         35.1M ± 1%        34.6M ± 1%  -1.43%  (p=0.000 n=10+10)
GoTypes                         72.8M ± 3%        73.3M ± 2%    ~     (p=0.724 n=10+10)
Compiler                         288M ± 3%         295M ± 4%    ~     (p=0.393 n=10+10)
SSA                              630M ± 1%         622M ± 1%  -1.18%  (p=0.001 n=10+10)
Flate                           26.0M ± 1%        26.2M ± 2%    ~     (p=0.493 n=10+10)
GoParser                        28.6M ± 1%        28.5M ± 2%    ~     (p=0.256 n=10+10)
Reflect                         55.5M ± 2%        55.4M ± 1%    ~     (p=0.436 n=10+10)
Tar                             33.0M ± 1%        32.8M ± 2%    ~     (p=0.075 n=10+10)
XML                             38.7M ± 1%        39.0M ± 1%    ~     (p=0.053 n=9+10)
LinkCompiler                     164M ± 1%         164M ± 1%  -0.27%  (p=0.029 n=10+10)
ExternalLinkCompiler             174M ± 0%         173M ± 0%  -0.33%  (p=0.002 n=9+10)
LinkWithoutDebugCompiler         137M ± 0%         136M ± 2%    ~     (p=0.825 n=9+10)

Change-Id: I9ecf2a10024513abef8fbfbe519e44e0b29b6167
Reviewed-on: https://go-review.googlesource.com/c/go/+/242258
Trust: Martin Möhrmann <[email protected]>
Trust: Michael Knyszek <[email protected]>
Run-TryBot: Martin Möhrmann <[email protected]>
TryBot-Result: Go Bot <[email protected]>
Reviewed-by: Michael Knyszek <[email protected]>
Reviewed-by: Keith Randall <[email protected]>
  • Loading branch information
@martisch
martisch committed Sep 14, 2020
1 parent c0c396b commit 14c7caa
Show file tree
Hide file tree
Showing 4 changed files with 174 additions and 89 deletions.
112 changes: 96 additions & 16 deletions src/runtime/mbitmap.go
View file
Original file line number Diff line number Diff line change
Expand Up @@ -30,10 +30,9 @@
// indicates scanning can ignore the rest of the allocation.
//
// The 2-bit entries are split when written into the byte, so that the top half
// of the byte contains 4 high bits and the bottom half contains 4 low (pointer)
// bits.
// This form allows a copy from the 1-bit to the 4-bit form to keep the
// pointer bits contiguous, instead of having to space them out.
// of the byte contains 4 high (scan) bits and the bottom half contains 4 low
// (pointer) bits. This form allows a copy from the 1-bit to the 4-bit form to
// keep the pointer bits contiguous, instead of having to space them out.
//
// The code makes use of the fact that the zero value for a heap
// bitmap means scalar/dead. This property must be preserved when
Expand Down Expand Up @@ -816,6 +815,12 @@ func (s *mspan) countAlloc() int {
func heapBitsSetType(x, size, dataSize uintptr, typ *_type) {
const doubleCheck = false // slow but helpful; enable to test modifications to this code

const (
mask1 = bitPointer | bitScan // 00010001
mask2 = bitPointer | bitScan | mask1<<heapBitsShift // 00110011
mask3 = bitPointer | bitScan | mask2<<heapBitsShift // 01110111
)

// dataSize is always size rounded up to the next malloc size class,
// except in the case of allocating a defer block, in which case
// size is sizeof(_defer{}) (at least 6 words) and dataSize may be
Expand Down Expand Up @@ -844,11 +849,12 @@ func heapBitsSetType(x, size, dataSize uintptr, typ *_type) {
h := heapBitsForAddr(x)
ptrmask := typ.gcdata // start of 1-bit pointer mask (or GC program, handled below)

// Heap bitmap bits for 2-word object are only 4 bits,
// so also shared with objects next to it.
// This is called out as a special case primarily for 32-bit systems,
// so that on 32-bit systems the code below can assume all objects
// are 4-word aligned (because they're all 16-byte aligned).
// 2-word objects only have 4 bitmap bits and 3-word objects only have 6 bitmap bits.
// Therefore, these objects share a heap bitmap byte with the objects next to them.
// These are called out as a special case primarily so the code below can assume all
// objects are at least 4 words long and that their bitmaps start either at the beginning
// of a bitmap byte, or half-way in (h.shift of 0 and 2 respectively).

if size == 2*sys.PtrSize {
if typ.size == sys.PtrSize {
// We're allocating a block big enough to hold two pointers.
Expand All @@ -865,7 +871,7 @@ func heapBitsSetType(x, size, dataSize uintptr, typ *_type) {
*h.bitp &^= (bitPointer | bitScan | (bitPointer|bitScan)<<heapBitsShift) << h.shift
*h.bitp |= (bitPointer | bitScan) << h.shift
} else {
// 2-element slice of pointer.
// 2-element array of pointer.
*h.bitp |= (bitPointer | bitScan | (bitPointer|bitScan)<<heapBitsShift) << h.shift
}
return
Expand All @@ -886,6 +892,70 @@ func heapBitsSetType(x, size, dataSize uintptr, typ *_type) {
*h.bitp &^= (bitPointer | bitScan | ((bitPointer | bitScan) << heapBitsShift)) << h.shift
*h.bitp |= uint8(hb << h.shift)
return
} else if size == 3*sys.PtrSize {
b := uint8(*ptrmask)
if doubleCheck {
if b == 0 {
println("runtime: invalid type ", typ.string())
throw("heapBitsSetType: called with non-pointer type")
}
if sys.PtrSize != 8 {
throw("heapBitsSetType: unexpected 3 pointer wide size class on 32 bit")
}
if typ.kind&kindGCProg != 0 {
throw("heapBitsSetType: unexpected GC prog for 3 pointer wide size class")
}
if typ.size == 2*sys.PtrSize {
print("runtime: heapBitsSetType size=", size, " but typ.size=", typ.size, "\n")
throw("heapBitsSetType: inconsistent object sizes")
}
}
if typ.size == sys.PtrSize {
// The type contains a pointer otherwise heapBitsSetType wouldn't have been called.
// Since the type is only 1 pointer wide and contains a pointer, its gcdata must be exactly 1.
if doubleCheck && *typ.gcdata != 1 {
print("runtime: heapBitsSetType size=", size, " typ.size=", typ.size, "but *typ.gcdata", *typ.gcdata, "\n")
throw("heapBitsSetType: unexpected gcdata for 1 pointer wide type size in 3 pointer wide size class")
}
// 3 element array of pointers. Unrolling ptrmask 3 times into p yields 00000111.
b = 7
}

hb := b & 7
// Set bitScan bits for all pointers.
hb |= hb << wordsPerBitmapByte
// First bitScan bit is always set since the type contains pointers.
hb |= bitScan
// Second bitScan bit needs to also be set if the third bitScan bit is set.
hb |= hb & (bitScan << (2 * heapBitsShift)) >> 1

// For h.shift > 1 heap bits cross a byte boundary and need to be written part
// to h.bitp and part to the next h.bitp.
switch h.shift {
case 0:
*h.bitp &^= mask3 << 0
*h.bitp |= hb << 0
case 1:
*h.bitp &^= mask3 << 1
*h.bitp |= hb << 1
case 2:
*h.bitp &^= mask2 << 2
*h.bitp |= (hb & mask2) << 2
// Two words written to the first byte.
// Advance two words to get to the next byte.
h = h.next().next()
*h.bitp &^= mask1
*h.bitp |= (hb >> 2) & mask1
case 3:
*h.bitp &^= mask1 << 3
*h.bitp |= (hb & mask1) << 3
// One word written to the first byte.
// Advance one word to get to the next byte.
h = h.next()
*h.bitp &^= mask2
*h.bitp |= (hb >> 1) & mask2
}
return
}

// Copy from 1-bit ptrmask into 2-bit bitmap.
Expand Down Expand Up @@ -1079,7 +1149,7 @@ func heapBitsSetType(x, size, dataSize uintptr, typ *_type) {
// word must be set to scan since there are pointers
// somewhere in the object.
// In all following words, we set the scan/dead
// appropriately to indicate that the object contains
// appropriately to indicate that the object continues
// to the next 2-bit entry in the bitmap.
//
// We set four bits at a time here, but if the object
Expand All @@ -1095,12 +1165,22 @@ func heapBitsSetType(x, size, dataSize uintptr, typ *_type) {
b >>= 4
nb -= 4

case sys.PtrSize == 8 && h.shift == 2:
case h.shift == 2:
// Ptrmask and heap bitmap are misaligned.
//
// On 32 bit architectures only the 6-word object that corresponds
// to a 24 bytes size class can start with h.shift of 2 here since
// all other non 16 byte aligned size classes have been handled by
// special code paths at the beginning of heapBitsSetType on 32 bit.
//
// Many size classes are only 16 byte aligned. On 64 bit architectures
// this results in a heap bitmap position starting with a h.shift of 2.
//
// The bits for the first two words are in a byte shared
// with another object, so we must be careful with the bits
// already there.
// We took care of 1-word and 2-word objects above,
//
// We took care of 1-word, 2-word, and 3-word objects above,
// so this is at least a 6-word object.
hb = (b & (bitPointer | bitPointer<<heapBitsShift)) << (2 * heapBitsShift)
hb |= bitScan << (2 * heapBitsShift)
Expand All @@ -1113,7 +1193,7 @@ func heapBitsSetType(x, size, dataSize uintptr, typ *_type) {
*hbitp |= uint8(hb)
hbitp = add1(hbitp)
if w += 2; w >= nw {
// We know that there is more data, because we handled 2-word objects above.
// We know that there is more data, because we handled 2-word and 3-word objects above.
// This must be at least a 6-word object. If we're out of pointer words,
// mark no scan in next bitmap byte and finish.
hb = 0
Expand Down Expand Up @@ -1248,12 +1328,12 @@ Phase4:
// Handle the first byte specially if it's shared. See
// Phase 1 for why this is the only special case we need.
if doubleCheck {
if !(h.shift == 0 || (sys.PtrSize == 8 && h.shift == 2)) {
if !(h.shift == 0 || h.shift == 2) {
print("x=", x, " size=", size, " cnw=", h.shift, "\n")
throw("bad start shift")
}
}
if sys.PtrSize == 8 && h.shift == 2 {
if h.shift == 2 {
*h.bitp = *h.bitp&^((bitPointer|bitScan|(bitPointer|bitScan)<<heapBitsShift)<<(2*heapBitsShift)) | *src
h = h.next().next()
cnw -= 2
Expand Down
6 changes: 3 additions & 3 deletions src/runtime/mksizeclasses.go
View file
Original file line number Diff line number Diff line change
Expand Up @@ -110,8 +110,8 @@ func makeClasses() []class {
align = 256
} else if size >= 128 {
align = size / 8
} else if size >= 16 {
align = 16 // required for x86 SSE instructions, if we want to use them
} else if size >= 32 {
align = 16 // heap bitmaps assume 16 byte alignment for allocations >= 32 bytes.
}
}
if !powerOfTwo(align) {
Expand Down Expand Up @@ -157,7 +157,7 @@ func makeClasses() []class {
}
}

if len(classes) != 67 {
if len(classes) != 68 {
panic("number of size classes has changed")
}

Expand Down
4 changes: 4 additions & 0 deletions src/runtime/mstats.go
View file
Original file line number Diff line number Diff line change
Expand Up @@ -78,6 +78,10 @@ type mstats struct {
nfree uint64
}

// Add an uint32 for even number of size classes to align below fields
// to 64 bits for atomic operations on 32 bit platforms.
_ [1 - _NumSizeClasses%2]uint32

// Statistics below here are not exported to MemStats directly.

last_gc_nanotime uint64 // last gc (monotonic time)
Expand Down
Loading

0 comments on commit 14c7caa

Please sign in to comment.