Documentation/static-keys.txt

   1 ===========
   2 Static Keys
   3 ===========
   4
   5 .. warning::
   6
   7    DEPRECATED API:
   8
   9    The use of 'struct static_key' directly, is now DEPRECATED. In addition
  10    static_key_{true,false}() is also DEPRECATED. IE DO NOT use the following::
  11
  12         struct static_key false = STATIC_KEY_INIT_FALSE;
  13         struct static_key true = STATIC_KEY_INIT_TRUE;
  14         static_key_true()
  15         static_key_false()
  16
  17    The updated API replacements are::
  18
  19         DEFINE_STATIC_KEY_TRUE(key);
  20         DEFINE_STATIC_KEY_FALSE(key);
  21         DEFINE_STATIC_KEY_ARRAY_TRUE(keys, count);
  22         DEFINE_STATIC_KEY_ARRAY_FALSE(keys, count);
  23         static_branch_likely()
  24         static_branch_unlikely()
  25
  26 Abstract
  27 ========
  28
  29 Static keys allows the inclusion of seldom used features in
  30 performance-sensitive fast-path kernel code, via a GCC feature and a code
  31 patching technique. A quick example::
  32
  33         DEFINE_STATIC_KEY_FALSE(key);
  34
  35         ...
  36
  37         if (static_branch_unlikely(&key))
  38                 do unlikely code
  39         else
  40                 do likely code
  41
  42         ...
  43         static_branch_enable(&key);
  44         ...
  45         static_branch_disable(&key);
  46         ...
  47
  48 The static_branch_unlikely() branch will be generated into the code with as little
  49 impact to the likely code path as possible.
  50
  51
  52 Motivation
  53 ==========
  54
  55
  56 Currently, tracepoints are implemented using a conditional branch. The
  57 conditional check requires checking a global variable for each tracepoint.
  58 Although the overhead of this check is small, it increases when the memory
  59 cache comes under pressure (memory cache lines for these global variables may
  60 be shared with other memory accesses). As we increase the number of tracepoints
  61 in the kernel this overhead may become more of an issue. In addition,
  62 tracepoints are often dormant (disabled) and provide no direct kernel
  63 functionality. Thus, it is highly desirable to reduce their impact as much as
  64 possible. Although tracepoints are the original motivation for this work, other
  65 kernel code paths should be able to make use of the static keys facility.
  66
  67
  68 Solution
  69 ========
  70
  71
  72 gcc (v4.5) adds a new 'asm goto' statement that allows branching to a label:
  73
  74 http://gcc.gnu.org/ml/gcc-patches/2009-07/msg01556.html
  75
  76 Using the 'asm goto', we can create branches that are either taken or not taken
  77 by default, without the need to check memory. Then, at run-time, we can patch
  78 the branch site to change the branch direction.
  79
  80 For example, if we have a simple branch that is disabled by default::
  81
  82         if (static_branch_unlikely(&key))
  83                 printk("I am the true branch\n");
  84
  85 Thus, by default the 'printk' will not be emitted. And the code generated will
  86 consist of a single atomic 'no-op' instruction (5 bytes on x86), in the
  87 straight-line code path. When the branch is 'flipped', we will patch the
  88 'no-op' in the straight-line codepath with a 'jump' instruction to the
  89 out-of-line true branch. Thus, changing branch direction is expensive but
  90 branch selection is basically 'free'. That is the basic tradeoff of this
  91 optimization.
  92
  93 This lowlevel patching mechanism is called 'jump label patching', and it gives
  94 the basis for the static keys facility.
  95
  96 Static key label API, usage and examples
  97 ========================================
  98
  99
 100 In order to make use of this optimization you must first define a key::
 101
 102         DEFINE_STATIC_KEY_TRUE(key);
 103
 104 or::
 105
 106         DEFINE_STATIC_KEY_FALSE(key);
 107
 108
 109 The key must be global, that is, it can't be allocated on the stack or dynamically
 110 allocated at run-time.
 111
 112 The key is then used in code as::
 113
 114         if (static_branch_unlikely(&key))
 115                 do unlikely code
 116         else
 117                 do likely code
 118
 119 Or::
 120
 121         if (static_branch_likely(&key))
 122                 do likely code
 123         else
 124                 do unlikely code
 125
 126 Keys defined via DEFINE_STATIC_KEY_TRUE(), or DEFINE_STATIC_KEY_FALSE, may
 127 be used in either static_branch_likely() or static_branch_unlikely()
 128 statements.
 129
 130 Branch(es) can be set true via::
 131
 132         static_branch_enable(&key);
 133
 134 or false via::
 135
 136         static_branch_disable(&key);
 137
 138 The branch(es) can then be switched via reference counts::
 139
 140         static_branch_inc(&key);
 141         ...
 142         static_branch_dec(&key);
 143
 144 Thus, 'static_branch_inc()' means 'make the branch true', and
 145 'static_branch_dec()' means 'make the branch false' with appropriate
 146 reference counting. For example, if the key is initialized true, a
 147 static_branch_dec(), will switch the branch to false. And a subsequent
 148 static_branch_inc(), will change the branch back to true. Likewise, if the
 149 key is initialized false, a 'static_branch_inc()', will change the branch to
 150 true. And then a 'static_branch_dec()', will again make the branch false.
 151
 152 Where an array of keys is required, it can be defined as::
 153
 154         DEFINE_STATIC_KEY_ARRAY_TRUE(keys, count);
 155
 156 or::
 157
 158         DEFINE_STATIC_KEY_ARRAY_FALSE(keys, count);
 159
 160 4) Architecture level code patching interface, 'jump labels'
 161
 162
 163 There are a few functions and macros that architectures must implement in order
 164 to take advantage of this optimization. If there is no architecture support, we
 165 simply fall back to a traditional, load, test, and jump sequence. Also, the
 166 struct jump_entry table must be at least 4-byte aligned because the
 167 static_key->entry field makes use of the two least significant bits.
 168
 169 * ``select HAVE_ARCH_JUMP_LABEL``,
 170     see: arch/x86/Kconfig
 171
 172 * ``#define JUMP_LABEL_NOP_SIZE``,
 173     see: arch/x86/include/asm/jump_label.h
 174
 175 * ``__always_inline bool arch_static_branch(struct static_key *key, bool branch)``,
 176     see: arch/x86/include/asm/jump_label.h
 177
 178 * ``__always_inline bool arch_static_branch_jump(struct static_key *key, bool branch)``,
 179     see: arch/x86/include/asm/jump_label.h
 180
 181 * ``void arch_jump_label_transform(struct jump_entry *entry, enum jump_label_type type)``,
 182     see: arch/x86/kernel/jump_label.c
 183
 184 * ``__init_or_module void arch_jump_label_transform_static(struct jump_entry *entry, enum jump_label_type type)``,
 185     see: arch/x86/kernel/jump_label.c
 186
 187 * ``struct jump_entry``,
 188     see: arch/x86/include/asm/jump_label.h
 189
 190
 191 5) Static keys / jump label analysis, results (x86_64):
 192
 193
 194 As an example, let's add the following branch to 'getppid()', such that the
 195 system call now looks like::
 196
 197   SYSCALL_DEFINE0(getppid)
 198   {
 199         int pid;
 200
 201   +     if (static_branch_unlikely(&key))
 202   +             printk("I am the true branch\n");
 203
 204         rcu_read_lock();
 205         pid = task_tgid_vnr(rcu_dereference(current->real_parent));
 206         rcu_read_unlock();
 207
 208         return pid;
 209   }
 210
 211 The resulting instructions with jump labels generated by GCC is::
 212
 213   ffffffff81044290 <sys_getppid>:
 214   ffffffff81044290:       55                      push   %rbp
 215   ffffffff81044291:       48 89 e5                mov    %rsp,%rbp
 216   ffffffff81044294:       e9 00 00 00 00          jmpq   ffffffff81044299 <sys_getppid+0x9>
 217   ffffffff81044299:       65 48 8b 04 25 c0 b6    mov    %gs:0xb6c0,%rax
 218   ffffffff810442a0:       00 00
 219   ffffffff810442a2:       48 8b 80 80 02 00 00    mov    0x280(%rax),%rax
 220   ffffffff810442a9:       48 8b 80 b0 02 00 00    mov    0x2b0(%rax),%rax
 221   ffffffff810442b0:       48 8b b8 e8 02 00 00    mov    0x2e8(%rax),%rdi
 222   ffffffff810442b7:       e8 f4 d9 00 00          callq  ffffffff81051cb0 <pid_vnr>
 223   ffffffff810442bc:       5d                      pop    %rbp
 224   ffffffff810442bd:       48 98                   cltq
 225   ffffffff810442bf:       c3                      retq
 226   ffffffff810442c0:       48 c7 c7 e3 54 98 81    mov    $0xffffffff819854e3,%rdi
 227   ffffffff810442c7:       31 c0                   xor    %eax,%eax
 228   ffffffff810442c9:       e8 71 13 6d 00          callq  ffffffff8171563f <printk>
 229   ffffffff810442ce:       eb c9                   jmp    ffffffff81044299 <sys_getppid+0x9>
 230
 231 Without the jump label optimization it looks like::
 232
 233   ffffffff810441f0 <sys_getppid>:
 234   ffffffff810441f0:       8b 05 8a 52 d8 00       mov    0xd8528a(%rip),%eax        # ffffffff81dc9480 <key>
 235   ffffffff810441f6:       55                      push   %rbp
 236   ffffffff810441f7:       48 89 e5                mov    %rsp,%rbp
 237   ffffffff810441fa:       85 c0                   test   %eax,%eax
 238   ffffffff810441fc:       75 27                   jne    ffffffff81044225 <sys_getppid+0x35>
 239   ffffffff810441fe:       65 48 8b 04 25 c0 b6    mov    %gs:0xb6c0,%rax
 240   ffffffff81044205:       00 00
 241   ffffffff81044207:       48 8b 80 80 02 00 00    mov    0x280(%rax),%rax
 242   ffffffff8104420e:       48 8b 80 b0 02 00 00    mov    0x2b0(%rax),%rax
 243   ffffffff81044215:       48 8b b8 e8 02 00 00    mov    0x2e8(%rax),%rdi
 244   ffffffff8104421c:       e8 2f da 00 00          callq  ffffffff81051c50 <pid_vnr>
 245   ffffffff81044221:       5d                      pop    %rbp
 246   ffffffff81044222:       48 98                   cltq
 247   ffffffff81044224:       c3                      retq
 248   ffffffff81044225:       48 c7 c7 13 53 98 81    mov    $0xffffffff81985313,%rdi
 249   ffffffff8104422c:       31 c0                   xor    %eax,%eax
 250   ffffffff8104422e:       e8 60 0f 6d 00          callq  ffffffff81715193 <printk>
 251   ffffffff81044233:       eb c9                   jmp    ffffffff810441fe <sys_getppid+0xe>
 252   ffffffff81044235:       66 66 2e 0f 1f 84 00    data32 nopw %cs:0x0(%rax,%rax,1)
 253   ffffffff8104423c:       00 00 00 00
 254
 255 Thus, the disable jump label case adds a 'mov', 'test' and 'jne' instruction
 256 vs. the jump label case just has a 'no-op' or 'jmp 0'. (The jmp 0, is patched
 257 to a 5 byte atomic no-op instruction at boot-time.) Thus, the disabled jump
 258 label case adds::
 259
 260   6 (mov) + 2 (test) + 2 (jne) = 10 - 5 (5 byte jump 0) = 5 addition bytes.
 261
 262 If we then include the padding bytes, the jump label code saves, 16 total bytes
 263 of instruction memory for this small function. In this case the non-jump label
 264 function is 80 bytes long. Thus, we have saved 20% of the instruction
 265 footprint. We can in fact improve this even further, since the 5-byte no-op
 266 really can be a 2-byte no-op since we can reach the branch with a 2-byte jmp.
 267 However, we have not yet implemented optimal no-op sizes (they are currently
 268 hard-coded).
 269
 270 Since there are a number of static key API uses in the scheduler paths,
 271 'pipe-test' (also known as 'perf bench sched pipe') can be used to show the
 272 performance improvement. Testing done on 3.3.0-rc2:
 273
 274 jump label disabled::
 275
 276  Performance counter stats for 'bash -c /tmp/pipe-test' (50 runs):
 277
 278         855.700314 task-clock                #    0.534 CPUs utilized            ( +-  0.11% )
 279            200,003 context-switches          #    0.234 M/sec                    ( +-  0.00% )
 280                  0 CPU-migrations            #    0.000 M/sec                    ( +- 39.58% )
 281                487 page-faults               #    0.001 M/sec                    ( +-  0.02% )
 282      1,474,374,262 cycles                    #    1.723 GHz                      ( +-  0.17% )
 283    <not supported> stalled-cycles-frontend
 284    <not supported> stalled-cycles-backend
 285      1,178,049,567 instructions              #    0.80  insns per cycle          ( +-  0.06% )
 286        208,368,926 branches                  #  243.507 M/sec                    ( +-  0.06% )
 287          5,569,188 branch-misses             #    2.67% of all branches          ( +-  0.54% )
 288
 289        1.601607384 seconds time elapsed                                          ( +-  0.07% )
 290
 291 jump label enabled::
 292
 293  Performance counter stats for 'bash -c /tmp/pipe-test' (50 runs):
 294
 295         841.043185 task-clock                #    0.533 CPUs utilized            ( +-  0.12% )
 296            200,004 context-switches          #    0.238 M/sec                    ( +-  0.00% )
 297                  0 CPU-migrations            #    0.000 M/sec                    ( +- 40.87% )
 298                487 page-faults               #    0.001 M/sec                    ( +-  0.05% )
 299      1,432,559,428 cycles                    #    1.703 GHz                      ( +-  0.18% )
 300    <not supported> stalled-cycles-frontend
 301    <not supported> stalled-cycles-backend
 302      1,175,363,994 instructions              #    0.82  insns per cycle          ( +-  0.04% )
 303        206,859,359 branches                  #  245.956 M/sec                    ( +-  0.04% )
 304          4,884,119 branch-misses             #    2.36% of all branches          ( +-  0.85% )
 305
 306        1.579384366 seconds time elapsed
 307
 308 The percentage of saved branches is .7%, and we've saved 12% on
 309 'branch-misses'. This is where we would expect to get the most savings, since
 310 this optimization is about reducing the number of branches. In addition, we've
 311 saved .2% on instructions, and 2.8% on cycles and 1.4% on elapsed time.