1
0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-10-18 10:32:48 +02:00
llvm-mirror/lib/MC/MCPseudoProbe.cpp
Hongtao Yu 85e4f6f241 [CSSPGO] Pseudo probe encoding and emission.
This change implements pseudo probe encoding and emission for CSSPGO. Please see RFC here for more context: https://groups.google.com/g/llvm-dev/c/1p1rdYbL93s

Pseudo probes are in the form of intrinsic calls on IR/MIR but they do not turn into any machine instructions. Instead they are emitted into the binary as a piece of data in standalone sections.  The probe-specific sections are not needed to be loaded into memory at execution time, thus they do not incur a runtime overhead. 

**ELF object emission**

The binary data to emit are organized as two ELF sections, i.e, the `.pseudo_probe_desc` section and the `.pseudo_probe` section. The `.pseudo_probe_desc` section stores a function descriptor for each function and the `.pseudo_probe` section stores the actual probes, each fo which corresponds to an IR basic block or an IR function callsite. A function descriptor is stored as a module-level metadata during the compilation and is serialized into the object file during object emission.

Both the probe descriptors and pseudo probes can be emitted into a separate ELF section per function to leverage the linker for deduplication.  A `.pseudo_probe` section shares the same COMDAT group with the function code so that when the function is dead, the probes are dead and disposed too. On the contrary, a `.pseudo_probe_desc` section has its own COMDAT group. This is because even if a function is dead, its probes may be inlined into other functions and its descriptor is still needed by the profile generation tool.

The format of `.pseudo_probe_desc` section looks like:

```
.section   .pseudo_probe_desc,"",@progbits
.quad   6309742469962978389  // Func GUID
.quad   4294967295           // Func Hash
.byte   9                    // Length of func name
.ascii  "_Z5funcAi"          // Func name
.quad   7102633082150537521
.quad   138828622701
.byte   12
.ascii  "_Z8funcLeafi"
.quad   446061515086924981
.quad   4294967295
.byte   9
.ascii  "_Z5funcBi"
.quad   -2016976694713209516
.quad   72617220756
.byte   7
.ascii  "_Z3fibi"
```

For each `.pseudoprobe` section, the encoded binary data consists of a single function record corresponding to an outlined function (i.e, a function with a code entry in the `.text` section). A function record has the following format :

```
FUNCTION BODY (one for each outlined function present in the text section)
    GUID (uint64)
        GUID of the function
    NPROBES (ULEB128)
        Number of probes originating from this function.
    NUM_INLINED_FUNCTIONS (ULEB128)
        Number of callees inlined into this function, aka number of
        first-level inlinees
    PROBE RECORDS
        A list of NPROBES entries. Each entry contains:
          INDEX (ULEB128)
          TYPE (uint4)
            0 - block probe, 1 - indirect call, 2 - direct call
          ATTRIBUTE (uint3)
            reserved
          ADDRESS_TYPE (uint1)
            0 - code address, 1 - address delta
          CODE_ADDRESS (uint64 or ULEB128)
            code address or address delta, depending on ADDRESS_TYPE
    INLINED FUNCTION RECORDS
        A list of NUM_INLINED_FUNCTIONS entries describing each of the inlined
        callees.  Each record contains:
          INLINE SITE
            GUID of the inlinee (uint64)
            ID of the callsite probe (ULEB128)
          FUNCTION BODY
            A FUNCTION BODY entry describing the inlined function.
```

To support building a context-sensitive profile, probes from inlinees are grouped by their inline contexts. An inline context is logically a call path through which a callee function lands in a caller function. The probe emitter builds an inline tree based on the debug metadata for each outlined function in the form of a trie tree. A tree root is the outlined function. Each tree edge stands for a callsite where inlining happens. Pseudo probes originating from an inlinee function are stored in a tree node and the tree path starting from the root all the way down to the tree node is the inline context of the probes. The emission happens on the whole tree top-down recursively. Probes of a tree node will be emitted altogether with their direct parent edge. Since a pseudo probe corresponds to a real code address, for size savings, the address is encoded as a delta from the previous probe except for the first probe. Variant-sized integer encoding, aka LEB128, is used for address delta and probe index.

**Assembling**

Pseudo probes can be printed as assembly directives alternatively. This allows for good assembly code readability and also provides a view of how optimizations and pseudo probes affect each other, especially helpful for diff time assembly analysis.

A pseudo probe directive has the following operands in order: function GUID, probe index, probe type, probe attributes and inline context. The directive is generated by the compiler and can be parsed by the assembler to form an encoded `.pseudoprobe` section in the object file.

A example assembly looks like:

```
foo2: # @foo2
# %bb.0: # %bb0
pushq %rax
testl %edi, %edi
.pseudoprobe 837061429793323041 1 0 0
je .LBB1_1
# %bb.2: # %bb2
.pseudoprobe 837061429793323041 6 2 0
callq foo
.pseudoprobe 837061429793323041 3 0 0
.pseudoprobe 837061429793323041 4 0 0
popq %rax
retq
.LBB1_1: # %bb1
.pseudoprobe 837061429793323041 5 1 0
callq *%rsi
.pseudoprobe 837061429793323041 2 0 0
.pseudoprobe 837061429793323041 4 0 0
popq %rax
retq
# -- End function
.section .pseudo_probe_desc,"",@progbits
.quad 6699318081062747564
.quad 72617220756
.byte 3
.ascii "foo"
.quad 837061429793323041
.quad 281547593931412
.byte 4
.ascii "foo2"
```

With inlining turned on, the assembly may look different around %bb2 with an inlined probe:

```
# %bb.2:                                # %bb2
.pseudoprobe    837061429793323041 3 0
.pseudoprobe    6699318081062747564 1 0 @ 837061429793323041:6
.pseudoprobe    837061429793323041 4 0
popq    %rax
retq
```

**Disassembling**

We have a disassembling tool (llvm-profgen) that can display disassembly alongside with pseudo probes. So far it only supports ELF executable file.

An example disassembly looks like:

```
00000000002011a0 <foo2>:
  2011a0: 50                    push   rax
  2011a1: 85 ff                 test   edi,edi
  [Probe]:  FUNC: foo2  Index: 1  Type: Block
  2011a3: 74 02                 je     2011a7 <foo2+0x7>
  [Probe]:  FUNC: foo2  Index: 3  Type: Block
  [Probe]:  FUNC: foo2  Index: 4  Type: Block
  [Probe]:  FUNC: foo   Index: 1  Type: Block  Inlined: @ foo2:6
  2011a5: 58                    pop    rax
  2011a6: c3                    ret
  [Probe]:  FUNC: foo2  Index: 2  Type: Block
  2011a7: bf 01 00 00 00        mov    edi,0x1
  [Probe]:  FUNC: foo2  Index: 5  Type: IndirectCall
  2011ac: ff d6                 call   rsi
  [Probe]:  FUNC: foo2  Index: 4  Type: Block
  2011ae: 58                    pop    rax
  2011af: c3                    ret
```

Reviewed By: wmi

Differential Revision: https://reviews.llvm.org/D91878
2020-12-10 17:29:28 -08:00

214 lines
6.9 KiB
C++

//===- lib/MC/MCPseudoProbe.cpp - Pseudo probe encoding support ----------===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
#include "llvm/MC/MCPseudoProbe.h"
#include "llvm/MC/MCAsmInfo.h"
#include "llvm/MC/MCContext.h"
#include "llvm/MC/MCObjectFileInfo.h"
#include "llvm/MC/MCObjectStreamer.h"
#include "llvm/MC/MCStreamer.h"
#define DEBUG_TYPE "mcpseudoprobe"
using namespace llvm;
#ifndef NDEBUG
int MCPseudoProbeTable::DdgPrintIndent = 0;
#endif
static const MCExpr *buildSymbolDiff(MCObjectStreamer *MCOS, const MCSymbol *A,
const MCSymbol *B) {
MCContext &Context = MCOS->getContext();
MCSymbolRefExpr::VariantKind Variant = MCSymbolRefExpr::VK_None;
const MCExpr *ARef = MCSymbolRefExpr::create(A, Variant, Context);
const MCExpr *BRef = MCSymbolRefExpr::create(B, Variant, Context);
const MCExpr *AddrDelta =
MCBinaryExpr::create(MCBinaryExpr::Sub, ARef, BRef, Context);
return AddrDelta;
}
void MCPseudoProbe::emit(MCObjectStreamer *MCOS,
const MCPseudoProbe *LastProbe) const {
// Emit Index
MCOS->emitULEB128IntValue(Index);
// Emit Type and the flag:
// Type (bit 0 to 3), with bit 4 to 6 for attributes.
// Flag (bit 7, 0 - code address, 1 - address delta). This indicates whether
// the following field is a symbolic code address or an address delta.
assert(Type <= 0xF && "Probe type too big to encode, exceeding 15");
assert(Attributes <= 0x7 &&
"Probe attributes too big to encode, exceeding 7");
uint8_t PackedType = Type | (Attributes << 4);
uint8_t Flag = LastProbe ? ((int8_t)MCPseudoProbeFlag::AddressDelta << 7) : 0;
MCOS->emitInt8(Flag | PackedType);
if (LastProbe) {
// Emit the delta between the address label and LastProbe.
const MCExpr *AddrDelta =
buildSymbolDiff(MCOS, Label, LastProbe->getLabel());
int64_t Delta;
if (AddrDelta->evaluateAsAbsolute(Delta, MCOS->getAssemblerPtr())) {
MCOS->emitSLEB128IntValue(Delta);
} else {
MCOS->insert(new MCPseudoProbeAddrFragment(AddrDelta));
}
} else {
// Emit label as a symbolic code address.
MCOS->emitSymbolValue(
Label, MCOS->getContext().getAsmInfo()->getCodePointerSize());
}
LLVM_DEBUG({
dbgs().indent(MCPseudoProbeTable::DdgPrintIndent);
dbgs() << "Probe: " << Index << "\n";
});
}
MCPseudoProbeInlineTree::~MCPseudoProbeInlineTree() {
for (auto &Inlinee : Inlinees)
delete Inlinee.second;
}
MCPseudoProbeInlineTree *
MCPseudoProbeInlineTree::getOrAddNode(InlineSite Site) {
auto Iter = Inlinees.find(Site);
if (Iter == Inlinees.end()) {
auto *Node = new MCPseudoProbeInlineTree(std::get<0>(Site));
Inlinees[Site] = Node;
return Node;
} else {
return Iter->second;
}
}
void MCPseudoProbeInlineTree::addPseudoProbe(
const MCPseudoProbe &Probe, const MCPseudoProbeInlineStack &InlineStack) {
// The function should not be called on the root.
assert(isRoot() && "Should not be called on root");
// When it comes here, the input look like:
// Probe: GUID of C, ...
// InlineStack: [88, A], [66, B]
// which means, Function A inlines function B at call site with a probe id of
// 88, and B inlines C at probe 66. The tri-tree expects a tree path like {[0,
// A], [88, B], [66, C]} to locate the tree node where the probe should be
// added. Note that the edge [0, A] means A is the top-level function we are
// emitting probes for.
// Make a [0, A] edge.
// An empty inline stack means the function that the probe originates from
// is a top-level function.
InlineSite Top;
if (InlineStack.empty()) {
Top = InlineSite(Probe.getGuid(), 0);
} else {
Top = InlineSite(std::get<0>(InlineStack.front()), 0);
}
auto *Cur = getOrAddNode(Top);
// Make interior edges by walking the inline stack. Once it's done, Cur should
// point to the node that the probe originates from.
if (!InlineStack.empty()) {
auto Iter = InlineStack.begin();
auto Index = std::get<1>(*Iter);
Iter++;
for (; Iter != InlineStack.end(); Iter++) {
// Make an edge by using the previous probe id and current GUID.
Cur = Cur->getOrAddNode(InlineSite(std::get<0>(*Iter), Index));
Index = std::get<1>(*Iter);
}
Cur = Cur->getOrAddNode(InlineSite(Probe.getGuid(), Index));
}
Cur->Probes.push_back(Probe);
}
void MCPseudoProbeInlineTree::emit(MCObjectStreamer *MCOS,
const MCPseudoProbe *&LastProbe) {
LLVM_DEBUG({
dbgs().indent(MCPseudoProbeTable::DdgPrintIndent);
dbgs() << "Group [\n";
MCPseudoProbeTable::DdgPrintIndent += 2;
});
// Emit probes grouped by GUID.
if (Guid != 0) {
LLVM_DEBUG({
dbgs().indent(MCPseudoProbeTable::DdgPrintIndent);
dbgs() << "GUID: " << Guid << "\n";
});
// Emit Guid
MCOS->emitInt64(Guid);
// Emit number of probes in this node
MCOS->emitULEB128IntValue(Probes.size());
// Emit number of direct inlinees
MCOS->emitULEB128IntValue(Inlinees.size());
// Emit probes in this group
for (const auto &Probe : Probes) {
Probe.emit(MCOS, LastProbe);
LastProbe = &Probe;
}
} else {
assert(Probes.empty() && "Root should not have probes");
}
// Emit descendent
for (const auto &Inlinee : Inlinees) {
if (Guid) {
// Emit probe index
MCOS->emitULEB128IntValue(std::get<1>(Inlinee.first));
LLVM_DEBUG({
dbgs().indent(MCPseudoProbeTable::DdgPrintIndent);
dbgs() << "InlineSite: " << std::get<1>(Inlinee.first) << "\n";
});
}
// Emit the group
Inlinee.second->emit(MCOS, LastProbe);
}
LLVM_DEBUG({
MCPseudoProbeTable::DdgPrintIndent -= 2;
dbgs().indent(MCPseudoProbeTable::DdgPrintIndent);
dbgs() << "]\n";
});
}
void MCPseudoProbeSection::emit(MCObjectStreamer *MCOS) {
MCContext &Ctx = MCOS->getContext();
for (auto &ProbeSec : MCProbeDivisions) {
const MCPseudoProbe *LastProbe = nullptr;
if (auto *S =
Ctx.getObjectFileInfo()->getPseudoProbeSection(ProbeSec.first)) {
// Switch to the .pseudoprobe section or a comdat group.
MCOS->SwitchSection(S);
// Emit probes grouped by GUID.
ProbeSec.second.emit(MCOS, LastProbe);
}
}
}
//
// This emits the pseudo probe tables.
//
void MCPseudoProbeTable::emit(MCObjectStreamer *MCOS) {
MCContext &Ctx = MCOS->getContext();
auto &ProbeTable = Ctx.getMCPseudoProbeTable();
// Bail out early so we don't switch to the pseudo_probe section needlessly
// and in doing so create an unnecessary (if empty) section.
auto &ProbeSections = ProbeTable.getProbeSections();
if (ProbeSections.empty())
return;
LLVM_DEBUG(MCPseudoProbeTable::DdgPrintIndent = 0);
// Put out the probe.
ProbeSections.emit(MCOS);
}