Expose basic coverage information to userspace through kcov interface.

In Linux, a kernel configuration is set that compiles the kernel with a
custom function that is called at the beginning of every basic block, which
updates the memory-mapped coverage information. The Go coverage tool does not
allow us to inject arbitrary instructions into basic blocks, but it does
provide data that we can convert to a kcov-like format and transfer them to
userspace through a memory mapping.

Note that this is not a strict implementation of kcov, which is especially
tricky to do because we do not have the same coverage tools available in Go
that that are available for the actual Linux kernel. In Linux, a kernel
configuration is set that compiles the kernel with a custom function that is
called at the beginning of every basic block to write program counters to the
kcov memory mapping. In Go, however, coverage tools only give us a count of
basic blocks as they are executed. Every time we return to userspace, we
collect the coverage information and write out PCs for each block that was
executed, providing userspace with the illusion that the kcov data is always
up to date. For convenience, we also generate a unique synthetic PC for each
block instead of using actual PCs. Finally, we do not provide thread-specific
coverage data (each kcov instance only contains PCs executed by the thread
owning it); instead, we will supply data for any file specified by --
instrumentation_filter.

Also, fix issue in nogo that was causing pkg/coverage:coverage_nogo
compilation to fail.

PiperOrigin-RevId: 328426526

2 files changed, 189 insertions, 0 deletions

diff --git a/pkg/coverage/BUILD b/pkg/coverage/BUILD
new file mode 100644
index 000000000..a198e8028
--- /dev/null
+++ b/pkg/coverage/BUILD
@@ -0,0 +1,14 @@
+load("//tools:defs.bzl", "go_library")
+
+package(licenses = ["notice"])
+
+go_library(
+    name = "coverage",
+    srcs = ["coverage.go"],
+    visibility = ["//:sandbox"],
+    deps = [
+        "//pkg/sync",
+        "//pkg/usermem",
+        "@io_bazel_rules_go//go/tools/coverdata",
+    ],
+)
diff --git a/pkg/coverage/coverage.go b/pkg/coverage/coverage.go
new file mode 100644
index 000000000..6831adcce
--- /dev/null
+++ b/pkg/coverage/coverage.go
@@ -0,0 +1,175 @@
+// Copyright 2020 The gVisor Authors.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+// Package coverage provides an interface through which Go coverage data can
+// be collected, converted to kcov format, and exposed to userspace.
+//
+// Coverage can be enabled by calling bazel {build,test} with
+// --collect_coverage_data and --instrumentation_filter with the desired
+// coverage surface. This causes bazel to use the Go cover tool manually to
+// generate instrumented files. It injects a hook that registers all coverage
+// data with the coverdata package.
+package coverage
+
+import (
+	"fmt"
+	"io"
+	"sort"
+	"sync/atomic"
+
+	"gvisor.dev/gvisor/pkg/sync"
+	"gvisor.dev/gvisor/pkg/usermem"
+
+	"github.com/bazelbuild/rules_go/go/tools/coverdata"
+)
+
+// KcovAvailable returns whether the kcov coverage interface is available. It is
+// available as long as coverage is enabled for some files.
+func KcovAvailable() bool {
+	return len(coverdata.Cover.Blocks) > 0
+}
+
+// coverageMu must be held while accessing coverdata.Cover. This prevents
+// concurrent reads/writes from multiple threads collecting coverage data.
+var coverageMu sync.RWMutex
+
+// once ensures that globalData is only initialized once.
+var once sync.Once
+
+var globalData struct {
+	// files is the set of covered files sorted by filename. It is calculated at
+	// startup.
+	files []string
+
+	// syntheticPCs are a set of PCs calculated at startup, where the PC
+	// at syntheticPCs[i][j] corresponds to file i, block j.
+	syntheticPCs [][]uint64
+}
+
+// ClearCoverageData clears existing coverage data.
+func ClearCoverageData() {
+	coverageMu.Lock()
+	defer coverageMu.Unlock()
+	for _, counters := range coverdata.Cover.Counters {
+		for index := 0; index < len(counters); index++ {
+			atomic.StoreUint32(&counters[index], 0)
+		}
+	}
+}
+
+var coveragePool = sync.Pool{
+	New: func() interface{} {
+		return make([]byte, 0)
+	},
+}
+
+// ConsumeCoverageData builds and writes the collection of covered PCs. It
+// returns the number of bytes written.
+//
+// In Linux, a kernel configuration is set that compiles the kernel with a
+// custom function that is called at the beginning of every basic block, which
+// updates the memory-mapped coverage information. The Go coverage tool does not
+// allow us to inject arbitrary instructions into basic blocks, but it does
+// provide data that we can convert to a kcov-like format and transfer them to
+// userspace through a memory mapping.
+//
+// Note that this is not a strict implementation of kcov, which is especially
+// tricky to do because we do not have the same coverage tools available in Go
+// that that are available for the actual Linux kernel. In Linux, a kernel
+// configuration is set that compiles the kernel with a custom function that is
+// called at the beginning of every basic block to write program counters to the
+// kcov memory mapping. In Go, however, coverage tools only give us a count of
+// basic blocks as they are executed. Every time we return to userspace, we
+// collect the coverage information and write out PCs for each block that was
+// executed, providing userspace with the illusion that the kcov data is always
+// up to date. For convenience, we also generate a unique synthetic PC for each
+// block instead of using actual PCs. Finally, we do not provide thread-specific
+// coverage data (each kcov instance only contains PCs executed by the thread
+// owning it); instead, we will supply data for any file specified by --
+// instrumentation_filter.
+//
+// Note that we "consume", i.e. clear, coverdata when this function is run, to
+// ensure that each event is only reported once.
+//
+// TODO(b/160639712): evaluate whether it is ok to reset the global coverage
+// data every time this function is run. We could technically have each thread
+// store a local snapshot against which we compare the most recent coverdata so
+// that separate threads do not affect each other's view of the data.
+func ConsumeCoverageData(w io.Writer) int {
+	once.Do(initCoverageData)
+
+	coverageMu.Lock()
+	defer coverageMu.Unlock()
+
+	total := 0
+	var pcBuffer [8]byte
+	for fileIndex, file := range globalData.files {
+		counters := coverdata.Cover.Counters[file]
+		for index := 0; index < len(counters); index++ {
+			val := atomic.SwapUint32(&counters[index], 0)
+			if val != 0 {
+				// Calculate the synthetic PC.
+				pc := globalData.syntheticPCs[fileIndex][index]
+
+				usermem.ByteOrder.PutUint64(pcBuffer[:], pc)
+				n, err := w.Write(pcBuffer[:])
+				if err != nil {
+					if err == io.EOF {
+						// Simply stop writing if we encounter EOF; it's ok if we attempted to
+						// write more than we can hold.
+						return total + n
+					}
+					panic(fmt.Sprintf("Internal error writing PCs to kcov area: %v", err))
+				}
+				total += n
+			}
+		}
+	}
+
+	if total == 0 {
+		// An empty profile indicates that coverage is not enabled, in which case
+		// there shouldn't be any task work registered.
+		panic("kcov task work is registered, but no coverage data was found")
+	}
+	return total
+}
+
+// initCoverageData initializes globalData. It should only be called once,
+// before any kcov data is written.
+func initCoverageData() {
+	// First, order all files. Then calculate synthetic PCs for every block
+	// (using the well-defined ordering for files as well).
+	for file := range coverdata.Cover.Blocks {
+		globalData.files = append(globalData.files, file)
+	}
+	sort.Strings(globalData.files)
+
+	// nextSyntheticPC is the first PC that we generate for a block.
+	//
+	// This uses a standard-looking kernel range for simplicity.
+	//
+	// FIXME(b/160639712): This is only necessary because syzkaller requires
+	// addresses in the kernel range. If we can remove this constraint, then we
+	// should be able to use the actual addresses.
+	var nextSyntheticPC uint64 = 0xffffffff80000000
+	for _, file := range globalData.files {
+		blocks := coverdata.Cover.Blocks[file]
+		thisFile := make([]uint64, 0, len(blocks))
+		for range blocks {
+			thisFile = append(thisFile, nextSyntheticPC)
+			nextSyntheticPC++ // Advance.
+		}
+		globalData.syntheticPCs = append(globalData.syntheticPCs, thisFile)
+	}
+}