summaryrefslogtreecommitdiffhomepage
path: root/pkg/sentry/kernel
diff options
context:
space:
mode:
Diffstat (limited to 'pkg/sentry/kernel')
-rw-r--r--pkg/sentry/kernel/BUILD247
-rw-r--r--pkg/sentry/kernel/README.md108
-rw-r--r--pkg/sentry/kernel/auth/BUILD70
-rwxr-xr-xpkg/sentry/kernel/auth/atomicptr_credentials_unsafe.go37
-rwxr-xr-xpkg/sentry/kernel/auth/auth_state_autogen.go164
-rwxr-xr-xpkg/sentry/kernel/auth/id_map_range.go62
-rwxr-xr-xpkg/sentry/kernel/auth/id_map_set.go1270
-rw-r--r--pkg/sentry/kernel/contexttest/BUILD18
-rw-r--r--pkg/sentry/kernel/contexttest/contexttest.go40
-rw-r--r--pkg/sentry/kernel/epoll/BUILD52
-rwxr-xr-xpkg/sentry/kernel/epoll/epoll_list.go173
-rwxr-xr-xpkg/sentry/kernel/epoll/epoll_state_autogen.go99
-rw-r--r--pkg/sentry/kernel/epoll/epoll_test.go54
-rw-r--r--pkg/sentry/kernel/eventfd/BUILD35
-rwxr-xr-xpkg/sentry/kernel/eventfd/eventfd_state_autogen.go27
-rw-r--r--pkg/sentry/kernel/eventfd/eventfd_test.go78
-rw-r--r--pkg/sentry/kernel/fasync/BUILD18
-rwxr-xr-xpkg/sentry/kernel/fasync/fasync_state_autogen.go32
-rw-r--r--pkg/sentry/kernel/fd_table_test.go228
-rw-r--r--pkg/sentry/kernel/futex/BUILD59
-rwxr-xr-xpkg/sentry/kernel/futex/atomicptr_bucket_unsafe.go37
-rwxr-xr-xpkg/sentry/kernel/futex/futex_state_autogen.go75
-rw-r--r--pkg/sentry/kernel/futex/futex_test.go530
-rwxr-xr-xpkg/sentry/kernel/futex/waiter_list.go173
-rw-r--r--pkg/sentry/kernel/g3doc/run_states.dot99
-rw-r--r--pkg/sentry/kernel/g3doc/run_states.pngbin234152 -> 0 bytes
-rwxr-xr-xpkg/sentry/kernel/kernel_state_autogen.go1210
-rw-r--r--pkg/sentry/kernel/memevent/BUILD40
-rwxr-xr-xpkg/sentry/kernel/memevent/memevent_state_autogen.go4
-rwxr-xr-x[-rw-r--r--]pkg/sentry/kernel/memevent/memory_events.go0
-rw-r--r--pkg/sentry/kernel/memevent/memory_events.proto29
-rwxr-xr-xpkg/sentry/kernel/memevent/memory_events_go_proto/memory_events.pb.go88
-rwxr-xr-xpkg/sentry/kernel/pending_signals_list.go173
-rw-r--r--pkg/sentry/kernel/pipe/BUILD69
-rwxr-xr-xpkg/sentry/kernel/pipe/buffer_list.go173
-rw-r--r--pkg/sentry/kernel/pipe/buffer_test.go32
-rw-r--r--pkg/sentry/kernel/pipe/node_test.go320
-rwxr-xr-xpkg/sentry/kernel/pipe/pipe_state_autogen.go132
-rw-r--r--pkg/sentry/kernel/pipe/pipe_test.go139
-rwxr-xr-x[-rw-r--r--]pkg/sentry/kernel/pipe/pipe_util.go0
-rwxr-xr-x[-rw-r--r--]pkg/sentry/kernel/pipe/vfs.go0
-rwxr-xr-xpkg/sentry/kernel/process_group_list.go173
-rw-r--r--pkg/sentry/kernel/sched/BUILD21
-rw-r--r--pkg/sentry/kernel/sched/cpuset_test.go44
-rwxr-xr-xpkg/sentry/kernel/sched/sched_state_autogen.go4
-rw-r--r--pkg/sentry/kernel/semaphore/BUILD51
-rwxr-xr-xpkg/sentry/kernel/semaphore/semaphore_state_autogen.go115
-rw-r--r--pkg/sentry/kernel/semaphore/semaphore_test.go172
-rwxr-xr-xpkg/sentry/kernel/semaphore/waiter_list.go173
-rwxr-xr-xpkg/sentry/kernel/seqatomic_taskgoroutineschedinfo_unsafe.go55
-rwxr-xr-xpkg/sentry/kernel/session_list.go173
-rw-r--r--pkg/sentry/kernel/shm/BUILD30
-rwxr-xr-xpkg/sentry/kernel/shm/shm_state_autogen.go74
-rw-r--r--pkg/sentry/kernel/signalfd/BUILD23
-rwxr-xr-x[-rw-r--r--]pkg/sentry/kernel/signalfd/signalfd.go0
-rwxr-xr-xpkg/sentry/kernel/signalfd/signalfd_state_autogen.go24
-rwxr-xr-xpkg/sentry/kernel/socket_list.go173
-rw-r--r--pkg/sentry/kernel/table_test.go110
-rwxr-xr-xpkg/sentry/kernel/task_list.go173
-rw-r--r--pkg/sentry/kernel/task_test.go69
-rw-r--r--pkg/sentry/kernel/time/BUILD20
-rwxr-xr-xpkg/sentry/kernel/time/time_state_autogen.go56
-rw-r--r--pkg/sentry/kernel/timekeeper_test.go156
-rwxr-xr-x[-rw-r--r--]pkg/sentry/kernel/tty.go0
-rw-r--r--pkg/sentry/kernel/uncaught_signal.proto37
-rwxr-xr-xpkg/sentry/kernel/uncaught_signal_go_proto/uncaught_signal.pb.go119
66 files changed, 5241 insertions, 2998 deletions
diff --git a/pkg/sentry/kernel/BUILD b/pkg/sentry/kernel/BUILD
deleted file mode 100644
index ac85ba0c8..000000000
--- a/pkg/sentry/kernel/BUILD
+++ /dev/null
@@ -1,247 +0,0 @@
-load("@io_bazel_rules_go//proto:def.bzl", "go_proto_library")
-load("@io_bazel_rules_go//go:def.bzl", "go_test")
-load("@rules_cc//cc:defs.bzl", "cc_proto_library")
-load("//tools/go_generics:defs.bzl", "go_template_instance")
-load("//tools/go_stateify:defs.bzl", "go_library")
-
-package(licenses = ["notice"])
-
-go_template_instance(
- name = "pending_signals_list",
- out = "pending_signals_list.go",
- package = "kernel",
- prefix = "pendingSignal",
- template = "//pkg/ilist:generic_list",
- types = {
- "Element": "*pendingSignal",
- "Linker": "*pendingSignal",
- },
-)
-
-go_template_instance(
- name = "process_group_list",
- out = "process_group_list.go",
- package = "kernel",
- prefix = "processGroup",
- template = "//pkg/ilist:generic_list",
- types = {
- "Element": "*ProcessGroup",
- "Linker": "*ProcessGroup",
- },
-)
-
-go_template_instance(
- name = "seqatomic_taskgoroutineschedinfo",
- out = "seqatomic_taskgoroutineschedinfo_unsafe.go",
- package = "kernel",
- suffix = "TaskGoroutineSchedInfo",
- template = "//pkg/sync:generic_seqatomic",
- types = {
- "Value": "TaskGoroutineSchedInfo",
- },
-)
-
-go_template_instance(
- name = "session_list",
- out = "session_list.go",
- package = "kernel",
- prefix = "session",
- template = "//pkg/ilist:generic_list",
- types = {
- "Element": "*Session",
- "Linker": "*Session",
- },
-)
-
-go_template_instance(
- name = "task_list",
- out = "task_list.go",
- package = "kernel",
- prefix = "task",
- template = "//pkg/ilist:generic_list",
- types = {
- "Element": "*Task",
- "Linker": "*Task",
- },
-)
-
-go_template_instance(
- name = "socket_list",
- out = "socket_list.go",
- package = "kernel",
- prefix = "socket",
- template = "//pkg/ilist:generic_list",
- types = {
- "Element": "*SocketEntry",
- "Linker": "*SocketEntry",
- },
-)
-
-proto_library(
- name = "uncaught_signal_proto",
- srcs = ["uncaught_signal.proto"],
- visibility = ["//visibility:public"],
- deps = ["//pkg/sentry/arch:registers_proto"],
-)
-
-cc_proto_library(
- name = "uncaught_signal_cc_proto",
- visibility = ["//visibility:public"],
- deps = [":uncaught_signal_proto"],
-)
-
-go_proto_library(
- name = "uncaught_signal_go_proto",
- importpath = "gvisor.dev/gvisor/pkg/sentry/kernel/uncaught_signal_go_proto",
- proto = ":uncaught_signal_proto",
- visibility = ["//visibility:public"],
- deps = ["//pkg/sentry/arch:registers_go_proto"],
-)
-
-go_library(
- name = "kernel",
- srcs = [
- "abstract_socket_namespace.go",
- "context.go",
- "fd_table.go",
- "fd_table_unsafe.go",
- "fs_context.go",
- "ipc_namespace.go",
- "kernel.go",
- "kernel_state.go",
- "pending_signals.go",
- "pending_signals_list.go",
- "pending_signals_state.go",
- "posixtimer.go",
- "process_group_list.go",
- "ptrace.go",
- "ptrace_amd64.go",
- "ptrace_arm64.go",
- "rseq.go",
- "seccomp.go",
- "seqatomic_taskgoroutineschedinfo_unsafe.go",
- "session_list.go",
- "sessions.go",
- "signal.go",
- "signal_handlers.go",
- "socket_list.go",
- "syscalls.go",
- "syscalls_state.go",
- "syslog.go",
- "task.go",
- "task_acct.go",
- "task_block.go",
- "task_clone.go",
- "task_context.go",
- "task_exec.go",
- "task_exit.go",
- "task_futex.go",
- "task_identity.go",
- "task_list.go",
- "task_log.go",
- "task_net.go",
- "task_run.go",
- "task_sched.go",
- "task_signals.go",
- "task_start.go",
- "task_stop.go",
- "task_syscall.go",
- "task_usermem.go",
- "thread_group.go",
- "threads.go",
- "timekeeper.go",
- "timekeeper_state.go",
- "tty.go",
- "uts_namespace.go",
- "vdso.go",
- "version.go",
- ],
- importpath = "gvisor.dev/gvisor/pkg/sentry/kernel",
- imports = [
- "gvisor.dev/gvisor/pkg/bpf",
- "gvisor.dev/gvisor/pkg/sentry/device",
- "gvisor.dev/gvisor/pkg/tcpip",
- ],
- visibility = ["//:sandbox"],
- deps = [
- ":uncaught_signal_go_proto",
- "//pkg/abi",
- "//pkg/abi/linux",
- "//pkg/amutex",
- "//pkg/binary",
- "//pkg/bits",
- "//pkg/bpf",
- "//pkg/cpuid",
- "//pkg/eventchannel",
- "//pkg/log",
- "//pkg/metric",
- "//pkg/refs",
- "//pkg/secio",
- "//pkg/sentry/arch",
- "//pkg/sentry/context",
- "//pkg/sentry/device",
- "//pkg/sentry/fs",
- "//pkg/sentry/fs/lock",
- "//pkg/sentry/fs/timerfd",
- "//pkg/sentry/hostcpu",
- "//pkg/sentry/inet",
- "//pkg/sentry/kernel/auth",
- "//pkg/sentry/kernel/epoll",
- "//pkg/sentry/kernel/futex",
- "//pkg/sentry/kernel/sched",
- "//pkg/sentry/kernel/semaphore",
- "//pkg/sentry/kernel/shm",
- "//pkg/sentry/kernel/time",
- "//pkg/sentry/limits",
- "//pkg/sentry/loader",
- "//pkg/sentry/memmap",
- "//pkg/sentry/mm",
- "//pkg/sentry/pgalloc",
- "//pkg/sentry/platform",
- "//pkg/sentry/safemem",
- "//pkg/sentry/socket/netlink/port",
- "//pkg/sentry/socket/unix/transport",
- "//pkg/sentry/time",
- "//pkg/sentry/unimpl",
- "//pkg/sentry/unimpl:unimplemented_syscall_go_proto",
- "//pkg/sentry/uniqueid",
- "//pkg/sentry/usage",
- "//pkg/sentry/usermem",
- "//pkg/state",
- "//pkg/state/statefile",
- "//pkg/sync",
- "//pkg/syserr",
- "//pkg/syserror",
- "//pkg/tcpip",
- "//pkg/tcpip/stack",
- "//pkg/waiter",
- ],
-)
-
-go_test(
- name = "kernel_test",
- size = "small",
- srcs = [
- "fd_table_test.go",
- "table_test.go",
- "task_test.go",
- "timekeeper_test.go",
- ],
- embed = [":kernel"],
- deps = [
- "//pkg/abi",
- "//pkg/sentry/arch",
- "//pkg/sentry/context",
- "//pkg/sentry/context/contexttest",
- "//pkg/sentry/fs",
- "//pkg/sentry/fs/filetest",
- "//pkg/sentry/kernel/sched",
- "//pkg/sentry/limits",
- "//pkg/sentry/pgalloc",
- "//pkg/sentry/time",
- "//pkg/sentry/usage",
- "//pkg/sentry/usermem",
- "//pkg/sync",
- "//pkg/syserror",
- ],
-)
diff --git a/pkg/sentry/kernel/README.md b/pkg/sentry/kernel/README.md
deleted file mode 100644
index 427311be8..000000000
--- a/pkg/sentry/kernel/README.md
+++ /dev/null
@@ -1,108 +0,0 @@
-This package contains:
-
-- A (partial) emulation of the "core Linux kernel", which governs task
- execution and scheduling, system call dispatch, and signal handling. See
- below for details.
-
-- The top-level interface for the sentry's Linux kernel emulation in general,
- used by the `main` function of all versions of the sentry. This interface
- revolves around the `Env` type (defined in `kernel.go`).
-
-# Background
-
-In Linux, each schedulable context is referred to interchangeably as a "task" or
-"thread". Tasks can be divided into userspace and kernel tasks. In the sentry,
-scheduling is managed by the Go runtime, so each schedulable context is a
-goroutine; only "userspace" (application) contexts are referred to as tasks, and
-represented by Task objects. (From this point forward, "task" refers to the
-sentry's notion of a task unless otherwise specified.)
-
-At a high level, Linux application threads can be thought of as repeating a "run
-loop":
-
-- Some amount of application code is executed in userspace.
-
-- A trap (explicit syscall invocation, hardware interrupt or exception, etc.)
- causes control flow to switch to the kernel.
-
-- Some amount of kernel code is executed in kernelspace, e.g. to handle the
- cause of the trap.
-
-- The kernel "returns from the trap" into application code.
-
-Analogously, each task in the sentry is associated with a *task goroutine* that
-executes that task's run loop (`Task.run` in `task_run.go`). However, the
-sentry's task run loop differs in structure in order to support saving execution
-state to, and resuming execution from, checkpoints.
-
-While in kernelspace, a Linux thread can be descheduled (cease execution) in a
-variety of ways:
-
-- It can yield or be preempted, becoming temporarily descheduled but still
- runnable. At present, the sentry delegates scheduling of runnable threads to
- the Go runtime.
-
-- It can exit, becoming permanently descheduled. The sentry's equivalent is
- returning from `Task.run`, terminating the task goroutine.
-
-- It can enter interruptible sleep, a state in which it can be woken by a
- caller-defined wakeup or the receipt of a signal. In the sentry,
- interruptible sleep (which is ambiguously referred to as *blocking*) is
- implemented by making all events that can end blocking (including signal
- notifications) communicated via Go channels and using `select` to multiplex
- wakeup sources; see `task_block.go`.
-
-- It can enter uninterruptible sleep, a state in which it can only be woken by
- a caller-defined wakeup. Killable sleep is a closely related variant in
- which the task can also be woken by SIGKILL. (These definitions also include
- Linux's "group-stopped" (`TASK_STOPPED`) and "ptrace-stopped"
- (`TASK_TRACED`) states.)
-
-To maximize compatibility with Linux, sentry checkpointing appears as a spurious
-signal-delivery interrupt on all tasks; interrupted system calls return `EINTR`
-or are automatically restarted as usual. However, these semantics require that
-uninterruptible and killable sleeps do not appear to be interrupted. In other
-words, the state of the task, including its progress through the interrupted
-operation, must be preserved by checkpointing. For many such sleeps, the wakeup
-condition is application-controlled, making it infeasible to wait for the sleep
-to end before checkpointing. Instead, we must support checkpointing progress
-through sleeping operations.
-
-# Implementation
-
-We break the task's control flow graph into *states*, delimited by:
-
-1. Points where uninterruptible and killable sleeps may occur. For example,
- there exists a state boundary between signal dequeueing and signal delivery
- because there may be an intervening ptrace signal-delivery-stop.
-
-2. Points where sleep-induced branches may "rejoin" normal execution. For
- example, the syscall exit state exists because it can be reached immediately
- following a synchronous syscall, or after a task that is sleeping in
- `execve()` or `vfork()` resumes execution.
-
-3. Points containing large branches. This is strictly for organizational
- purposes. For example, the state that processes interrupt-signaled
- conditions is kept separate from the main "app" state to reduce the size of
- the latter.
-
-4. `SyscallReinvoke`, which does not correspond to anything in Linux, and
- exists solely to serve the autosave feature.
-
-![dot -Tpng -Goverlap=false -orun_states.png run_states.dot](g3doc/run_states.png "Task control flow graph")
-
-States before which a stop may occur are represented as implementations of the
-`taskRunState` interface named `run(state)`, allowing them to be saved and
-restored. States that cannot be immediately preceded by a stop are simply `Task`
-methods named `do(state)`.
-
-Conditions that can require task goroutines to cease execution for unknown
-lengths of time are called *stops*. Stops are divided into *internal stops*,
-which are stops whose start and end conditions are implemented within the
-sentry, and *external stops*, which are stops whose start and end conditions are
-not known to the sentry. Hence all uninterruptible and killable sleeps are
-internal stops, and the existence of a pending checkpoint operation is an
-external stop. Internal stops are reified into instances of the `TaskStop` type,
-while external stops are merely counted. The task run loop alternates between
-checking for stops and advancing the task's state. This allows checkpointing to
-hold tasks in a stopped state while waiting for all tasks in the system to stop.
diff --git a/pkg/sentry/kernel/auth/BUILD b/pkg/sentry/kernel/auth/BUILD
deleted file mode 100644
index 1aa72fa47..000000000
--- a/pkg/sentry/kernel/auth/BUILD
+++ /dev/null
@@ -1,70 +0,0 @@
-load("//tools/go_generics:defs.bzl", "go_template_instance")
-load("//tools/go_stateify:defs.bzl", "go_library")
-
-package(licenses = ["notice"])
-
-go_template_instance(
- name = "atomicptr_credentials",
- out = "atomicptr_credentials_unsafe.go",
- package = "auth",
- suffix = "Credentials",
- template = "//pkg/sync:generic_atomicptr",
- types = {
- "Value": "Credentials",
- },
-)
-
-go_template_instance(
- name = "id_map_range",
- out = "id_map_range.go",
- package = "auth",
- prefix = "idMap",
- template = "//pkg/segment:generic_range",
- types = {
- "T": "uint32",
- },
-)
-
-go_template_instance(
- name = "id_map_set",
- out = "id_map_set.go",
- consts = {
- "minDegree": "3",
- },
- package = "auth",
- prefix = "idMap",
- template = "//pkg/segment:generic_set",
- types = {
- "Key": "uint32",
- "Range": "idMapRange",
- "Value": "uint32",
- "Functions": "idMapFunctions",
- },
-)
-
-go_library(
- name = "auth",
- srcs = [
- "atomicptr_credentials_unsafe.go",
- "auth.go",
- "capability_set.go",
- "context.go",
- "credentials.go",
- "id.go",
- "id_map.go",
- "id_map_functions.go",
- "id_map_range.go",
- "id_map_set.go",
- "user_namespace.go",
- ],
- importpath = "gvisor.dev/gvisor/pkg/sentry/kernel/auth",
- visibility = ["//pkg/sentry:internal"],
- deps = [
- "//pkg/abi/linux",
- "//pkg/bits",
- "//pkg/log",
- "//pkg/sentry/context",
- "//pkg/sync",
- "//pkg/syserror",
- ],
-)
diff --git a/pkg/sentry/kernel/auth/atomicptr_credentials_unsafe.go b/pkg/sentry/kernel/auth/atomicptr_credentials_unsafe.go
new file mode 100755
index 000000000..4535c958f
--- /dev/null
+++ b/pkg/sentry/kernel/auth/atomicptr_credentials_unsafe.go
@@ -0,0 +1,37 @@
+package auth
+
+import (
+ "sync/atomic"
+ "unsafe"
+)
+
+// An AtomicPtr is a pointer to a value of type Value that can be atomically
+// loaded and stored. The zero value of an AtomicPtr represents nil.
+//
+// Note that copying AtomicPtr by value performs a non-atomic read of the
+// stored pointer, which is unsafe if Store() can be called concurrently; in
+// this case, do `dst.Store(src.Load())` instead.
+//
+// +stateify savable
+type AtomicPtrCredentials struct {
+ ptr unsafe.Pointer `state:".(*Credentials)"`
+}
+
+func (p *AtomicPtrCredentials) savePtr() *Credentials {
+ return p.Load()
+}
+
+func (p *AtomicPtrCredentials) loadPtr(v *Credentials) {
+ p.Store(v)
+}
+
+// Load returns the value set by the most recent Store. It returns nil if there
+// has been no previous call to Store.
+func (p *AtomicPtrCredentials) Load() *Credentials {
+ return (*Credentials)(atomic.LoadPointer(&p.ptr))
+}
+
+// Store sets the value returned by Load to x.
+func (p *AtomicPtrCredentials) Store(x *Credentials) {
+ atomic.StorePointer(&p.ptr, (unsafe.Pointer)(x))
+}
diff --git a/pkg/sentry/kernel/auth/auth_state_autogen.go b/pkg/sentry/kernel/auth/auth_state_autogen.go
new file mode 100755
index 000000000..ef5f8d957
--- /dev/null
+++ b/pkg/sentry/kernel/auth/auth_state_autogen.go
@@ -0,0 +1,164 @@
+// automatically generated by stateify.
+
+package auth
+
+import (
+ "gvisor.dev/gvisor/pkg/state"
+)
+
+func (x *AtomicPtrCredentials) beforeSave() {}
+func (x *AtomicPtrCredentials) save(m state.Map) {
+ x.beforeSave()
+ var ptr *Credentials = x.savePtr()
+ m.SaveValue("ptr", ptr)
+}
+
+func (x *AtomicPtrCredentials) afterLoad() {}
+func (x *AtomicPtrCredentials) load(m state.Map) {
+ m.LoadValue("ptr", new(*Credentials), func(y interface{}) { x.loadPtr(y.(*Credentials)) })
+}
+
+func (x *Credentials) beforeSave() {}
+func (x *Credentials) save(m state.Map) {
+ x.beforeSave()
+ m.Save("RealKUID", &x.RealKUID)
+ m.Save("EffectiveKUID", &x.EffectiveKUID)
+ m.Save("SavedKUID", &x.SavedKUID)
+ m.Save("RealKGID", &x.RealKGID)
+ m.Save("EffectiveKGID", &x.EffectiveKGID)
+ m.Save("SavedKGID", &x.SavedKGID)
+ m.Save("ExtraKGIDs", &x.ExtraKGIDs)
+ m.Save("PermittedCaps", &x.PermittedCaps)
+ m.Save("InheritableCaps", &x.InheritableCaps)
+ m.Save("EffectiveCaps", &x.EffectiveCaps)
+ m.Save("BoundingCaps", &x.BoundingCaps)
+ m.Save("KeepCaps", &x.KeepCaps)
+ m.Save("UserNamespace", &x.UserNamespace)
+}
+
+func (x *Credentials) afterLoad() {}
+func (x *Credentials) load(m state.Map) {
+ m.Load("RealKUID", &x.RealKUID)
+ m.Load("EffectiveKUID", &x.EffectiveKUID)
+ m.Load("SavedKUID", &x.SavedKUID)
+ m.Load("RealKGID", &x.RealKGID)
+ m.Load("EffectiveKGID", &x.EffectiveKGID)
+ m.Load("SavedKGID", &x.SavedKGID)
+ m.Load("ExtraKGIDs", &x.ExtraKGIDs)
+ m.Load("PermittedCaps", &x.PermittedCaps)
+ m.Load("InheritableCaps", &x.InheritableCaps)
+ m.Load("EffectiveCaps", &x.EffectiveCaps)
+ m.Load("BoundingCaps", &x.BoundingCaps)
+ m.Load("KeepCaps", &x.KeepCaps)
+ m.Load("UserNamespace", &x.UserNamespace)
+}
+
+func (x *IDMapEntry) beforeSave() {}
+func (x *IDMapEntry) save(m state.Map) {
+ x.beforeSave()
+ m.Save("FirstID", &x.FirstID)
+ m.Save("FirstParentID", &x.FirstParentID)
+ m.Save("Length", &x.Length)
+}
+
+func (x *IDMapEntry) afterLoad() {}
+func (x *IDMapEntry) load(m state.Map) {
+ m.Load("FirstID", &x.FirstID)
+ m.Load("FirstParentID", &x.FirstParentID)
+ m.Load("Length", &x.Length)
+}
+
+func (x *idMapRange) beforeSave() {}
+func (x *idMapRange) save(m state.Map) {
+ x.beforeSave()
+ m.Save("Start", &x.Start)
+ m.Save("End", &x.End)
+}
+
+func (x *idMapRange) afterLoad() {}
+func (x *idMapRange) load(m state.Map) {
+ m.Load("Start", &x.Start)
+ m.Load("End", &x.End)
+}
+
+func (x *idMapSet) beforeSave() {}
+func (x *idMapSet) save(m state.Map) {
+ x.beforeSave()
+ var root *idMapSegmentDataSlices = x.saveRoot()
+ m.SaveValue("root", root)
+}
+
+func (x *idMapSet) afterLoad() {}
+func (x *idMapSet) load(m state.Map) {
+ m.LoadValue("root", new(*idMapSegmentDataSlices), func(y interface{}) { x.loadRoot(y.(*idMapSegmentDataSlices)) })
+}
+
+func (x *idMapnode) beforeSave() {}
+func (x *idMapnode) save(m state.Map) {
+ x.beforeSave()
+ m.Save("nrSegments", &x.nrSegments)
+ m.Save("parent", &x.parent)
+ m.Save("parentIndex", &x.parentIndex)
+ m.Save("hasChildren", &x.hasChildren)
+ m.Save("keys", &x.keys)
+ m.Save("values", &x.values)
+ m.Save("children", &x.children)
+}
+
+func (x *idMapnode) afterLoad() {}
+func (x *idMapnode) load(m state.Map) {
+ m.Load("nrSegments", &x.nrSegments)
+ m.Load("parent", &x.parent)
+ m.Load("parentIndex", &x.parentIndex)
+ m.Load("hasChildren", &x.hasChildren)
+ m.Load("keys", &x.keys)
+ m.Load("values", &x.values)
+ m.Load("children", &x.children)
+}
+
+func (x *idMapSegmentDataSlices) beforeSave() {}
+func (x *idMapSegmentDataSlices) save(m state.Map) {
+ x.beforeSave()
+ m.Save("Start", &x.Start)
+ m.Save("End", &x.End)
+ m.Save("Values", &x.Values)
+}
+
+func (x *idMapSegmentDataSlices) afterLoad() {}
+func (x *idMapSegmentDataSlices) load(m state.Map) {
+ m.Load("Start", &x.Start)
+ m.Load("End", &x.End)
+ m.Load("Values", &x.Values)
+}
+
+func (x *UserNamespace) beforeSave() {}
+func (x *UserNamespace) save(m state.Map) {
+ x.beforeSave()
+ m.Save("parent", &x.parent)
+ m.Save("owner", &x.owner)
+ m.Save("uidMapFromParent", &x.uidMapFromParent)
+ m.Save("uidMapToParent", &x.uidMapToParent)
+ m.Save("gidMapFromParent", &x.gidMapFromParent)
+ m.Save("gidMapToParent", &x.gidMapToParent)
+}
+
+func (x *UserNamespace) afterLoad() {}
+func (x *UserNamespace) load(m state.Map) {
+ m.Load("parent", &x.parent)
+ m.Load("owner", &x.owner)
+ m.Load("uidMapFromParent", &x.uidMapFromParent)
+ m.Load("uidMapToParent", &x.uidMapToParent)
+ m.Load("gidMapFromParent", &x.gidMapFromParent)
+ m.Load("gidMapToParent", &x.gidMapToParent)
+}
+
+func init() {
+ state.Register("auth.AtomicPtrCredentials", (*AtomicPtrCredentials)(nil), state.Fns{Save: (*AtomicPtrCredentials).save, Load: (*AtomicPtrCredentials).load})
+ state.Register("auth.Credentials", (*Credentials)(nil), state.Fns{Save: (*Credentials).save, Load: (*Credentials).load})
+ state.Register("auth.IDMapEntry", (*IDMapEntry)(nil), state.Fns{Save: (*IDMapEntry).save, Load: (*IDMapEntry).load})
+ state.Register("auth.idMapRange", (*idMapRange)(nil), state.Fns{Save: (*idMapRange).save, Load: (*idMapRange).load})
+ state.Register("auth.idMapSet", (*idMapSet)(nil), state.Fns{Save: (*idMapSet).save, Load: (*idMapSet).load})
+ state.Register("auth.idMapnode", (*idMapnode)(nil), state.Fns{Save: (*idMapnode).save, Load: (*idMapnode).load})
+ state.Register("auth.idMapSegmentDataSlices", (*idMapSegmentDataSlices)(nil), state.Fns{Save: (*idMapSegmentDataSlices).save, Load: (*idMapSegmentDataSlices).load})
+ state.Register("auth.UserNamespace", (*UserNamespace)(nil), state.Fns{Save: (*UserNamespace).save, Load: (*UserNamespace).load})
+}
diff --git a/pkg/sentry/kernel/auth/id_map_range.go b/pkg/sentry/kernel/auth/id_map_range.go
new file mode 100755
index 000000000..833fa3518
--- /dev/null
+++ b/pkg/sentry/kernel/auth/id_map_range.go
@@ -0,0 +1,62 @@
+package auth
+
+// A Range represents a contiguous range of T.
+//
+// +stateify savable
+type idMapRange struct {
+ // Start is the inclusive start of the range.
+ Start uint32
+
+ // End is the exclusive end of the range.
+ End uint32
+}
+
+// WellFormed returns true if r.Start <= r.End. All other methods on a Range
+// require that the Range is well-formed.
+func (r idMapRange) WellFormed() bool {
+ return r.Start <= r.End
+}
+
+// Length returns the length of the range.
+func (r idMapRange) Length() uint32 {
+ return r.End - r.Start
+}
+
+// Contains returns true if r contains x.
+func (r idMapRange) Contains(x uint32) bool {
+ return r.Start <= x && x < r.End
+}
+
+// Overlaps returns true if r and r2 overlap.
+func (r idMapRange) Overlaps(r2 idMapRange) bool {
+ return r.Start < r2.End && r2.Start < r.End
+}
+
+// IsSupersetOf returns true if r is a superset of r2; that is, the range r2 is
+// contained within r.
+func (r idMapRange) IsSupersetOf(r2 idMapRange) bool {
+ return r.Start <= r2.Start && r.End >= r2.End
+}
+
+// Intersect returns a range consisting of the intersection between r and r2.
+// If r and r2 do not overlap, Intersect returns a range with unspecified
+// bounds, but for which Length() == 0.
+func (r idMapRange) Intersect(r2 idMapRange) idMapRange {
+ if r.Start < r2.Start {
+ r.Start = r2.Start
+ }
+ if r.End > r2.End {
+ r.End = r2.End
+ }
+ if r.End < r.Start {
+ r.End = r.Start
+ }
+ return r
+}
+
+// CanSplitAt returns true if it is legal to split a segment spanning the range
+// r at x; that is, splitting at x would produce two ranges, both of which have
+// non-zero length.
+func (r idMapRange) CanSplitAt(x uint32) bool {
+ return r.Contains(x) && r.Start < x
+}
diff --git a/pkg/sentry/kernel/auth/id_map_set.go b/pkg/sentry/kernel/auth/id_map_set.go
new file mode 100755
index 000000000..73a17f281
--- /dev/null
+++ b/pkg/sentry/kernel/auth/id_map_set.go
@@ -0,0 +1,1270 @@
+package auth
+
+import (
+ "bytes"
+ "fmt"
+)
+
+const (
+ // minDegree is the minimum degree of an internal node in a Set B-tree.
+ //
+ // - Any non-root node has at least minDegree-1 segments.
+ //
+ // - Any non-root internal (non-leaf) node has at least minDegree children.
+ //
+ // - The root node may have fewer than minDegree-1 segments, but it may
+ // only have 0 segments if the tree is empty.
+ //
+ // Our implementation requires minDegree >= 3. Higher values of minDegree
+ // usually improve performance, but increase memory usage for small sets.
+ idMapminDegree = 3
+
+ idMapmaxDegree = 2 * idMapminDegree
+)
+
+// A Set is a mapping of segments with non-overlapping Range keys. The zero
+// value for a Set is an empty set. Set values are not safely movable nor
+// copyable. Set is thread-compatible.
+//
+// +stateify savable
+type idMapSet struct {
+ root idMapnode `state:".(*idMapSegmentDataSlices)"`
+}
+
+// IsEmpty returns true if the set contains no segments.
+func (s *idMapSet) IsEmpty() bool {
+ return s.root.nrSegments == 0
+}
+
+// IsEmptyRange returns true iff no segments in the set overlap the given
+// range. This is semantically equivalent to s.SpanRange(r) == 0, but may be
+// more efficient.
+func (s *idMapSet) IsEmptyRange(r idMapRange) bool {
+ switch {
+ case r.Length() < 0:
+ panic(fmt.Sprintf("invalid range %v", r))
+ case r.Length() == 0:
+ return true
+ }
+ _, gap := s.Find(r.Start)
+ if !gap.Ok() {
+ return false
+ }
+ return r.End <= gap.End()
+}
+
+// Span returns the total size of all segments in the set.
+func (s *idMapSet) Span() uint32 {
+ var sz uint32
+ for seg := s.FirstSegment(); seg.Ok(); seg = seg.NextSegment() {
+ sz += seg.Range().Length()
+ }
+ return sz
+}
+
+// SpanRange returns the total size of the intersection of segments in the set
+// with the given range.
+func (s *idMapSet) SpanRange(r idMapRange) uint32 {
+ switch {
+ case r.Length() < 0:
+ panic(fmt.Sprintf("invalid range %v", r))
+ case r.Length() == 0:
+ return 0
+ }
+ var sz uint32
+ for seg := s.LowerBoundSegment(r.Start); seg.Ok() && seg.Start() < r.End; seg = seg.NextSegment() {
+ sz += seg.Range().Intersect(r).Length()
+ }
+ return sz
+}
+
+// FirstSegment returns the first segment in the set. If the set is empty,
+// FirstSegment returns a terminal iterator.
+func (s *idMapSet) FirstSegment() idMapIterator {
+ if s.root.nrSegments == 0 {
+ return idMapIterator{}
+ }
+ return s.root.firstSegment()
+}
+
+// LastSegment returns the last segment in the set. If the set is empty,
+// LastSegment returns a terminal iterator.
+func (s *idMapSet) LastSegment() idMapIterator {
+ if s.root.nrSegments == 0 {
+ return idMapIterator{}
+ }
+ return s.root.lastSegment()
+}
+
+// FirstGap returns the first gap in the set.
+func (s *idMapSet) FirstGap() idMapGapIterator {
+ n := &s.root
+ for n.hasChildren {
+ n = n.children[0]
+ }
+ return idMapGapIterator{n, 0}
+}
+
+// LastGap returns the last gap in the set.
+func (s *idMapSet) LastGap() idMapGapIterator {
+ n := &s.root
+ for n.hasChildren {
+ n = n.children[n.nrSegments]
+ }
+ return idMapGapIterator{n, n.nrSegments}
+}
+
+// Find returns the segment or gap whose range contains the given key. If a
+// segment is found, the returned Iterator is non-terminal and the
+// returned GapIterator is terminal. Otherwise, the returned Iterator is
+// terminal and the returned GapIterator is non-terminal.
+func (s *idMapSet) Find(key uint32) (idMapIterator, idMapGapIterator) {
+ n := &s.root
+ for {
+
+ lower := 0
+ upper := n.nrSegments
+ for lower < upper {
+ i := lower + (upper-lower)/2
+ if r := n.keys[i]; key < r.End {
+ if key >= r.Start {
+ return idMapIterator{n, i}, idMapGapIterator{}
+ }
+ upper = i
+ } else {
+ lower = i + 1
+ }
+ }
+ i := lower
+ if !n.hasChildren {
+ return idMapIterator{}, idMapGapIterator{n, i}
+ }
+ n = n.children[i]
+ }
+}
+
+// FindSegment returns the segment whose range contains the given key. If no
+// such segment exists, FindSegment returns a terminal iterator.
+func (s *idMapSet) FindSegment(key uint32) idMapIterator {
+ seg, _ := s.Find(key)
+ return seg
+}
+
+// LowerBoundSegment returns the segment with the lowest range that contains a
+// key greater than or equal to min. If no such segment exists,
+// LowerBoundSegment returns a terminal iterator.
+func (s *idMapSet) LowerBoundSegment(min uint32) idMapIterator {
+ seg, gap := s.Find(min)
+ if seg.Ok() {
+ return seg
+ }
+ return gap.NextSegment()
+}
+
+// UpperBoundSegment returns the segment with the highest range that contains a
+// key less than or equal to max. If no such segment exists, UpperBoundSegment
+// returns a terminal iterator.
+func (s *idMapSet) UpperBoundSegment(max uint32) idMapIterator {
+ seg, gap := s.Find(max)
+ if seg.Ok() {
+ return seg
+ }
+ return gap.PrevSegment()
+}
+
+// FindGap returns the gap containing the given key. If no such gap exists
+// (i.e. the set contains a segment containing that key), FindGap returns a
+// terminal iterator.
+func (s *idMapSet) FindGap(key uint32) idMapGapIterator {
+ _, gap := s.Find(key)
+ return gap
+}
+
+// LowerBoundGap returns the gap with the lowest range that is greater than or
+// equal to min.
+func (s *idMapSet) LowerBoundGap(min uint32) idMapGapIterator {
+ seg, gap := s.Find(min)
+ if gap.Ok() {
+ return gap
+ }
+ return seg.NextGap()
+}
+
+// UpperBoundGap returns the gap with the highest range that is less than or
+// equal to max.
+func (s *idMapSet) UpperBoundGap(max uint32) idMapGapIterator {
+ seg, gap := s.Find(max)
+ if gap.Ok() {
+ return gap
+ }
+ return seg.PrevGap()
+}
+
+// Add inserts the given segment into the set and returns true. If the new
+// segment can be merged with adjacent segments, Add will do so. If the new
+// segment would overlap an existing segment, Add returns false. If Add
+// succeeds, all existing iterators are invalidated.
+func (s *idMapSet) Add(r idMapRange, val uint32) bool {
+ if r.Length() <= 0 {
+ panic(fmt.Sprintf("invalid segment range %v", r))
+ }
+ gap := s.FindGap(r.Start)
+ if !gap.Ok() {
+ return false
+ }
+ if r.End > gap.End() {
+ return false
+ }
+ s.Insert(gap, r, val)
+ return true
+}
+
+// AddWithoutMerging inserts the given segment into the set and returns true.
+// If it would overlap an existing segment, AddWithoutMerging does nothing and
+// returns false. If AddWithoutMerging succeeds, all existing iterators are
+// invalidated.
+func (s *idMapSet) AddWithoutMerging(r idMapRange, val uint32) bool {
+ if r.Length() <= 0 {
+ panic(fmt.Sprintf("invalid segment range %v", r))
+ }
+ gap := s.FindGap(r.Start)
+ if !gap.Ok() {
+ return false
+ }
+ if r.End > gap.End() {
+ return false
+ }
+ s.InsertWithoutMergingUnchecked(gap, r, val)
+ return true
+}
+
+// Insert inserts the given segment into the given gap. If the new segment can
+// be merged with adjacent segments, Insert will do so. Insert returns an
+// iterator to the segment containing the inserted value (which may have been
+// merged with other values). All existing iterators (including gap, but not
+// including the returned iterator) are invalidated.
+//
+// If the gap cannot accommodate the segment, or if r is invalid, Insert panics.
+//
+// Insert is semantically equivalent to a InsertWithoutMerging followed by a
+// Merge, but may be more efficient. Note that there is no unchecked variant of
+// Insert since Insert must retrieve and inspect gap's predecessor and
+// successor segments regardless.
+func (s *idMapSet) Insert(gap idMapGapIterator, r idMapRange, val uint32) idMapIterator {
+ if r.Length() <= 0 {
+ panic(fmt.Sprintf("invalid segment range %v", r))
+ }
+ prev, next := gap.PrevSegment(), gap.NextSegment()
+ if prev.Ok() && prev.End() > r.Start {
+ panic(fmt.Sprintf("new segment %v overlaps predecessor %v", r, prev.Range()))
+ }
+ if next.Ok() && next.Start() < r.End {
+ panic(fmt.Sprintf("new segment %v overlaps successor %v", r, next.Range()))
+ }
+ if prev.Ok() && prev.End() == r.Start {
+ if mval, ok := (idMapFunctions{}).Merge(prev.Range(), prev.Value(), r, val); ok {
+ prev.SetEndUnchecked(r.End)
+ prev.SetValue(mval)
+ if next.Ok() && next.Start() == r.End {
+ val = mval
+ if mval, ok := (idMapFunctions{}).Merge(prev.Range(), val, next.Range(), next.Value()); ok {
+ prev.SetEndUnchecked(next.End())
+ prev.SetValue(mval)
+ return s.Remove(next).PrevSegment()
+ }
+ }
+ return prev
+ }
+ }
+ if next.Ok() && next.Start() == r.End {
+ if mval, ok := (idMapFunctions{}).Merge(r, val, next.Range(), next.Value()); ok {
+ next.SetStartUnchecked(r.Start)
+ next.SetValue(mval)
+ return next
+ }
+ }
+ return s.InsertWithoutMergingUnchecked(gap, r, val)
+}
+
+// InsertWithoutMerging inserts the given segment into the given gap and
+// returns an iterator to the inserted segment. All existing iterators
+// (including gap, but not including the returned iterator) are invalidated.
+//
+// If the gap cannot accommodate the segment, or if r is invalid,
+// InsertWithoutMerging panics.
+func (s *idMapSet) InsertWithoutMerging(gap idMapGapIterator, r idMapRange, val uint32) idMapIterator {
+ if r.Length() <= 0 {
+ panic(fmt.Sprintf("invalid segment range %v", r))
+ }
+ if gr := gap.Range(); !gr.IsSupersetOf(r) {
+ panic(fmt.Sprintf("cannot insert segment range %v into gap range %v", r, gr))
+ }
+ return s.InsertWithoutMergingUnchecked(gap, r, val)
+}
+
+// InsertWithoutMergingUnchecked inserts the given segment into the given gap
+// and returns an iterator to the inserted segment. All existing iterators
+// (including gap, but not including the returned iterator) are invalidated.
+//
+// Preconditions: r.Start >= gap.Start(); r.End <= gap.End().
+func (s *idMapSet) InsertWithoutMergingUnchecked(gap idMapGapIterator, r idMapRange, val uint32) idMapIterator {
+ gap = gap.node.rebalanceBeforeInsert(gap)
+ copy(gap.node.keys[gap.index+1:], gap.node.keys[gap.index:gap.node.nrSegments])
+ copy(gap.node.values[gap.index+1:], gap.node.values[gap.index:gap.node.nrSegments])
+ gap.node.keys[gap.index] = r
+ gap.node.values[gap.index] = val
+ gap.node.nrSegments++
+ return idMapIterator{gap.node, gap.index}
+}
+
+// Remove removes the given segment and returns an iterator to the vacated gap.
+// All existing iterators (including seg, but not including the returned
+// iterator) are invalidated.
+func (s *idMapSet) Remove(seg idMapIterator) idMapGapIterator {
+
+ if seg.node.hasChildren {
+
+ victim := seg.PrevSegment()
+
+ seg.SetRangeUnchecked(victim.Range())
+ seg.SetValue(victim.Value())
+ return s.Remove(victim).NextGap()
+ }
+ copy(seg.node.keys[seg.index:], seg.node.keys[seg.index+1:seg.node.nrSegments])
+ copy(seg.node.values[seg.index:], seg.node.values[seg.index+1:seg.node.nrSegments])
+ idMapFunctions{}.ClearValue(&seg.node.values[seg.node.nrSegments-1])
+ seg.node.nrSegments--
+ return seg.node.rebalanceAfterRemove(idMapGapIterator{seg.node, seg.index})
+}
+
+// RemoveAll removes all segments from the set. All existing iterators are
+// invalidated.
+func (s *idMapSet) RemoveAll() {
+ s.root = idMapnode{}
+}
+
+// RemoveRange removes all segments in the given range. An iterator to the
+// newly formed gap is returned, and all existing iterators are invalidated.
+func (s *idMapSet) RemoveRange(r idMapRange) idMapGapIterator {
+ seg, gap := s.Find(r.Start)
+ if seg.Ok() {
+ seg = s.Isolate(seg, r)
+ gap = s.Remove(seg)
+ }
+ for seg = gap.NextSegment(); seg.Ok() && seg.Start() < r.End; seg = gap.NextSegment() {
+ seg = s.Isolate(seg, r)
+ gap = s.Remove(seg)
+ }
+ return gap
+}
+
+// Merge attempts to merge two neighboring segments. If successful, Merge
+// returns an iterator to the merged segment, and all existing iterators are
+// invalidated. Otherwise, Merge returns a terminal iterator.
+//
+// If first is not the predecessor of second, Merge panics.
+func (s *idMapSet) Merge(first, second idMapIterator) idMapIterator {
+ if first.NextSegment() != second {
+ panic(fmt.Sprintf("attempt to merge non-neighboring segments %v, %v", first.Range(), second.Range()))
+ }
+ return s.MergeUnchecked(first, second)
+}
+
+// MergeUnchecked attempts to merge two neighboring segments. If successful,
+// MergeUnchecked returns an iterator to the merged segment, and all existing
+// iterators are invalidated. Otherwise, MergeUnchecked returns a terminal
+// iterator.
+//
+// Precondition: first is the predecessor of second: first.NextSegment() ==
+// second, first == second.PrevSegment().
+func (s *idMapSet) MergeUnchecked(first, second idMapIterator) idMapIterator {
+ if first.End() == second.Start() {
+ if mval, ok := (idMapFunctions{}).Merge(first.Range(), first.Value(), second.Range(), second.Value()); ok {
+
+ first.SetEndUnchecked(second.End())
+ first.SetValue(mval)
+ return s.Remove(second).PrevSegment()
+ }
+ }
+ return idMapIterator{}
+}
+
+// MergeAll attempts to merge all adjacent segments in the set. All existing
+// iterators are invalidated.
+func (s *idMapSet) MergeAll() {
+ seg := s.FirstSegment()
+ if !seg.Ok() {
+ return
+ }
+ next := seg.NextSegment()
+ for next.Ok() {
+ if mseg := s.MergeUnchecked(seg, next); mseg.Ok() {
+ seg, next = mseg, mseg.NextSegment()
+ } else {
+ seg, next = next, next.NextSegment()
+ }
+ }
+}
+
+// MergeRange attempts to merge all adjacent segments that contain a key in the
+// specific range. All existing iterators are invalidated.
+func (s *idMapSet) MergeRange(r idMapRange) {
+ seg := s.LowerBoundSegment(r.Start)
+ if !seg.Ok() {
+ return
+ }
+ next := seg.NextSegment()
+ for next.Ok() && next.Range().Start < r.End {
+ if mseg := s.MergeUnchecked(seg, next); mseg.Ok() {
+ seg, next = mseg, mseg.NextSegment()
+ } else {
+ seg, next = next, next.NextSegment()
+ }
+ }
+}
+
+// MergeAdjacent attempts to merge the segment containing r.Start with its
+// predecessor, and the segment containing r.End-1 with its successor.
+func (s *idMapSet) MergeAdjacent(r idMapRange) {
+ first := s.FindSegment(r.Start)
+ if first.Ok() {
+ if prev := first.PrevSegment(); prev.Ok() {
+ s.Merge(prev, first)
+ }
+ }
+ last := s.FindSegment(r.End - 1)
+ if last.Ok() {
+ if next := last.NextSegment(); next.Ok() {
+ s.Merge(last, next)
+ }
+ }
+}
+
+// Split splits the given segment at the given key and returns iterators to the
+// two resulting segments. All existing iterators (including seg, but not
+// including the returned iterators) are invalidated.
+//
+// If the segment cannot be split at split (because split is at the start or
+// end of the segment's range, so splitting would produce a segment with zero
+// length, or because split falls outside the segment's range altogether),
+// Split panics.
+func (s *idMapSet) Split(seg idMapIterator, split uint32) (idMapIterator, idMapIterator) {
+ if !seg.Range().CanSplitAt(split) {
+ panic(fmt.Sprintf("can't split %v at %v", seg.Range(), split))
+ }
+ return s.SplitUnchecked(seg, split)
+}
+
+// SplitUnchecked splits the given segment at the given key and returns
+// iterators to the two resulting segments. All existing iterators (including
+// seg, but not including the returned iterators) are invalidated.
+//
+// Preconditions: seg.Start() < key < seg.End().
+func (s *idMapSet) SplitUnchecked(seg idMapIterator, split uint32) (idMapIterator, idMapIterator) {
+ val1, val2 := (idMapFunctions{}).Split(seg.Range(), seg.Value(), split)
+ end2 := seg.End()
+ seg.SetEndUnchecked(split)
+ seg.SetValue(val1)
+ seg2 := s.InsertWithoutMergingUnchecked(seg.NextGap(), idMapRange{split, end2}, val2)
+
+ return seg2.PrevSegment(), seg2
+}
+
+// SplitAt splits the segment straddling split, if one exists. SplitAt returns
+// true if a segment was split and false otherwise. If SplitAt splits a
+// segment, all existing iterators are invalidated.
+func (s *idMapSet) SplitAt(split uint32) bool {
+ if seg := s.FindSegment(split); seg.Ok() && seg.Range().CanSplitAt(split) {
+ s.SplitUnchecked(seg, split)
+ return true
+ }
+ return false
+}
+
+// Isolate ensures that the given segment's range does not escape r by
+// splitting at r.Start and r.End if necessary, and returns an updated iterator
+// to the bounded segment. All existing iterators (including seg, but not
+// including the returned iterators) are invalidated.
+func (s *idMapSet) Isolate(seg idMapIterator, r idMapRange) idMapIterator {
+ if seg.Range().CanSplitAt(r.Start) {
+ _, seg = s.SplitUnchecked(seg, r.Start)
+ }
+ if seg.Range().CanSplitAt(r.End) {
+ seg, _ = s.SplitUnchecked(seg, r.End)
+ }
+ return seg
+}
+
+// ApplyContiguous applies a function to a contiguous range of segments,
+// splitting if necessary. The function is applied until the first gap is
+// encountered, at which point the gap is returned. If the function is applied
+// across the entire range, a terminal gap is returned. All existing iterators
+// are invalidated.
+//
+// N.B. The Iterator must not be invalidated by the function.
+func (s *idMapSet) ApplyContiguous(r idMapRange, fn func(seg idMapIterator)) idMapGapIterator {
+ seg, gap := s.Find(r.Start)
+ if !seg.Ok() {
+ return gap
+ }
+ for {
+ seg = s.Isolate(seg, r)
+ fn(seg)
+ if seg.End() >= r.End {
+ return idMapGapIterator{}
+ }
+ gap = seg.NextGap()
+ if !gap.IsEmpty() {
+ return gap
+ }
+ seg = gap.NextSegment()
+ if !seg.Ok() {
+
+ return idMapGapIterator{}
+ }
+ }
+}
+
+// +stateify savable
+type idMapnode struct {
+ // An internal binary tree node looks like:
+ //
+ // K
+ // / \
+ // Cl Cr
+ //
+ // where all keys in the subtree rooted by Cl (the left subtree) are less
+ // than K (the key of the parent node), and all keys in the subtree rooted
+ // by Cr (the right subtree) are greater than K.
+ //
+ // An internal B-tree node's indexes work out to look like:
+ //
+ // K0 K1 K2 ... Kn-1
+ // / \/ \/ \ ... / \
+ // C0 C1 C2 C3 ... Cn-1 Cn
+ //
+ // where n is nrSegments.
+ nrSegments int
+
+ // parent is a pointer to this node's parent. If this node is root, parent
+ // is nil.
+ parent *idMapnode
+
+ // parentIndex is the index of this node in parent.children.
+ parentIndex int
+
+ // Flag for internal nodes that is technically redundant with "children[0]
+ // != nil", but is stored in the first cache line. "hasChildren" rather
+ // than "isLeaf" because false must be the correct value for an empty root.
+ hasChildren bool
+
+ // Nodes store keys and values in separate arrays to maximize locality in
+ // the common case (scanning keys for lookup).
+ keys [idMapmaxDegree - 1]idMapRange
+ values [idMapmaxDegree - 1]uint32
+ children [idMapmaxDegree]*idMapnode
+}
+
+// firstSegment returns the first segment in the subtree rooted by n.
+//
+// Preconditions: n.nrSegments != 0.
+func (n *idMapnode) firstSegment() idMapIterator {
+ for n.hasChildren {
+ n = n.children[0]
+ }
+ return idMapIterator{n, 0}
+}
+
+// lastSegment returns the last segment in the subtree rooted by n.
+//
+// Preconditions: n.nrSegments != 0.
+func (n *idMapnode) lastSegment() idMapIterator {
+ for n.hasChildren {
+ n = n.children[n.nrSegments]
+ }
+ return idMapIterator{n, n.nrSegments - 1}
+}
+
+func (n *idMapnode) prevSibling() *idMapnode {
+ if n.parent == nil || n.parentIndex == 0 {
+ return nil
+ }
+ return n.parent.children[n.parentIndex-1]
+}
+
+func (n *idMapnode) nextSibling() *idMapnode {
+ if n.parent == nil || n.parentIndex == n.parent.nrSegments {
+ return nil
+ }
+ return n.parent.children[n.parentIndex+1]
+}
+
+// rebalanceBeforeInsert splits n and its ancestors if they are full, as
+// required for insertion, and returns an updated iterator to the position
+// represented by gap.
+func (n *idMapnode) rebalanceBeforeInsert(gap idMapGapIterator) idMapGapIterator {
+ if n.parent != nil {
+ gap = n.parent.rebalanceBeforeInsert(gap)
+ }
+ if n.nrSegments < idMapmaxDegree-1 {
+ return gap
+ }
+ if n.parent == nil {
+
+ left := &idMapnode{
+ nrSegments: idMapminDegree - 1,
+ parent: n,
+ parentIndex: 0,
+ hasChildren: n.hasChildren,
+ }
+ right := &idMapnode{
+ nrSegments: idMapminDegree - 1,
+ parent: n,
+ parentIndex: 1,
+ hasChildren: n.hasChildren,
+ }
+ copy(left.keys[:idMapminDegree-1], n.keys[:idMapminDegree-1])
+ copy(left.values[:idMapminDegree-1], n.values[:idMapminDegree-1])
+ copy(right.keys[:idMapminDegree-1], n.keys[idMapminDegree:])
+ copy(right.values[:idMapminDegree-1], n.values[idMapminDegree:])
+ n.keys[0], n.values[0] = n.keys[idMapminDegree-1], n.values[idMapminDegree-1]
+ idMapzeroValueSlice(n.values[1:])
+ if n.hasChildren {
+ copy(left.children[:idMapminDegree], n.children[:idMapminDegree])
+ copy(right.children[:idMapminDegree], n.children[idMapminDegree:])
+ idMapzeroNodeSlice(n.children[2:])
+ for i := 0; i < idMapminDegree; i++ {
+ left.children[i].parent = left
+ left.children[i].parentIndex = i
+ right.children[i].parent = right
+ right.children[i].parentIndex = i
+ }
+ }
+ n.nrSegments = 1
+ n.hasChildren = true
+ n.children[0] = left
+ n.children[1] = right
+ if gap.node != n {
+ return gap
+ }
+ if gap.index < idMapminDegree {
+ return idMapGapIterator{left, gap.index}
+ }
+ return idMapGapIterator{right, gap.index - idMapminDegree}
+ }
+
+ copy(n.parent.keys[n.parentIndex+1:], n.parent.keys[n.parentIndex:n.parent.nrSegments])
+ copy(n.parent.values[n.parentIndex+1:], n.parent.values[n.parentIndex:n.parent.nrSegments])
+ n.parent.keys[n.parentIndex], n.parent.values[n.parentIndex] = n.keys[idMapminDegree-1], n.values[idMapminDegree-1]
+ copy(n.parent.children[n.parentIndex+2:], n.parent.children[n.parentIndex+1:n.parent.nrSegments+1])
+ for i := n.parentIndex + 2; i < n.parent.nrSegments+2; i++ {
+ n.parent.children[i].parentIndex = i
+ }
+ sibling := &idMapnode{
+ nrSegments: idMapminDegree - 1,
+ parent: n.parent,
+ parentIndex: n.parentIndex + 1,
+ hasChildren: n.hasChildren,
+ }
+ n.parent.children[n.parentIndex+1] = sibling
+ n.parent.nrSegments++
+ copy(sibling.keys[:idMapminDegree-1], n.keys[idMapminDegree:])
+ copy(sibling.values[:idMapminDegree-1], n.values[idMapminDegree:])
+ idMapzeroValueSlice(n.values[idMapminDegree-1:])
+ if n.hasChildren {
+ copy(sibling.children[:idMapminDegree], n.children[idMapminDegree:])
+ idMapzeroNodeSlice(n.children[idMapminDegree:])
+ for i := 0; i < idMapminDegree; i++ {
+ sibling.children[i].parent = sibling
+ sibling.children[i].parentIndex = i
+ }
+ }
+ n.nrSegments = idMapminDegree - 1
+
+ if gap.node != n {
+ return gap
+ }
+ if gap.index < idMapminDegree {
+ return gap
+ }
+ return idMapGapIterator{sibling, gap.index - idMapminDegree}
+}
+
+// rebalanceAfterRemove "unsplits" n and its ancestors if they are deficient
+// (contain fewer segments than required by B-tree invariants), as required for
+// removal, and returns an updated iterator to the position represented by gap.
+//
+// Precondition: n is the only node in the tree that may currently violate a
+// B-tree invariant.
+func (n *idMapnode) rebalanceAfterRemove(gap idMapGapIterator) idMapGapIterator {
+ for {
+ if n.nrSegments >= idMapminDegree-1 {
+ return gap
+ }
+ if n.parent == nil {
+
+ return gap
+ }
+
+ if sibling := n.prevSibling(); sibling != nil && sibling.nrSegments >= idMapminDegree {
+ copy(n.keys[1:], n.keys[:n.nrSegments])
+ copy(n.values[1:], n.values[:n.nrSegments])
+ n.keys[0] = n.parent.keys[n.parentIndex-1]
+ n.values[0] = n.parent.values[n.parentIndex-1]
+ n.parent.keys[n.parentIndex-1] = sibling.keys[sibling.nrSegments-1]
+ n.parent.values[n.parentIndex-1] = sibling.values[sibling.nrSegments-1]
+ idMapFunctions{}.ClearValue(&sibling.values[sibling.nrSegments-1])
+ if n.hasChildren {
+ copy(n.children[1:], n.children[:n.nrSegments+1])
+ n.children[0] = sibling.children[sibling.nrSegments]
+ sibling.children[sibling.nrSegments] = nil
+ n.children[0].parent = n
+ n.children[0].parentIndex = 0
+ for i := 1; i < n.nrSegments+2; i++ {
+ n.children[i].parentIndex = i
+ }
+ }
+ n.nrSegments++
+ sibling.nrSegments--
+ if gap.node == sibling && gap.index == sibling.nrSegments {
+ return idMapGapIterator{n, 0}
+ }
+ if gap.node == n {
+ return idMapGapIterator{n, gap.index + 1}
+ }
+ return gap
+ }
+ if sibling := n.nextSibling(); sibling != nil && sibling.nrSegments >= idMapminDegree {
+ n.keys[n.nrSegments] = n.parent.keys[n.parentIndex]
+ n.values[n.nrSegments] = n.parent.values[n.parentIndex]
+ n.parent.keys[n.parentIndex] = sibling.keys[0]
+ n.parent.values[n.parentIndex] = sibling.values[0]
+ copy(sibling.keys[:sibling.nrSegments-1], sibling.keys[1:])
+ copy(sibling.values[:sibling.nrSegments-1], sibling.values[1:])
+ idMapFunctions{}.ClearValue(&sibling.values[sibling.nrSegments-1])
+ if n.hasChildren {
+ n.children[n.nrSegments+1] = sibling.children[0]
+ copy(sibling.children[:sibling.nrSegments], sibling.children[1:])
+ sibling.children[sibling.nrSegments] = nil
+ n.children[n.nrSegments+1].parent = n
+ n.children[n.nrSegments+1].parentIndex = n.nrSegments + 1
+ for i := 0; i < sibling.nrSegments; i++ {
+ sibling.children[i].parentIndex = i
+ }
+ }
+ n.nrSegments++
+ sibling.nrSegments--
+ if gap.node == sibling {
+ if gap.index == 0 {
+ return idMapGapIterator{n, n.nrSegments}
+ }
+ return idMapGapIterator{sibling, gap.index - 1}
+ }
+ return gap
+ }
+
+ p := n.parent
+ if p.nrSegments == 1 {
+
+ left, right := p.children[0], p.children[1]
+ p.nrSegments = left.nrSegments + right.nrSegments + 1
+ p.hasChildren = left.hasChildren
+ p.keys[left.nrSegments] = p.keys[0]
+ p.values[left.nrSegments] = p.values[0]
+ copy(p.keys[:left.nrSegments], left.keys[:left.nrSegments])
+ copy(p.values[:left.nrSegments], left.values[:left.nrSegments])
+ copy(p.keys[left.nrSegments+1:], right.keys[:right.nrSegments])
+ copy(p.values[left.nrSegments+1:], right.values[:right.nrSegments])
+ if left.hasChildren {
+ copy(p.children[:left.nrSegments+1], left.children[:left.nrSegments+1])
+ copy(p.children[left.nrSegments+1:], right.children[:right.nrSegments+1])
+ for i := 0; i < p.nrSegments+1; i++ {
+ p.children[i].parent = p
+ p.children[i].parentIndex = i
+ }
+ } else {
+ p.children[0] = nil
+ p.children[1] = nil
+ }
+ if gap.node == left {
+ return idMapGapIterator{p, gap.index}
+ }
+ if gap.node == right {
+ return idMapGapIterator{p, gap.index + left.nrSegments + 1}
+ }
+ return gap
+ }
+ // Merge n and either sibling, along with the segment separating the
+ // two, into whichever of the two nodes comes first. This is the
+ // reverse of the non-root splitting case in
+ // node.rebalanceBeforeInsert.
+ var left, right *idMapnode
+ if n.parentIndex > 0 {
+ left = n.prevSibling()
+ right = n
+ } else {
+ left = n
+ right = n.nextSibling()
+ }
+
+ if gap.node == right {
+ gap = idMapGapIterator{left, gap.index + left.nrSegments + 1}
+ }
+ left.keys[left.nrSegments] = p.keys[left.parentIndex]
+ left.values[left.nrSegments] = p.values[left.parentIndex]
+ copy(left.keys[left.nrSegments+1:], right.keys[:right.nrSegments])
+ copy(left.values[left.nrSegments+1:], right.values[:right.nrSegments])
+ if left.hasChildren {
+ copy(left.children[left.nrSegments+1:], right.children[:right.nrSegments+1])
+ for i := left.nrSegments + 1; i < left.nrSegments+right.nrSegments+2; i++ {
+ left.children[i].parent = left
+ left.children[i].parentIndex = i
+ }
+ }
+ left.nrSegments += right.nrSegments + 1
+ copy(p.keys[left.parentIndex:], p.keys[left.parentIndex+1:p.nrSegments])
+ copy(p.values[left.parentIndex:], p.values[left.parentIndex+1:p.nrSegments])
+ idMapFunctions{}.ClearValue(&p.values[p.nrSegments-1])
+ copy(p.children[left.parentIndex+1:], p.children[left.parentIndex+2:p.nrSegments+1])
+ for i := 0; i < p.nrSegments; i++ {
+ p.children[i].parentIndex = i
+ }
+ p.children[p.nrSegments] = nil
+ p.nrSegments--
+
+ n = p
+ }
+}
+
+// A Iterator is conceptually one of:
+//
+// - A pointer to a segment in a set; or
+//
+// - A terminal iterator, which is a sentinel indicating that the end of
+// iteration has been reached.
+//
+// Iterators are copyable values and are meaningfully equality-comparable. The
+// zero value of Iterator is a terminal iterator.
+//
+// Unless otherwise specified, any mutation of a set invalidates all existing
+// iterators into the set.
+type idMapIterator struct {
+ // node is the node containing the iterated segment. If the iterator is
+ // terminal, node is nil.
+ node *idMapnode
+
+ // index is the index of the segment in node.keys/values.
+ index int
+}
+
+// Ok returns true if the iterator is not terminal. All other methods are only
+// valid for non-terminal iterators.
+func (seg idMapIterator) Ok() bool {
+ return seg.node != nil
+}
+
+// Range returns the iterated segment's range key.
+func (seg idMapIterator) Range() idMapRange {
+ return seg.node.keys[seg.index]
+}
+
+// Start is equivalent to Range().Start, but should be preferred if only the
+// start of the range is needed.
+func (seg idMapIterator) Start() uint32 {
+ return seg.node.keys[seg.index].Start
+}
+
+// End is equivalent to Range().End, but should be preferred if only the end of
+// the range is needed.
+func (seg idMapIterator) End() uint32 {
+ return seg.node.keys[seg.index].End
+}
+
+// SetRangeUnchecked mutates the iterated segment's range key. This operation
+// does not invalidate any iterators.
+//
+// Preconditions:
+//
+// - r.Length() > 0.
+//
+// - The new range must not overlap an existing one: If seg.NextSegment().Ok(),
+// then r.end <= seg.NextSegment().Start(); if seg.PrevSegment().Ok(), then
+// r.start >= seg.PrevSegment().End().
+func (seg idMapIterator) SetRangeUnchecked(r idMapRange) {
+ seg.node.keys[seg.index] = r
+}
+
+// SetRange mutates the iterated segment's range key. If the new range would
+// cause the iterated segment to overlap another segment, or if the new range
+// is invalid, SetRange panics. This operation does not invalidate any
+// iterators.
+func (seg idMapIterator) SetRange(r idMapRange) {
+ if r.Length() <= 0 {
+ panic(fmt.Sprintf("invalid segment range %v", r))
+ }
+ if prev := seg.PrevSegment(); prev.Ok() && r.Start < prev.End() {
+ panic(fmt.Sprintf("new segment range %v overlaps segment range %v", r, prev.Range()))
+ }
+ if next := seg.NextSegment(); next.Ok() && r.End > next.Start() {
+ panic(fmt.Sprintf("new segment range %v overlaps segment range %v", r, next.Range()))
+ }
+ seg.SetRangeUnchecked(r)
+}
+
+// SetStartUnchecked mutates the iterated segment's start. This operation does
+// not invalidate any iterators.
+//
+// Preconditions: The new start must be valid: start < seg.End(); if
+// seg.PrevSegment().Ok(), then start >= seg.PrevSegment().End().
+func (seg idMapIterator) SetStartUnchecked(start uint32) {
+ seg.node.keys[seg.index].Start = start
+}
+
+// SetStart mutates the iterated segment's start. If the new start value would
+// cause the iterated segment to overlap another segment, or would result in an
+// invalid range, SetStart panics. This operation does not invalidate any
+// iterators.
+func (seg idMapIterator) SetStart(start uint32) {
+ if start >= seg.End() {
+ panic(fmt.Sprintf("new start %v would invalidate segment range %v", start, seg.Range()))
+ }
+ if prev := seg.PrevSegment(); prev.Ok() && start < prev.End() {
+ panic(fmt.Sprintf("new start %v would cause segment range %v to overlap segment range %v", start, seg.Range(), prev.Range()))
+ }
+ seg.SetStartUnchecked(start)
+}
+
+// SetEndUnchecked mutates the iterated segment's end. This operation does not
+// invalidate any iterators.
+//
+// Preconditions: The new end must be valid: end > seg.Start(); if
+// seg.NextSegment().Ok(), then end <= seg.NextSegment().Start().
+func (seg idMapIterator) SetEndUnchecked(end uint32) {
+ seg.node.keys[seg.index].End = end
+}
+
+// SetEnd mutates the iterated segment's end. If the new end value would cause
+// the iterated segment to overlap another segment, or would result in an
+// invalid range, SetEnd panics. This operation does not invalidate any
+// iterators.
+func (seg idMapIterator) SetEnd(end uint32) {
+ if end <= seg.Start() {
+ panic(fmt.Sprintf("new end %v would invalidate segment range %v", end, seg.Range()))
+ }
+ if next := seg.NextSegment(); next.Ok() && end > next.Start() {
+ panic(fmt.Sprintf("new end %v would cause segment range %v to overlap segment range %v", end, seg.Range(), next.Range()))
+ }
+ seg.SetEndUnchecked(end)
+}
+
+// Value returns a copy of the iterated segment's value.
+func (seg idMapIterator) Value() uint32 {
+ return seg.node.values[seg.index]
+}
+
+// ValuePtr returns a pointer to the iterated segment's value. The pointer is
+// invalidated if the iterator is invalidated. This operation does not
+// invalidate any iterators.
+func (seg idMapIterator) ValuePtr() *uint32 {
+ return &seg.node.values[seg.index]
+}
+
+// SetValue mutates the iterated segment's value. This operation does not
+// invalidate any iterators.
+func (seg idMapIterator) SetValue(val uint32) {
+ seg.node.values[seg.index] = val
+}
+
+// PrevSegment returns the iterated segment's predecessor. If there is no
+// preceding segment, PrevSegment returns a terminal iterator.
+func (seg idMapIterator) PrevSegment() idMapIterator {
+ if seg.node.hasChildren {
+ return seg.node.children[seg.index].lastSegment()
+ }
+ if seg.index > 0 {
+ return idMapIterator{seg.node, seg.index - 1}
+ }
+ if seg.node.parent == nil {
+ return idMapIterator{}
+ }
+ return idMapsegmentBeforePosition(seg.node.parent, seg.node.parentIndex)
+}
+
+// NextSegment returns the iterated segment's successor. If there is no
+// succeeding segment, NextSegment returns a terminal iterator.
+func (seg idMapIterator) NextSegment() idMapIterator {
+ if seg.node.hasChildren {
+ return seg.node.children[seg.index+1].firstSegment()
+ }
+ if seg.index < seg.node.nrSegments-1 {
+ return idMapIterator{seg.node, seg.index + 1}
+ }
+ if seg.node.parent == nil {
+ return idMapIterator{}
+ }
+ return idMapsegmentAfterPosition(seg.node.parent, seg.node.parentIndex)
+}
+
+// PrevGap returns the gap immediately before the iterated segment.
+func (seg idMapIterator) PrevGap() idMapGapIterator {
+ if seg.node.hasChildren {
+
+ return seg.node.children[seg.index].lastSegment().NextGap()
+ }
+ return idMapGapIterator{seg.node, seg.index}
+}
+
+// NextGap returns the gap immediately after the iterated segment.
+func (seg idMapIterator) NextGap() idMapGapIterator {
+ if seg.node.hasChildren {
+ return seg.node.children[seg.index+1].firstSegment().PrevGap()
+ }
+ return idMapGapIterator{seg.node, seg.index + 1}
+}
+
+// PrevNonEmpty returns the iterated segment's predecessor if it is adjacent,
+// or the gap before the iterated segment otherwise. If seg.Start() ==
+// Functions.MinKey(), PrevNonEmpty will return two terminal iterators.
+// Otherwise, exactly one of the iterators returned by PrevNonEmpty will be
+// non-terminal.
+func (seg idMapIterator) PrevNonEmpty() (idMapIterator, idMapGapIterator) {
+ gap := seg.PrevGap()
+ if gap.Range().Length() != 0 {
+ return idMapIterator{}, gap
+ }
+ return gap.PrevSegment(), idMapGapIterator{}
+}
+
+// NextNonEmpty returns the iterated segment's successor if it is adjacent, or
+// the gap after the iterated segment otherwise. If seg.End() ==
+// Functions.MaxKey(), NextNonEmpty will return two terminal iterators.
+// Otherwise, exactly one of the iterators returned by NextNonEmpty will be
+// non-terminal.
+func (seg idMapIterator) NextNonEmpty() (idMapIterator, idMapGapIterator) {
+ gap := seg.NextGap()
+ if gap.Range().Length() != 0 {
+ return idMapIterator{}, gap
+ }
+ return gap.NextSegment(), idMapGapIterator{}
+}
+
+// A GapIterator is conceptually one of:
+//
+// - A pointer to a position between two segments, before the first segment, or
+// after the last segment in a set, called a *gap*; or
+//
+// - A terminal iterator, which is a sentinel indicating that the end of
+// iteration has been reached.
+//
+// Note that the gap between two adjacent segments exists (iterators to it are
+// non-terminal), but has a length of zero. GapIterator.IsEmpty returns true
+// for such gaps. An empty set contains a single gap, spanning the entire range
+// of the set's keys.
+//
+// GapIterators are copyable values and are meaningfully equality-comparable.
+// The zero value of GapIterator is a terminal iterator.
+//
+// Unless otherwise specified, any mutation of a set invalidates all existing
+// iterators into the set.
+type idMapGapIterator struct {
+ // The representation of a GapIterator is identical to that of an Iterator,
+ // except that index corresponds to positions between segments in the same
+ // way as for node.children (see comment for node.nrSegments).
+ node *idMapnode
+ index int
+}
+
+// Ok returns true if the iterator is not terminal. All other methods are only
+// valid for non-terminal iterators.
+func (gap idMapGapIterator) Ok() bool {
+ return gap.node != nil
+}
+
+// Range returns the range spanned by the iterated gap.
+func (gap idMapGapIterator) Range() idMapRange {
+ return idMapRange{gap.Start(), gap.End()}
+}
+
+// Start is equivalent to Range().Start, but should be preferred if only the
+// start of the range is needed.
+func (gap idMapGapIterator) Start() uint32 {
+ if ps := gap.PrevSegment(); ps.Ok() {
+ return ps.End()
+ }
+ return idMapFunctions{}.MinKey()
+}
+
+// End is equivalent to Range().End, but should be preferred if only the end of
+// the range is needed.
+func (gap idMapGapIterator) End() uint32 {
+ if ns := gap.NextSegment(); ns.Ok() {
+ return ns.Start()
+ }
+ return idMapFunctions{}.MaxKey()
+}
+
+// IsEmpty returns true if the iterated gap is empty (that is, the "gap" is
+// between two adjacent segments.)
+func (gap idMapGapIterator) IsEmpty() bool {
+ return gap.Range().Length() == 0
+}
+
+// PrevSegment returns the segment immediately before the iterated gap. If no
+// such segment exists, PrevSegment returns a terminal iterator.
+func (gap idMapGapIterator) PrevSegment() idMapIterator {
+ return idMapsegmentBeforePosition(gap.node, gap.index)
+}
+
+// NextSegment returns the segment immediately after the iterated gap. If no
+// such segment exists, NextSegment returns a terminal iterator.
+func (gap idMapGapIterator) NextSegment() idMapIterator {
+ return idMapsegmentAfterPosition(gap.node, gap.index)
+}
+
+// PrevGap returns the iterated gap's predecessor. If no such gap exists,
+// PrevGap returns a terminal iterator.
+func (gap idMapGapIterator) PrevGap() idMapGapIterator {
+ seg := gap.PrevSegment()
+ if !seg.Ok() {
+ return idMapGapIterator{}
+ }
+ return seg.PrevGap()
+}
+
+// NextGap returns the iterated gap's successor. If no such gap exists, NextGap
+// returns a terminal iterator.
+func (gap idMapGapIterator) NextGap() idMapGapIterator {
+ seg := gap.NextSegment()
+ if !seg.Ok() {
+ return idMapGapIterator{}
+ }
+ return seg.NextGap()
+}
+
+// segmentBeforePosition returns the predecessor segment of the position given
+// by n.children[i], which may or may not contain a child. If no such segment
+// exists, segmentBeforePosition returns a terminal iterator.
+func idMapsegmentBeforePosition(n *idMapnode, i int) idMapIterator {
+ for i == 0 {
+ if n.parent == nil {
+ return idMapIterator{}
+ }
+ n, i = n.parent, n.parentIndex
+ }
+ return idMapIterator{n, i - 1}
+}
+
+// segmentAfterPosition returns the successor segment of the position given by
+// n.children[i], which may or may not contain a child. If no such segment
+// exists, segmentAfterPosition returns a terminal iterator.
+func idMapsegmentAfterPosition(n *idMapnode, i int) idMapIterator {
+ for i == n.nrSegments {
+ if n.parent == nil {
+ return idMapIterator{}
+ }
+ n, i = n.parent, n.parentIndex
+ }
+ return idMapIterator{n, i}
+}
+
+func idMapzeroValueSlice(slice []uint32) {
+
+ for i := range slice {
+ idMapFunctions{}.ClearValue(&slice[i])
+ }
+}
+
+func idMapzeroNodeSlice(slice []*idMapnode) {
+ for i := range slice {
+ slice[i] = nil
+ }
+}
+
+// String stringifies a Set for debugging.
+func (s *idMapSet) String() string {
+ return s.root.String()
+}
+
+// String stringifies a node (and all of its children) for debugging.
+func (n *idMapnode) String() string {
+ var buf bytes.Buffer
+ n.writeDebugString(&buf, "")
+ return buf.String()
+}
+
+func (n *idMapnode) writeDebugString(buf *bytes.Buffer, prefix string) {
+ if n.hasChildren != (n.nrSegments > 0 && n.children[0] != nil) {
+ buf.WriteString(prefix)
+ buf.WriteString(fmt.Sprintf("WARNING: inconsistent value of hasChildren: got %v, want %v\n", n.hasChildren, !n.hasChildren))
+ }
+ for i := 0; i < n.nrSegments; i++ {
+ if child := n.children[i]; child != nil {
+ cprefix := fmt.Sprintf("%s- % 3d ", prefix, i)
+ if child.parent != n || child.parentIndex != i {
+ buf.WriteString(cprefix)
+ buf.WriteString(fmt.Sprintf("WARNING: inconsistent linkage to parent: got (%p, %d), want (%p, %d)\n", child.parent, child.parentIndex, n, i))
+ }
+ child.writeDebugString(buf, fmt.Sprintf("%s- % 3d ", prefix, i))
+ }
+ buf.WriteString(prefix)
+ buf.WriteString(fmt.Sprintf("- % 3d: %v => %v\n", i, n.keys[i], n.values[i]))
+ }
+ if child := n.children[n.nrSegments]; child != nil {
+ child.writeDebugString(buf, fmt.Sprintf("%s- % 3d ", prefix, n.nrSegments))
+ }
+}
+
+// SegmentDataSlices represents segments from a set as slices of start, end, and
+// values. SegmentDataSlices is primarily used as an intermediate representation
+// for save/restore and the layout here is optimized for that.
+//
+// +stateify savable
+type idMapSegmentDataSlices struct {
+ Start []uint32
+ End []uint32
+ Values []uint32
+}
+
+// ExportSortedSlice returns a copy of all segments in the given set, in ascending
+// key order.
+func (s *idMapSet) ExportSortedSlices() *idMapSegmentDataSlices {
+ var sds idMapSegmentDataSlices
+ for seg := s.FirstSegment(); seg.Ok(); seg = seg.NextSegment() {
+ sds.Start = append(sds.Start, seg.Start())
+ sds.End = append(sds.End, seg.End())
+ sds.Values = append(sds.Values, seg.Value())
+ }
+ sds.Start = sds.Start[:len(sds.Start):len(sds.Start)]
+ sds.End = sds.End[:len(sds.End):len(sds.End)]
+ sds.Values = sds.Values[:len(sds.Values):len(sds.Values)]
+ return &sds
+}
+
+// ImportSortedSlice initializes the given set from the given slice.
+//
+// Preconditions: s must be empty. sds must represent a valid set (the segments
+// in sds must have valid lengths that do not overlap). The segments in sds
+// must be sorted in ascending key order.
+func (s *idMapSet) ImportSortedSlices(sds *idMapSegmentDataSlices) error {
+ if !s.IsEmpty() {
+ return fmt.Errorf("cannot import into non-empty set %v", s)
+ }
+ gap := s.FirstGap()
+ for i := range sds.Start {
+ r := idMapRange{sds.Start[i], sds.End[i]}
+ if !gap.Range().IsSupersetOf(r) {
+ return fmt.Errorf("segment overlaps a preceding segment or is incorrectly sorted: [%d, %d) => %v", sds.Start[i], sds.End[i], sds.Values[i])
+ }
+ gap = s.InsertWithoutMerging(gap, r, sds.Values[i]).NextGap()
+ }
+ return nil
+}
+func (s *idMapSet) saveRoot() *idMapSegmentDataSlices {
+ return s.ExportSortedSlices()
+}
+
+func (s *idMapSet) loadRoot(sds *idMapSegmentDataSlices) {
+ if err := s.ImportSortedSlices(sds); err != nil {
+ panic(err)
+ }
+}
diff --git a/pkg/sentry/kernel/contexttest/BUILD b/pkg/sentry/kernel/contexttest/BUILD
deleted file mode 100644
index 3a88a585c..000000000
--- a/pkg/sentry/kernel/contexttest/BUILD
+++ /dev/null
@@ -1,18 +0,0 @@
-load("//tools/go_stateify:defs.bzl", "go_library")
-
-package(licenses = ["notice"])
-
-go_library(
- name = "contexttest",
- testonly = 1,
- srcs = ["contexttest.go"],
- importpath = "gvisor.dev/gvisor/pkg/sentry/kernel/contexttest",
- visibility = ["//pkg/sentry:internal"],
- deps = [
- "//pkg/sentry/context",
- "//pkg/sentry/context/contexttest",
- "//pkg/sentry/kernel",
- "//pkg/sentry/pgalloc",
- "//pkg/sentry/platform",
- ],
-)
diff --git a/pkg/sentry/kernel/contexttest/contexttest.go b/pkg/sentry/kernel/contexttest/contexttest.go
deleted file mode 100644
index 82f9d8922..000000000
--- a/pkg/sentry/kernel/contexttest/contexttest.go
+++ /dev/null
@@ -1,40 +0,0 @@
-// Copyright 2018 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 contexttest provides a test context.Context which includes
-// a dummy kernel pointing to a valid platform.
-package contexttest
-
-import (
- "testing"
-
- "gvisor.dev/gvisor/pkg/sentry/context"
- "gvisor.dev/gvisor/pkg/sentry/context/contexttest"
- "gvisor.dev/gvisor/pkg/sentry/kernel"
- "gvisor.dev/gvisor/pkg/sentry/pgalloc"
- "gvisor.dev/gvisor/pkg/sentry/platform"
-)
-
-// Context returns a Context that may be used in tests. Uses ptrace as the
-// platform.Platform, and provides a stub kernel that only serves to point to
-// the platform.
-func Context(tb testing.TB) context.Context {
- ctx := contexttest.Context(tb)
- k := &kernel.Kernel{
- Platform: platform.FromContext(ctx),
- }
- k.SetMemoryFile(pgalloc.MemoryFileFromContext(ctx))
- ctx.(*contexttest.TestContext).RegisterValue(kernel.CtxKernel, k)
- return ctx
-}
diff --git a/pkg/sentry/kernel/epoll/BUILD b/pkg/sentry/kernel/epoll/BUILD
deleted file mode 100644
index c47f6b6fc..000000000
--- a/pkg/sentry/kernel/epoll/BUILD
+++ /dev/null
@@ -1,52 +0,0 @@
-load("@io_bazel_rules_go//go:def.bzl", "go_test")
-load("//tools/go_generics:defs.bzl", "go_template_instance")
-load("//tools/go_stateify:defs.bzl", "go_library")
-
-package(licenses = ["notice"])
-
-go_template_instance(
- name = "epoll_list",
- out = "epoll_list.go",
- package = "epoll",
- prefix = "pollEntry",
- template = "//pkg/ilist:generic_list",
- types = {
- "Element": "*pollEntry",
- "Linker": "*pollEntry",
- },
-)
-
-go_library(
- name = "epoll",
- srcs = [
- "epoll.go",
- "epoll_list.go",
- "epoll_state.go",
- ],
- importpath = "gvisor.dev/gvisor/pkg/sentry/kernel/epoll",
- visibility = ["//pkg/sentry:internal"],
- deps = [
- "//pkg/refs",
- "//pkg/sentry/context",
- "//pkg/sentry/fs",
- "//pkg/sentry/fs/anon",
- "//pkg/sentry/fs/fsutil",
- "//pkg/sentry/usermem",
- "//pkg/sync",
- "//pkg/waiter",
- ],
-)
-
-go_test(
- name = "epoll_test",
- size = "small",
- srcs = [
- "epoll_test.go",
- ],
- embed = [":epoll"],
- deps = [
- "//pkg/sentry/context/contexttest",
- "//pkg/sentry/fs/filetest",
- "//pkg/waiter",
- ],
-)
diff --git a/pkg/sentry/kernel/epoll/epoll_list.go b/pkg/sentry/kernel/epoll/epoll_list.go
new file mode 100755
index 000000000..94d5c9e57
--- /dev/null
+++ b/pkg/sentry/kernel/epoll/epoll_list.go
@@ -0,0 +1,173 @@
+package epoll
+
+// ElementMapper provides an identity mapping by default.
+//
+// This can be replaced to provide a struct that maps elements to linker
+// objects, if they are not the same. An ElementMapper is not typically
+// required if: Linker is left as is, Element is left as is, or Linker and
+// Element are the same type.
+type pollEntryElementMapper struct{}
+
+// linkerFor maps an Element to a Linker.
+//
+// This default implementation should be inlined.
+//
+//go:nosplit
+func (pollEntryElementMapper) linkerFor(elem *pollEntry) *pollEntry { return elem }
+
+// List is an intrusive list. Entries can be added to or removed from the list
+// in O(1) time and with no additional memory allocations.
+//
+// The zero value for List is an empty list ready to use.
+//
+// To iterate over a list (where l is a List):
+// for e := l.Front(); e != nil; e = e.Next() {
+// // do something with e.
+// }
+//
+// +stateify savable
+type pollEntryList struct {
+ head *pollEntry
+ tail *pollEntry
+}
+
+// Reset resets list l to the empty state.
+func (l *pollEntryList) Reset() {
+ l.head = nil
+ l.tail = nil
+}
+
+// Empty returns true iff the list is empty.
+func (l *pollEntryList) Empty() bool {
+ return l.head == nil
+}
+
+// Front returns the first element of list l or nil.
+func (l *pollEntryList) Front() *pollEntry {
+ return l.head
+}
+
+// Back returns the last element of list l or nil.
+func (l *pollEntryList) Back() *pollEntry {
+ return l.tail
+}
+
+// PushFront inserts the element e at the front of list l.
+func (l *pollEntryList) PushFront(e *pollEntry) {
+ pollEntryElementMapper{}.linkerFor(e).SetNext(l.head)
+ pollEntryElementMapper{}.linkerFor(e).SetPrev(nil)
+
+ if l.head != nil {
+ pollEntryElementMapper{}.linkerFor(l.head).SetPrev(e)
+ } else {
+ l.tail = e
+ }
+
+ l.head = e
+}
+
+// PushBack inserts the element e at the back of list l.
+func (l *pollEntryList) PushBack(e *pollEntry) {
+ pollEntryElementMapper{}.linkerFor(e).SetNext(nil)
+ pollEntryElementMapper{}.linkerFor(e).SetPrev(l.tail)
+
+ if l.tail != nil {
+ pollEntryElementMapper{}.linkerFor(l.tail).SetNext(e)
+ } else {
+ l.head = e
+ }
+
+ l.tail = e
+}
+
+// PushBackList inserts list m at the end of list l, emptying m.
+func (l *pollEntryList) PushBackList(m *pollEntryList) {
+ if l.head == nil {
+ l.head = m.head
+ l.tail = m.tail
+ } else if m.head != nil {
+ pollEntryElementMapper{}.linkerFor(l.tail).SetNext(m.head)
+ pollEntryElementMapper{}.linkerFor(m.head).SetPrev(l.tail)
+
+ l.tail = m.tail
+ }
+
+ m.head = nil
+ m.tail = nil
+}
+
+// InsertAfter inserts e after b.
+func (l *pollEntryList) InsertAfter(b, e *pollEntry) {
+ a := pollEntryElementMapper{}.linkerFor(b).Next()
+ pollEntryElementMapper{}.linkerFor(e).SetNext(a)
+ pollEntryElementMapper{}.linkerFor(e).SetPrev(b)
+ pollEntryElementMapper{}.linkerFor(b).SetNext(e)
+
+ if a != nil {
+ pollEntryElementMapper{}.linkerFor(a).SetPrev(e)
+ } else {
+ l.tail = e
+ }
+}
+
+// InsertBefore inserts e before a.
+func (l *pollEntryList) InsertBefore(a, e *pollEntry) {
+ b := pollEntryElementMapper{}.linkerFor(a).Prev()
+ pollEntryElementMapper{}.linkerFor(e).SetNext(a)
+ pollEntryElementMapper{}.linkerFor(e).SetPrev(b)
+ pollEntryElementMapper{}.linkerFor(a).SetPrev(e)
+
+ if b != nil {
+ pollEntryElementMapper{}.linkerFor(b).SetNext(e)
+ } else {
+ l.head = e
+ }
+}
+
+// Remove removes e from l.
+func (l *pollEntryList) Remove(e *pollEntry) {
+ prev := pollEntryElementMapper{}.linkerFor(e).Prev()
+ next := pollEntryElementMapper{}.linkerFor(e).Next()
+
+ if prev != nil {
+ pollEntryElementMapper{}.linkerFor(prev).SetNext(next)
+ } else {
+ l.head = next
+ }
+
+ if next != nil {
+ pollEntryElementMapper{}.linkerFor(next).SetPrev(prev)
+ } else {
+ l.tail = prev
+ }
+}
+
+// Entry is a default implementation of Linker. Users can add anonymous fields
+// of this type to their structs to make them automatically implement the
+// methods needed by List.
+//
+// +stateify savable
+type pollEntryEntry struct {
+ next *pollEntry
+ prev *pollEntry
+}
+
+// Next returns the entry that follows e in the list.
+func (e *pollEntryEntry) Next() *pollEntry {
+ return e.next
+}
+
+// Prev returns the entry that precedes e in the list.
+func (e *pollEntryEntry) Prev() *pollEntry {
+ return e.prev
+}
+
+// SetNext assigns 'entry' as the entry that follows e in the list.
+func (e *pollEntryEntry) SetNext(elem *pollEntry) {
+ e.next = elem
+}
+
+// SetPrev assigns 'entry' as the entry that precedes e in the list.
+func (e *pollEntryEntry) SetPrev(elem *pollEntry) {
+ e.prev = elem
+}
diff --git a/pkg/sentry/kernel/epoll/epoll_state_autogen.go b/pkg/sentry/kernel/epoll/epoll_state_autogen.go
new file mode 100755
index 000000000..c738a9355
--- /dev/null
+++ b/pkg/sentry/kernel/epoll/epoll_state_autogen.go
@@ -0,0 +1,99 @@
+// automatically generated by stateify.
+
+package epoll
+
+import (
+ "gvisor.dev/gvisor/pkg/state"
+)
+
+func (x *FileIdentifier) beforeSave() {}
+func (x *FileIdentifier) save(m state.Map) {
+ x.beforeSave()
+ m.Save("File", &x.File)
+ m.Save("Fd", &x.Fd)
+}
+
+func (x *FileIdentifier) afterLoad() {}
+func (x *FileIdentifier) load(m state.Map) {
+ m.LoadWait("File", &x.File)
+ m.Load("Fd", &x.Fd)
+}
+
+func (x *pollEntry) beforeSave() {}
+func (x *pollEntry) save(m state.Map) {
+ x.beforeSave()
+ m.Save("pollEntryEntry", &x.pollEntryEntry)
+ m.Save("id", &x.id)
+ m.Save("userData", &x.userData)
+ m.Save("mask", &x.mask)
+ m.Save("flags", &x.flags)
+ m.Save("epoll", &x.epoll)
+}
+
+func (x *pollEntry) load(m state.Map) {
+ m.Load("pollEntryEntry", &x.pollEntryEntry)
+ m.LoadWait("id", &x.id)
+ m.Load("userData", &x.userData)
+ m.Load("mask", &x.mask)
+ m.Load("flags", &x.flags)
+ m.Load("epoll", &x.epoll)
+ m.AfterLoad(x.afterLoad)
+}
+
+func (x *EventPoll) beforeSave() {}
+func (x *EventPoll) save(m state.Map) {
+ x.beforeSave()
+ if !state.IsZeroValue(x.FilePipeSeek) { m.Failf("FilePipeSeek is %v, expected zero", x.FilePipeSeek) }
+ if !state.IsZeroValue(x.FileNotDirReaddir) { m.Failf("FileNotDirReaddir is %v, expected zero", x.FileNotDirReaddir) }
+ if !state.IsZeroValue(x.FileNoFsync) { m.Failf("FileNoFsync is %v, expected zero", x.FileNoFsync) }
+ if !state.IsZeroValue(x.FileNoopFlush) { m.Failf("FileNoopFlush is %v, expected zero", x.FileNoopFlush) }
+ if !state.IsZeroValue(x.FileNoIoctl) { m.Failf("FileNoIoctl is %v, expected zero", x.FileNoIoctl) }
+ if !state.IsZeroValue(x.FileNoMMap) { m.Failf("FileNoMMap is %v, expected zero", x.FileNoMMap) }
+ if !state.IsZeroValue(x.Queue) { m.Failf("Queue is %v, expected zero", x.Queue) }
+ m.Save("files", &x.files)
+ m.Save("readyList", &x.readyList)
+ m.Save("waitingList", &x.waitingList)
+ m.Save("disabledList", &x.disabledList)
+}
+
+func (x *EventPoll) load(m state.Map) {
+ m.Load("files", &x.files)
+ m.Load("readyList", &x.readyList)
+ m.Load("waitingList", &x.waitingList)
+ m.Load("disabledList", &x.disabledList)
+ m.AfterLoad(x.afterLoad)
+}
+
+func (x *pollEntryList) beforeSave() {}
+func (x *pollEntryList) save(m state.Map) {
+ x.beforeSave()
+ m.Save("head", &x.head)
+ m.Save("tail", &x.tail)
+}
+
+func (x *pollEntryList) afterLoad() {}
+func (x *pollEntryList) load(m state.Map) {
+ m.Load("head", &x.head)
+ m.Load("tail", &x.tail)
+}
+
+func (x *pollEntryEntry) beforeSave() {}
+func (x *pollEntryEntry) save(m state.Map) {
+ x.beforeSave()
+ m.Save("next", &x.next)
+ m.Save("prev", &x.prev)
+}
+
+func (x *pollEntryEntry) afterLoad() {}
+func (x *pollEntryEntry) load(m state.Map) {
+ m.Load("next", &x.next)
+ m.Load("prev", &x.prev)
+}
+
+func init() {
+ state.Register("epoll.FileIdentifier", (*FileIdentifier)(nil), state.Fns{Save: (*FileIdentifier).save, Load: (*FileIdentifier).load})
+ state.Register("epoll.pollEntry", (*pollEntry)(nil), state.Fns{Save: (*pollEntry).save, Load: (*pollEntry).load})
+ state.Register("epoll.EventPoll", (*EventPoll)(nil), state.Fns{Save: (*EventPoll).save, Load: (*EventPoll).load})
+ state.Register("epoll.pollEntryList", (*pollEntryList)(nil), state.Fns{Save: (*pollEntryList).save, Load: (*pollEntryList).load})
+ state.Register("epoll.pollEntryEntry", (*pollEntryEntry)(nil), state.Fns{Save: (*pollEntryEntry).save, Load: (*pollEntryEntry).load})
+}
diff --git a/pkg/sentry/kernel/epoll/epoll_test.go b/pkg/sentry/kernel/epoll/epoll_test.go
deleted file mode 100644
index 4a20d4c82..000000000
--- a/pkg/sentry/kernel/epoll/epoll_test.go
+++ /dev/null
@@ -1,54 +0,0 @@
-// Copyright 2018 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 epoll
-
-import (
- "testing"
-
- "gvisor.dev/gvisor/pkg/sentry/context/contexttest"
- "gvisor.dev/gvisor/pkg/sentry/fs/filetest"
- "gvisor.dev/gvisor/pkg/waiter"
-)
-
-func TestFileDestroyed(t *testing.T) {
- f := filetest.NewTestFile(t)
- id := FileIdentifier{f, 12}
-
- efile := NewEventPoll(contexttest.Context(t))
- e := efile.FileOperations.(*EventPoll)
- if err := e.AddEntry(id, 0, waiter.EventIn, [2]int32{}); err != nil {
- t.Fatalf("addEntry failed: %v", err)
- }
-
- // Check that we get an event reported twice in a row.
- evt := e.ReadEvents(1)
- if len(evt) != 1 {
- t.Fatalf("Unexpected number of ready events: want %v, got %v", 1, len(evt))
- }
-
- evt = e.ReadEvents(1)
- if len(evt) != 1 {
- t.Fatalf("Unexpected number of ready events: want %v, got %v", 1, len(evt))
- }
-
- // Destroy the file. Check that we get no more events.
- f.DecRef()
-
- evt = e.ReadEvents(1)
- if len(evt) != 0 {
- t.Fatalf("Unexpected number of ready events: want %v, got %v", 0, len(evt))
- }
-
-}
diff --git a/pkg/sentry/kernel/eventfd/BUILD b/pkg/sentry/kernel/eventfd/BUILD
deleted file mode 100644
index c831fbab2..000000000
--- a/pkg/sentry/kernel/eventfd/BUILD
+++ /dev/null
@@ -1,35 +0,0 @@
-load("@io_bazel_rules_go//go:def.bzl", "go_test")
-load("//tools/go_stateify:defs.bzl", "go_library")
-
-package(licenses = ["notice"])
-
-go_library(
- name = "eventfd",
- srcs = ["eventfd.go"],
- importpath = "gvisor.dev/gvisor/pkg/sentry/kernel/eventfd",
- visibility = ["//pkg/sentry:internal"],
- deps = [
- "//pkg/abi/linux",
- "//pkg/fdnotifier",
- "//pkg/sentry/context",
- "//pkg/sentry/fs",
- "//pkg/sentry/fs/anon",
- "//pkg/sentry/fs/fsutil",
- "//pkg/sentry/usermem",
- "//pkg/sync",
- "//pkg/syserror",
- "//pkg/waiter",
- ],
-)
-
-go_test(
- name = "eventfd_test",
- size = "small",
- srcs = ["eventfd_test.go"],
- embed = [":eventfd"],
- deps = [
- "//pkg/sentry/context/contexttest",
- "//pkg/sentry/usermem",
- "//pkg/waiter",
- ],
-)
diff --git a/pkg/sentry/kernel/eventfd/eventfd_state_autogen.go b/pkg/sentry/kernel/eventfd/eventfd_state_autogen.go
new file mode 100755
index 000000000..f6e94b521
--- /dev/null
+++ b/pkg/sentry/kernel/eventfd/eventfd_state_autogen.go
@@ -0,0 +1,27 @@
+// automatically generated by stateify.
+
+package eventfd
+
+import (
+ "gvisor.dev/gvisor/pkg/state"
+)
+
+func (x *EventOperations) beforeSave() {}
+func (x *EventOperations) save(m state.Map) {
+ x.beforeSave()
+ if !state.IsZeroValue(x.wq) { m.Failf("wq is %v, expected zero", x.wq) }
+ m.Save("val", &x.val)
+ m.Save("semMode", &x.semMode)
+ m.Save("hostfd", &x.hostfd)
+}
+
+func (x *EventOperations) afterLoad() {}
+func (x *EventOperations) load(m state.Map) {
+ m.Load("val", &x.val)
+ m.Load("semMode", &x.semMode)
+ m.Load("hostfd", &x.hostfd)
+}
+
+func init() {
+ state.Register("eventfd.EventOperations", (*EventOperations)(nil), state.Fns{Save: (*EventOperations).save, Load: (*EventOperations).load})
+}
diff --git a/pkg/sentry/kernel/eventfd/eventfd_test.go b/pkg/sentry/kernel/eventfd/eventfd_test.go
deleted file mode 100644
index 018c7f3ef..000000000
--- a/pkg/sentry/kernel/eventfd/eventfd_test.go
+++ /dev/null
@@ -1,78 +0,0 @@
-// Copyright 2018 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 eventfd
-
-import (
- "testing"
-
- "gvisor.dev/gvisor/pkg/sentry/context/contexttest"
- "gvisor.dev/gvisor/pkg/sentry/usermem"
- "gvisor.dev/gvisor/pkg/waiter"
-)
-
-func TestEventfd(t *testing.T) {
- initVals := []uint64{
- 0,
- // Using a non-zero initial value verifies that writing to an
- // eventfd signals when the eventfd's counter was already
- // non-zero.
- 343,
- }
-
- for _, initVal := range initVals {
- ctx := contexttest.Context(t)
-
- // Make a new event that is writable.
- event := New(ctx, initVal, false)
-
- // Register a callback for a write event.
- w, ch := waiter.NewChannelEntry(nil)
- event.EventRegister(&w, waiter.EventIn)
- defer event.EventUnregister(&w)
-
- data := []byte("00000124")
- // Create and submit a write request.
- n, err := event.Writev(ctx, usermem.BytesIOSequence(data))
- if err != nil {
- t.Fatal(err)
- }
- if n != 8 {
- t.Errorf("eventfd.write wrote %d bytes, not full int64", n)
- }
-
- // Check if the callback fired due to the write event.
- select {
- case <-ch:
- default:
- t.Errorf("Didn't get notified of EventIn after write")
- }
- }
-}
-
-func TestEventfdStat(t *testing.T) {
- ctx := contexttest.Context(t)
-
- // Make a new event that is writable.
- event := New(ctx, 0, false)
-
- // Create and submit an stat request.
- uattr, err := event.Dirent.Inode.UnstableAttr(ctx)
- if err != nil {
- t.Fatalf("eventfd stat request failed: %v", err)
- }
- if uattr.Size != 0 {
- t.Fatal("EventFD size should be 0")
- }
-}
diff --git a/pkg/sentry/kernel/fasync/BUILD b/pkg/sentry/kernel/fasync/BUILD
deleted file mode 100644
index 6b36bc63e..000000000
--- a/pkg/sentry/kernel/fasync/BUILD
+++ /dev/null
@@ -1,18 +0,0 @@
-load("//tools/go_stateify:defs.bzl", "go_library")
-
-package(licenses = ["notice"])
-
-go_library(
- name = "fasync",
- srcs = ["fasync.go"],
- importpath = "gvisor.dev/gvisor/pkg/sentry/kernel/fasync",
- visibility = ["//:sandbox"],
- deps = [
- "//pkg/abi/linux",
- "//pkg/sentry/fs",
- "//pkg/sentry/kernel",
- "//pkg/sentry/kernel/auth",
- "//pkg/sync",
- "//pkg/waiter",
- ],
-)
diff --git a/pkg/sentry/kernel/fasync/fasync_state_autogen.go b/pkg/sentry/kernel/fasync/fasync_state_autogen.go
new file mode 100755
index 000000000..c17daa8a1
--- /dev/null
+++ b/pkg/sentry/kernel/fasync/fasync_state_autogen.go
@@ -0,0 +1,32 @@
+// automatically generated by stateify.
+
+package fasync
+
+import (
+ "gvisor.dev/gvisor/pkg/state"
+)
+
+func (x *FileAsync) beforeSave() {}
+func (x *FileAsync) save(m state.Map) {
+ x.beforeSave()
+ m.Save("e", &x.e)
+ m.Save("requester", &x.requester)
+ m.Save("registered", &x.registered)
+ m.Save("recipientPG", &x.recipientPG)
+ m.Save("recipientTG", &x.recipientTG)
+ m.Save("recipientT", &x.recipientT)
+}
+
+func (x *FileAsync) afterLoad() {}
+func (x *FileAsync) load(m state.Map) {
+ m.Load("e", &x.e)
+ m.Load("requester", &x.requester)
+ m.Load("registered", &x.registered)
+ m.Load("recipientPG", &x.recipientPG)
+ m.Load("recipientTG", &x.recipientTG)
+ m.Load("recipientT", &x.recipientT)
+}
+
+func init() {
+ state.Register("fasync.FileAsync", (*FileAsync)(nil), state.Fns{Save: (*FileAsync).save, Load: (*FileAsync).load})
+}
diff --git a/pkg/sentry/kernel/fd_table_test.go b/pkg/sentry/kernel/fd_table_test.go
deleted file mode 100644
index eccb7d1e7..000000000
--- a/pkg/sentry/kernel/fd_table_test.go
+++ /dev/null
@@ -1,228 +0,0 @@
-// Copyright 2018 Google LLC
-//
-// 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 kernel
-
-import (
- "runtime"
- "testing"
-
- "gvisor.dev/gvisor/pkg/sentry/context"
- "gvisor.dev/gvisor/pkg/sentry/context/contexttest"
- "gvisor.dev/gvisor/pkg/sentry/fs"
- "gvisor.dev/gvisor/pkg/sentry/fs/filetest"
- "gvisor.dev/gvisor/pkg/sentry/limits"
- "gvisor.dev/gvisor/pkg/sync"
-)
-
-const (
- // maxFD is the maximum FD to try to create in the map.
- //
- // This number of open files has been seen in the wild.
- maxFD = 2 * 1024
-)
-
-func runTest(t testing.TB, fn func(ctx context.Context, fdTable *FDTable, file *fs.File, limitSet *limits.LimitSet)) {
- t.Helper() // Don't show in stacks.
-
- // Create the limits and context.
- limitSet := limits.NewLimitSet()
- limitSet.Set(limits.NumberOfFiles, limits.Limit{maxFD, maxFD}, true)
- ctx := contexttest.WithLimitSet(contexttest.Context(t), limitSet)
-
- // Create a test file.;
- file := filetest.NewTestFile(t)
-
- // Create the table.
- fdTable := new(FDTable)
- fdTable.init()
-
- // Run the test.
- fn(ctx, fdTable, file, limitSet)
-}
-
-// TestFDTableMany allocates maxFD FDs, i.e. maxes out the FDTable, until there
-// is no room, then makes sure that NewFDAt works and also that if we remove
-// one and add one that works too.
-func TestFDTableMany(t *testing.T) {
- runTest(t, func(ctx context.Context, fdTable *FDTable, file *fs.File, _ *limits.LimitSet) {
- for i := 0; i < maxFD; i++ {
- if _, err := fdTable.NewFDs(ctx, 0, []*fs.File{file}, FDFlags{}); err != nil {
- t.Fatalf("Allocated %v FDs but wanted to allocate %v", i, maxFD)
- }
- }
-
- if _, err := fdTable.NewFDs(ctx, 0, []*fs.File{file}, FDFlags{}); err == nil {
- t.Fatalf("fdTable.NewFDs(0, r) in full map: got nil, wanted error")
- }
-
- if err := fdTable.NewFDAt(ctx, 1, file, FDFlags{}); err != nil {
- t.Fatalf("fdTable.NewFDAt(1, r, FDFlags{}): got %v, wanted nil", err)
- }
-
- i := int32(2)
- fdTable.Remove(i)
- if fds, err := fdTable.NewFDs(ctx, 0, []*fs.File{file}, FDFlags{}); err != nil || fds[0] != i {
- t.Fatalf("Allocated %v FDs but wanted to allocate %v: %v", i, maxFD, err)
- }
- })
-}
-
-func TestFDTableOverLimit(t *testing.T) {
- runTest(t, func(ctx context.Context, fdTable *FDTable, file *fs.File, _ *limits.LimitSet) {
- if _, err := fdTable.NewFDs(ctx, maxFD, []*fs.File{file}, FDFlags{}); err == nil {
- t.Fatalf("fdTable.NewFDs(maxFD, f): got nil, wanted error")
- }
-
- if _, err := fdTable.NewFDs(ctx, maxFD-2, []*fs.File{file, file, file}, FDFlags{}); err == nil {
- t.Fatalf("fdTable.NewFDs(maxFD-2, {f,f,f}): got nil, wanted error")
- }
-
- if fds, err := fdTable.NewFDs(ctx, maxFD-3, []*fs.File{file, file, file}, FDFlags{}); err != nil {
- t.Fatalf("fdTable.NewFDs(maxFD-3, {f,f,f}): got %v, wanted nil", err)
- } else {
- for _, fd := range fds {
- fdTable.Remove(fd)
- }
- }
-
- if fds, err := fdTable.NewFDs(ctx, maxFD-1, []*fs.File{file}, FDFlags{}); err != nil || fds[0] != maxFD-1 {
- t.Fatalf("fdTable.NewFDAt(1, r, FDFlags{}): got %v, wanted nil", err)
- }
-
- if fds, err := fdTable.NewFDs(ctx, 0, []*fs.File{file}, FDFlags{}); err != nil {
- t.Fatalf("Adding an FD to a resized map: got %v, want nil", err)
- } else if len(fds) != 1 || fds[0] != 0 {
- t.Fatalf("Added an FD to a resized map: got %v, want {1}", fds)
- }
- })
-}
-
-// TestFDTable does a set of simple tests to make sure simple adds, removes,
-// GetRefs, and DecRefs work. The ordering is just weird enough that a
-// table-driven approach seemed clumsy.
-func TestFDTable(t *testing.T) {
- runTest(t, func(ctx context.Context, fdTable *FDTable, file *fs.File, limitSet *limits.LimitSet) {
- // Cap the limit at one.
- limitSet.Set(limits.NumberOfFiles, limits.Limit{1, maxFD}, true)
-
- if _, err := fdTable.NewFDs(ctx, 0, []*fs.File{file}, FDFlags{}); err != nil {
- t.Fatalf("Adding an FD to an empty 1-size map: got %v, want nil", err)
- }
-
- if _, err := fdTable.NewFDs(ctx, 0, []*fs.File{file}, FDFlags{}); err == nil {
- t.Fatalf("Adding an FD to a filled 1-size map: got nil, wanted an error")
- }
-
- // Remove the previous limit.
- limitSet.Set(limits.NumberOfFiles, limits.Limit{maxFD, maxFD}, true)
-
- if fds, err := fdTable.NewFDs(ctx, 0, []*fs.File{file}, FDFlags{}); err != nil {
- t.Fatalf("Adding an FD to a resized map: got %v, want nil", err)
- } else if len(fds) != 1 || fds[0] != 1 {
- t.Fatalf("Added an FD to a resized map: got %v, want {1}", fds)
- }
-
- if err := fdTable.NewFDAt(ctx, 1, file, FDFlags{}); err != nil {
- t.Fatalf("Replacing FD 1 via fdTable.NewFDAt(1, r, FDFlags{}): got %v, wanted nil", err)
- }
-
- if err := fdTable.NewFDAt(ctx, maxFD+1, file, FDFlags{}); err == nil {
- t.Fatalf("Using an FD that was too large via fdTable.NewFDAt(%v, r, FDFlags{}): got nil, wanted an error", maxFD+1)
- }
-
- if ref, _ := fdTable.Get(1); ref == nil {
- t.Fatalf("fdTable.Get(1): got nil, wanted %v", file)
- }
-
- if ref, _ := fdTable.Get(2); ref != nil {
- t.Fatalf("fdTable.Get(2): got a %v, wanted nil", ref)
- }
-
- ref := fdTable.Remove(1)
- if ref == nil {
- t.Fatalf("fdTable.Remove(1) for an existing FD: failed, want success")
- }
- ref.DecRef()
-
- if ref := fdTable.Remove(1); ref != nil {
- t.Fatalf("r.Remove(1) for a removed FD: got success, want failure")
- }
- })
-}
-
-func TestDescriptorFlags(t *testing.T) {
- runTest(t, func(ctx context.Context, fdTable *FDTable, file *fs.File, _ *limits.LimitSet) {
- if err := fdTable.NewFDAt(ctx, 2, file, FDFlags{CloseOnExec: true}); err != nil {
- t.Fatalf("fdTable.NewFDAt(2, r, FDFlags{}): got %v, wanted nil", err)
- }
-
- newFile, flags := fdTable.Get(2)
- if newFile == nil {
- t.Fatalf("fdTable.Get(2): got a %v, wanted nil", newFile)
- }
-
- if !flags.CloseOnExec {
- t.Fatalf("new File flags %v don't match original %d\n", flags, 0)
- }
- })
-}
-
-func BenchmarkFDLookupAndDecRef(b *testing.B) {
- b.StopTimer() // Setup.
-
- runTest(b, func(ctx context.Context, fdTable *FDTable, file *fs.File, _ *limits.LimitSet) {
- fds, err := fdTable.NewFDs(ctx, 0, []*fs.File{file, file, file, file, file}, FDFlags{})
- if err != nil {
- b.Fatalf("fdTable.NewFDs: got %v, wanted nil", err)
- }
-
- b.StartTimer() // Benchmark.
- for i := 0; i < b.N; i++ {
- tf, _ := fdTable.Get(fds[i%len(fds)])
- tf.DecRef()
- }
- })
-}
-
-func BenchmarkFDLookupAndDecRefConcurrent(b *testing.B) {
- b.StopTimer() // Setup.
-
- runTest(b, func(ctx context.Context, fdTable *FDTable, file *fs.File, _ *limits.LimitSet) {
- fds, err := fdTable.NewFDs(ctx, 0, []*fs.File{file, file, file, file, file}, FDFlags{})
- if err != nil {
- b.Fatalf("fdTable.NewFDs: got %v, wanted nil", err)
- }
-
- concurrency := runtime.GOMAXPROCS(0)
- if concurrency < 4 {
- concurrency = 4
- }
- each := b.N / concurrency
-
- b.StartTimer() // Benchmark.
- var wg sync.WaitGroup
- for i := 0; i < concurrency; i++ {
- wg.Add(1)
- go func() {
- defer wg.Done()
- for i := 0; i < each; i++ {
- tf, _ := fdTable.Get(fds[i%len(fds)])
- tf.DecRef()
- }
- }()
- }
- wg.Wait()
- })
-}
diff --git a/pkg/sentry/kernel/futex/BUILD b/pkg/sentry/kernel/futex/BUILD
deleted file mode 100644
index 50db443ce..000000000
--- a/pkg/sentry/kernel/futex/BUILD
+++ /dev/null
@@ -1,59 +0,0 @@
-load("@io_bazel_rules_go//go:def.bzl", "go_test")
-load("//tools/go_generics:defs.bzl", "go_template_instance")
-load("//tools/go_stateify:defs.bzl", "go_library")
-
-package(licenses = ["notice"])
-
-go_template_instance(
- name = "atomicptr_bucket",
- out = "atomicptr_bucket_unsafe.go",
- package = "futex",
- suffix = "Bucket",
- template = "//pkg/sync:generic_atomicptr",
- types = {
- "Value": "bucket",
- },
-)
-
-go_template_instance(
- name = "waiter_list",
- out = "waiter_list.go",
- package = "futex",
- prefix = "waiter",
- template = "//pkg/ilist:generic_list",
- types = {
- "Element": "*Waiter",
- "Linker": "*Waiter",
- },
-)
-
-go_library(
- name = "futex",
- srcs = [
- "atomicptr_bucket_unsafe.go",
- "futex.go",
- "waiter_list.go",
- ],
- importpath = "gvisor.dev/gvisor/pkg/sentry/kernel/futex",
- visibility = ["//pkg/sentry:internal"],
- deps = [
- "//pkg/abi/linux",
- "//pkg/log",
- "//pkg/sentry/context",
- "//pkg/sentry/memmap",
- "//pkg/sentry/usermem",
- "//pkg/sync",
- "//pkg/syserror",
- ],
-)
-
-go_test(
- name = "futex_test",
- size = "small",
- srcs = ["futex_test.go"],
- embed = [":futex"],
- deps = [
- "//pkg/sentry/usermem",
- "//pkg/sync",
- ],
-)
diff --git a/pkg/sentry/kernel/futex/atomicptr_bucket_unsafe.go b/pkg/sentry/kernel/futex/atomicptr_bucket_unsafe.go
new file mode 100755
index 000000000..d3fdf09b0
--- /dev/null
+++ b/pkg/sentry/kernel/futex/atomicptr_bucket_unsafe.go
@@ -0,0 +1,37 @@
+package futex
+
+import (
+ "sync/atomic"
+ "unsafe"
+)
+
+// An AtomicPtr is a pointer to a value of type Value that can be atomically
+// loaded and stored. The zero value of an AtomicPtr represents nil.
+//
+// Note that copying AtomicPtr by value performs a non-atomic read of the
+// stored pointer, which is unsafe if Store() can be called concurrently; in
+// this case, do `dst.Store(src.Load())` instead.
+//
+// +stateify savable
+type AtomicPtrBucket struct {
+ ptr unsafe.Pointer `state:".(*bucket)"`
+}
+
+func (p *AtomicPtrBucket) savePtr() *bucket {
+ return p.Load()
+}
+
+func (p *AtomicPtrBucket) loadPtr(v *bucket) {
+ p.Store(v)
+}
+
+// Load returns the value set by the most recent Store. It returns nil if there
+// has been no previous call to Store.
+func (p *AtomicPtrBucket) Load() *bucket {
+ return (*bucket)(atomic.LoadPointer(&p.ptr))
+}
+
+// Store sets the value returned by Load to x.
+func (p *AtomicPtrBucket) Store(x *bucket) {
+ atomic.StorePointer(&p.ptr, (unsafe.Pointer)(x))
+}
diff --git a/pkg/sentry/kernel/futex/futex_state_autogen.go b/pkg/sentry/kernel/futex/futex_state_autogen.go
new file mode 100755
index 000000000..b2ff39f17
--- /dev/null
+++ b/pkg/sentry/kernel/futex/futex_state_autogen.go
@@ -0,0 +1,75 @@
+// automatically generated by stateify.
+
+package futex
+
+import (
+ "gvisor.dev/gvisor/pkg/state"
+)
+
+func (x *AtomicPtrBucket) beforeSave() {}
+func (x *AtomicPtrBucket) save(m state.Map) {
+ x.beforeSave()
+ var ptr *bucket = x.savePtr()
+ m.SaveValue("ptr", ptr)
+}
+
+func (x *AtomicPtrBucket) afterLoad() {}
+func (x *AtomicPtrBucket) load(m state.Map) {
+ m.LoadValue("ptr", new(*bucket), func(y interface{}) { x.loadPtr(y.(*bucket)) })
+}
+
+func (x *bucket) beforeSave() {}
+func (x *bucket) save(m state.Map) {
+ x.beforeSave()
+ if !state.IsZeroValue(x.waiters) { m.Failf("waiters is %v, expected zero", x.waiters) }
+}
+
+func (x *bucket) afterLoad() {}
+func (x *bucket) load(m state.Map) {
+}
+
+func (x *Manager) beforeSave() {}
+func (x *Manager) save(m state.Map) {
+ x.beforeSave()
+ if !state.IsZeroValue(x.privateBuckets) { m.Failf("privateBuckets is %v, expected zero", x.privateBuckets) }
+ m.Save("sharedBucket", &x.sharedBucket)
+}
+
+func (x *Manager) afterLoad() {}
+func (x *Manager) load(m state.Map) {
+ m.Load("sharedBucket", &x.sharedBucket)
+}
+
+func (x *waiterList) beforeSave() {}
+func (x *waiterList) save(m state.Map) {
+ x.beforeSave()
+ m.Save("head", &x.head)
+ m.Save("tail", &x.tail)
+}
+
+func (x *waiterList) afterLoad() {}
+func (x *waiterList) load(m state.Map) {
+ m.Load("head", &x.head)
+ m.Load("tail", &x.tail)
+}
+
+func (x *waiterEntry) beforeSave() {}
+func (x *waiterEntry) save(m state.Map) {
+ x.beforeSave()
+ m.Save("next", &x.next)
+ m.Save("prev", &x.prev)
+}
+
+func (x *waiterEntry) afterLoad() {}
+func (x *waiterEntry) load(m state.Map) {
+ m.Load("next", &x.next)
+ m.Load("prev", &x.prev)
+}
+
+func init() {
+ state.Register("futex.AtomicPtrBucket", (*AtomicPtrBucket)(nil), state.Fns{Save: (*AtomicPtrBucket).save, Load: (*AtomicPtrBucket).load})
+ state.Register("futex.bucket", (*bucket)(nil), state.Fns{Save: (*bucket).save, Load: (*bucket).load})
+ state.Register("futex.Manager", (*Manager)(nil), state.Fns{Save: (*Manager).save, Load: (*Manager).load})
+ state.Register("futex.waiterList", (*waiterList)(nil), state.Fns{Save: (*waiterList).save, Load: (*waiterList).load})
+ state.Register("futex.waiterEntry", (*waiterEntry)(nil), state.Fns{Save: (*waiterEntry).save, Load: (*waiterEntry).load})
+}
diff --git a/pkg/sentry/kernel/futex/futex_test.go b/pkg/sentry/kernel/futex/futex_test.go
deleted file mode 100644
index c23126ca5..000000000
--- a/pkg/sentry/kernel/futex/futex_test.go
+++ /dev/null
@@ -1,530 +0,0 @@
-// Copyright 2018 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 futex
-
-import (
- "math"
- "runtime"
- "sync/atomic"
- "syscall"
- "testing"
- "unsafe"
-
- "gvisor.dev/gvisor/pkg/sentry/usermem"
- "gvisor.dev/gvisor/pkg/sync"
-)
-
-// testData implements the Target interface, and allows us to
-// treat the address passed for futex operations as an index in
-// a byte slice for testing simplicity.
-type testData []byte
-
-const sizeofInt32 = 4
-
-func newTestData(size uint) testData {
- return make([]byte, size)
-}
-
-func (t testData) SwapUint32(addr usermem.Addr, new uint32) (uint32, error) {
- val := atomic.SwapUint32((*uint32)(unsafe.Pointer(&t[addr])), new)
- return val, nil
-}
-
-func (t testData) CompareAndSwapUint32(addr usermem.Addr, old, new uint32) (uint32, error) {
- if atomic.CompareAndSwapUint32((*uint32)(unsafe.Pointer(&t[addr])), old, new) {
- return old, nil
- }
- return atomic.LoadUint32((*uint32)(unsafe.Pointer(&t[addr]))), nil
-}
-
-func (t testData) LoadUint32(addr usermem.Addr) (uint32, error) {
- return atomic.LoadUint32((*uint32)(unsafe.Pointer(&t[addr]))), nil
-}
-
-func (t testData) GetSharedKey(addr usermem.Addr) (Key, error) {
- return Key{
- Kind: KindSharedMappable,
- Offset: uint64(addr),
- }, nil
-}
-
-func futexKind(private bool) string {
- if private {
- return "private"
- }
- return "shared"
-}
-
-func newPreparedTestWaiter(t *testing.T, m *Manager, ta Target, addr usermem.Addr, private bool, val uint32, bitmask uint32) *Waiter {
- w := NewWaiter()
- if err := m.WaitPrepare(w, ta, addr, private, val, bitmask); err != nil {
- t.Fatalf("WaitPrepare failed: %v", err)
- }
- return w
-}
-
-func TestFutexWake(t *testing.T) {
- for _, private := range []bool{false, true} {
- t.Run(futexKind(private), func(t *testing.T) {
- m := NewManager()
- d := newTestData(sizeofInt32)
-
- // Start waiting for wakeup.
- w := newPreparedTestWaiter(t, m, d, 0, private, 0, ^uint32(0))
- defer m.WaitComplete(w)
-
- // Perform a wakeup.
- if n, err := m.Wake(d, 0, private, ^uint32(0), 1); err != nil || n != 1 {
- t.Errorf("Wake: got (%d, %v), wanted (1, nil)", n, err)
- }
-
- // Expect the waiter to have been woken.
- if !w.woken() {
- t.Error("waiter not woken")
- }
- })
- }
-}
-
-func TestFutexWakeBitmask(t *testing.T) {
- for _, private := range []bool{false, true} {
- t.Run(futexKind(private), func(t *testing.T) {
- m := NewManager()
- d := newTestData(sizeofInt32)
-
- // Start waiting for wakeup.
- w := newPreparedTestWaiter(t, m, d, 0, private, 0, 0x0000ffff)
- defer m.WaitComplete(w)
-
- // Perform a wakeup using the wrong bitmask.
- if n, err := m.Wake(d, 0, private, 0xffff0000, 1); err != nil || n != 0 {
- t.Errorf("Wake with non-matching bitmask: got (%d, %v), wanted (0, nil)", n, err)
- }
-
- // Expect the waiter to still be waiting.
- if w.woken() {
- t.Error("waiter woken unexpectedly")
- }
-
- // Perform a wakeup using the right bitmask.
- if n, err := m.Wake(d, 0, private, 0x00000001, 1); err != nil || n != 1 {
- t.Errorf("Wake with matching bitmask: got (%d, %v), wanted (1, nil)", n, err)
- }
-
- // Expect that the waiter was woken.
- if !w.woken() {
- t.Error("waiter not woken")
- }
- })
- }
-}
-
-func TestFutexWakeTwo(t *testing.T) {
- for _, private := range []bool{false, true} {
- t.Run(futexKind(private), func(t *testing.T) {
- m := NewManager()
- d := newTestData(sizeofInt32)
-
- // Start three waiters waiting for wakeup.
- var ws [3]*Waiter
- for i := range ws {
- ws[i] = newPreparedTestWaiter(t, m, d, 0, private, 0, ^uint32(0))
- defer m.WaitComplete(ws[i])
- }
-
- // Perform two wakeups.
- const wakeups = 2
- if n, err := m.Wake(d, 0, private, ^uint32(0), 2); err != nil || n != wakeups {
- t.Errorf("Wake: got (%d, %v), wanted (%d, nil)", n, err, wakeups)
- }
-
- // Expect that exactly two waiters were woken.
- // We don't get guarantees about exactly which two,
- // (although we expect them to be w1 and w2).
- awake := 0
- for i := range ws {
- if ws[i].woken() {
- awake++
- }
- }
- if awake != wakeups {
- t.Errorf("got %d woken waiters, wanted %d", awake, wakeups)
- }
- })
- }
-}
-
-func TestFutexWakeUnrelated(t *testing.T) {
- for _, private := range []bool{false, true} {
- t.Run(futexKind(private), func(t *testing.T) {
- m := NewManager()
- d := newTestData(2 * sizeofInt32)
-
- // Start two waiters waiting for wakeup on different addresses.
- w1 := newPreparedTestWaiter(t, m, d, 0*sizeofInt32, private, 0, ^uint32(0))
- defer m.WaitComplete(w1)
- w2 := newPreparedTestWaiter(t, m, d, 1*sizeofInt32, private, 0, ^uint32(0))
- defer m.WaitComplete(w2)
-
- // Perform two wakeups on the second address.
- if n, err := m.Wake(d, 1*sizeofInt32, private, ^uint32(0), 2); err != nil || n != 1 {
- t.Errorf("Wake: got (%d, %v), wanted (1, nil)", n, err)
- }
-
- // Expect that only the second waiter was woken.
- if w1.woken() {
- t.Error("w1 woken unexpectedly")
- }
- if !w2.woken() {
- t.Error("w2 not woken")
- }
- })
- }
-}
-
-func TestWakeOpEmpty(t *testing.T) {
- for _, private := range []bool{false, true} {
- t.Run(futexKind(private), func(t *testing.T) {
- m := NewManager()
- d := newTestData(2 * sizeofInt32)
-
- // Perform wakeups with no waiters.
- if n, err := m.WakeOp(d, 0, sizeofInt32, private, 10, 10, 0); err != nil || n != 0 {
- t.Fatalf("WakeOp: got (%d, %v), wanted (0, nil)", n, err)
- }
- })
- }
-}
-
-func TestWakeOpFirstNonEmpty(t *testing.T) {
- for _, private := range []bool{false, true} {
- t.Run(futexKind(private), func(t *testing.T) {
- m := NewManager()
- d := newTestData(8)
-
- // Add two waiters on address 0.
- w1 := newPreparedTestWaiter(t, m, d, 0, private, 0, ^uint32(0))
- defer m.WaitComplete(w1)
- w2 := newPreparedTestWaiter(t, m, d, 0, private, 0, ^uint32(0))
- defer m.WaitComplete(w2)
-
- // Perform 10 wakeups on address 0.
- if n, err := m.WakeOp(d, 0, sizeofInt32, private, 10, 0, 0); err != nil || n != 2 {
- t.Errorf("WakeOp: got (%d, %v), wanted (2, nil)", n, err)
- }
-
- // Expect that both waiters were woken.
- if !w1.woken() {
- t.Error("w1 not woken")
- }
- if !w2.woken() {
- t.Error("w2 not woken")
- }
- })
- }
-}
-
-func TestWakeOpSecondNonEmpty(t *testing.T) {
- for _, private := range []bool{false, true} {
- t.Run(futexKind(private), func(t *testing.T) {
- m := NewManager()
- d := newTestData(8)
-
- // Add two waiters on address sizeofInt32.
- w1 := newPreparedTestWaiter(t, m, d, sizeofInt32, private, 0, ^uint32(0))
- defer m.WaitComplete(w1)
- w2 := newPreparedTestWaiter(t, m, d, sizeofInt32, private, 0, ^uint32(0))
- defer m.WaitComplete(w2)
-
- // Perform 10 wakeups on address sizeofInt32 (contingent on
- // d.Op(0), which should succeed).
- if n, err := m.WakeOp(d, 0, sizeofInt32, private, 0, 10, 0); err != nil || n != 2 {
- t.Errorf("WakeOp: got (%d, %v), wanted (2, nil)", n, err)
- }
-
- // Expect that both waiters were woken.
- if !w1.woken() {
- t.Error("w1 not woken")
- }
- if !w2.woken() {
- t.Error("w2 not woken")
- }
- })
- }
-}
-
-func TestWakeOpSecondNonEmptyFailingOp(t *testing.T) {
- for _, private := range []bool{false, true} {
- t.Run(futexKind(private), func(t *testing.T) {
- m := NewManager()
- d := newTestData(8)
-
- // Add two waiters on address sizeofInt32.
- w1 := newPreparedTestWaiter(t, m, d, sizeofInt32, private, 0, ^uint32(0))
- defer m.WaitComplete(w1)
- w2 := newPreparedTestWaiter(t, m, d, sizeofInt32, private, 0, ^uint32(0))
- defer m.WaitComplete(w2)
-
- // Perform 10 wakeups on address sizeofInt32 (contingent on
- // d.Op(1), which should fail).
- if n, err := m.WakeOp(d, 0, sizeofInt32, private, 0, 10, 1); err != nil || n != 0 {
- t.Errorf("WakeOp: got (%d, %v), wanted (0, nil)", n, err)
- }
-
- // Expect that neither waiter was woken.
- if w1.woken() {
- t.Error("w1 woken unexpectedly")
- }
- if w2.woken() {
- t.Error("w2 woken unexpectedly")
- }
- })
- }
-}
-
-func TestWakeOpAllNonEmpty(t *testing.T) {
- for _, private := range []bool{false, true} {
- t.Run(futexKind(private), func(t *testing.T) {
- m := NewManager()
- d := newTestData(8)
-
- // Add two waiters on address 0.
- w1 := newPreparedTestWaiter(t, m, d, 0, private, 0, ^uint32(0))
- defer m.WaitComplete(w1)
- w2 := newPreparedTestWaiter(t, m, d, 0, private, 0, ^uint32(0))
- defer m.WaitComplete(w2)
-
- // Add two waiters on address sizeofInt32.
- w3 := newPreparedTestWaiter(t, m, d, sizeofInt32, private, 0, ^uint32(0))
- defer m.WaitComplete(w3)
- w4 := newPreparedTestWaiter(t, m, d, sizeofInt32, private, 0, ^uint32(0))
- defer m.WaitComplete(w4)
-
- // Perform 10 wakeups on address 0 (unconditionally), and 10
- // wakeups on address sizeofInt32 (contingent on d.Op(0), which
- // should succeed).
- if n, err := m.WakeOp(d, 0, sizeofInt32, private, 10, 10, 0); err != nil || n != 4 {
- t.Errorf("WakeOp: got (%d, %v), wanted (4, nil)", n, err)
- }
-
- // Expect that all waiters were woken.
- if !w1.woken() {
- t.Error("w1 not woken")
- }
- if !w2.woken() {
- t.Error("w2 not woken")
- }
- if !w3.woken() {
- t.Error("w3 not woken")
- }
- if !w4.woken() {
- t.Error("w4 not woken")
- }
- })
- }
-}
-
-func TestWakeOpAllNonEmptyFailingOp(t *testing.T) {
- for _, private := range []bool{false, true} {
- t.Run(futexKind(private), func(t *testing.T) {
- m := NewManager()
- d := newTestData(8)
-
- // Add two waiters on address 0.
- w1 := newPreparedTestWaiter(t, m, d, 0, private, 0, ^uint32(0))
- defer m.WaitComplete(w1)
- w2 := newPreparedTestWaiter(t, m, d, 0, private, 0, ^uint32(0))
- defer m.WaitComplete(w2)
-
- // Add two waiters on address sizeofInt32.
- w3 := newPreparedTestWaiter(t, m, d, sizeofInt32, private, 0, ^uint32(0))
- defer m.WaitComplete(w3)
- w4 := newPreparedTestWaiter(t, m, d, sizeofInt32, private, 0, ^uint32(0))
- defer m.WaitComplete(w4)
-
- // Perform 10 wakeups on address 0 (unconditionally), and 10
- // wakeups on address sizeofInt32 (contingent on d.Op(1), which
- // should fail).
- if n, err := m.WakeOp(d, 0, sizeofInt32, private, 10, 10, 1); err != nil || n != 2 {
- t.Errorf("WakeOp: got (%d, %v), wanted (2, nil)", n, err)
- }
-
- // Expect that only the first two waiters were woken.
- if !w1.woken() {
- t.Error("w1 not woken")
- }
- if !w2.woken() {
- t.Error("w2 not woken")
- }
- if w3.woken() {
- t.Error("w3 woken unexpectedly")
- }
- if w4.woken() {
- t.Error("w4 woken unexpectedly")
- }
- })
- }
-}
-
-func TestWakeOpSameAddress(t *testing.T) {
- for _, private := range []bool{false, true} {
- t.Run(futexKind(private), func(t *testing.T) {
- m := NewManager()
- d := newTestData(8)
-
- // Add four waiters on address 0.
- var ws [4]*Waiter
- for i := range ws {
- ws[i] = newPreparedTestWaiter(t, m, d, 0, private, 0, ^uint32(0))
- defer m.WaitComplete(ws[i])
- }
-
- // Perform 1 wakeup on address 0 (unconditionally), and 1 wakeup
- // on address 0 (contingent on d.Op(0), which should succeed).
- const wakeups = 2
- if n, err := m.WakeOp(d, 0, 0, private, 1, 1, 0); err != nil || n != wakeups {
- t.Errorf("WakeOp: got (%d, %v), wanted (%d, nil)", n, err, wakeups)
- }
-
- // Expect that exactly two waiters were woken.
- awake := 0
- for i := range ws {
- if ws[i].woken() {
- awake++
- }
- }
- if awake != wakeups {
- t.Errorf("got %d woken waiters, wanted %d", awake, wakeups)
- }
- })
- }
-}
-
-func TestWakeOpSameAddressFailingOp(t *testing.T) {
- for _, private := range []bool{false, true} {
- t.Run(futexKind(private), func(t *testing.T) {
- m := NewManager()
- d := newTestData(8)
-
- // Add four waiters on address 0.
- var ws [4]*Waiter
- for i := range ws {
- ws[i] = newPreparedTestWaiter(t, m, d, 0, private, 0, ^uint32(0))
- defer m.WaitComplete(ws[i])
- }
-
- // Perform 1 wakeup on address 0 (unconditionally), and 1 wakeup
- // on address 0 (contingent on d.Op(1), which should fail).
- const wakeups = 1
- if n, err := m.WakeOp(d, 0, 0, private, 1, 1, 1); err != nil || n != wakeups {
- t.Errorf("WakeOp: got (%d, %v), wanted (%d, nil)", n, err, wakeups)
- }
-
- // Expect that exactly one waiter was woken.
- awake := 0
- for i := range ws {
- if ws[i].woken() {
- awake++
- }
- }
- if awake != wakeups {
- t.Errorf("got %d woken waiters, wanted %d", awake, wakeups)
- }
- })
- }
-}
-
-const (
- testMutexSize = sizeofInt32
- testMutexLocked uint32 = 1
- testMutexUnlocked uint32 = 0
-)
-
-// testMutex ties together a testData slice, an address, and a
-// futex manager in order to implement the sync.Locker interface.
-// Beyond being used as a Locker, this is a simple mechanism for
-// changing the underlying values for simpler tests.
-type testMutex struct {
- a usermem.Addr
- d testData
- m *Manager
-}
-
-func newTestMutex(addr usermem.Addr, d testData, m *Manager) *testMutex {
- return &testMutex{a: addr, d: d, m: m}
-}
-
-// Lock acquires the testMutex.
-// This may wait for it to be available via the futex manager.
-func (t *testMutex) Lock() {
- for {
- // Attempt to grab the lock.
- if atomic.CompareAndSwapUint32(
- (*uint32)(unsafe.Pointer(&t.d[t.a])),
- testMutexUnlocked,
- testMutexLocked) {
- // Lock held.
- return
- }
-
- // Wait for it to be "not locked".
- w := NewWaiter()
- err := t.m.WaitPrepare(w, t.d, t.a, true, testMutexLocked, ^uint32(0))
- if err == syscall.EAGAIN {
- continue
- }
- if err != nil {
- // Should never happen.
- panic("WaitPrepare returned unexpected error: " + err.Error())
- }
- <-w.C
- t.m.WaitComplete(w)
- }
-}
-
-// Unlock releases the testMutex.
-// This will notify any waiters via the futex manager.
-func (t *testMutex) Unlock() {
- // Unlock.
- atomic.StoreUint32((*uint32)(unsafe.Pointer(&t.d[t.a])), testMutexUnlocked)
-
- // Notify all waiters.
- t.m.Wake(t.d, t.a, true, ^uint32(0), math.MaxInt32)
-}
-
-// This function was shamelessly stolen from mutex_test.go.
-func HammerMutex(l sync.Locker, loops int, cdone chan bool) {
- for i := 0; i < loops; i++ {
- l.Lock()
- runtime.Gosched()
- l.Unlock()
- }
- cdone <- true
-}
-
-func TestMutexStress(t *testing.T) {
- m := NewManager()
- d := newTestData(testMutexSize)
- tm := newTestMutex(0*testMutexSize, d, m)
- c := make(chan bool)
-
- for i := 0; i < 10; i++ {
- go HammerMutex(tm, 1000, c)
- }
-
- for i := 0; i < 10; i++ {
- <-c
- }
-}
diff --git a/pkg/sentry/kernel/futex/waiter_list.go b/pkg/sentry/kernel/futex/waiter_list.go
new file mode 100755
index 000000000..cca5c4721
--- /dev/null
+++ b/pkg/sentry/kernel/futex/waiter_list.go
@@ -0,0 +1,173 @@
+package futex
+
+// ElementMapper provides an identity mapping by default.
+//
+// This can be replaced to provide a struct that maps elements to linker
+// objects, if they are not the same. An ElementMapper is not typically
+// required if: Linker is left as is, Element is left as is, or Linker and
+// Element are the same type.
+type waiterElementMapper struct{}
+
+// linkerFor maps an Element to a Linker.
+//
+// This default implementation should be inlined.
+//
+//go:nosplit
+func (waiterElementMapper) linkerFor(elem *Waiter) *Waiter { return elem }
+
+// List is an intrusive list. Entries can be added to or removed from the list
+// in O(1) time and with no additional memory allocations.
+//
+// The zero value for List is an empty list ready to use.
+//
+// To iterate over a list (where l is a List):
+// for e := l.Front(); e != nil; e = e.Next() {
+// // do something with e.
+// }
+//
+// +stateify savable
+type waiterList struct {
+ head *Waiter
+ tail *Waiter
+}
+
+// Reset resets list l to the empty state.
+func (l *waiterList) Reset() {
+ l.head = nil
+ l.tail = nil
+}
+
+// Empty returns true iff the list is empty.
+func (l *waiterList) Empty() bool {
+ return l.head == nil
+}
+
+// Front returns the first element of list l or nil.
+func (l *waiterList) Front() *Waiter {
+ return l.head
+}
+
+// Back returns the last element of list l or nil.
+func (l *waiterList) Back() *Waiter {
+ return l.tail
+}
+
+// PushFront inserts the element e at the front of list l.
+func (l *waiterList) PushFront(e *Waiter) {
+ waiterElementMapper{}.linkerFor(e).SetNext(l.head)
+ waiterElementMapper{}.linkerFor(e).SetPrev(nil)
+
+ if l.head != nil {
+ waiterElementMapper{}.linkerFor(l.head).SetPrev(e)
+ } else {
+ l.tail = e
+ }
+
+ l.head = e
+}
+
+// PushBack inserts the element e at the back of list l.
+func (l *waiterList) PushBack(e *Waiter) {
+ waiterElementMapper{}.linkerFor(e).SetNext(nil)
+ waiterElementMapper{}.linkerFor(e).SetPrev(l.tail)
+
+ if l.tail != nil {
+ waiterElementMapper{}.linkerFor(l.tail).SetNext(e)
+ } else {
+ l.head = e
+ }
+
+ l.tail = e
+}
+
+// PushBackList inserts list m at the end of list l, emptying m.
+func (l *waiterList) PushBackList(m *waiterList) {
+ if l.head == nil {
+ l.head = m.head
+ l.tail = m.tail
+ } else if m.head != nil {
+ waiterElementMapper{}.linkerFor(l.tail).SetNext(m.head)
+ waiterElementMapper{}.linkerFor(m.head).SetPrev(l.tail)
+
+ l.tail = m.tail
+ }
+
+ m.head = nil
+ m.tail = nil
+}
+
+// InsertAfter inserts e after b.
+func (l *waiterList) InsertAfter(b, e *Waiter) {
+ a := waiterElementMapper{}.linkerFor(b).Next()
+ waiterElementMapper{}.linkerFor(e).SetNext(a)
+ waiterElementMapper{}.linkerFor(e).SetPrev(b)
+ waiterElementMapper{}.linkerFor(b).SetNext(e)
+
+ if a != nil {
+ waiterElementMapper{}.linkerFor(a).SetPrev(e)
+ } else {
+ l.tail = e
+ }
+}
+
+// InsertBefore inserts e before a.
+func (l *waiterList) InsertBefore(a, e *Waiter) {
+ b := waiterElementMapper{}.linkerFor(a).Prev()
+ waiterElementMapper{}.linkerFor(e).SetNext(a)
+ waiterElementMapper{}.linkerFor(e).SetPrev(b)
+ waiterElementMapper{}.linkerFor(a).SetPrev(e)
+
+ if b != nil {
+ waiterElementMapper{}.linkerFor(b).SetNext(e)
+ } else {
+ l.head = e
+ }
+}
+
+// Remove removes e from l.
+func (l *waiterList) Remove(e *Waiter) {
+ prev := waiterElementMapper{}.linkerFor(e).Prev()
+ next := waiterElementMapper{}.linkerFor(e).Next()
+
+ if prev != nil {
+ waiterElementMapper{}.linkerFor(prev).SetNext(next)
+ } else {
+ l.head = next
+ }
+
+ if next != nil {
+ waiterElementMapper{}.linkerFor(next).SetPrev(prev)
+ } else {
+ l.tail = prev
+ }
+}
+
+// Entry is a default implementation of Linker. Users can add anonymous fields
+// of this type to their structs to make them automatically implement the
+// methods needed by List.
+//
+// +stateify savable
+type waiterEntry struct {
+ next *Waiter
+ prev *Waiter
+}
+
+// Next returns the entry that follows e in the list.
+func (e *waiterEntry) Next() *Waiter {
+ return e.next
+}
+
+// Prev returns the entry that precedes e in the list.
+func (e *waiterEntry) Prev() *Waiter {
+ return e.prev
+}
+
+// SetNext assigns 'entry' as the entry that follows e in the list.
+func (e *waiterEntry) SetNext(elem *Waiter) {
+ e.next = elem
+}
+
+// SetPrev assigns 'entry' as the entry that precedes e in the list.
+func (e *waiterEntry) SetPrev(elem *Waiter) {
+ e.prev = elem
+}
diff --git a/pkg/sentry/kernel/g3doc/run_states.dot b/pkg/sentry/kernel/g3doc/run_states.dot
deleted file mode 100644
index 7861fe1f5..000000000
--- a/pkg/sentry/kernel/g3doc/run_states.dot
+++ /dev/null
@@ -1,99 +0,0 @@
-digraph {
- subgraph {
- App;
- }
- subgraph {
- Interrupt;
- InterruptAfterSignalDeliveryStop;
- }
- subgraph {
- Syscall;
- SyscallAfterPtraceEventSeccomp;
- SyscallEnter;
- SyscallAfterSyscallEnterStop;
- SyscallAfterSysemuStop;
- SyscallInvoke;
- SyscallAfterPtraceEventClone;
- SyscallAfterExecStop;
- SyscallAfterVforkStop;
- SyscallReinvoke;
- SyscallExit;
- }
- subgraph {
- Vsyscall;
- VsyscallAfterPtraceEventSeccomp;
- VsyscallInvoke;
- }
- subgraph {
- Exit;
- ExitMain; // leave thread group, release resources, reparent children, kill PID namespace and wait if TGID 1
- ExitNotify; // signal parent/tracer, become waitable
- ExitDone; // represented by t.runState == nil
- }
-
- // Task exit
- Exit -> ExitMain;
- ExitMain -> ExitNotify;
- ExitNotify -> ExitDone;
-
- // Execution of untrusted application code
- App -> App;
-
- // Interrupts (usually signal delivery)
- App -> Interrupt;
- Interrupt -> Interrupt; // if other interrupt conditions may still apply
- Interrupt -> Exit; // if killed
-
- // Syscalls
- App -> Syscall;
- Syscall -> SyscallEnter;
- SyscallEnter -> SyscallInvoke;
- SyscallInvoke -> SyscallExit;
- SyscallExit -> App;
-
- // exit, exit_group
- SyscallInvoke -> Exit;
-
- // execve
- SyscallInvoke -> SyscallAfterExecStop;
- SyscallAfterExecStop -> SyscallExit;
- SyscallAfterExecStop -> App; // fatal signal pending
-
- // vfork
- SyscallInvoke -> SyscallAfterVforkStop;
- SyscallAfterVforkStop -> SyscallExit;
-
- // Vsyscalls
- App -> Vsyscall;
- Vsyscall -> VsyscallInvoke;
- Vsyscall -> App; // fault while reading return address from stack
- VsyscallInvoke -> App;
-
- // ptrace-specific branches
- Interrupt -> InterruptAfterSignalDeliveryStop;
- InterruptAfterSignalDeliveryStop -> Interrupt;
- SyscallEnter -> SyscallAfterSyscallEnterStop;
- SyscallAfterSyscallEnterStop -> SyscallInvoke;
- SyscallAfterSyscallEnterStop -> SyscallExit; // skipped by tracer
- SyscallAfterSyscallEnterStop -> App; // fatal signal pending
- SyscallEnter -> SyscallAfterSysemuStop;
- SyscallAfterSysemuStop -> SyscallExit;
- SyscallAfterSysemuStop -> App; // fatal signal pending
- SyscallInvoke -> SyscallAfterPtraceEventClone;
- SyscallAfterPtraceEventClone -> SyscallExit;
- SyscallAfterPtraceEventClone -> SyscallAfterVforkStop;
-
- // seccomp
- Syscall -> App; // SECCOMP_RET_TRAP, SECCOMP_RET_ERRNO, SECCOMP_RET_KILL, SECCOMP_RET_TRACE without tracer
- Syscall -> SyscallAfterPtraceEventSeccomp; // SECCOMP_RET_TRACE
- SyscallAfterPtraceEventSeccomp -> SyscallEnter;
- SyscallAfterPtraceEventSeccomp -> SyscallExit; // skipped by tracer
- SyscallAfterPtraceEventSeccomp -> App; // fatal signal pending
- Vsyscall -> VsyscallAfterPtraceEventSeccomp;
- VsyscallAfterPtraceEventSeccomp -> VsyscallInvoke;
- VsyscallAfterPtraceEventSeccomp -> App;
-
- // Autosave
- SyscallInvoke -> SyscallReinvoke;
- SyscallReinvoke -> SyscallInvoke;
-}
diff --git a/pkg/sentry/kernel/g3doc/run_states.png b/pkg/sentry/kernel/g3doc/run_states.png
deleted file mode 100644
index b63b60f02..000000000
--- a/pkg/sentry/kernel/g3doc/run_states.png
+++ /dev/null
Binary files differ
diff --git a/pkg/sentry/kernel/kernel_state_autogen.go b/pkg/sentry/kernel/kernel_state_autogen.go
new file mode 100755
index 000000000..2c32826c5
--- /dev/null
+++ b/pkg/sentry/kernel/kernel_state_autogen.go
@@ -0,0 +1,1210 @@
+// automatically generated by stateify.
+
+package kernel
+
+import (
+ "gvisor.dev/gvisor/pkg/state"
+ "gvisor.dev/gvisor/pkg/bpf"
+ "gvisor.dev/gvisor/pkg/sentry/device"
+ "gvisor.dev/gvisor/pkg/tcpip"
+)
+
+func (x *abstractEndpoint) beforeSave() {}
+func (x *abstractEndpoint) save(m state.Map) {
+ x.beforeSave()
+ m.Save("ep", &x.ep)
+ m.Save("wr", &x.wr)
+ m.Save("name", &x.name)
+ m.Save("ns", &x.ns)
+}
+
+func (x *abstractEndpoint) afterLoad() {}
+func (x *abstractEndpoint) load(m state.Map) {
+ m.Load("ep", &x.ep)
+ m.Load("wr", &x.wr)
+ m.Load("name", &x.name)
+ m.Load("ns", &x.ns)
+}
+
+func (x *AbstractSocketNamespace) beforeSave() {}
+func (x *AbstractSocketNamespace) save(m state.Map) {
+ x.beforeSave()
+ m.Save("endpoints", &x.endpoints)
+}
+
+func (x *AbstractSocketNamespace) afterLoad() {}
+func (x *AbstractSocketNamespace) load(m state.Map) {
+ m.Load("endpoints", &x.endpoints)
+}
+
+func (x *FDFlags) beforeSave() {}
+func (x *FDFlags) save(m state.Map) {
+ x.beforeSave()
+ m.Save("CloseOnExec", &x.CloseOnExec)
+}
+
+func (x *FDFlags) afterLoad() {}
+func (x *FDFlags) load(m state.Map) {
+ m.Load("CloseOnExec", &x.CloseOnExec)
+}
+
+func (x *descriptor) beforeSave() {}
+func (x *descriptor) save(m state.Map) {
+ x.beforeSave()
+ m.Save("file", &x.file)
+ m.Save("flags", &x.flags)
+}
+
+func (x *descriptor) afterLoad() {}
+func (x *descriptor) load(m state.Map) {
+ m.Load("file", &x.file)
+ m.Load("flags", &x.flags)
+}
+
+func (x *FDTable) beforeSave() {}
+func (x *FDTable) save(m state.Map) {
+ x.beforeSave()
+ var descriptorTable map[int32]descriptor = x.saveDescriptorTable()
+ m.SaveValue("descriptorTable", descriptorTable)
+ m.Save("AtomicRefCount", &x.AtomicRefCount)
+ m.Save("k", &x.k)
+ m.Save("uid", &x.uid)
+ m.Save("next", &x.next)
+ m.Save("used", &x.used)
+}
+
+func (x *FDTable) afterLoad() {}
+func (x *FDTable) load(m state.Map) {
+ m.Load("AtomicRefCount", &x.AtomicRefCount)
+ m.Load("k", &x.k)
+ m.Load("uid", &x.uid)
+ m.Load("next", &x.next)
+ m.Load("used", &x.used)
+ m.LoadValue("descriptorTable", new(map[int32]descriptor), func(y interface{}) { x.loadDescriptorTable(y.(map[int32]descriptor)) })
+}
+
+func (x *FSContext) beforeSave() {}
+func (x *FSContext) save(m state.Map) {
+ x.beforeSave()
+ m.Save("AtomicRefCount", &x.AtomicRefCount)
+ m.Save("root", &x.root)
+ m.Save("cwd", &x.cwd)
+ m.Save("umask", &x.umask)
+}
+
+func (x *FSContext) afterLoad() {}
+func (x *FSContext) load(m state.Map) {
+ m.Load("AtomicRefCount", &x.AtomicRefCount)
+ m.Load("root", &x.root)
+ m.Load("cwd", &x.cwd)
+ m.Load("umask", &x.umask)
+}
+
+func (x *IPCNamespace) beforeSave() {}
+func (x *IPCNamespace) save(m state.Map) {
+ x.beforeSave()
+ m.Save("userNS", &x.userNS)
+ m.Save("semaphores", &x.semaphores)
+ m.Save("shms", &x.shms)
+}
+
+func (x *IPCNamespace) afterLoad() {}
+func (x *IPCNamespace) load(m state.Map) {
+ m.Load("userNS", &x.userNS)
+ m.Load("semaphores", &x.semaphores)
+ m.Load("shms", &x.shms)
+}
+
+func (x *Kernel) beforeSave() {}
+func (x *Kernel) save(m state.Map) {
+ x.beforeSave()
+ var danglingEndpoints []tcpip.Endpoint = x.saveDanglingEndpoints()
+ m.SaveValue("danglingEndpoints", danglingEndpoints)
+ var deviceRegistry *device.Registry = x.saveDeviceRegistry()
+ m.SaveValue("deviceRegistry", deviceRegistry)
+ m.Save("featureSet", &x.featureSet)
+ m.Save("timekeeper", &x.timekeeper)
+ m.Save("tasks", &x.tasks)
+ m.Save("rootUserNamespace", &x.rootUserNamespace)
+ m.Save("applicationCores", &x.applicationCores)
+ m.Save("useHostCores", &x.useHostCores)
+ m.Save("extraAuxv", &x.extraAuxv)
+ m.Save("vdso", &x.vdso)
+ m.Save("rootUTSNamespace", &x.rootUTSNamespace)
+ m.Save("rootIPCNamespace", &x.rootIPCNamespace)
+ m.Save("rootAbstractSocketNamespace", &x.rootAbstractSocketNamespace)
+ m.Save("futexes", &x.futexes)
+ m.Save("globalInit", &x.globalInit)
+ m.Save("realtimeClock", &x.realtimeClock)
+ m.Save("monotonicClock", &x.monotonicClock)
+ m.Save("syslog", &x.syslog)
+ m.Save("runningTasks", &x.runningTasks)
+ m.Save("cpuClock", &x.cpuClock)
+ m.Save("cpuClockTickerDisabled", &x.cpuClockTickerDisabled)
+ m.Save("cpuClockTickerSetting", &x.cpuClockTickerSetting)
+ m.Save("fdMapUids", &x.fdMapUids)
+ m.Save("uniqueID", &x.uniqueID)
+ m.Save("nextInotifyCookie", &x.nextInotifyCookie)
+ m.Save("netlinkPorts", &x.netlinkPorts)
+ m.Save("sockets", &x.sockets)
+ m.Save("nextSocketEntry", &x.nextSocketEntry)
+ m.Save("DirentCacheLimiter", &x.DirentCacheLimiter)
+}
+
+func (x *Kernel) afterLoad() {}
+func (x *Kernel) load(m state.Map) {
+ m.Load("featureSet", &x.featureSet)
+ m.Load("timekeeper", &x.timekeeper)
+ m.Load("tasks", &x.tasks)
+ m.Load("rootUserNamespace", &x.rootUserNamespace)
+ m.Load("applicationCores", &x.applicationCores)
+ m.Load("useHostCores", &x.useHostCores)
+ m.Load("extraAuxv", &x.extraAuxv)
+ m.Load("vdso", &x.vdso)
+ m.Load("rootUTSNamespace", &x.rootUTSNamespace)
+ m.Load("rootIPCNamespace", &x.rootIPCNamespace)
+ m.Load("rootAbstractSocketNamespace", &x.rootAbstractSocketNamespace)
+ m.Load("futexes", &x.futexes)
+ m.Load("globalInit", &x.globalInit)
+ m.Load("realtimeClock", &x.realtimeClock)
+ m.Load("monotonicClock", &x.monotonicClock)
+ m.Load("syslog", &x.syslog)
+ m.Load("runningTasks", &x.runningTasks)
+ m.Load("cpuClock", &x.cpuClock)
+ m.Load("cpuClockTickerDisabled", &x.cpuClockTickerDisabled)
+ m.Load("cpuClockTickerSetting", &x.cpuClockTickerSetting)
+ m.Load("fdMapUids", &x.fdMapUids)
+ m.Load("uniqueID", &x.uniqueID)
+ m.Load("nextInotifyCookie", &x.nextInotifyCookie)
+ m.Load("netlinkPorts", &x.netlinkPorts)
+ m.Load("sockets", &x.sockets)
+ m.Load("nextSocketEntry", &x.nextSocketEntry)
+ m.Load("DirentCacheLimiter", &x.DirentCacheLimiter)
+ m.LoadValue("danglingEndpoints", new([]tcpip.Endpoint), func(y interface{}) { x.loadDanglingEndpoints(y.([]tcpip.Endpoint)) })
+ m.LoadValue("deviceRegistry", new(*device.Registry), func(y interface{}) { x.loadDeviceRegistry(y.(*device.Registry)) })
+}
+
+func (x *SocketEntry) beforeSave() {}
+func (x *SocketEntry) save(m state.Map) {
+ x.beforeSave()
+ m.Save("socketEntry", &x.socketEntry)
+ m.Save("k", &x.k)
+ m.Save("Sock", &x.Sock)
+ m.Save("ID", &x.ID)
+}
+
+func (x *SocketEntry) afterLoad() {}
+func (x *SocketEntry) load(m state.Map) {
+ m.Load("socketEntry", &x.socketEntry)
+ m.Load("k", &x.k)
+ m.Load("Sock", &x.Sock)
+ m.Load("ID", &x.ID)
+}
+
+func (x *pendingSignals) beforeSave() {}
+func (x *pendingSignals) save(m state.Map) {
+ x.beforeSave()
+ var signals []savedPendingSignal = x.saveSignals()
+ m.SaveValue("signals", signals)
+}
+
+func (x *pendingSignals) afterLoad() {}
+func (x *pendingSignals) load(m state.Map) {
+ m.LoadValue("signals", new([]savedPendingSignal), func(y interface{}) { x.loadSignals(y.([]savedPendingSignal)) })
+}
+
+func (x *pendingSignalQueue) beforeSave() {}
+func (x *pendingSignalQueue) save(m state.Map) {
+ x.beforeSave()
+ m.Save("pendingSignalList", &x.pendingSignalList)
+ m.Save("length", &x.length)
+}
+
+func (x *pendingSignalQueue) afterLoad() {}
+func (x *pendingSignalQueue) load(m state.Map) {
+ m.Load("pendingSignalList", &x.pendingSignalList)
+ m.Load("length", &x.length)
+}
+
+func (x *pendingSignal) beforeSave() {}
+func (x *pendingSignal) save(m state.Map) {
+ x.beforeSave()
+ m.Save("pendingSignalEntry", &x.pendingSignalEntry)
+ m.Save("SignalInfo", &x.SignalInfo)
+ m.Save("timer", &x.timer)
+}
+
+func (x *pendingSignal) afterLoad() {}
+func (x *pendingSignal) load(m state.Map) {
+ m.Load("pendingSignalEntry", &x.pendingSignalEntry)
+ m.Load("SignalInfo", &x.SignalInfo)
+ m.Load("timer", &x.timer)
+}
+
+func (x *pendingSignalList) beforeSave() {}
+func (x *pendingSignalList) save(m state.Map) {
+ x.beforeSave()
+ m.Save("head", &x.head)
+ m.Save("tail", &x.tail)
+}
+
+func (x *pendingSignalList) afterLoad() {}
+func (x *pendingSignalList) load(m state.Map) {
+ m.Load("head", &x.head)
+ m.Load("tail", &x.tail)
+}
+
+func (x *pendingSignalEntry) beforeSave() {}
+func (x *pendingSignalEntry) save(m state.Map) {
+ x.beforeSave()
+ m.Save("next", &x.next)
+ m.Save("prev", &x.prev)
+}
+
+func (x *pendingSignalEntry) afterLoad() {}
+func (x *pendingSignalEntry) load(m state.Map) {
+ m.Load("next", &x.next)
+ m.Load("prev", &x.prev)
+}
+
+func (x *savedPendingSignal) beforeSave() {}
+func (x *savedPendingSignal) save(m state.Map) {
+ x.beforeSave()
+ m.Save("si", &x.si)
+ m.Save("timer", &x.timer)
+}
+
+func (x *savedPendingSignal) afterLoad() {}
+func (x *savedPendingSignal) load(m state.Map) {
+ m.Load("si", &x.si)
+ m.Load("timer", &x.timer)
+}
+
+func (x *IntervalTimer) beforeSave() {}
+func (x *IntervalTimer) save(m state.Map) {
+ x.beforeSave()
+ m.Save("timer", &x.timer)
+ m.Save("target", &x.target)
+ m.Save("signo", &x.signo)
+ m.Save("id", &x.id)
+ m.Save("sigval", &x.sigval)
+ m.Save("group", &x.group)
+ m.Save("sigpending", &x.sigpending)
+ m.Save("sigorphan", &x.sigorphan)
+ m.Save("overrunCur", &x.overrunCur)
+ m.Save("overrunLast", &x.overrunLast)
+}
+
+func (x *IntervalTimer) afterLoad() {}
+func (x *IntervalTimer) load(m state.Map) {
+ m.Load("timer", &x.timer)
+ m.Load("target", &x.target)
+ m.Load("signo", &x.signo)
+ m.Load("id", &x.id)
+ m.Load("sigval", &x.sigval)
+ m.Load("group", &x.group)
+ m.Load("sigpending", &x.sigpending)
+ m.Load("sigorphan", &x.sigorphan)
+ m.Load("overrunCur", &x.overrunCur)
+ m.Load("overrunLast", &x.overrunLast)
+}
+
+func (x *processGroupList) beforeSave() {}
+func (x *processGroupList) save(m state.Map) {
+ x.beforeSave()
+ m.Save("head", &x.head)
+ m.Save("tail", &x.tail)
+}
+
+func (x *processGroupList) afterLoad() {}
+func (x *processGroupList) load(m state.Map) {
+ m.Load("head", &x.head)
+ m.Load("tail", &x.tail)
+}
+
+func (x *processGroupEntry) beforeSave() {}
+func (x *processGroupEntry) save(m state.Map) {
+ x.beforeSave()
+ m.Save("next", &x.next)
+ m.Save("prev", &x.prev)
+}
+
+func (x *processGroupEntry) afterLoad() {}
+func (x *processGroupEntry) load(m state.Map) {
+ m.Load("next", &x.next)
+ m.Load("prev", &x.prev)
+}
+
+func (x *ptraceOptions) beforeSave() {}
+func (x *ptraceOptions) save(m state.Map) {
+ x.beforeSave()
+ m.Save("ExitKill", &x.ExitKill)
+ m.Save("SysGood", &x.SysGood)
+ m.Save("TraceClone", &x.TraceClone)
+ m.Save("TraceExec", &x.TraceExec)
+ m.Save("TraceExit", &x.TraceExit)
+ m.Save("TraceFork", &x.TraceFork)
+ m.Save("TraceSeccomp", &x.TraceSeccomp)
+ m.Save("TraceVfork", &x.TraceVfork)
+ m.Save("TraceVforkDone", &x.TraceVforkDone)
+}
+
+func (x *ptraceOptions) afterLoad() {}
+func (x *ptraceOptions) load(m state.Map) {
+ m.Load("ExitKill", &x.ExitKill)
+ m.Load("SysGood", &x.SysGood)
+ m.Load("TraceClone", &x.TraceClone)
+ m.Load("TraceExec", &x.TraceExec)
+ m.Load("TraceExit", &x.TraceExit)
+ m.Load("TraceFork", &x.TraceFork)
+ m.Load("TraceSeccomp", &x.TraceSeccomp)
+ m.Load("TraceVfork", &x.TraceVfork)
+ m.Load("TraceVforkDone", &x.TraceVforkDone)
+}
+
+func (x *ptraceStop) beforeSave() {}
+func (x *ptraceStop) save(m state.Map) {
+ x.beforeSave()
+ m.Save("frozen", &x.frozen)
+ m.Save("listen", &x.listen)
+}
+
+func (x *ptraceStop) afterLoad() {}
+func (x *ptraceStop) load(m state.Map) {
+ m.Load("frozen", &x.frozen)
+ m.Load("listen", &x.listen)
+}
+
+func (x *OldRSeqCriticalRegion) beforeSave() {}
+func (x *OldRSeqCriticalRegion) save(m state.Map) {
+ x.beforeSave()
+ m.Save("CriticalSection", &x.CriticalSection)
+ m.Save("Restart", &x.Restart)
+}
+
+func (x *OldRSeqCriticalRegion) afterLoad() {}
+func (x *OldRSeqCriticalRegion) load(m state.Map) {
+ m.Load("CriticalSection", &x.CriticalSection)
+ m.Load("Restart", &x.Restart)
+}
+
+func (x *sessionList) beforeSave() {}
+func (x *sessionList) save(m state.Map) {
+ x.beforeSave()
+ m.Save("head", &x.head)
+ m.Save("tail", &x.tail)
+}
+
+func (x *sessionList) afterLoad() {}
+func (x *sessionList) load(m state.Map) {
+ m.Load("head", &x.head)
+ m.Load("tail", &x.tail)
+}
+
+func (x *sessionEntry) beforeSave() {}
+func (x *sessionEntry) save(m state.Map) {
+ x.beforeSave()
+ m.Save("next", &x.next)
+ m.Save("prev", &x.prev)
+}
+
+func (x *sessionEntry) afterLoad() {}
+func (x *sessionEntry) load(m state.Map) {
+ m.Load("next", &x.next)
+ m.Load("prev", &x.prev)
+}
+
+func (x *Session) beforeSave() {}
+func (x *Session) save(m state.Map) {
+ x.beforeSave()
+ m.Save("refs", &x.refs)
+ m.Save("leader", &x.leader)
+ m.Save("id", &x.id)
+ m.Save("foreground", &x.foreground)
+ m.Save("processGroups", &x.processGroups)
+ m.Save("sessionEntry", &x.sessionEntry)
+}
+
+func (x *Session) afterLoad() {}
+func (x *Session) load(m state.Map) {
+ m.Load("refs", &x.refs)
+ m.Load("leader", &x.leader)
+ m.Load("id", &x.id)
+ m.Load("foreground", &x.foreground)
+ m.Load("processGroups", &x.processGroups)
+ m.Load("sessionEntry", &x.sessionEntry)
+}
+
+func (x *ProcessGroup) beforeSave() {}
+func (x *ProcessGroup) save(m state.Map) {
+ x.beforeSave()
+ m.Save("refs", &x.refs)
+ m.Save("originator", &x.originator)
+ m.Save("id", &x.id)
+ m.Save("session", &x.session)
+ m.Save("ancestors", &x.ancestors)
+ m.Save("processGroupEntry", &x.processGroupEntry)
+}
+
+func (x *ProcessGroup) afterLoad() {}
+func (x *ProcessGroup) load(m state.Map) {
+ m.Load("refs", &x.refs)
+ m.Load("originator", &x.originator)
+ m.Load("id", &x.id)
+ m.Load("session", &x.session)
+ m.Load("ancestors", &x.ancestors)
+ m.Load("processGroupEntry", &x.processGroupEntry)
+}
+
+func (x *SignalHandlers) beforeSave() {}
+func (x *SignalHandlers) save(m state.Map) {
+ x.beforeSave()
+ m.Save("actions", &x.actions)
+}
+
+func (x *SignalHandlers) afterLoad() {}
+func (x *SignalHandlers) load(m state.Map) {
+ m.Load("actions", &x.actions)
+}
+
+func (x *socketList) beforeSave() {}
+func (x *socketList) save(m state.Map) {
+ x.beforeSave()
+ m.Save("head", &x.head)
+ m.Save("tail", &x.tail)
+}
+
+func (x *socketList) afterLoad() {}
+func (x *socketList) load(m state.Map) {
+ m.Load("head", &x.head)
+ m.Load("tail", &x.tail)
+}
+
+func (x *socketEntry) beforeSave() {}
+func (x *socketEntry) save(m state.Map) {
+ x.beforeSave()
+ m.Save("next", &x.next)
+ m.Save("prev", &x.prev)
+}
+
+func (x *socketEntry) afterLoad() {}
+func (x *socketEntry) load(m state.Map) {
+ m.Load("next", &x.next)
+ m.Load("prev", &x.prev)
+}
+
+func (x *SyscallTable) beforeSave() {}
+func (x *SyscallTable) save(m state.Map) {
+ x.beforeSave()
+ m.Save("OS", &x.OS)
+ m.Save("Arch", &x.Arch)
+}
+
+func (x *SyscallTable) load(m state.Map) {
+ m.LoadWait("OS", &x.OS)
+ m.LoadWait("Arch", &x.Arch)
+ m.AfterLoad(x.afterLoad)
+}
+
+func (x *syslog) beforeSave() {}
+func (x *syslog) save(m state.Map) {
+ x.beforeSave()
+ m.Save("msg", &x.msg)
+}
+
+func (x *syslog) afterLoad() {}
+func (x *syslog) load(m state.Map) {
+ m.Load("msg", &x.msg)
+}
+
+func (x *Task) beforeSave() {}
+func (x *Task) save(m state.Map) {
+ x.beforeSave()
+ if !state.IsZeroValue(x.signalQueue) { m.Failf("signalQueue is %v, expected zero", x.signalQueue) }
+ var ptraceTracer *Task = x.savePtraceTracer()
+ m.SaveValue("ptraceTracer", ptraceTracer)
+ var syscallFilters []bpf.Program = x.saveSyscallFilters()
+ m.SaveValue("syscallFilters", syscallFilters)
+ m.Save("taskNode", &x.taskNode)
+ m.Save("runState", &x.runState)
+ m.Save("haveSyscallReturn", &x.haveSyscallReturn)
+ m.Save("gosched", &x.gosched)
+ m.Save("yieldCount", &x.yieldCount)
+ m.Save("pendingSignals", &x.pendingSignals)
+ m.Save("signalMask", &x.signalMask)
+ m.Save("realSignalMask", &x.realSignalMask)
+ m.Save("haveSavedSignalMask", &x.haveSavedSignalMask)
+ m.Save("savedSignalMask", &x.savedSignalMask)
+ m.Save("signalStack", &x.signalStack)
+ m.Save("groupStopPending", &x.groupStopPending)
+ m.Save("groupStopAcknowledged", &x.groupStopAcknowledged)
+ m.Save("trapStopPending", &x.trapStopPending)
+ m.Save("trapNotifyPending", &x.trapNotifyPending)
+ m.Save("stop", &x.stop)
+ m.Save("exitStatus", &x.exitStatus)
+ m.Save("syscallRestartBlock", &x.syscallRestartBlock)
+ m.Save("k", &x.k)
+ m.Save("containerID", &x.containerID)
+ m.Save("tc", &x.tc)
+ m.Save("fsContext", &x.fsContext)
+ m.Save("fdTable", &x.fdTable)
+ m.Save("vforkParent", &x.vforkParent)
+ m.Save("exitState", &x.exitState)
+ m.Save("exitTracerNotified", &x.exitTracerNotified)
+ m.Save("exitTracerAcked", &x.exitTracerAcked)
+ m.Save("exitParentNotified", &x.exitParentNotified)
+ m.Save("exitParentAcked", &x.exitParentAcked)
+ m.Save("ptraceTracees", &x.ptraceTracees)
+ m.Save("ptraceSeized", &x.ptraceSeized)
+ m.Save("ptraceOpts", &x.ptraceOpts)
+ m.Save("ptraceSyscallMode", &x.ptraceSyscallMode)
+ m.Save("ptraceSinglestep", &x.ptraceSinglestep)
+ m.Save("ptraceCode", &x.ptraceCode)
+ m.Save("ptraceSiginfo", &x.ptraceSiginfo)
+ m.Save("ptraceEventMsg", &x.ptraceEventMsg)
+ m.Save("ioUsage", &x.ioUsage)
+ m.Save("creds", &x.creds)
+ m.Save("utsns", &x.utsns)
+ m.Save("ipcns", &x.ipcns)
+ m.Save("abstractSockets", &x.abstractSockets)
+ m.Save("parentDeathSignal", &x.parentDeathSignal)
+ m.Save("cleartid", &x.cleartid)
+ m.Save("allowedCPUMask", &x.allowedCPUMask)
+ m.Save("cpu", &x.cpu)
+ m.Save("niceness", &x.niceness)
+ m.Save("numaPolicy", &x.numaPolicy)
+ m.Save("numaNodeMask", &x.numaNodeMask)
+ m.Save("netns", &x.netns)
+ m.Save("rseqCPU", &x.rseqCPU)
+ m.Save("oldRSeqCPUAddr", &x.oldRSeqCPUAddr)
+ m.Save("rseqAddr", &x.rseqAddr)
+ m.Save("rseqSignature", &x.rseqSignature)
+ m.Save("startTime", &x.startTime)
+}
+
+func (x *Task) load(m state.Map) {
+ m.Load("taskNode", &x.taskNode)
+ m.Load("runState", &x.runState)
+ m.Load("haveSyscallReturn", &x.haveSyscallReturn)
+ m.Load("gosched", &x.gosched)
+ m.Load("yieldCount", &x.yieldCount)
+ m.Load("pendingSignals", &x.pendingSignals)
+ m.Load("signalMask", &x.signalMask)
+ m.Load("realSignalMask", &x.realSignalMask)
+ m.Load("haveSavedSignalMask", &x.haveSavedSignalMask)
+ m.Load("savedSignalMask", &x.savedSignalMask)
+ m.Load("signalStack", &x.signalStack)
+ m.Load("groupStopPending", &x.groupStopPending)
+ m.Load("groupStopAcknowledged", &x.groupStopAcknowledged)
+ m.Load("trapStopPending", &x.trapStopPending)
+ m.Load("trapNotifyPending", &x.trapNotifyPending)
+ m.Load("stop", &x.stop)
+ m.Load("exitStatus", &x.exitStatus)
+ m.Load("syscallRestartBlock", &x.syscallRestartBlock)
+ m.Load("k", &x.k)
+ m.Load("containerID", &x.containerID)
+ m.Load("tc", &x.tc)
+ m.Load("fsContext", &x.fsContext)
+ m.Load("fdTable", &x.fdTable)
+ m.Load("vforkParent", &x.vforkParent)
+ m.Load("exitState", &x.exitState)
+ m.Load("exitTracerNotified", &x.exitTracerNotified)
+ m.Load("exitTracerAcked", &x.exitTracerAcked)
+ m.Load("exitParentNotified", &x.exitParentNotified)
+ m.Load("exitParentAcked", &x.exitParentAcked)
+ m.Load("ptraceTracees", &x.ptraceTracees)
+ m.Load("ptraceSeized", &x.ptraceSeized)
+ m.Load("ptraceOpts", &x.ptraceOpts)
+ m.Load("ptraceSyscallMode", &x.ptraceSyscallMode)
+ m.Load("ptraceSinglestep", &x.ptraceSinglestep)
+ m.Load("ptraceCode", &x.ptraceCode)
+ m.Load("ptraceSiginfo", &x.ptraceSiginfo)
+ m.Load("ptraceEventMsg", &x.ptraceEventMsg)
+ m.Load("ioUsage", &x.ioUsage)
+ m.Load("creds", &x.creds)
+ m.Load("utsns", &x.utsns)
+ m.Load("ipcns", &x.ipcns)
+ m.Load("abstractSockets", &x.abstractSockets)
+ m.Load("parentDeathSignal", &x.parentDeathSignal)
+ m.Load("cleartid", &x.cleartid)
+ m.Load("allowedCPUMask", &x.allowedCPUMask)
+ m.Load("cpu", &x.cpu)
+ m.Load("niceness", &x.niceness)
+ m.Load("numaPolicy", &x.numaPolicy)
+ m.Load("numaNodeMask", &x.numaNodeMask)
+ m.Load("netns", &x.netns)
+ m.Load("rseqCPU", &x.rseqCPU)
+ m.Load("oldRSeqCPUAddr", &x.oldRSeqCPUAddr)
+ m.Load("rseqAddr", &x.rseqAddr)
+ m.Load("rseqSignature", &x.rseqSignature)
+ m.Load("startTime", &x.startTime)
+ m.LoadValue("ptraceTracer", new(*Task), func(y interface{}) { x.loadPtraceTracer(y.(*Task)) })
+ m.LoadValue("syscallFilters", new([]bpf.Program), func(y interface{}) { x.loadSyscallFilters(y.([]bpf.Program)) })
+ m.AfterLoad(x.afterLoad)
+}
+
+func (x *runSyscallAfterPtraceEventClone) beforeSave() {}
+func (x *runSyscallAfterPtraceEventClone) save(m state.Map) {
+ x.beforeSave()
+ m.Save("vforkChild", &x.vforkChild)
+ m.Save("vforkChildTID", &x.vforkChildTID)
+}
+
+func (x *runSyscallAfterPtraceEventClone) afterLoad() {}
+func (x *runSyscallAfterPtraceEventClone) load(m state.Map) {
+ m.Load("vforkChild", &x.vforkChild)
+ m.Load("vforkChildTID", &x.vforkChildTID)
+}
+
+func (x *runSyscallAfterVforkStop) beforeSave() {}
+func (x *runSyscallAfterVforkStop) save(m state.Map) {
+ x.beforeSave()
+ m.Save("childTID", &x.childTID)
+}
+
+func (x *runSyscallAfterVforkStop) afterLoad() {}
+func (x *runSyscallAfterVforkStop) load(m state.Map) {
+ m.Load("childTID", &x.childTID)
+}
+
+func (x *vforkStop) beforeSave() {}
+func (x *vforkStop) save(m state.Map) {
+ x.beforeSave()
+}
+
+func (x *vforkStop) afterLoad() {}
+func (x *vforkStop) load(m state.Map) {
+}
+
+func (x *TaskContext) beforeSave() {}
+func (x *TaskContext) save(m state.Map) {
+ x.beforeSave()
+ m.Save("Name", &x.Name)
+ m.Save("Arch", &x.Arch)
+ m.Save("MemoryManager", &x.MemoryManager)
+ m.Save("fu", &x.fu)
+ m.Save("st", &x.st)
+}
+
+func (x *TaskContext) afterLoad() {}
+func (x *TaskContext) load(m state.Map) {
+ m.Load("Name", &x.Name)
+ m.Load("Arch", &x.Arch)
+ m.Load("MemoryManager", &x.MemoryManager)
+ m.Load("fu", &x.fu)
+ m.Load("st", &x.st)
+}
+
+func (x *execStop) beforeSave() {}
+func (x *execStop) save(m state.Map) {
+ x.beforeSave()
+}
+
+func (x *execStop) afterLoad() {}
+func (x *execStop) load(m state.Map) {
+}
+
+func (x *runSyscallAfterExecStop) beforeSave() {}
+func (x *runSyscallAfterExecStop) save(m state.Map) {
+ x.beforeSave()
+ m.Save("tc", &x.tc)
+}
+
+func (x *runSyscallAfterExecStop) afterLoad() {}
+func (x *runSyscallAfterExecStop) load(m state.Map) {
+ m.Load("tc", &x.tc)
+}
+
+func (x *ExitStatus) beforeSave() {}
+func (x *ExitStatus) save(m state.Map) {
+ x.beforeSave()
+ m.Save("Code", &x.Code)
+ m.Save("Signo", &x.Signo)
+}
+
+func (x *ExitStatus) afterLoad() {}
+func (x *ExitStatus) load(m state.Map) {
+ m.Load("Code", &x.Code)
+ m.Load("Signo", &x.Signo)
+}
+
+func (x *runExit) beforeSave() {}
+func (x *runExit) save(m state.Map) {
+ x.beforeSave()
+}
+
+func (x *runExit) afterLoad() {}
+func (x *runExit) load(m state.Map) {
+}
+
+func (x *runExitMain) beforeSave() {}
+func (x *runExitMain) save(m state.Map) {
+ x.beforeSave()
+}
+
+func (x *runExitMain) afterLoad() {}
+func (x *runExitMain) load(m state.Map) {
+}
+
+func (x *runExitNotify) beforeSave() {}
+func (x *runExitNotify) save(m state.Map) {
+ x.beforeSave()
+}
+
+func (x *runExitNotify) afterLoad() {}
+func (x *runExitNotify) load(m state.Map) {
+}
+
+func (x *taskList) beforeSave() {}
+func (x *taskList) save(m state.Map) {
+ x.beforeSave()
+ m.Save("head", &x.head)
+ m.Save("tail", &x.tail)
+}
+
+func (x *taskList) afterLoad() {}
+func (x *taskList) load(m state.Map) {
+ m.Load("head", &x.head)
+ m.Load("tail", &x.tail)
+}
+
+func (x *taskEntry) beforeSave() {}
+func (x *taskEntry) save(m state.Map) {
+ x.beforeSave()
+ m.Save("next", &x.next)
+ m.Save("prev", &x.prev)
+}
+
+func (x *taskEntry) afterLoad() {}
+func (x *taskEntry) load(m state.Map) {
+ m.Load("next", &x.next)
+ m.Load("prev", &x.prev)
+}
+
+func (x *runApp) beforeSave() {}
+func (x *runApp) save(m state.Map) {
+ x.beforeSave()
+}
+
+func (x *runApp) afterLoad() {}
+func (x *runApp) load(m state.Map) {
+}
+
+func (x *TaskGoroutineSchedInfo) beforeSave() {}
+func (x *TaskGoroutineSchedInfo) save(m state.Map) {
+ x.beforeSave()
+ m.Save("Timestamp", &x.Timestamp)
+ m.Save("State", &x.State)
+ m.Save("UserTicks", &x.UserTicks)
+ m.Save("SysTicks", &x.SysTicks)
+}
+
+func (x *TaskGoroutineSchedInfo) afterLoad() {}
+func (x *TaskGoroutineSchedInfo) load(m state.Map) {
+ m.Load("Timestamp", &x.Timestamp)
+ m.Load("State", &x.State)
+ m.Load("UserTicks", &x.UserTicks)
+ m.Load("SysTicks", &x.SysTicks)
+}
+
+func (x *taskClock) beforeSave() {}
+func (x *taskClock) save(m state.Map) {
+ x.beforeSave()
+ m.Save("t", &x.t)
+ m.Save("includeSys", &x.includeSys)
+}
+
+func (x *taskClock) afterLoad() {}
+func (x *taskClock) load(m state.Map) {
+ m.Load("t", &x.t)
+ m.Load("includeSys", &x.includeSys)
+}
+
+func (x *tgClock) beforeSave() {}
+func (x *tgClock) save(m state.Map) {
+ x.beforeSave()
+ m.Save("tg", &x.tg)
+ m.Save("includeSys", &x.includeSys)
+}
+
+func (x *tgClock) afterLoad() {}
+func (x *tgClock) load(m state.Map) {
+ m.Load("tg", &x.tg)
+ m.Load("includeSys", &x.includeSys)
+}
+
+func (x *groupStop) beforeSave() {}
+func (x *groupStop) save(m state.Map) {
+ x.beforeSave()
+}
+
+func (x *groupStop) afterLoad() {}
+func (x *groupStop) load(m state.Map) {
+}
+
+func (x *runInterrupt) beforeSave() {}
+func (x *runInterrupt) save(m state.Map) {
+ x.beforeSave()
+}
+
+func (x *runInterrupt) afterLoad() {}
+func (x *runInterrupt) load(m state.Map) {
+}
+
+func (x *runInterruptAfterSignalDeliveryStop) beforeSave() {}
+func (x *runInterruptAfterSignalDeliveryStop) save(m state.Map) {
+ x.beforeSave()
+}
+
+func (x *runInterruptAfterSignalDeliveryStop) afterLoad() {}
+func (x *runInterruptAfterSignalDeliveryStop) load(m state.Map) {
+}
+
+func (x *runSyscallAfterSyscallEnterStop) beforeSave() {}
+func (x *runSyscallAfterSyscallEnterStop) save(m state.Map) {
+ x.beforeSave()
+}
+
+func (x *runSyscallAfterSyscallEnterStop) afterLoad() {}
+func (x *runSyscallAfterSyscallEnterStop) load(m state.Map) {
+}
+
+func (x *runSyscallAfterSysemuStop) beforeSave() {}
+func (x *runSyscallAfterSysemuStop) save(m state.Map) {
+ x.beforeSave()
+}
+
+func (x *runSyscallAfterSysemuStop) afterLoad() {}
+func (x *runSyscallAfterSysemuStop) load(m state.Map) {
+}
+
+func (x *runSyscallReinvoke) beforeSave() {}
+func (x *runSyscallReinvoke) save(m state.Map) {
+ x.beforeSave()
+}
+
+func (x *runSyscallReinvoke) afterLoad() {}
+func (x *runSyscallReinvoke) load(m state.Map) {
+}
+
+func (x *runSyscallExit) beforeSave() {}
+func (x *runSyscallExit) save(m state.Map) {
+ x.beforeSave()
+}
+
+func (x *runSyscallExit) afterLoad() {}
+func (x *runSyscallExit) load(m state.Map) {
+}
+
+func (x *ThreadGroup) beforeSave() {}
+func (x *ThreadGroup) save(m state.Map) {
+ x.beforeSave()
+ var oldRSeqCritical *OldRSeqCriticalRegion = x.saveOldRSeqCritical()
+ m.SaveValue("oldRSeqCritical", oldRSeqCritical)
+ m.Save("threadGroupNode", &x.threadGroupNode)
+ m.Save("signalHandlers", &x.signalHandlers)
+ m.Save("pendingSignals", &x.pendingSignals)
+ m.Save("groupStopDequeued", &x.groupStopDequeued)
+ m.Save("groupStopSignal", &x.groupStopSignal)
+ m.Save("groupStopPendingCount", &x.groupStopPendingCount)
+ m.Save("groupStopComplete", &x.groupStopComplete)
+ m.Save("groupStopWaitable", &x.groupStopWaitable)
+ m.Save("groupContNotify", &x.groupContNotify)
+ m.Save("groupContInterrupted", &x.groupContInterrupted)
+ m.Save("groupContWaitable", &x.groupContWaitable)
+ m.Save("exiting", &x.exiting)
+ m.Save("exitStatus", &x.exitStatus)
+ m.Save("terminationSignal", &x.terminationSignal)
+ m.Save("itimerRealTimer", &x.itimerRealTimer)
+ m.Save("itimerVirtSetting", &x.itimerVirtSetting)
+ m.Save("itimerProfSetting", &x.itimerProfSetting)
+ m.Save("rlimitCPUSoftSetting", &x.rlimitCPUSoftSetting)
+ m.Save("cpuTimersEnabled", &x.cpuTimersEnabled)
+ m.Save("timers", &x.timers)
+ m.Save("nextTimerID", &x.nextTimerID)
+ m.Save("exitedCPUStats", &x.exitedCPUStats)
+ m.Save("childCPUStats", &x.childCPUStats)
+ m.Save("ioUsage", &x.ioUsage)
+ m.Save("maxRSS", &x.maxRSS)
+ m.Save("childMaxRSS", &x.childMaxRSS)
+ m.Save("limits", &x.limits)
+ m.Save("processGroup", &x.processGroup)
+ m.Save("execed", &x.execed)
+ m.Save("mounts", &x.mounts)
+ m.Save("tty", &x.tty)
+}
+
+func (x *ThreadGroup) afterLoad() {}
+func (x *ThreadGroup) load(m state.Map) {
+ m.Load("threadGroupNode", &x.threadGroupNode)
+ m.Load("signalHandlers", &x.signalHandlers)
+ m.Load("pendingSignals", &x.pendingSignals)
+ m.Load("groupStopDequeued", &x.groupStopDequeued)
+ m.Load("groupStopSignal", &x.groupStopSignal)
+ m.Load("groupStopPendingCount", &x.groupStopPendingCount)
+ m.Load("groupStopComplete", &x.groupStopComplete)
+ m.Load("groupStopWaitable", &x.groupStopWaitable)
+ m.Load("groupContNotify", &x.groupContNotify)
+ m.Load("groupContInterrupted", &x.groupContInterrupted)
+ m.Load("groupContWaitable", &x.groupContWaitable)
+ m.Load("exiting", &x.exiting)
+ m.Load("exitStatus", &x.exitStatus)
+ m.Load("terminationSignal", &x.terminationSignal)
+ m.Load("itimerRealTimer", &x.itimerRealTimer)
+ m.Load("itimerVirtSetting", &x.itimerVirtSetting)
+ m.Load("itimerProfSetting", &x.itimerProfSetting)
+ m.Load("rlimitCPUSoftSetting", &x.rlimitCPUSoftSetting)
+ m.Load("cpuTimersEnabled", &x.cpuTimersEnabled)
+ m.Load("timers", &x.timers)
+ m.Load("nextTimerID", &x.nextTimerID)
+ m.Load("exitedCPUStats", &x.exitedCPUStats)
+ m.Load("childCPUStats", &x.childCPUStats)
+ m.Load("ioUsage", &x.ioUsage)
+ m.Load("maxRSS", &x.maxRSS)
+ m.Load("childMaxRSS", &x.childMaxRSS)
+ m.Load("limits", &x.limits)
+ m.Load("processGroup", &x.processGroup)
+ m.Load("execed", &x.execed)
+ m.Load("mounts", &x.mounts)
+ m.Load("tty", &x.tty)
+ m.LoadValue("oldRSeqCritical", new(*OldRSeqCriticalRegion), func(y interface{}) { x.loadOldRSeqCritical(y.(*OldRSeqCriticalRegion)) })
+}
+
+func (x *itimerRealListener) beforeSave() {}
+func (x *itimerRealListener) save(m state.Map) {
+ x.beforeSave()
+ m.Save("tg", &x.tg)
+}
+
+func (x *itimerRealListener) afterLoad() {}
+func (x *itimerRealListener) load(m state.Map) {
+ m.Load("tg", &x.tg)
+}
+
+func (x *TaskSet) beforeSave() {}
+func (x *TaskSet) save(m state.Map) {
+ x.beforeSave()
+ m.Save("Root", &x.Root)
+ m.Save("sessions", &x.sessions)
+}
+
+func (x *TaskSet) afterLoad() {}
+func (x *TaskSet) load(m state.Map) {
+ m.Load("Root", &x.Root)
+ m.Load("sessions", &x.sessions)
+}
+
+func (x *PIDNamespace) beforeSave() {}
+func (x *PIDNamespace) save(m state.Map) {
+ x.beforeSave()
+ m.Save("owner", &x.owner)
+ m.Save("parent", &x.parent)
+ m.Save("userns", &x.userns)
+ m.Save("last", &x.last)
+ m.Save("tasks", &x.tasks)
+ m.Save("tids", &x.tids)
+ m.Save("tgids", &x.tgids)
+ m.Save("sessions", &x.sessions)
+ m.Save("sids", &x.sids)
+ m.Save("processGroups", &x.processGroups)
+ m.Save("pgids", &x.pgids)
+ m.Save("exiting", &x.exiting)
+}
+
+func (x *PIDNamespace) afterLoad() {}
+func (x *PIDNamespace) load(m state.Map) {
+ m.Load("owner", &x.owner)
+ m.Load("parent", &x.parent)
+ m.Load("userns", &x.userns)
+ m.Load("last", &x.last)
+ m.Load("tasks", &x.tasks)
+ m.Load("tids", &x.tids)
+ m.Load("tgids", &x.tgids)
+ m.Load("sessions", &x.sessions)
+ m.Load("sids", &x.sids)
+ m.Load("processGroups", &x.processGroups)
+ m.Load("pgids", &x.pgids)
+ m.Load("exiting", &x.exiting)
+}
+
+func (x *threadGroupNode) beforeSave() {}
+func (x *threadGroupNode) save(m state.Map) {
+ x.beforeSave()
+ m.Save("pidns", &x.pidns)
+ m.Save("leader", &x.leader)
+ m.Save("execing", &x.execing)
+ m.Save("tasks", &x.tasks)
+ m.Save("tasksCount", &x.tasksCount)
+ m.Save("liveTasks", &x.liveTasks)
+ m.Save("activeTasks", &x.activeTasks)
+}
+
+func (x *threadGroupNode) afterLoad() {}
+func (x *threadGroupNode) load(m state.Map) {
+ m.Load("pidns", &x.pidns)
+ m.Load("leader", &x.leader)
+ m.Load("execing", &x.execing)
+ m.Load("tasks", &x.tasks)
+ m.Load("tasksCount", &x.tasksCount)
+ m.Load("liveTasks", &x.liveTasks)
+ m.Load("activeTasks", &x.activeTasks)
+}
+
+func (x *taskNode) beforeSave() {}
+func (x *taskNode) save(m state.Map) {
+ x.beforeSave()
+ m.Save("tg", &x.tg)
+ m.Save("taskEntry", &x.taskEntry)
+ m.Save("parent", &x.parent)
+ m.Save("children", &x.children)
+ m.Save("childPIDNamespace", &x.childPIDNamespace)
+}
+
+func (x *taskNode) afterLoad() {}
+func (x *taskNode) load(m state.Map) {
+ m.LoadWait("tg", &x.tg)
+ m.Load("taskEntry", &x.taskEntry)
+ m.Load("parent", &x.parent)
+ m.Load("children", &x.children)
+ m.Load("childPIDNamespace", &x.childPIDNamespace)
+}
+
+func (x *Timekeeper) save(m state.Map) {
+ x.beforeSave()
+ m.Save("bootTime", &x.bootTime)
+ m.Save("saveMonotonic", &x.saveMonotonic)
+ m.Save("saveRealtime", &x.saveRealtime)
+ m.Save("params", &x.params)
+}
+
+func (x *Timekeeper) load(m state.Map) {
+ m.Load("bootTime", &x.bootTime)
+ m.Load("saveMonotonic", &x.saveMonotonic)
+ m.Load("saveRealtime", &x.saveRealtime)
+ m.Load("params", &x.params)
+ m.AfterLoad(x.afterLoad)
+}
+
+func (x *timekeeperClock) beforeSave() {}
+func (x *timekeeperClock) save(m state.Map) {
+ x.beforeSave()
+ m.Save("tk", &x.tk)
+ m.Save("c", &x.c)
+}
+
+func (x *timekeeperClock) afterLoad() {}
+func (x *timekeeperClock) load(m state.Map) {
+ m.Load("tk", &x.tk)
+ m.Load("c", &x.c)
+}
+
+func (x *TTY) beforeSave() {}
+func (x *TTY) save(m state.Map) {
+ x.beforeSave()
+ m.Save("Index", &x.Index)
+ m.Save("tg", &x.tg)
+}
+
+func (x *TTY) afterLoad() {}
+func (x *TTY) load(m state.Map) {
+ m.Load("Index", &x.Index)
+ m.Load("tg", &x.tg)
+}
+
+func (x *UTSNamespace) beforeSave() {}
+func (x *UTSNamespace) save(m state.Map) {
+ x.beforeSave()
+ m.Save("hostName", &x.hostName)
+ m.Save("domainName", &x.domainName)
+ m.Save("userns", &x.userns)
+}
+
+func (x *UTSNamespace) afterLoad() {}
+func (x *UTSNamespace) load(m state.Map) {
+ m.Load("hostName", &x.hostName)
+ m.Load("domainName", &x.domainName)
+ m.Load("userns", &x.userns)
+}
+
+func (x *VDSOParamPage) beforeSave() {}
+func (x *VDSOParamPage) save(m state.Map) {
+ x.beforeSave()
+ m.Save("mfp", &x.mfp)
+ m.Save("fr", &x.fr)
+ m.Save("seq", &x.seq)
+}
+
+func (x *VDSOParamPage) afterLoad() {}
+func (x *VDSOParamPage) load(m state.Map) {
+ m.Load("mfp", &x.mfp)
+ m.Load("fr", &x.fr)
+ m.Load("seq", &x.seq)
+}
+
+func init() {
+ state.Register("kernel.abstractEndpoint", (*abstractEndpoint)(nil), state.Fns{Save: (*abstractEndpoint).save, Load: (*abstractEndpoint).load})
+ state.Register("kernel.AbstractSocketNamespace", (*AbstractSocketNamespace)(nil), state.Fns{Save: (*AbstractSocketNamespace).save, Load: (*AbstractSocketNamespace).load})
+ state.Register("kernel.FDFlags", (*FDFlags)(nil), state.Fns{Save: (*FDFlags).save, Load: (*FDFlags).load})
+ state.Register("kernel.descriptor", (*descriptor)(nil), state.Fns{Save: (*descriptor).save, Load: (*descriptor).load})
+ state.Register("kernel.FDTable", (*FDTable)(nil), state.Fns{Save: (*FDTable).save, Load: (*FDTable).load})
+ state.Register("kernel.FSContext", (*FSContext)(nil), state.Fns{Save: (*FSContext).save, Load: (*FSContext).load})
+ state.Register("kernel.IPCNamespace", (*IPCNamespace)(nil), state.Fns{Save: (*IPCNamespace).save, Load: (*IPCNamespace).load})
+ state.Register("kernel.Kernel", (*Kernel)(nil), state.Fns{Save: (*Kernel).save, Load: (*Kernel).load})
+ state.Register("kernel.SocketEntry", (*SocketEntry)(nil), state.Fns{Save: (*SocketEntry).save, Load: (*SocketEntry).load})
+ state.Register("kernel.pendingSignals", (*pendingSignals)(nil), state.Fns{Save: (*pendingSignals).save, Load: (*pendingSignals).load})
+ state.Register("kernel.pendingSignalQueue", (*pendingSignalQueue)(nil), state.Fns{Save: (*pendingSignalQueue).save, Load: (*pendingSignalQueue).load})
+ state.Register("kernel.pendingSignal", (*pendingSignal)(nil), state.Fns{Save: (*pendingSignal).save, Load: (*pendingSignal).load})
+ state.Register("kernel.pendingSignalList", (*pendingSignalList)(nil), state.Fns{Save: (*pendingSignalList).save, Load: (*pendingSignalList).load})
+ state.Register("kernel.pendingSignalEntry", (*pendingSignalEntry)(nil), state.Fns{Save: (*pendingSignalEntry).save, Load: (*pendingSignalEntry).load})
+ state.Register("kernel.savedPendingSignal", (*savedPendingSignal)(nil), state.Fns{Save: (*savedPendingSignal).save, Load: (*savedPendingSignal).load})
+ state.Register("kernel.IntervalTimer", (*IntervalTimer)(nil), state.Fns{Save: (*IntervalTimer).save, Load: (*IntervalTimer).load})
+ state.Register("kernel.processGroupList", (*processGroupList)(nil), state.Fns{Save: (*processGroupList).save, Load: (*processGroupList).load})
+ state.Register("kernel.processGroupEntry", (*processGroupEntry)(nil), state.Fns{Save: (*processGroupEntry).save, Load: (*processGroupEntry).load})
+ state.Register("kernel.ptraceOptions", (*ptraceOptions)(nil), state.Fns{Save: (*ptraceOptions).save, Load: (*ptraceOptions).load})
+ state.Register("kernel.ptraceStop", (*ptraceStop)(nil), state.Fns{Save: (*ptraceStop).save, Load: (*ptraceStop).load})
+ state.Register("kernel.OldRSeqCriticalRegion", (*OldRSeqCriticalRegion)(nil), state.Fns{Save: (*OldRSeqCriticalRegion).save, Load: (*OldRSeqCriticalRegion).load})
+ state.Register("kernel.sessionList", (*sessionList)(nil), state.Fns{Save: (*sessionList).save, Load: (*sessionList).load})
+ state.Register("kernel.sessionEntry", (*sessionEntry)(nil), state.Fns{Save: (*sessionEntry).save, Load: (*sessionEntry).load})
+ state.Register("kernel.Session", (*Session)(nil), state.Fns{Save: (*Session).save, Load: (*Session).load})
+ state.Register("kernel.ProcessGroup", (*ProcessGroup)(nil), state.Fns{Save: (*ProcessGroup).save, Load: (*ProcessGroup).load})
+ state.Register("kernel.SignalHandlers", (*SignalHandlers)(nil), state.Fns{Save: (*SignalHandlers).save, Load: (*SignalHandlers).load})
+ state.Register("kernel.socketList", (*socketList)(nil), state.Fns{Save: (*socketList).save, Load: (*socketList).load})
+ state.Register("kernel.socketEntry", (*socketEntry)(nil), state.Fns{Save: (*socketEntry).save, Load: (*socketEntry).load})
+ state.Register("kernel.SyscallTable", (*SyscallTable)(nil), state.Fns{Save: (*SyscallTable).save, Load: (*SyscallTable).load})
+ state.Register("kernel.syslog", (*syslog)(nil), state.Fns{Save: (*syslog).save, Load: (*syslog).load})
+ state.Register("kernel.Task", (*Task)(nil), state.Fns{Save: (*Task).save, Load: (*Task).load})
+ state.Register("kernel.runSyscallAfterPtraceEventClone", (*runSyscallAfterPtraceEventClone)(nil), state.Fns{Save: (*runSyscallAfterPtraceEventClone).save, Load: (*runSyscallAfterPtraceEventClone).load})
+ state.Register("kernel.runSyscallAfterVforkStop", (*runSyscallAfterVforkStop)(nil), state.Fns{Save: (*runSyscallAfterVforkStop).save, Load: (*runSyscallAfterVforkStop).load})
+ state.Register("kernel.vforkStop", (*vforkStop)(nil), state.Fns{Save: (*vforkStop).save, Load: (*vforkStop).load})
+ state.Register("kernel.TaskContext", (*TaskContext)(nil), state.Fns{Save: (*TaskContext).save, Load: (*TaskContext).load})
+ state.Register("kernel.execStop", (*execStop)(nil), state.Fns{Save: (*execStop).save, Load: (*execStop).load})
+ state.Register("kernel.runSyscallAfterExecStop", (*runSyscallAfterExecStop)(nil), state.Fns{Save: (*runSyscallAfterExecStop).save, Load: (*runSyscallAfterExecStop).load})
+ state.Register("kernel.ExitStatus", (*ExitStatus)(nil), state.Fns{Save: (*ExitStatus).save, Load: (*ExitStatus).load})
+ state.Register("kernel.runExit", (*runExit)(nil), state.Fns{Save: (*runExit).save, Load: (*runExit).load})
+ state.Register("kernel.runExitMain", (*runExitMain)(nil), state.Fns{Save: (*runExitMain).save, Load: (*runExitMain).load})
+ state.Register("kernel.runExitNotify", (*runExitNotify)(nil), state.Fns{Save: (*runExitNotify).save, Load: (*runExitNotify).load})
+ state.Register("kernel.taskList", (*taskList)(nil), state.Fns{Save: (*taskList).save, Load: (*taskList).load})
+ state.Register("kernel.taskEntry", (*taskEntry)(nil), state.Fns{Save: (*taskEntry).save, Load: (*taskEntry).load})
+ state.Register("kernel.runApp", (*runApp)(nil), state.Fns{Save: (*runApp).save, Load: (*runApp).load})
+ state.Register("kernel.TaskGoroutineSchedInfo", (*TaskGoroutineSchedInfo)(nil), state.Fns{Save: (*TaskGoroutineSchedInfo).save, Load: (*TaskGoroutineSchedInfo).load})
+ state.Register("kernel.taskClock", (*taskClock)(nil), state.Fns{Save: (*taskClock).save, Load: (*taskClock).load})
+ state.Register("kernel.tgClock", (*tgClock)(nil), state.Fns{Save: (*tgClock).save, Load: (*tgClock).load})
+ state.Register("kernel.groupStop", (*groupStop)(nil), state.Fns{Save: (*groupStop).save, Load: (*groupStop).load})
+ state.Register("kernel.runInterrupt", (*runInterrupt)(nil), state.Fns{Save: (*runInterrupt).save, Load: (*runInterrupt).load})
+ state.Register("kernel.runInterruptAfterSignalDeliveryStop", (*runInterruptAfterSignalDeliveryStop)(nil), state.Fns{Save: (*runInterruptAfterSignalDeliveryStop).save, Load: (*runInterruptAfterSignalDeliveryStop).load})
+ state.Register("kernel.runSyscallAfterSyscallEnterStop", (*runSyscallAfterSyscallEnterStop)(nil), state.Fns{Save: (*runSyscallAfterSyscallEnterStop).save, Load: (*runSyscallAfterSyscallEnterStop).load})
+ state.Register("kernel.runSyscallAfterSysemuStop", (*runSyscallAfterSysemuStop)(nil), state.Fns{Save: (*runSyscallAfterSysemuStop).save, Load: (*runSyscallAfterSysemuStop).load})
+ state.Register("kernel.runSyscallReinvoke", (*runSyscallReinvoke)(nil), state.Fns{Save: (*runSyscallReinvoke).save, Load: (*runSyscallReinvoke).load})
+ state.Register("kernel.runSyscallExit", (*runSyscallExit)(nil), state.Fns{Save: (*runSyscallExit).save, Load: (*runSyscallExit).load})
+ state.Register("kernel.ThreadGroup", (*ThreadGroup)(nil), state.Fns{Save: (*ThreadGroup).save, Load: (*ThreadGroup).load})
+ state.Register("kernel.itimerRealListener", (*itimerRealListener)(nil), state.Fns{Save: (*itimerRealListener).save, Load: (*itimerRealListener).load})
+ state.Register("kernel.TaskSet", (*TaskSet)(nil), state.Fns{Save: (*TaskSet).save, Load: (*TaskSet).load})
+ state.Register("kernel.PIDNamespace", (*PIDNamespace)(nil), state.Fns{Save: (*PIDNamespace).save, Load: (*PIDNamespace).load})
+ state.Register("kernel.threadGroupNode", (*threadGroupNode)(nil), state.Fns{Save: (*threadGroupNode).save, Load: (*threadGroupNode).load})
+ state.Register("kernel.taskNode", (*taskNode)(nil), state.Fns{Save: (*taskNode).save, Load: (*taskNode).load})
+ state.Register("kernel.Timekeeper", (*Timekeeper)(nil), state.Fns{Save: (*Timekeeper).save, Load: (*Timekeeper).load})
+ state.Register("kernel.timekeeperClock", (*timekeeperClock)(nil), state.Fns{Save: (*timekeeperClock).save, Load: (*timekeeperClock).load})
+ state.Register("kernel.TTY", (*TTY)(nil), state.Fns{Save: (*TTY).save, Load: (*TTY).load})
+ state.Register("kernel.UTSNamespace", (*UTSNamespace)(nil), state.Fns{Save: (*UTSNamespace).save, Load: (*UTSNamespace).load})
+ state.Register("kernel.VDSOParamPage", (*VDSOParamPage)(nil), state.Fns{Save: (*VDSOParamPage).save, Load: (*VDSOParamPage).load})
+}
diff --git a/pkg/sentry/kernel/memevent/BUILD b/pkg/sentry/kernel/memevent/BUILD
deleted file mode 100644
index 7f36252a9..000000000
--- a/pkg/sentry/kernel/memevent/BUILD
+++ /dev/null
@@ -1,40 +0,0 @@
-load("//tools/go_stateify:defs.bzl", "go_library")
-load("@io_bazel_rules_go//proto:def.bzl", "go_proto_library")
-load("@rules_cc//cc:defs.bzl", "cc_proto_library")
-
-package(licenses = ["notice"])
-
-go_library(
- name = "memevent",
- srcs = ["memory_events.go"],
- importpath = "gvisor.dev/gvisor/pkg/sentry/kernel/memevent",
- visibility = ["//:sandbox"],
- deps = [
- ":memory_events_go_proto",
- "//pkg/eventchannel",
- "//pkg/log",
- "//pkg/metric",
- "//pkg/sentry/kernel",
- "//pkg/sentry/usage",
- "//pkg/sync",
- ],
-)
-
-proto_library(
- name = "memory_events_proto",
- srcs = ["memory_events.proto"],
- visibility = ["//visibility:public"],
-)
-
-cc_proto_library(
- name = "memory_events_cc_proto",
- visibility = ["//visibility:public"],
- deps = [":memory_events_proto"],
-)
-
-go_proto_library(
- name = "memory_events_go_proto",
- importpath = "gvisor.dev/gvisor/pkg/sentry/kernel/memevent/memory_events_go_proto",
- proto = ":memory_events_proto",
- visibility = ["//visibility:public"],
-)
diff --git a/pkg/sentry/kernel/memevent/memevent_state_autogen.go b/pkg/sentry/kernel/memevent/memevent_state_autogen.go
new file mode 100755
index 000000000..8bfbba80f
--- /dev/null
+++ b/pkg/sentry/kernel/memevent/memevent_state_autogen.go
@@ -0,0 +1,4 @@
+// automatically generated by stateify.
+
+package memevent
+
diff --git a/pkg/sentry/kernel/memevent/memory_events.go b/pkg/sentry/kernel/memevent/memory_events.go
index 200565bb8..200565bb8 100644..100755
--- a/pkg/sentry/kernel/memevent/memory_events.go
+++ b/pkg/sentry/kernel/memevent/memory_events.go
diff --git a/pkg/sentry/kernel/memevent/memory_events.proto b/pkg/sentry/kernel/memevent/memory_events.proto
deleted file mode 100644
index bf8029ff5..000000000
--- a/pkg/sentry/kernel/memevent/memory_events.proto
+++ /dev/null
@@ -1,29 +0,0 @@
-// Copyright 2018 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.
-
-syntax = "proto3";
-
-package gvisor;
-
-// MemoryUsageEvent describes the memory usage of the sandbox at a single
-// instant in time. These messages are emitted periodically on the eventchannel.
-message MemoryUsageEvent {
- // The total memory usage of the sandboxed application in bytes, calculated
- // using the 'fast' method.
- uint64 total = 1;
-
- // Memory used to back memory-mapped regions for files in the application, in
- // bytes. This corresponds to the usage.MemoryKind.Mapped memory type.
- uint64 mapped = 2;
-}
diff --git a/pkg/sentry/kernel/memevent/memory_events_go_proto/memory_events.pb.go b/pkg/sentry/kernel/memevent/memory_events_go_proto/memory_events.pb.go
new file mode 100755
index 000000000..f8b857fa9
--- /dev/null
+++ b/pkg/sentry/kernel/memevent/memory_events_go_proto/memory_events.pb.go
@@ -0,0 +1,88 @@
+// Code generated by protoc-gen-go. DO NOT EDIT.
+// source: pkg/sentry/kernel/memevent/memory_events.proto
+
+package gvisor
+
+import (
+ fmt "fmt"
+ proto "github.com/golang/protobuf/proto"
+ math "math"
+)
+
+// Reference imports to suppress errors if they are not otherwise used.
+var _ = proto.Marshal
+var _ = fmt.Errorf
+var _ = math.Inf
+
+// This is a compile-time assertion to ensure that this generated file
+// is compatible with the proto package it is being compiled against.
+// A compilation error at this line likely means your copy of the
+// proto package needs to be updated.
+const _ = proto.ProtoPackageIsVersion3 // please upgrade the proto package
+
+type MemoryUsageEvent struct {
+ Total uint64 `protobuf:"varint,1,opt,name=total,proto3" json:"total,omitempty"`
+ Mapped uint64 `protobuf:"varint,2,opt,name=mapped,proto3" json:"mapped,omitempty"`
+ XXX_NoUnkeyedLiteral struct{} `json:"-"`
+ XXX_unrecognized []byte `json:"-"`
+ XXX_sizecache int32 `json:"-"`
+}
+
+func (m *MemoryUsageEvent) Reset() { *m = MemoryUsageEvent{} }
+func (m *MemoryUsageEvent) String() string { return proto.CompactTextString(m) }
+func (*MemoryUsageEvent) ProtoMessage() {}
+func (*MemoryUsageEvent) Descriptor() ([]byte, []int) {
+ return fileDescriptor_cd85fc8d1130e4b0, []int{0}
+}
+
+func (m *MemoryUsageEvent) XXX_Unmarshal(b []byte) error {
+ return xxx_messageInfo_MemoryUsageEvent.Unmarshal(m, b)
+}
+func (m *MemoryUsageEvent) XXX_Marshal(b []byte, deterministic bool) ([]byte, error) {
+ return xxx_messageInfo_MemoryUsageEvent.Marshal(b, m, deterministic)
+}
+func (m *MemoryUsageEvent) XXX_Merge(src proto.Message) {
+ xxx_messageInfo_MemoryUsageEvent.Merge(m, src)
+}
+func (m *MemoryUsageEvent) XXX_Size() int {
+ return xxx_messageInfo_MemoryUsageEvent.Size(m)
+}
+func (m *MemoryUsageEvent) XXX_DiscardUnknown() {
+ xxx_messageInfo_MemoryUsageEvent.DiscardUnknown(m)
+}
+
+var xxx_messageInfo_MemoryUsageEvent proto.InternalMessageInfo
+
+func (m *MemoryUsageEvent) GetTotal() uint64 {
+ if m != nil {
+ return m.Total
+ }
+ return 0
+}
+
+func (m *MemoryUsageEvent) GetMapped() uint64 {
+ if m != nil {
+ return m.Mapped
+ }
+ return 0
+}
+
+func init() {
+ proto.RegisterType((*MemoryUsageEvent)(nil), "gvisor.MemoryUsageEvent")
+}
+
+func init() {
+ proto.RegisterFile("pkg/sentry/kernel/memevent/memory_events.proto", fileDescriptor_cd85fc8d1130e4b0)
+}
+
+var fileDescriptor_cd85fc8d1130e4b0 = []byte{
+ // 128 bytes of a gzipped FileDescriptorProto
+ 0x1f, 0x8b, 0x08, 0x00, 0x00, 0x00, 0x00, 0x00, 0x02, 0xff, 0xe2, 0xd2, 0x2b, 0xc8, 0x4e, 0xd7,
+ 0x2f, 0x4e, 0xcd, 0x2b, 0x29, 0xaa, 0xd4, 0xcf, 0x4e, 0x2d, 0xca, 0x4b, 0xcd, 0xd1, 0xcf, 0x4d,
+ 0xcd, 0x4d, 0x2d, 0x4b, 0xcd, 0x2b, 0x01, 0x31, 0xf2, 0x8b, 0x2a, 0xe3, 0xc1, 0x9c, 0x62, 0xbd,
+ 0x82, 0xa2, 0xfc, 0x92, 0x7c, 0x21, 0xb6, 0xf4, 0xb2, 0xcc, 0xe2, 0xfc, 0x22, 0x25, 0x07, 0x2e,
+ 0x01, 0x5f, 0xb0, 0x74, 0x68, 0x71, 0x62, 0x7a, 0xaa, 0x2b, 0x48, 0x89, 0x90, 0x08, 0x17, 0x6b,
+ 0x49, 0x7e, 0x49, 0x62, 0x8e, 0x04, 0xa3, 0x02, 0xa3, 0x06, 0x4b, 0x10, 0x84, 0x23, 0x24, 0xc6,
+ 0xc5, 0x96, 0x9b, 0x58, 0x50, 0x90, 0x9a, 0x22, 0xc1, 0x04, 0x16, 0x86, 0xf2, 0x92, 0xd8, 0xc0,
+ 0x06, 0x1a, 0x03, 0x02, 0x00, 0x00, 0xff, 0xff, 0x99, 0x31, 0x2f, 0x9d, 0x82, 0x00, 0x00, 0x00,
+}
diff --git a/pkg/sentry/kernel/pending_signals_list.go b/pkg/sentry/kernel/pending_signals_list.go
new file mode 100755
index 000000000..a3499371a
--- /dev/null
+++ b/pkg/sentry/kernel/pending_signals_list.go
@@ -0,0 +1,173 @@
+package kernel
+
+// ElementMapper provides an identity mapping by default.
+//
+// This can be replaced to provide a struct that maps elements to linker
+// objects, if they are not the same. An ElementMapper is not typically
+// required if: Linker is left as is, Element is left as is, or Linker and
+// Element are the same type.
+type pendingSignalElementMapper struct{}
+
+// linkerFor maps an Element to a Linker.
+//
+// This default implementation should be inlined.
+//
+//go:nosplit
+func (pendingSignalElementMapper) linkerFor(elem *pendingSignal) *pendingSignal { return elem }
+
+// List is an intrusive list. Entries can be added to or removed from the list
+// in O(1) time and with no additional memory allocations.
+//
+// The zero value for List is an empty list ready to use.
+//
+// To iterate over a list (where l is a List):
+// for e := l.Front(); e != nil; e = e.Next() {
+// // do something with e.
+// }
+//
+// +stateify savable
+type pendingSignalList struct {
+ head *pendingSignal
+ tail *pendingSignal
+}
+
+// Reset resets list l to the empty state.
+func (l *pendingSignalList) Reset() {
+ l.head = nil
+ l.tail = nil
+}
+
+// Empty returns true iff the list is empty.
+func (l *pendingSignalList) Empty() bool {
+ return l.head == nil
+}
+
+// Front returns the first element of list l or nil.
+func (l *pendingSignalList) Front() *pendingSignal {
+ return l.head
+}
+
+// Back returns the last element of list l or nil.
+func (l *pendingSignalList) Back() *pendingSignal {
+ return l.tail
+}
+
+// PushFront inserts the element e at the front of list l.
+func (l *pendingSignalList) PushFront(e *pendingSignal) {
+ pendingSignalElementMapper{}.linkerFor(e).SetNext(l.head)
+ pendingSignalElementMapper{}.linkerFor(e).SetPrev(nil)
+
+ if l.head != nil {
+ pendingSignalElementMapper{}.linkerFor(l.head).SetPrev(e)
+ } else {
+ l.tail = e
+ }
+
+ l.head = e
+}
+
+// PushBack inserts the element e at the back of list l.
+func (l *pendingSignalList) PushBack(e *pendingSignal) {
+ pendingSignalElementMapper{}.linkerFor(e).SetNext(nil)
+ pendingSignalElementMapper{}.linkerFor(e).SetPrev(l.tail)
+
+ if l.tail != nil {
+ pendingSignalElementMapper{}.linkerFor(l.tail).SetNext(e)
+ } else {
+ l.head = e
+ }
+
+ l.tail = e
+}
+
+// PushBackList inserts list m at the end of list l, emptying m.
+func (l *pendingSignalList) PushBackList(m *pendingSignalList) {
+ if l.head == nil {
+ l.head = m.head
+ l.tail = m.tail
+ } else if m.head != nil {
+ pendingSignalElementMapper{}.linkerFor(l.tail).SetNext(m.head)
+ pendingSignalElementMapper{}.linkerFor(m.head).SetPrev(l.tail)
+
+ l.tail = m.tail
+ }
+
+ m.head = nil
+ m.tail = nil
+}
+
+// InsertAfter inserts e after b.
+func (l *pendingSignalList) InsertAfter(b, e *pendingSignal) {
+ a := pendingSignalElementMapper{}.linkerFor(b).Next()
+ pendingSignalElementMapper{}.linkerFor(e).SetNext(a)
+ pendingSignalElementMapper{}.linkerFor(e).SetPrev(b)
+ pendingSignalElementMapper{}.linkerFor(b).SetNext(e)
+
+ if a != nil {
+ pendingSignalElementMapper{}.linkerFor(a).SetPrev(e)
+ } else {
+ l.tail = e
+ }
+}
+
+// InsertBefore inserts e before a.
+func (l *pendingSignalList) InsertBefore(a, e *pendingSignal) {
+ b := pendingSignalElementMapper{}.linkerFor(a).Prev()
+ pendingSignalElementMapper{}.linkerFor(e).SetNext(a)
+ pendingSignalElementMapper{}.linkerFor(e).SetPrev(b)
+ pendingSignalElementMapper{}.linkerFor(a).SetPrev(e)
+
+ if b != nil {
+ pendingSignalElementMapper{}.linkerFor(b).SetNext(e)
+ } else {
+ l.head = e
+ }
+}
+
+// Remove removes e from l.
+func (l *pendingSignalList) Remove(e *pendingSignal) {
+ prev := pendingSignalElementMapper{}.linkerFor(e).Prev()
+ next := pendingSignalElementMapper{}.linkerFor(e).Next()
+
+ if prev != nil {
+ pendingSignalElementMapper{}.linkerFor(prev).SetNext(next)
+ } else {
+ l.head = next
+ }
+
+ if next != nil {
+ pendingSignalElementMapper{}.linkerFor(next).SetPrev(prev)
+ } else {
+ l.tail = prev
+ }
+}
+
+// Entry is a default implementation of Linker. Users can add anonymous fields
+// of this type to their structs to make them automatically implement the
+// methods needed by List.
+//
+// +stateify savable
+type pendingSignalEntry struct {
+ next *pendingSignal
+ prev *pendingSignal
+}
+
+// Next returns the entry that follows e in the list.
+func (e *pendingSignalEntry) Next() *pendingSignal {
+ return e.next
+}
+
+// Prev returns the entry that precedes e in the list.
+func (e *pendingSignalEntry) Prev() *pendingSignal {
+ return e.prev
+}
+
+// SetNext assigns 'entry' as the entry that follows e in the list.
+func (e *pendingSignalEntry) SetNext(elem *pendingSignal) {
+ e.next = elem
+}
+
+// SetPrev assigns 'entry' as the entry that precedes e in the list.
+func (e *pendingSignalEntry) SetPrev(elem *pendingSignal) {
+ e.prev = elem
+}
diff --git a/pkg/sentry/kernel/pipe/BUILD b/pkg/sentry/kernel/pipe/BUILD
deleted file mode 100644
index 5eeaeff66..000000000
--- a/pkg/sentry/kernel/pipe/BUILD
+++ /dev/null
@@ -1,69 +0,0 @@
-load("@io_bazel_rules_go//go:def.bzl", "go_test")
-load("//tools/go_generics:defs.bzl", "go_template_instance")
-load("//tools/go_stateify:defs.bzl", "go_library")
-
-package(licenses = ["notice"])
-
-go_template_instance(
- name = "buffer_list",
- out = "buffer_list.go",
- package = "pipe",
- prefix = "buffer",
- template = "//pkg/ilist:generic_list",
- types = {
- "Element": "*buffer",
- "Linker": "*buffer",
- },
-)
-
-go_library(
- name = "pipe",
- srcs = [
- "buffer.go",
- "buffer_list.go",
- "device.go",
- "node.go",
- "pipe.go",
- "pipe_util.go",
- "reader.go",
- "reader_writer.go",
- "vfs.go",
- "writer.go",
- ],
- importpath = "gvisor.dev/gvisor/pkg/sentry/kernel/pipe",
- visibility = ["//pkg/sentry:internal"],
- deps = [
- "//pkg/abi/linux",
- "//pkg/amutex",
- "//pkg/sentry/arch",
- "//pkg/sentry/context",
- "//pkg/sentry/device",
- "//pkg/sentry/fs",
- "//pkg/sentry/fs/fsutil",
- "//pkg/sentry/safemem",
- "//pkg/sentry/usermem",
- "//pkg/sentry/vfs",
- "//pkg/sync",
- "//pkg/syserror",
- "//pkg/waiter",
- ],
-)
-
-go_test(
- name = "pipe_test",
- size = "small",
- srcs = [
- "buffer_test.go",
- "node_test.go",
- "pipe_test.go",
- ],
- embed = [":pipe"],
- deps = [
- "//pkg/sentry/context",
- "//pkg/sentry/context/contexttest",
- "//pkg/sentry/fs",
- "//pkg/sentry/usermem",
- "//pkg/syserror",
- "//pkg/waiter",
- ],
-)
diff --git a/pkg/sentry/kernel/pipe/buffer_list.go b/pkg/sentry/kernel/pipe/buffer_list.go
new file mode 100755
index 000000000..42ec78788
--- /dev/null
+++ b/pkg/sentry/kernel/pipe/buffer_list.go
@@ -0,0 +1,173 @@
+package pipe
+
+// ElementMapper provides an identity mapping by default.
+//
+// This can be replaced to provide a struct that maps elements to linker
+// objects, if they are not the same. An ElementMapper is not typically
+// required if: Linker is left as is, Element is left as is, or Linker and
+// Element are the same type.
+type bufferElementMapper struct{}
+
+// linkerFor maps an Element to a Linker.
+//
+// This default implementation should be inlined.
+//
+//go:nosplit
+func (bufferElementMapper) linkerFor(elem *buffer) *buffer { return elem }
+
+// List is an intrusive list. Entries can be added to or removed from the list
+// in O(1) time and with no additional memory allocations.
+//
+// The zero value for List is an empty list ready to use.
+//
+// To iterate over a list (where l is a List):
+// for e := l.Front(); e != nil; e = e.Next() {
+// // do something with e.
+// }
+//
+// +stateify savable
+type bufferList struct {
+ head *buffer
+ tail *buffer
+}
+
+// Reset resets list l to the empty state.
+func (l *bufferList) Reset() {
+ l.head = nil
+ l.tail = nil
+}
+
+// Empty returns true iff the list is empty.
+func (l *bufferList) Empty() bool {
+ return l.head == nil
+}
+
+// Front returns the first element of list l or nil.
+func (l *bufferList) Front() *buffer {
+ return l.head
+}
+
+// Back returns the last element of list l or nil.
+func (l *bufferList) Back() *buffer {
+ return l.tail
+}
+
+// PushFront inserts the element e at the front of list l.
+func (l *bufferList) PushFront(e *buffer) {
+ bufferElementMapper{}.linkerFor(e).SetNext(l.head)
+ bufferElementMapper{}.linkerFor(e).SetPrev(nil)
+
+ if l.head != nil {
+ bufferElementMapper{}.linkerFor(l.head).SetPrev(e)
+ } else {
+ l.tail = e
+ }
+
+ l.head = e
+}
+
+// PushBack inserts the element e at the back of list l.
+func (l *bufferList) PushBack(e *buffer) {
+ bufferElementMapper{}.linkerFor(e).SetNext(nil)
+ bufferElementMapper{}.linkerFor(e).SetPrev(l.tail)
+
+ if l.tail != nil {
+ bufferElementMapper{}.linkerFor(l.tail).SetNext(e)
+ } else {
+ l.head = e
+ }
+
+ l.tail = e
+}
+
+// PushBackList inserts list m at the end of list l, emptying m.
+func (l *bufferList) PushBackList(m *bufferList) {
+ if l.head == nil {
+ l.head = m.head
+ l.tail = m.tail
+ } else if m.head != nil {
+ bufferElementMapper{}.linkerFor(l.tail).SetNext(m.head)
+ bufferElementMapper{}.linkerFor(m.head).SetPrev(l.tail)
+
+ l.tail = m.tail
+ }
+
+ m.head = nil
+ m.tail = nil
+}
+
+// InsertAfter inserts e after b.
+func (l *bufferList) InsertAfter(b, e *buffer) {
+ a := bufferElementMapper{}.linkerFor(b).Next()
+ bufferElementMapper{}.linkerFor(e).SetNext(a)
+ bufferElementMapper{}.linkerFor(e).SetPrev(b)
+ bufferElementMapper{}.linkerFor(b).SetNext(e)
+
+ if a != nil {
+ bufferElementMapper{}.linkerFor(a).SetPrev(e)
+ } else {
+ l.tail = e
+ }
+}
+
+// InsertBefore inserts e before a.
+func (l *bufferList) InsertBefore(a, e *buffer) {
+ b := bufferElementMapper{}.linkerFor(a).Prev()
+ bufferElementMapper{}.linkerFor(e).SetNext(a)
+ bufferElementMapper{}.linkerFor(e).SetPrev(b)
+ bufferElementMapper{}.linkerFor(a).SetPrev(e)
+
+ if b != nil {
+ bufferElementMapper{}.linkerFor(b).SetNext(e)
+ } else {
+ l.head = e
+ }
+}
+
+// Remove removes e from l.
+func (l *bufferList) Remove(e *buffer) {
+ prev := bufferElementMapper{}.linkerFor(e).Prev()
+ next := bufferElementMapper{}.linkerFor(e).Next()
+
+ if prev != nil {
+ bufferElementMapper{}.linkerFor(prev).SetNext(next)
+ } else {
+ l.head = next
+ }
+
+ if next != nil {
+ bufferElementMapper{}.linkerFor(next).SetPrev(prev)
+ } else {
+ l.tail = prev
+ }
+}
+
+// Entry is a default implementation of Linker. Users can add anonymous fields
+// of this type to their structs to make them automatically implement the
+// methods needed by List.
+//
+// +stateify savable
+type bufferEntry struct {
+ next *buffer
+ prev *buffer
+}
+
+// Next returns the entry that follows e in the list.
+func (e *bufferEntry) Next() *buffer {
+ return e.next
+}
+
+// Prev returns the entry that precedes e in the list.
+func (e *bufferEntry) Prev() *buffer {
+ return e.prev
+}
+
+// SetNext assigns 'entry' as the entry that follows e in the list.
+func (e *bufferEntry) SetNext(elem *buffer) {
+ e.next = elem
+}
+
+// SetPrev assigns 'entry' as the entry that precedes e in the list.
+func (e *bufferEntry) SetPrev(elem *buffer) {
+ e.prev = elem
+}
diff --git a/pkg/sentry/kernel/pipe/buffer_test.go b/pkg/sentry/kernel/pipe/buffer_test.go
deleted file mode 100644
index ee1b90115..000000000
--- a/pkg/sentry/kernel/pipe/buffer_test.go
+++ /dev/null
@@ -1,32 +0,0 @@
-// Copyright 2019 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 pipe
-
-import (
- "testing"
- "unsafe"
-
- "gvisor.dev/gvisor/pkg/sentry/usermem"
-)
-
-func TestBufferSize(t *testing.T) {
- bufferSize := unsafe.Sizeof(buffer{})
- if bufferSize < usermem.PageSize {
- t.Errorf("buffer is less than a page")
- }
- if bufferSize > (2 * usermem.PageSize) {
- t.Errorf("buffer is greater than two pages")
- }
-}
diff --git a/pkg/sentry/kernel/pipe/node_test.go b/pkg/sentry/kernel/pipe/node_test.go
deleted file mode 100644
index 16fa80abe..000000000
--- a/pkg/sentry/kernel/pipe/node_test.go
+++ /dev/null
@@ -1,320 +0,0 @@
-// Copyright 2018 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 pipe
-
-import (
- "testing"
- "time"
-
- "gvisor.dev/gvisor/pkg/sentry/context"
- "gvisor.dev/gvisor/pkg/sentry/context/contexttest"
- "gvisor.dev/gvisor/pkg/sentry/fs"
- "gvisor.dev/gvisor/pkg/sentry/usermem"
- "gvisor.dev/gvisor/pkg/syserror"
-)
-
-type sleeper struct {
- context.Context
- ch chan struct{}
-}
-
-func newSleeperContext(t *testing.T) context.Context {
- return &sleeper{
- Context: contexttest.Context(t),
- ch: make(chan struct{}),
- }
-}
-
-func (s *sleeper) SleepStart() <-chan struct{} {
- return s.ch
-}
-
-func (s *sleeper) SleepFinish(bool) {
-}
-
-func (s *sleeper) Cancel() {
- s.ch <- struct{}{}
-}
-
-func (s *sleeper) Interrupted() bool {
- return len(s.ch) != 0
-}
-
-type openResult struct {
- *fs.File
- error
-}
-
-var perms fs.FilePermissions = fs.FilePermissions{
- User: fs.PermMask{Read: true, Write: true},
-}
-
-func testOpenOrDie(ctx context.Context, t *testing.T, n fs.InodeOperations, flags fs.FileFlags, doneChan chan<- struct{}) (*fs.File, error) {
- inode := fs.NewMockInode(ctx, fs.NewMockMountSource(nil), fs.StableAttr{Type: fs.Pipe})
- d := fs.NewDirent(ctx, inode, "pipe")
- file, err := n.GetFile(ctx, d, flags)
- if err != nil {
- t.Fatalf("open with flags %+v failed: %v", flags, err)
- }
- if doneChan != nil {
- doneChan <- struct{}{}
- }
- return file, err
-}
-
-func testOpen(ctx context.Context, t *testing.T, n fs.InodeOperations, flags fs.FileFlags, resChan chan<- openResult) (*fs.File, error) {
- inode := fs.NewMockInode(ctx, fs.NewMockMountSource(nil), fs.StableAttr{Type: fs.Pipe})
- d := fs.NewDirent(ctx, inode, "pipe")
- file, err := n.GetFile(ctx, d, flags)
- if resChan != nil {
- resChan <- openResult{file, err}
- }
- return file, err
-}
-
-func newNamedPipe(t *testing.T) *Pipe {
- return NewPipe(true, DefaultPipeSize, usermem.PageSize)
-}
-
-func newAnonPipe(t *testing.T) *Pipe {
- return NewPipe(false, DefaultPipeSize, usermem.PageSize)
-}
-
-// assertRecvBlocks ensures that a recv attempt on c blocks for at least
-// blockDuration. This is useful for checking that a goroutine that is supposed
-// to be executing a blocking operation is actually blocking.
-func assertRecvBlocks(t *testing.T, c <-chan struct{}, blockDuration time.Duration, failMsg string) {
- select {
- case <-c:
- t.Fatalf(failMsg)
- case <-time.After(blockDuration):
- // Ok, blocked for the required duration.
- }
-}
-
-func TestReadOpenBlocksForWriteOpen(t *testing.T) {
- ctx := newSleeperContext(t)
- f := NewInodeOperations(ctx, perms, newNamedPipe(t))
-
- rDone := make(chan struct{})
- go testOpenOrDie(ctx, t, f, fs.FileFlags{Read: true}, rDone)
-
- // Verify that the open for read is blocking.
- assertRecvBlocks(t, rDone, time.Millisecond*100,
- "open for read not blocking with no writers")
-
- wDone := make(chan struct{})
- go testOpenOrDie(ctx, t, f, fs.FileFlags{Write: true}, wDone)
-
- <-wDone
- <-rDone
-}
-
-func TestWriteOpenBlocksForReadOpen(t *testing.T) {
- ctx := newSleeperContext(t)
- f := NewInodeOperations(ctx, perms, newNamedPipe(t))
-
- wDone := make(chan struct{})
- go testOpenOrDie(ctx, t, f, fs.FileFlags{Write: true}, wDone)
-
- // Verify that the open for write is blocking
- assertRecvBlocks(t, wDone, time.Millisecond*100,
- "open for write not blocking with no readers")
-
- rDone := make(chan struct{})
- go testOpenOrDie(ctx, t, f, fs.FileFlags{Read: true}, rDone)
-
- <-rDone
- <-wDone
-}
-
-func TestMultipleWriteOpenDoesntCountAsReadOpen(t *testing.T) {
- ctx := newSleeperContext(t)
- f := NewInodeOperations(ctx, perms, newNamedPipe(t))
-
- rDone1 := make(chan struct{})
- rDone2 := make(chan struct{})
- go testOpenOrDie(ctx, t, f, fs.FileFlags{Read: true}, rDone1)
- go testOpenOrDie(ctx, t, f, fs.FileFlags{Read: true}, rDone2)
-
- assertRecvBlocks(t, rDone1, time.Millisecond*100,
- "open for read didn't block with no writers")
- assertRecvBlocks(t, rDone2, time.Millisecond*100,
- "open for read didn't block with no writers")
-
- wDone := make(chan struct{})
- go testOpenOrDie(ctx, t, f, fs.FileFlags{Write: true}, wDone)
-
- <-wDone
- <-rDone2
- <-rDone1
-}
-
-func TestClosedReaderBlocksWriteOpen(t *testing.T) {
- ctx := newSleeperContext(t)
- f := NewInodeOperations(ctx, perms, newNamedPipe(t))
-
- rFile, _ := testOpenOrDie(ctx, t, f, fs.FileFlags{Read: true, NonBlocking: true}, nil)
- rFile.DecRef()
-
- wDone := make(chan struct{})
- // This open for write should block because the reader is now gone.
- go testOpenOrDie(ctx, t, f, fs.FileFlags{Write: true}, wDone)
- assertRecvBlocks(t, wDone, time.Millisecond*100,
- "open for write didn't block with no concurrent readers")
-
- // Open for read again. This should unblock the open for write.
- rDone := make(chan struct{})
- go testOpenOrDie(ctx, t, f, fs.FileFlags{Read: true}, rDone)
-
- <-rDone
- <-wDone
-}
-
-func TestReadWriteOpenNeverBlocks(t *testing.T) {
- ctx := newSleeperContext(t)
- f := NewInodeOperations(ctx, perms, newNamedPipe(t))
-
- rwDone := make(chan struct{})
- // Open for read-write never wait for a reader or writer, even if the
- // nonblocking flag is not set.
- go testOpenOrDie(ctx, t, f, fs.FileFlags{Read: true, Write: true, NonBlocking: false}, rwDone)
- <-rwDone
-}
-
-func TestReadWriteOpenUnblocksReadOpen(t *testing.T) {
- ctx := newSleeperContext(t)
- f := NewInodeOperations(ctx, perms, newNamedPipe(t))
-
- rDone := make(chan struct{})
- go testOpenOrDie(ctx, t, f, fs.FileFlags{Read: true}, rDone)
-
- rwDone := make(chan struct{})
- go testOpenOrDie(ctx, t, f, fs.FileFlags{Read: true, Write: true}, rwDone)
-
- <-rwDone
- <-rDone
-}
-
-func TestReadWriteOpenUnblocksWriteOpen(t *testing.T) {
- ctx := newSleeperContext(t)
- f := NewInodeOperations(ctx, perms, newNamedPipe(t))
-
- wDone := make(chan struct{})
- go testOpenOrDie(ctx, t, f, fs.FileFlags{Write: true}, wDone)
-
- rwDone := make(chan struct{})
- go testOpenOrDie(ctx, t, f, fs.FileFlags{Read: true, Write: true}, rwDone)
-
- <-rwDone
- <-wDone
-}
-
-func TestBlockedOpenIsCancellable(t *testing.T) {
- ctx := newSleeperContext(t)
- f := NewInodeOperations(ctx, perms, newNamedPipe(t))
-
- done := make(chan openResult)
- go testOpen(ctx, t, f, fs.FileFlags{Read: true}, done)
- select {
- case <-done:
- t.Fatalf("open for read didn't block with no writers")
- case <-time.After(time.Millisecond * 100):
- // Ok.
- }
-
- ctx.(*sleeper).Cancel()
- // If the cancel on the sleeper didn't work, the open for read would never
- // return.
- res := <-done
- if res.error != syserror.ErrInterrupted {
- t.Fatalf("Cancellation didn't cause GetFile to return fs.ErrInterrupted, got %v.",
- res.error)
- }
-}
-
-func TestNonblockingReadOpenFileNoWriters(t *testing.T) {
- ctx := newSleeperContext(t)
- f := NewInodeOperations(ctx, perms, newNamedPipe(t))
-
- if _, err := testOpen(ctx, t, f, fs.FileFlags{Read: true, NonBlocking: true}, nil); err != nil {
- t.Fatalf("Nonblocking open for read failed with error %v.", err)
- }
-}
-
-func TestNonblockingWriteOpenFileNoReaders(t *testing.T) {
- ctx := newSleeperContext(t)
- f := NewInodeOperations(ctx, perms, newNamedPipe(t))
-
- if _, err := testOpen(ctx, t, f, fs.FileFlags{Write: true, NonBlocking: true}, nil); err != syserror.ENXIO {
- t.Fatalf("Nonblocking open for write failed unexpected error %v.", err)
- }
-}
-
-func TestNonBlockingReadOpenWithWriter(t *testing.T) {
- ctx := newSleeperContext(t)
- f := NewInodeOperations(ctx, perms, newNamedPipe(t))
-
- wDone := make(chan struct{})
- go testOpenOrDie(ctx, t, f, fs.FileFlags{Write: true}, wDone)
-
- // Open for write blocks since there are no readers yet.
- assertRecvBlocks(t, wDone, time.Millisecond*100,
- "Open for write didn't block with no reader.")
-
- if _, err := testOpen(ctx, t, f, fs.FileFlags{Read: true, NonBlocking: true}, nil); err != nil {
- t.Fatalf("Nonblocking open for read failed with error %v.", err)
- }
-
- // Open for write should now be unblocked.
- <-wDone
-}
-
-func TestNonBlockingWriteOpenWithReader(t *testing.T) {
- ctx := newSleeperContext(t)
- f := NewInodeOperations(ctx, perms, newNamedPipe(t))
-
- rDone := make(chan struct{})
- go testOpenOrDie(ctx, t, f, fs.FileFlags{Read: true}, rDone)
-
- // Open for write blocked, since no reader yet.
- assertRecvBlocks(t, rDone, time.Millisecond*100,
- "Open for reader didn't block with no writer.")
-
- if _, err := testOpen(ctx, t, f, fs.FileFlags{Write: true, NonBlocking: true}, nil); err != nil {
- t.Fatalf("Nonblocking open for write failed with error %v.", err)
- }
-
- // Open for write should now be unblocked.
- <-rDone
-}
-
-func TestAnonReadOpen(t *testing.T) {
- ctx := newSleeperContext(t)
- f := NewInodeOperations(ctx, perms, newAnonPipe(t))
-
- if _, err := testOpen(ctx, t, f, fs.FileFlags{Read: true}, nil); err != nil {
- t.Fatalf("open anon pipe for read failed: %v", err)
- }
-}
-
-func TestAnonWriteOpen(t *testing.T) {
- ctx := newSleeperContext(t)
- f := NewInodeOperations(ctx, perms, newAnonPipe(t))
-
- if _, err := testOpen(ctx, t, f, fs.FileFlags{Write: true}, nil); err != nil {
- t.Fatalf("open anon pipe for write failed: %v", err)
- }
-}
diff --git a/pkg/sentry/kernel/pipe/pipe_state_autogen.go b/pkg/sentry/kernel/pipe/pipe_state_autogen.go
new file mode 100755
index 000000000..cad035789
--- /dev/null
+++ b/pkg/sentry/kernel/pipe/pipe_state_autogen.go
@@ -0,0 +1,132 @@
+// automatically generated by stateify.
+
+package pipe
+
+import (
+ "gvisor.dev/gvisor/pkg/state"
+)
+
+func (x *buffer) beforeSave() {}
+func (x *buffer) save(m state.Map) {
+ x.beforeSave()
+ m.Save("data", &x.data)
+ m.Save("read", &x.read)
+ m.Save("write", &x.write)
+ m.Save("bufferEntry", &x.bufferEntry)
+}
+
+func (x *buffer) afterLoad() {}
+func (x *buffer) load(m state.Map) {
+ m.Load("data", &x.data)
+ m.Load("read", &x.read)
+ m.Load("write", &x.write)
+ m.Load("bufferEntry", &x.bufferEntry)
+}
+
+func (x *bufferList) beforeSave() {}
+func (x *bufferList) save(m state.Map) {
+ x.beforeSave()
+ m.Save("head", &x.head)
+ m.Save("tail", &x.tail)
+}
+
+func (x *bufferList) afterLoad() {}
+func (x *bufferList) load(m state.Map) {
+ m.Load("head", &x.head)
+ m.Load("tail", &x.tail)
+}
+
+func (x *bufferEntry) beforeSave() {}
+func (x *bufferEntry) save(m state.Map) {
+ x.beforeSave()
+ m.Save("next", &x.next)
+ m.Save("prev", &x.prev)
+}
+
+func (x *bufferEntry) afterLoad() {}
+func (x *bufferEntry) load(m state.Map) {
+ m.Load("next", &x.next)
+ m.Load("prev", &x.prev)
+}
+
+func (x *inodeOperations) beforeSave() {}
+func (x *inodeOperations) save(m state.Map) {
+ x.beforeSave()
+ m.Save("InodeSimpleAttributes", &x.InodeSimpleAttributes)
+ m.Save("p", &x.p)
+}
+
+func (x *inodeOperations) afterLoad() {}
+func (x *inodeOperations) load(m state.Map) {
+ m.Load("InodeSimpleAttributes", &x.InodeSimpleAttributes)
+ m.Load("p", &x.p)
+}
+
+func (x *Pipe) beforeSave() {}
+func (x *Pipe) save(m state.Map) {
+ x.beforeSave()
+ m.Save("isNamed", &x.isNamed)
+ m.Save("atomicIOBytes", &x.atomicIOBytes)
+ m.Save("readers", &x.readers)
+ m.Save("writers", &x.writers)
+ m.Save("data", &x.data)
+ m.Save("max", &x.max)
+ m.Save("size", &x.size)
+ m.Save("hadWriter", &x.hadWriter)
+}
+
+func (x *Pipe) afterLoad() {}
+func (x *Pipe) load(m state.Map) {
+ m.Load("isNamed", &x.isNamed)
+ m.Load("atomicIOBytes", &x.atomicIOBytes)
+ m.Load("readers", &x.readers)
+ m.Load("writers", &x.writers)
+ m.Load("data", &x.data)
+ m.Load("max", &x.max)
+ m.Load("size", &x.size)
+ m.Load("hadWriter", &x.hadWriter)
+}
+
+func (x *Reader) beforeSave() {}
+func (x *Reader) save(m state.Map) {
+ x.beforeSave()
+ m.Save("ReaderWriter", &x.ReaderWriter)
+}
+
+func (x *Reader) afterLoad() {}
+func (x *Reader) load(m state.Map) {
+ m.Load("ReaderWriter", &x.ReaderWriter)
+}
+
+func (x *ReaderWriter) beforeSave() {}
+func (x *ReaderWriter) save(m state.Map) {
+ x.beforeSave()
+ m.Save("Pipe", &x.Pipe)
+}
+
+func (x *ReaderWriter) afterLoad() {}
+func (x *ReaderWriter) load(m state.Map) {
+ m.Load("Pipe", &x.Pipe)
+}
+
+func (x *Writer) beforeSave() {}
+func (x *Writer) save(m state.Map) {
+ x.beforeSave()
+ m.Save("ReaderWriter", &x.ReaderWriter)
+}
+
+func (x *Writer) afterLoad() {}
+func (x *Writer) load(m state.Map) {
+ m.Load("ReaderWriter", &x.ReaderWriter)
+}
+
+func init() {
+ state.Register("pipe.buffer", (*buffer)(nil), state.Fns{Save: (*buffer).save, Load: (*buffer).load})
+ state.Register("pipe.bufferList", (*bufferList)(nil), state.Fns{Save: (*bufferList).save, Load: (*bufferList).load})
+ state.Register("pipe.bufferEntry", (*bufferEntry)(nil), state.Fns{Save: (*bufferEntry).save, Load: (*bufferEntry).load})
+ state.Register("pipe.inodeOperations", (*inodeOperations)(nil), state.Fns{Save: (*inodeOperations).save, Load: (*inodeOperations).load})
+ state.Register("pipe.Pipe", (*Pipe)(nil), state.Fns{Save: (*Pipe).save, Load: (*Pipe).load})
+ state.Register("pipe.Reader", (*Reader)(nil), state.Fns{Save: (*Reader).save, Load: (*Reader).load})
+ state.Register("pipe.ReaderWriter", (*ReaderWriter)(nil), state.Fns{Save: (*ReaderWriter).save, Load: (*ReaderWriter).load})
+ state.Register("pipe.Writer", (*Writer)(nil), state.Fns{Save: (*Writer).save, Load: (*Writer).load})
+}
diff --git a/pkg/sentry/kernel/pipe/pipe_test.go b/pkg/sentry/kernel/pipe/pipe_test.go
deleted file mode 100644
index e3a14b665..000000000
--- a/pkg/sentry/kernel/pipe/pipe_test.go
+++ /dev/null
@@ -1,139 +0,0 @@
-// Copyright 2018 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 pipe
-
-import (
- "bytes"
- "testing"
-
- "gvisor.dev/gvisor/pkg/sentry/context/contexttest"
- "gvisor.dev/gvisor/pkg/sentry/usermem"
- "gvisor.dev/gvisor/pkg/syserror"
- "gvisor.dev/gvisor/pkg/waiter"
-)
-
-func TestPipeRW(t *testing.T) {
- ctx := contexttest.Context(t)
- r, w := NewConnectedPipe(ctx, 65536, 4096)
- defer r.DecRef()
- defer w.DecRef()
-
- msg := []byte("here's some bytes")
- wantN := int64(len(msg))
- n, err := w.Writev(ctx, usermem.BytesIOSequence(msg))
- if n != wantN || err != nil {
- t.Fatalf("Writev: got (%d, %v), wanted (%d, nil)", n, err, wantN)
- }
-
- buf := make([]byte, len(msg))
- n, err = r.Readv(ctx, usermem.BytesIOSequence(buf))
- if n != wantN || err != nil || !bytes.Equal(buf, msg) {
- t.Fatalf("Readv: got (%d, %v) %q, wanted (%d, nil) %q", n, err, buf, wantN, msg)
- }
-}
-
-func TestPipeReadBlock(t *testing.T) {
- ctx := contexttest.Context(t)
- r, w := NewConnectedPipe(ctx, 65536, 4096)
- defer r.DecRef()
- defer w.DecRef()
-
- n, err := r.Readv(ctx, usermem.BytesIOSequence(make([]byte, 1)))
- if n != 0 || err != syserror.ErrWouldBlock {
- t.Fatalf("Readv: got (%d, %v), wanted (0, %v)", n, err, syserror.ErrWouldBlock)
- }
-}
-
-func TestPipeWriteBlock(t *testing.T) {
- const atomicIOBytes = 2
- const capacity = MinimumPipeSize
-
- ctx := contexttest.Context(t)
- r, w := NewConnectedPipe(ctx, capacity, atomicIOBytes)
- defer r.DecRef()
- defer w.DecRef()
-
- msg := make([]byte, capacity+1)
- n, err := w.Writev(ctx, usermem.BytesIOSequence(msg))
- if wantN, wantErr := int64(capacity), syserror.ErrWouldBlock; n != wantN || err != wantErr {
- t.Fatalf("Writev: got (%d, %v), wanted (%d, %v)", n, err, wantN, wantErr)
- }
-}
-
-func TestPipeWriteUntilEnd(t *testing.T) {
- const atomicIOBytes = 2
-
- ctx := contexttest.Context(t)
- r, w := NewConnectedPipe(ctx, atomicIOBytes, atomicIOBytes)
- defer r.DecRef()
- defer w.DecRef()
-
- msg := []byte("here's some bytes")
-
- wDone := make(chan struct{}, 0)
- rDone := make(chan struct{}, 0)
- defer func() {
- // Signal the reader to stop and wait until it does so.
- close(wDone)
- <-rDone
- }()
-
- go func() {
- defer close(rDone)
- // Read from r until done is closed.
- ctx := contexttest.Context(t)
- buf := make([]byte, len(msg)+1)
- dst := usermem.BytesIOSequence(buf)
- e, ch := waiter.NewChannelEntry(nil)
- r.EventRegister(&e, waiter.EventIn)
- defer r.EventUnregister(&e)
- for {
- n, err := r.Readv(ctx, dst)
- dst = dst.DropFirst64(n)
- if err == syserror.ErrWouldBlock {
- select {
- case <-ch:
- continue
- case <-wDone:
- // We expect to have 1 byte left in dst since len(buf) ==
- // len(msg)+1.
- if dst.NumBytes() != 1 || !bytes.Equal(buf[:len(msg)], msg) {
- t.Errorf("Reader: got %q (%d bytes remaining), wanted %q", buf, dst.NumBytes(), msg)
- }
- return
- }
- }
- if err != nil {
- t.Fatalf("Readv: got unexpected error %v", err)
- }
- }
- }()
-
- src := usermem.BytesIOSequence(msg)
- e, ch := waiter.NewChannelEntry(nil)
- w.EventRegister(&e, waiter.EventOut)
- defer w.EventUnregister(&e)
- for src.NumBytes() != 0 {
- n, err := w.Writev(ctx, src)
- src = src.DropFirst64(n)
- if err == syserror.ErrWouldBlock {
- <-ch
- continue
- }
- if err != nil {
- t.Fatalf("Writev: got (%d, %v)", n, err)
- }
- }
-}
diff --git a/pkg/sentry/kernel/pipe/pipe_util.go b/pkg/sentry/kernel/pipe/pipe_util.go
index 8394eb78b..8394eb78b 100644..100755
--- a/pkg/sentry/kernel/pipe/pipe_util.go
+++ b/pkg/sentry/kernel/pipe/pipe_util.go
diff --git a/pkg/sentry/kernel/pipe/vfs.go b/pkg/sentry/kernel/pipe/vfs.go
index 6f83e3cee..6f83e3cee 100644..100755
--- a/pkg/sentry/kernel/pipe/vfs.go
+++ b/pkg/sentry/kernel/pipe/vfs.go
diff --git a/pkg/sentry/kernel/process_group_list.go b/pkg/sentry/kernel/process_group_list.go
new file mode 100755
index 000000000..853145237
--- /dev/null
+++ b/pkg/sentry/kernel/process_group_list.go
@@ -0,0 +1,173 @@
+package kernel
+
+// ElementMapper provides an identity mapping by default.
+//
+// This can be replaced to provide a struct that maps elements to linker
+// objects, if they are not the same. An ElementMapper is not typically
+// required if: Linker is left as is, Element is left as is, or Linker and
+// Element are the same type.
+type processGroupElementMapper struct{}
+
+// linkerFor maps an Element to a Linker.
+//
+// This default implementation should be inlined.
+//
+//go:nosplit
+func (processGroupElementMapper) linkerFor(elem *ProcessGroup) *ProcessGroup { return elem }
+
+// List is an intrusive list. Entries can be added to or removed from the list
+// in O(1) time and with no additional memory allocations.
+//
+// The zero value for List is an empty list ready to use.
+//
+// To iterate over a list (where l is a List):
+// for e := l.Front(); e != nil; e = e.Next() {
+// // do something with e.
+// }
+//
+// +stateify savable
+type processGroupList struct {
+ head *ProcessGroup
+ tail *ProcessGroup
+}
+
+// Reset resets list l to the empty state.
+func (l *processGroupList) Reset() {
+ l.head = nil
+ l.tail = nil
+}
+
+// Empty returns true iff the list is empty.
+func (l *processGroupList) Empty() bool {
+ return l.head == nil
+}
+
+// Front returns the first element of list l or nil.
+func (l *processGroupList) Front() *ProcessGroup {
+ return l.head
+}
+
+// Back returns the last element of list l or nil.
+func (l *processGroupList) Back() *ProcessGroup {
+ return l.tail
+}
+
+// PushFront inserts the element e at the front of list l.
+func (l *processGroupList) PushFront(e *ProcessGroup) {
+ processGroupElementMapper{}.linkerFor(e).SetNext(l.head)
+ processGroupElementMapper{}.linkerFor(e).SetPrev(nil)
+
+ if l.head != nil {
+ processGroupElementMapper{}.linkerFor(l.head).SetPrev(e)
+ } else {
+ l.tail = e
+ }
+
+ l.head = e
+}
+
+// PushBack inserts the element e at the back of list l.
+func (l *processGroupList) PushBack(e *ProcessGroup) {
+ processGroupElementMapper{}.linkerFor(e).SetNext(nil)
+ processGroupElementMapper{}.linkerFor(e).SetPrev(l.tail)
+
+ if l.tail != nil {
+ processGroupElementMapper{}.linkerFor(l.tail).SetNext(e)
+ } else {
+ l.head = e
+ }
+
+ l.tail = e
+}
+
+// PushBackList inserts list m at the end of list l, emptying m.
+func (l *processGroupList) PushBackList(m *processGroupList) {
+ if l.head == nil {
+ l.head = m.head
+ l.tail = m.tail
+ } else if m.head != nil {
+ processGroupElementMapper{}.linkerFor(l.tail).SetNext(m.head)
+ processGroupElementMapper{}.linkerFor(m.head).SetPrev(l.tail)
+
+ l.tail = m.tail
+ }
+
+ m.head = nil
+ m.tail = nil
+}
+
+// InsertAfter inserts e after b.
+func (l *processGroupList) InsertAfter(b, e *ProcessGroup) {
+ a := processGroupElementMapper{}.linkerFor(b).Next()
+ processGroupElementMapper{}.linkerFor(e).SetNext(a)
+ processGroupElementMapper{}.linkerFor(e).SetPrev(b)
+ processGroupElementMapper{}.linkerFor(b).SetNext(e)
+
+ if a != nil {
+ processGroupElementMapper{}.linkerFor(a).SetPrev(e)
+ } else {
+ l.tail = e
+ }
+}
+
+// InsertBefore inserts e before a.
+func (l *processGroupList) InsertBefore(a, e *ProcessGroup) {
+ b := processGroupElementMapper{}.linkerFor(a).Prev()
+ processGroupElementMapper{}.linkerFor(e).SetNext(a)
+ processGroupElementMapper{}.linkerFor(e).SetPrev(b)
+ processGroupElementMapper{}.linkerFor(a).SetPrev(e)
+
+ if b != nil {
+ processGroupElementMapper{}.linkerFor(b).SetNext(e)
+ } else {
+ l.head = e
+ }
+}
+
+// Remove removes e from l.
+func (l *processGroupList) Remove(e *ProcessGroup) {
+ prev := processGroupElementMapper{}.linkerFor(e).Prev()
+ next := processGroupElementMapper{}.linkerFor(e).Next()
+
+ if prev != nil {
+ processGroupElementMapper{}.linkerFor(prev).SetNext(next)
+ } else {
+ l.head = next
+ }
+
+ if next != nil {
+ processGroupElementMapper{}.linkerFor(next).SetPrev(prev)
+ } else {
+ l.tail = prev
+ }
+}
+
+// Entry is a default implementation of Linker. Users can add anonymous fields
+// of this type to their structs to make them automatically implement the
+// methods needed by List.
+//
+// +stateify savable
+type processGroupEntry struct {
+ next *ProcessGroup
+ prev *ProcessGroup
+}
+
+// Next returns the entry that follows e in the list.
+func (e *processGroupEntry) Next() *ProcessGroup {
+ return e.next
+}
+
+// Prev returns the entry that precedes e in the list.
+func (e *processGroupEntry) Prev() *ProcessGroup {
+ return e.prev
+}
+
+// SetNext assigns 'entry' as the entry that follows e in the list.
+func (e *processGroupEntry) SetNext(elem *ProcessGroup) {
+ e.next = elem
+}
+
+// SetPrev assigns 'entry' as the entry that precedes e in the list.
+func (e *processGroupEntry) SetPrev(elem *ProcessGroup) {
+ e.prev = elem
+}
diff --git a/pkg/sentry/kernel/sched/BUILD b/pkg/sentry/kernel/sched/BUILD
deleted file mode 100644
index 98ea7a0d8..000000000
--- a/pkg/sentry/kernel/sched/BUILD
+++ /dev/null
@@ -1,21 +0,0 @@
-load("//tools/go_stateify:defs.bzl", "go_library")
-load("@io_bazel_rules_go//go:def.bzl", "go_test")
-
-package(licenses = ["notice"])
-
-go_library(
- name = "sched",
- srcs = [
- "cpuset.go",
- "sched.go",
- ],
- importpath = "gvisor.dev/gvisor/pkg/sentry/kernel/sched",
- visibility = ["//pkg/sentry:internal"],
-)
-
-go_test(
- name = "sched_test",
- size = "small",
- srcs = ["cpuset_test.go"],
- embed = [":sched"],
-)
diff --git a/pkg/sentry/kernel/sched/cpuset_test.go b/pkg/sentry/kernel/sched/cpuset_test.go
deleted file mode 100644
index 3af9f1197..000000000
--- a/pkg/sentry/kernel/sched/cpuset_test.go
+++ /dev/null
@@ -1,44 +0,0 @@
-// Copyright 2018 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 sched
-
-import (
- "testing"
-)
-
-func TestNumCPUs(t *testing.T) {
- for i := uint(0); i < 1024; i++ {
- c := NewCPUSet(i)
- for j := uint(0); j < i; j++ {
- c.Set(j)
- }
- n := c.NumCPUs()
- if n != i {
- t.Errorf("got wrong number of cpus %d, want %d", n, i)
- }
- }
-}
-
-func TestClearAbove(t *testing.T) {
- const n = 1024
- c := NewFullCPUSet(n)
- for i := uint(0); i < n; i++ {
- cpu := n - i
- c.ClearAbove(cpu)
- if got := c.NumCPUs(); got != cpu {
- t.Errorf("iteration %d: got %d cpus, wanted %d", i, got, cpu)
- }
- }
-}
diff --git a/pkg/sentry/kernel/sched/sched_state_autogen.go b/pkg/sentry/kernel/sched/sched_state_autogen.go
new file mode 100755
index 000000000..2a482732e
--- /dev/null
+++ b/pkg/sentry/kernel/sched/sched_state_autogen.go
@@ -0,0 +1,4 @@
+// automatically generated by stateify.
+
+package sched
+
diff --git a/pkg/sentry/kernel/semaphore/BUILD b/pkg/sentry/kernel/semaphore/BUILD
deleted file mode 100644
index 13a961594..000000000
--- a/pkg/sentry/kernel/semaphore/BUILD
+++ /dev/null
@@ -1,51 +0,0 @@
-load("@io_bazel_rules_go//go:def.bzl", "go_test")
-load("//tools/go_generics:defs.bzl", "go_template_instance")
-load("//tools/go_stateify:defs.bzl", "go_library")
-
-package(licenses = ["notice"])
-
-go_template_instance(
- name = "waiter_list",
- out = "waiter_list.go",
- package = "semaphore",
- prefix = "waiter",
- template = "//pkg/ilist:generic_list",
- types = {
- "Element": "*waiter",
- "Linker": "*waiter",
- },
-)
-
-go_library(
- name = "semaphore",
- srcs = [
- "semaphore.go",
- "waiter_list.go",
- ],
- importpath = "gvisor.dev/gvisor/pkg/sentry/kernel/semaphore",
- visibility = ["//pkg/sentry:internal"],
- deps = [
- "//pkg/abi/linux",
- "//pkg/log",
- "//pkg/sentry/context",
- "//pkg/sentry/fs",
- "//pkg/sentry/kernel/auth",
- "//pkg/sentry/kernel/time",
- "//pkg/sync",
- "//pkg/syserror",
- ],
-)
-
-go_test(
- name = "semaphore_test",
- size = "small",
- srcs = ["semaphore_test.go"],
- embed = [":semaphore"],
- deps = [
- "//pkg/abi/linux",
- "//pkg/sentry/context",
- "//pkg/sentry/context/contexttest",
- "//pkg/sentry/kernel/auth",
- "//pkg/syserror",
- ],
-)
diff --git a/pkg/sentry/kernel/semaphore/semaphore_state_autogen.go b/pkg/sentry/kernel/semaphore/semaphore_state_autogen.go
new file mode 100755
index 000000000..f225f6f75
--- /dev/null
+++ b/pkg/sentry/kernel/semaphore/semaphore_state_autogen.go
@@ -0,0 +1,115 @@
+// automatically generated by stateify.
+
+package semaphore
+
+import (
+ "gvisor.dev/gvisor/pkg/state"
+)
+
+func (x *Registry) beforeSave() {}
+func (x *Registry) save(m state.Map) {
+ x.beforeSave()
+ m.Save("userNS", &x.userNS)
+ m.Save("semaphores", &x.semaphores)
+ m.Save("lastIDUsed", &x.lastIDUsed)
+}
+
+func (x *Registry) afterLoad() {}
+func (x *Registry) load(m state.Map) {
+ m.Load("userNS", &x.userNS)
+ m.Load("semaphores", &x.semaphores)
+ m.Load("lastIDUsed", &x.lastIDUsed)
+}
+
+func (x *Set) beforeSave() {}
+func (x *Set) save(m state.Map) {
+ x.beforeSave()
+ m.Save("registry", &x.registry)
+ m.Save("ID", &x.ID)
+ m.Save("key", &x.key)
+ m.Save("creator", &x.creator)
+ m.Save("owner", &x.owner)
+ m.Save("perms", &x.perms)
+ m.Save("opTime", &x.opTime)
+ m.Save("changeTime", &x.changeTime)
+ m.Save("sems", &x.sems)
+ m.Save("dead", &x.dead)
+}
+
+func (x *Set) afterLoad() {}
+func (x *Set) load(m state.Map) {
+ m.Load("registry", &x.registry)
+ m.Load("ID", &x.ID)
+ m.Load("key", &x.key)
+ m.Load("creator", &x.creator)
+ m.Load("owner", &x.owner)
+ m.Load("perms", &x.perms)
+ m.Load("opTime", &x.opTime)
+ m.Load("changeTime", &x.changeTime)
+ m.Load("sems", &x.sems)
+ m.Load("dead", &x.dead)
+}
+
+func (x *sem) beforeSave() {}
+func (x *sem) save(m state.Map) {
+ x.beforeSave()
+ if !state.IsZeroValue(x.waiters) { m.Failf("waiters is %v, expected zero", x.waiters) }
+ m.Save("value", &x.value)
+ m.Save("pid", &x.pid)
+}
+
+func (x *sem) afterLoad() {}
+func (x *sem) load(m state.Map) {
+ m.Load("value", &x.value)
+ m.Load("pid", &x.pid)
+}
+
+func (x *waiter) beforeSave() {}
+func (x *waiter) save(m state.Map) {
+ x.beforeSave()
+ m.Save("waiterEntry", &x.waiterEntry)
+ m.Save("value", &x.value)
+ m.Save("ch", &x.ch)
+}
+
+func (x *waiter) afterLoad() {}
+func (x *waiter) load(m state.Map) {
+ m.Load("waiterEntry", &x.waiterEntry)
+ m.Load("value", &x.value)
+ m.Load("ch", &x.ch)
+}
+
+func (x *waiterList) beforeSave() {}
+func (x *waiterList) save(m state.Map) {
+ x.beforeSave()
+ m.Save("head", &x.head)
+ m.Save("tail", &x.tail)
+}
+
+func (x *waiterList) afterLoad() {}
+func (x *waiterList) load(m state.Map) {
+ m.Load("head", &x.head)
+ m.Load("tail", &x.tail)
+}
+
+func (x *waiterEntry) beforeSave() {}
+func (x *waiterEntry) save(m state.Map) {
+ x.beforeSave()
+ m.Save("next", &x.next)
+ m.Save("prev", &x.prev)
+}
+
+func (x *waiterEntry) afterLoad() {}
+func (x *waiterEntry) load(m state.Map) {
+ m.Load("next", &x.next)
+ m.Load("prev", &x.prev)
+}
+
+func init() {
+ state.Register("semaphore.Registry", (*Registry)(nil), state.Fns{Save: (*Registry).save, Load: (*Registry).load})
+ state.Register("semaphore.Set", (*Set)(nil), state.Fns{Save: (*Set).save, Load: (*Set).load})
+ state.Register("semaphore.sem", (*sem)(nil), state.Fns{Save: (*sem).save, Load: (*sem).load})
+ state.Register("semaphore.waiter", (*waiter)(nil), state.Fns{Save: (*waiter).save, Load: (*waiter).load})
+ state.Register("semaphore.waiterList", (*waiterList)(nil), state.Fns{Save: (*waiterList).save, Load: (*waiterList).load})
+ state.Register("semaphore.waiterEntry", (*waiterEntry)(nil), state.Fns{Save: (*waiterEntry).save, Load: (*waiterEntry).load})
+}
diff --git a/pkg/sentry/kernel/semaphore/semaphore_test.go b/pkg/sentry/kernel/semaphore/semaphore_test.go
deleted file mode 100644
index c235f6ca4..000000000
--- a/pkg/sentry/kernel/semaphore/semaphore_test.go
+++ /dev/null
@@ -1,172 +0,0 @@
-// Copyright 2018 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 semaphore
-
-import (
- "testing"
-
- "gvisor.dev/gvisor/pkg/abi/linux"
- "gvisor.dev/gvisor/pkg/sentry/context"
- "gvisor.dev/gvisor/pkg/sentry/context/contexttest"
- "gvisor.dev/gvisor/pkg/sentry/kernel/auth"
- "gvisor.dev/gvisor/pkg/syserror"
-)
-
-func executeOps(ctx context.Context, t *testing.T, set *Set, ops []linux.Sembuf, block bool) chan struct{} {
- ch, _, err := set.executeOps(ctx, ops, 123)
- if err != nil {
- t.Fatalf("ExecuteOps(ops) failed, err: %v, ops: %+v", err, ops)
- }
- if block {
- if ch == nil {
- t.Fatalf("ExecuteOps(ops) got: nil, expected: !nil, ops: %+v", ops)
- }
- if signalled(ch) {
- t.Fatalf("ExecuteOps(ops) channel should not have been signalled, ops: %+v", ops)
- }
- } else {
- if ch != nil {
- t.Fatalf("ExecuteOps(ops) got: %v, expected: nil, ops: %+v", ch, ops)
- }
- }
- return ch
-}
-
-func signalled(ch chan struct{}) bool {
- select {
- case <-ch:
- return true
- default:
- return false
- }
-}
-
-func TestBasic(t *testing.T) {
- ctx := contexttest.Context(t)
- set := &Set{ID: 123, sems: make([]sem, 1)}
- ops := []linux.Sembuf{
- {SemOp: 1},
- }
- executeOps(ctx, t, set, ops, false)
-
- ops[0].SemOp = -1
- executeOps(ctx, t, set, ops, false)
-
- ops[0].SemOp = -1
- ch1 := executeOps(ctx, t, set, ops, true)
-
- ops[0].SemOp = 1
- executeOps(ctx, t, set, ops, false)
- if !signalled(ch1) {
- t.Fatalf("ExecuteOps(ops) channel should not have been signalled, ops: %+v", ops)
- }
-}
-
-func TestWaitForZero(t *testing.T) {
- ctx := contexttest.Context(t)
- set := &Set{ID: 123, sems: make([]sem, 1)}
- ops := []linux.Sembuf{
- {SemOp: 0},
- }
- executeOps(ctx, t, set, ops, false)
-
- ops[0].SemOp = -2
- ch1 := executeOps(ctx, t, set, ops, true)
-
- ops[0].SemOp = 0
- executeOps(ctx, t, set, ops, false)
-
- ops[0].SemOp = 1
- executeOps(ctx, t, set, ops, false)
-
- ops[0].SemOp = 0
- chZero1 := executeOps(ctx, t, set, ops, true)
-
- ops[0].SemOp = 0
- chZero2 := executeOps(ctx, t, set, ops, true)
-
- ops[0].SemOp = 1
- executeOps(ctx, t, set, ops, false)
- if !signalled(ch1) {
- t.Fatalf("ExecuteOps(ops) channel should have been signalled, ops: %+v, set: %+v", ops, set)
- }
-
- ops[0].SemOp = -2
- executeOps(ctx, t, set, ops, false)
- if !signalled(chZero1) {
- t.Fatalf("ExecuteOps(ops) channel zero 1 should have been signalled, ops: %+v, set: %+v", ops, set)
- }
- if !signalled(chZero2) {
- t.Fatalf("ExecuteOps(ops) channel zero 2 should have been signalled, ops: %+v, set: %+v", ops, set)
- }
-}
-
-func TestNoWait(t *testing.T) {
- ctx := contexttest.Context(t)
- set := &Set{ID: 123, sems: make([]sem, 1)}
- ops := []linux.Sembuf{
- {SemOp: 1},
- }
- executeOps(ctx, t, set, ops, false)
-
- ops[0].SemOp = -2
- ops[0].SemFlg = linux.IPC_NOWAIT
- if _, _, err := set.executeOps(ctx, ops, 123); err != syserror.ErrWouldBlock {
- t.Fatalf("ExecuteOps(ops) wrong result, got: %v, expected: %v", err, syserror.ErrWouldBlock)
- }
-
- ops[0].SemOp = 0
- ops[0].SemFlg = linux.IPC_NOWAIT
- if _, _, err := set.executeOps(ctx, ops, 123); err != syserror.ErrWouldBlock {
- t.Fatalf("ExecuteOps(ops) wrong result, got: %v, expected: %v", err, syserror.ErrWouldBlock)
- }
-}
-
-func TestUnregister(t *testing.T) {
- ctx := contexttest.Context(t)
- r := NewRegistry(auth.NewRootUserNamespace())
- set, err := r.FindOrCreate(ctx, 123, 2, linux.FileMode(0x600), true, true, true)
- if err != nil {
- t.Fatalf("FindOrCreate() failed, err: %v", err)
- }
- if got := r.FindByID(set.ID); got.ID != set.ID {
- t.Fatalf("FindById(%d) failed, got: %+v, expected: %+v", set.ID, got, set)
- }
-
- ops := []linux.Sembuf{
- {SemOp: -1},
- }
- chs := make([]chan struct{}, 0, 5)
- for i := 0; i < 5; i++ {
- ch := executeOps(ctx, t, set, ops, true)
- chs = append(chs, ch)
- }
-
- creds := auth.CredentialsFromContext(ctx)
- if err := r.RemoveID(set.ID, creds); err != nil {
- t.Fatalf("RemoveID(%d) failed, err: %v", set.ID, err)
- }
- if !set.dead {
- t.Fatalf("set is not dead: %+v", set)
- }
- if got := r.FindByID(set.ID); got != nil {
- t.Fatalf("FindById(%d) failed, got: %+v, expected: nil", set.ID, got)
- }
- for i, ch := range chs {
- if !signalled(ch) {
- t.Fatalf("channel %d should have been signalled", i)
- }
- }
-}
diff --git a/pkg/sentry/kernel/semaphore/waiter_list.go b/pkg/sentry/kernel/semaphore/waiter_list.go
new file mode 100755
index 000000000..33e29fb55
--- /dev/null
+++ b/pkg/sentry/kernel/semaphore/waiter_list.go
@@ -0,0 +1,173 @@
+package semaphore
+
+// ElementMapper provides an identity mapping by default.
+//
+// This can be replaced to provide a struct that maps elements to linker
+// objects, if they are not the same. An ElementMapper is not typically
+// required if: Linker is left as is, Element is left as is, or Linker and
+// Element are the same type.
+type waiterElementMapper struct{}
+
+// linkerFor maps an Element to a Linker.
+//
+// This default implementation should be inlined.
+//
+//go:nosplit
+func (waiterElementMapper) linkerFor(elem *waiter) *waiter { return elem }
+
+// List is an intrusive list. Entries can be added to or removed from the list
+// in O(1) time and with no additional memory allocations.
+//
+// The zero value for List is an empty list ready to use.
+//
+// To iterate over a list (where l is a List):
+// for e := l.Front(); e != nil; e = e.Next() {
+// // do something with e.
+// }
+//
+// +stateify savable
+type waiterList struct {
+ head *waiter
+ tail *waiter
+}
+
+// Reset resets list l to the empty state.
+func (l *waiterList) Reset() {
+ l.head = nil
+ l.tail = nil
+}
+
+// Empty returns true iff the list is empty.
+func (l *waiterList) Empty() bool {
+ return l.head == nil
+}
+
+// Front returns the first element of list l or nil.
+func (l *waiterList) Front() *waiter {
+ return l.head
+}
+
+// Back returns the last element of list l or nil.
+func (l *waiterList) Back() *waiter {
+ return l.tail
+}
+
+// PushFront inserts the element e at the front of list l.
+func (l *waiterList) PushFront(e *waiter) {
+ waiterElementMapper{}.linkerFor(e).SetNext(l.head)
+ waiterElementMapper{}.linkerFor(e).SetPrev(nil)
+
+ if l.head != nil {
+ waiterElementMapper{}.linkerFor(l.head).SetPrev(e)
+ } else {
+ l.tail = e
+ }
+
+ l.head = e
+}
+
+// PushBack inserts the element e at the back of list l.
+func (l *waiterList) PushBack(e *waiter) {
+ waiterElementMapper{}.linkerFor(e).SetNext(nil)
+ waiterElementMapper{}.linkerFor(e).SetPrev(l.tail)
+
+ if l.tail != nil {
+ waiterElementMapper{}.linkerFor(l.tail).SetNext(e)
+ } else {
+ l.head = e
+ }
+
+ l.tail = e
+}
+
+// PushBackList inserts list m at the end of list l, emptying m.
+func (l *waiterList) PushBackList(m *waiterList) {
+ if l.head == nil {
+ l.head = m.head
+ l.tail = m.tail
+ } else if m.head != nil {
+ waiterElementMapper{}.linkerFor(l.tail).SetNext(m.head)
+ waiterElementMapper{}.linkerFor(m.head).SetPrev(l.tail)
+
+ l.tail = m.tail
+ }
+
+ m.head = nil
+ m.tail = nil
+}
+
+// InsertAfter inserts e after b.
+func (l *waiterList) InsertAfter(b, e *waiter) {
+ a := waiterElementMapper{}.linkerFor(b).Next()
+ waiterElementMapper{}.linkerFor(e).SetNext(a)
+ waiterElementMapper{}.linkerFor(e).SetPrev(b)
+ waiterElementMapper{}.linkerFor(b).SetNext(e)
+
+ if a != nil {
+ waiterElementMapper{}.linkerFor(a).SetPrev(e)
+ } else {
+ l.tail = e
+ }
+}
+
+// InsertBefore inserts e before a.
+func (l *waiterList) InsertBefore(a, e *waiter) {
+ b := waiterElementMapper{}.linkerFor(a).Prev()
+ waiterElementMapper{}.linkerFor(e).SetNext(a)
+ waiterElementMapper{}.linkerFor(e).SetPrev(b)
+ waiterElementMapper{}.linkerFor(a).SetPrev(e)
+
+ if b != nil {
+ waiterElementMapper{}.linkerFor(b).SetNext(e)
+ } else {
+ l.head = e
+ }
+}
+
+// Remove removes e from l.
+func (l *waiterList) Remove(e *waiter) {
+ prev := waiterElementMapper{}.linkerFor(e).Prev()
+ next := waiterElementMapper{}.linkerFor(e).Next()
+
+ if prev != nil {
+ waiterElementMapper{}.linkerFor(prev).SetNext(next)
+ } else {
+ l.head = next
+ }
+
+ if next != nil {
+ waiterElementMapper{}.linkerFor(next).SetPrev(prev)
+ } else {
+ l.tail = prev
+ }
+}
+
+// Entry is a default implementation of Linker. Users can add anonymous fields
+// of this type to their structs to make them automatically implement the
+// methods needed by List.
+//
+// +stateify savable
+type waiterEntry struct {
+ next *waiter
+ prev *waiter
+}
+
+// Next returns the entry that follows e in the list.
+func (e *waiterEntry) Next() *waiter {
+ return e.next
+}
+
+// Prev returns the entry that precedes e in the list.
+func (e *waiterEntry) Prev() *waiter {
+ return e.prev
+}
+
+// SetNext assigns 'entry' as the entry that follows e in the list.
+func (e *waiterEntry) SetNext(elem *waiter) {
+ e.next = elem
+}
+
+// SetPrev assigns 'entry' as the entry that precedes e in the list.
+func (e *waiterEntry) SetPrev(elem *waiter) {
+ e.prev = elem
+}
diff --git a/pkg/sentry/kernel/seqatomic_taskgoroutineschedinfo_unsafe.go b/pkg/sentry/kernel/seqatomic_taskgoroutineschedinfo_unsafe.go
new file mode 100755
index 000000000..950645965
--- /dev/null
+++ b/pkg/sentry/kernel/seqatomic_taskgoroutineschedinfo_unsafe.go
@@ -0,0 +1,55 @@
+package kernel
+
+import (
+ "fmt"
+ "reflect"
+ "strings"
+ "unsafe"
+
+ "gvisor.dev/gvisor/pkg/sync"
+)
+
+// SeqAtomicLoad returns a copy of *ptr, ensuring that the read does not race
+// with any writer critical sections in sc.
+func SeqAtomicLoadTaskGoroutineSchedInfo(sc *sync.SeqCount, ptr *TaskGoroutineSchedInfo) TaskGoroutineSchedInfo {
+ // This function doesn't use SeqAtomicTryLoad because doing so is
+ // measurably, significantly (~20%) slower; Go is awful at inlining.
+ var val TaskGoroutineSchedInfo
+ for {
+ epoch := sc.BeginRead()
+ if sync.RaceEnabled {
+
+ sync.Memmove(unsafe.Pointer(&val), unsafe.Pointer(ptr), unsafe.Sizeof(val))
+ } else {
+
+ val = *ptr
+ }
+ if sc.ReadOk(epoch) {
+ break
+ }
+ }
+ return val
+}
+
+// SeqAtomicTryLoad returns a copy of *ptr while in a reader critical section
+// in sc initiated by a call to sc.BeginRead() that returned epoch. If the read
+// would race with a writer critical section, SeqAtomicTryLoad returns
+// (unspecified, false).
+func SeqAtomicTryLoadTaskGoroutineSchedInfo(sc *sync.SeqCount, epoch sync.SeqCountEpoch, ptr *TaskGoroutineSchedInfo) (TaskGoroutineSchedInfo, bool) {
+ var val TaskGoroutineSchedInfo
+ if sync.RaceEnabled {
+ sync.Memmove(unsafe.Pointer(&val), unsafe.Pointer(ptr), unsafe.Sizeof(val))
+ } else {
+ val = *ptr
+ }
+ return val, sc.ReadOk(epoch)
+}
+
+func initTaskGoroutineSchedInfo() {
+ var val TaskGoroutineSchedInfo
+ typ := reflect.TypeOf(val)
+ name := typ.Name()
+ if ptrs := sync.PointersInType(typ, name); len(ptrs) != 0 {
+ panic(fmt.Sprintf("SeqAtomicLoad<%s> is invalid since values %s of type %s contain pointers:\n%s", typ, name, typ, strings.Join(ptrs, "\n")))
+ }
+}
diff --git a/pkg/sentry/kernel/session_list.go b/pkg/sentry/kernel/session_list.go
new file mode 100755
index 000000000..9ba27b164
--- /dev/null
+++ b/pkg/sentry/kernel/session_list.go
@@ -0,0 +1,173 @@
+package kernel
+
+// ElementMapper provides an identity mapping by default.
+//
+// This can be replaced to provide a struct that maps elements to linker
+// objects, if they are not the same. An ElementMapper is not typically
+// required if: Linker is left as is, Element is left as is, or Linker and
+// Element are the same type.
+type sessionElementMapper struct{}
+
+// linkerFor maps an Element to a Linker.
+//
+// This default implementation should be inlined.
+//
+//go:nosplit
+func (sessionElementMapper) linkerFor(elem *Session) *Session { return elem }
+
+// List is an intrusive list. Entries can be added to or removed from the list
+// in O(1) time and with no additional memory allocations.
+//
+// The zero value for List is an empty list ready to use.
+//
+// To iterate over a list (where l is a List):
+// for e := l.Front(); e != nil; e = e.Next() {
+// // do something with e.
+// }
+//
+// +stateify savable
+type sessionList struct {
+ head *Session
+ tail *Session
+}
+
+// Reset resets list l to the empty state.
+func (l *sessionList) Reset() {
+ l.head = nil
+ l.tail = nil
+}
+
+// Empty returns true iff the list is empty.
+func (l *sessionList) Empty() bool {
+ return l.head == nil
+}
+
+// Front returns the first element of list l or nil.
+func (l *sessionList) Front() *Session {
+ return l.head
+}
+
+// Back returns the last element of list l or nil.
+func (l *sessionList) Back() *Session {
+ return l.tail
+}
+
+// PushFront inserts the element e at the front of list l.
+func (l *sessionList) PushFront(e *Session) {
+ sessionElementMapper{}.linkerFor(e).SetNext(l.head)
+ sessionElementMapper{}.linkerFor(e).SetPrev(nil)
+
+ if l.head != nil {
+ sessionElementMapper{}.linkerFor(l.head).SetPrev(e)
+ } else {
+ l.tail = e
+ }
+
+ l.head = e
+}
+
+// PushBack inserts the element e at the back of list l.
+func (l *sessionList) PushBack(e *Session) {
+ sessionElementMapper{}.linkerFor(e).SetNext(nil)
+ sessionElementMapper{}.linkerFor(e).SetPrev(l.tail)
+
+ if l.tail != nil {
+ sessionElementMapper{}.linkerFor(l.tail).SetNext(e)
+ } else {
+ l.head = e
+ }
+
+ l.tail = e
+}
+
+// PushBackList inserts list m at the end of list l, emptying m.
+func (l *sessionList) PushBackList(m *sessionList) {
+ if l.head == nil {
+ l.head = m.head
+ l.tail = m.tail
+ } else if m.head != nil {
+ sessionElementMapper{}.linkerFor(l.tail).SetNext(m.head)
+ sessionElementMapper{}.linkerFor(m.head).SetPrev(l.tail)
+
+ l.tail = m.tail
+ }
+
+ m.head = nil
+ m.tail = nil
+}
+
+// InsertAfter inserts e after b.
+func (l *sessionList) InsertAfter(b, e *Session) {
+ a := sessionElementMapper{}.linkerFor(b).Next()
+ sessionElementMapper{}.linkerFor(e).SetNext(a)
+ sessionElementMapper{}.linkerFor(e).SetPrev(b)
+ sessionElementMapper{}.linkerFor(b).SetNext(e)
+
+ if a != nil {
+ sessionElementMapper{}.linkerFor(a).SetPrev(e)
+ } else {
+ l.tail = e
+ }
+}
+
+// InsertBefore inserts e before a.
+func (l *sessionList) InsertBefore(a, e *Session) {
+ b := sessionElementMapper{}.linkerFor(a).Prev()
+ sessionElementMapper{}.linkerFor(e).SetNext(a)
+ sessionElementMapper{}.linkerFor(e).SetPrev(b)
+ sessionElementMapper{}.linkerFor(a).SetPrev(e)
+
+ if b != nil {
+ sessionElementMapper{}.linkerFor(b).SetNext(e)
+ } else {
+ l.head = e
+ }
+}
+
+// Remove removes e from l.
+func (l *sessionList) Remove(e *Session) {
+ prev := sessionElementMapper{}.linkerFor(e).Prev()
+ next := sessionElementMapper{}.linkerFor(e).Next()
+
+ if prev != nil {
+ sessionElementMapper{}.linkerFor(prev).SetNext(next)
+ } else {
+ l.head = next
+ }
+
+ if next != nil {
+ sessionElementMapper{}.linkerFor(next).SetPrev(prev)
+ } else {
+ l.tail = prev
+ }
+}
+
+// Entry is a default implementation of Linker. Users can add anonymous fields
+// of this type to their structs to make them automatically implement the
+// methods needed by List.
+//
+// +stateify savable
+type sessionEntry struct {
+ next *Session
+ prev *Session
+}
+
+// Next returns the entry that follows e in the list.
+func (e *sessionEntry) Next() *Session {
+ return e.next
+}
+
+// Prev returns the entry that precedes e in the list.
+func (e *sessionEntry) Prev() *Session {
+ return e.prev
+}
+
+// SetNext assigns 'entry' as the entry that follows e in the list.
+func (e *sessionEntry) SetNext(elem *Session) {
+ e.next = elem
+}
+
+// SetPrev assigns 'entry' as the entry that precedes e in the list.
+func (e *sessionEntry) SetPrev(elem *Session) {
+ e.prev = elem
+}
diff --git a/pkg/sentry/kernel/shm/BUILD b/pkg/sentry/kernel/shm/BUILD
deleted file mode 100644
index 7321b22ed..000000000
--- a/pkg/sentry/kernel/shm/BUILD
+++ /dev/null
@@ -1,30 +0,0 @@
-load("//tools/go_stateify:defs.bzl", "go_library")
-
-package(licenses = ["notice"])
-
-go_library(
- name = "shm",
- srcs = [
- "device.go",
- "shm.go",
- ],
- importpath = "gvisor.dev/gvisor/pkg/sentry/kernel/shm",
- visibility = ["//pkg/sentry:internal"],
- deps = [
- "//pkg/abi/linux",
- "//pkg/log",
- "//pkg/refs",
- "//pkg/sentry/context",
- "//pkg/sentry/device",
- "//pkg/sentry/fs",
- "//pkg/sentry/kernel/auth",
- "//pkg/sentry/kernel/time",
- "//pkg/sentry/memmap",
- "//pkg/sentry/pgalloc",
- "//pkg/sentry/platform",
- "//pkg/sentry/usage",
- "//pkg/sentry/usermem",
- "//pkg/sync",
- "//pkg/syserror",
- ],
-)
diff --git a/pkg/sentry/kernel/shm/shm_state_autogen.go b/pkg/sentry/kernel/shm/shm_state_autogen.go
new file mode 100755
index 000000000..4fd44bb6a
--- /dev/null
+++ b/pkg/sentry/kernel/shm/shm_state_autogen.go
@@ -0,0 +1,74 @@
+// automatically generated by stateify.
+
+package shm
+
+import (
+ "gvisor.dev/gvisor/pkg/state"
+)
+
+func (x *Registry) beforeSave() {}
+func (x *Registry) save(m state.Map) {
+ x.beforeSave()
+ m.Save("userNS", &x.userNS)
+ m.Save("shms", &x.shms)
+ m.Save("keysToShms", &x.keysToShms)
+ m.Save("totalPages", &x.totalPages)
+ m.Save("lastIDUsed", &x.lastIDUsed)
+}
+
+func (x *Registry) afterLoad() {}
+func (x *Registry) load(m state.Map) {
+ m.Load("userNS", &x.userNS)
+ m.Load("shms", &x.shms)
+ m.Load("keysToShms", &x.keysToShms)
+ m.Load("totalPages", &x.totalPages)
+ m.Load("lastIDUsed", &x.lastIDUsed)
+}
+
+func (x *Shm) beforeSave() {}
+func (x *Shm) save(m state.Map) {
+ x.beforeSave()
+ m.Save("AtomicRefCount", &x.AtomicRefCount)
+ m.Save("mfp", &x.mfp)
+ m.Save("registry", &x.registry)
+ m.Save("ID", &x.ID)
+ m.Save("creator", &x.creator)
+ m.Save("size", &x.size)
+ m.Save("effectiveSize", &x.effectiveSize)
+ m.Save("fr", &x.fr)
+ m.Save("key", &x.key)
+ m.Save("perms", &x.perms)
+ m.Save("owner", &x.owner)
+ m.Save("attachTime", &x.attachTime)
+ m.Save("detachTime", &x.detachTime)
+ m.Save("changeTime", &x.changeTime)
+ m.Save("creatorPID", &x.creatorPID)
+ m.Save("lastAttachDetachPID", &x.lastAttachDetachPID)
+ m.Save("pendingDestruction", &x.pendingDestruction)
+}
+
+func (x *Shm) afterLoad() {}
+func (x *Shm) load(m state.Map) {
+ m.Load("AtomicRefCount", &x.AtomicRefCount)
+ m.Load("mfp", &x.mfp)
+ m.Load("registry", &x.registry)
+ m.Load("ID", &x.ID)
+ m.Load("creator", &x.creator)
+ m.Load("size", &x.size)
+ m.Load("effectiveSize", &x.effectiveSize)
+ m.Load("fr", &x.fr)
+ m.Load("key", &x.key)
+ m.Load("perms", &x.perms)
+ m.Load("owner", &x.owner)
+ m.Load("attachTime", &x.attachTime)
+ m.Load("detachTime", &x.detachTime)
+ m.Load("changeTime", &x.changeTime)
+ m.Load("creatorPID", &x.creatorPID)
+ m.Load("lastAttachDetachPID", &x.lastAttachDetachPID)
+ m.Load("pendingDestruction", &x.pendingDestruction)
+}
+
+func init() {
+ state.Register("shm.Registry", (*Registry)(nil), state.Fns{Save: (*Registry).save, Load: (*Registry).load})
+ state.Register("shm.Shm", (*Shm)(nil), state.Fns{Save: (*Shm).save, Load: (*Shm).load})
+}
diff --git a/pkg/sentry/kernel/signalfd/BUILD b/pkg/sentry/kernel/signalfd/BUILD
deleted file mode 100644
index 89e4d84b1..000000000
--- a/pkg/sentry/kernel/signalfd/BUILD
+++ /dev/null
@@ -1,23 +0,0 @@
-load("//tools/go_stateify:defs.bzl", "go_library")
-
-package(licenses = ["notice"])
-
-go_library(
- name = "signalfd",
- srcs = ["signalfd.go"],
- importpath = "gvisor.dev/gvisor/pkg/sentry/kernel/signalfd",
- visibility = ["//pkg/sentry:internal"],
- deps = [
- "//pkg/abi/linux",
- "//pkg/binary",
- "//pkg/sentry/context",
- "//pkg/sentry/fs",
- "//pkg/sentry/fs/anon",
- "//pkg/sentry/fs/fsutil",
- "//pkg/sentry/kernel",
- "//pkg/sentry/usermem",
- "//pkg/sync",
- "//pkg/syserror",
- "//pkg/waiter",
- ],
-)
diff --git a/pkg/sentry/kernel/signalfd/signalfd.go b/pkg/sentry/kernel/signalfd/signalfd.go
index 28be4a939..28be4a939 100644..100755
--- a/pkg/sentry/kernel/signalfd/signalfd.go
+++ b/pkg/sentry/kernel/signalfd/signalfd.go
diff --git a/pkg/sentry/kernel/signalfd/signalfd_state_autogen.go b/pkg/sentry/kernel/signalfd/signalfd_state_autogen.go
new file mode 100755
index 000000000..66f09886d
--- /dev/null
+++ b/pkg/sentry/kernel/signalfd/signalfd_state_autogen.go
@@ -0,0 +1,24 @@
+// automatically generated by stateify.
+
+package signalfd
+
+import (
+ "gvisor.dev/gvisor/pkg/state"
+)
+
+func (x *SignalOperations) beforeSave() {}
+func (x *SignalOperations) save(m state.Map) {
+ x.beforeSave()
+ m.Save("target", &x.target)
+ m.Save("mask", &x.mask)
+}
+
+func (x *SignalOperations) afterLoad() {}
+func (x *SignalOperations) load(m state.Map) {
+ m.Load("target", &x.target)
+ m.Load("mask", &x.mask)
+}
+
+func init() {
+ state.Register("signalfd.SignalOperations", (*SignalOperations)(nil), state.Fns{Save: (*SignalOperations).save, Load: (*SignalOperations).load})
+}
diff --git a/pkg/sentry/kernel/socket_list.go b/pkg/sentry/kernel/socket_list.go
new file mode 100755
index 000000000..aed0a555e
--- /dev/null
+++ b/pkg/sentry/kernel/socket_list.go
@@ -0,0 +1,173 @@
+package kernel
+
+// ElementMapper provides an identity mapping by default.
+//
+// This can be replaced to provide a struct that maps elements to linker
+// objects, if they are not the same. An ElementMapper is not typically
+// required if: Linker is left as is, Element is left as is, or Linker and
+// Element are the same type.
+type socketElementMapper struct{}
+
+// linkerFor maps an Element to a Linker.
+//
+// This default implementation should be inlined.
+//
+//go:nosplit
+func (socketElementMapper) linkerFor(elem *SocketEntry) *SocketEntry { return elem }
+
+// List is an intrusive list. Entries can be added to or removed from the list
+// in O(1) time and with no additional memory allocations.
+//
+// The zero value for List is an empty list ready to use.
+//
+// To iterate over a list (where l is a List):
+// for e := l.Front(); e != nil; e = e.Next() {
+// // do something with e.
+// }
+//
+// +stateify savable
+type socketList struct {
+ head *SocketEntry
+ tail *SocketEntry
+}
+
+// Reset resets list l to the empty state.
+func (l *socketList) Reset() {
+ l.head = nil
+ l.tail = nil
+}
+
+// Empty returns true iff the list is empty.
+func (l *socketList) Empty() bool {
+ return l.head == nil
+}
+
+// Front returns the first element of list l or nil.
+func (l *socketList) Front() *SocketEntry {
+ return l.head
+}
+
+// Back returns the last element of list l or nil.
+func (l *socketList) Back() *SocketEntry {
+ return l.tail
+}
+
+// PushFront inserts the element e at the front of list l.
+func (l *socketList) PushFront(e *SocketEntry) {
+ socketElementMapper{}.linkerFor(e).SetNext(l.head)
+ socketElementMapper{}.linkerFor(e).SetPrev(nil)
+
+ if l.head != nil {
+ socketElementMapper{}.linkerFor(l.head).SetPrev(e)
+ } else {
+ l.tail = e
+ }
+
+ l.head = e
+}
+
+// PushBack inserts the element e at the back of list l.
+func (l *socketList) PushBack(e *SocketEntry) {
+ socketElementMapper{}.linkerFor(e).SetNext(nil)
+ socketElementMapper{}.linkerFor(e).SetPrev(l.tail)
+
+ if l.tail != nil {
+ socketElementMapper{}.linkerFor(l.tail).SetNext(e)
+ } else {
+ l.head = e
+ }
+
+ l.tail = e
+}
+
+// PushBackList inserts list m at the end of list l, emptying m.
+func (l *socketList) PushBackList(m *socketList) {
+ if l.head == nil {
+ l.head = m.head
+ l.tail = m.tail
+ } else if m.head != nil {
+ socketElementMapper{}.linkerFor(l.tail).SetNext(m.head)
+ socketElementMapper{}.linkerFor(m.head).SetPrev(l.tail)
+
+ l.tail = m.tail
+ }
+
+ m.head = nil
+ m.tail = nil
+}
+
+// InsertAfter inserts e after b.
+func (l *socketList) InsertAfter(b, e *SocketEntry) {
+ a := socketElementMapper{}.linkerFor(b).Next()
+ socketElementMapper{}.linkerFor(e).SetNext(a)
+ socketElementMapper{}.linkerFor(e).SetPrev(b)
+ socketElementMapper{}.linkerFor(b).SetNext(e)
+
+ if a != nil {
+ socketElementMapper{}.linkerFor(a).SetPrev(e)
+ } else {
+ l.tail = e
+ }
+}
+
+// InsertBefore inserts e before a.
+func (l *socketList) InsertBefore(a, e *SocketEntry) {
+ b := socketElementMapper{}.linkerFor(a).Prev()
+ socketElementMapper{}.linkerFor(e).SetNext(a)
+ socketElementMapper{}.linkerFor(e).SetPrev(b)
+ socketElementMapper{}.linkerFor(a).SetPrev(e)
+
+ if b != nil {
+ socketElementMapper{}.linkerFor(b).SetNext(e)
+ } else {
+ l.head = e
+ }
+}
+
+// Remove removes e from l.
+func (l *socketList) Remove(e *SocketEntry) {
+ prev := socketElementMapper{}.linkerFor(e).Prev()
+ next := socketElementMapper{}.linkerFor(e).Next()
+
+ if prev != nil {
+ socketElementMapper{}.linkerFor(prev).SetNext(next)
+ } else {
+ l.head = next
+ }
+
+ if next != nil {
+ socketElementMapper{}.linkerFor(next).SetPrev(prev)
+ } else {
+ l.tail = prev
+ }
+}
+
+// Entry is a default implementation of Linker. Users can add anonymous fields
+// of this type to their structs to make them automatically implement the
+// methods needed by List.
+//
+// +stateify savable
+type socketEntry struct {
+ next *SocketEntry
+ prev *SocketEntry
+}
+
+// Next returns the entry that follows e in the list.
+func (e *socketEntry) Next() *SocketEntry {
+ return e.next
+}
+
+// Prev returns the entry that precedes e in the list.
+func (e *socketEntry) Prev() *SocketEntry {
+ return e.prev
+}
+
+// SetNext assigns 'entry' as the entry that follows e in the list.
+func (e *socketEntry) SetNext(elem *SocketEntry) {
+ e.next = elem
+}
+
+// SetPrev assigns 'entry' as the entry that precedes e in the list.
+func (e *socketEntry) SetPrev(elem *SocketEntry) {
+ e.prev = elem
+}
diff --git a/pkg/sentry/kernel/table_test.go b/pkg/sentry/kernel/table_test.go
deleted file mode 100644
index 32cf47e05..000000000
--- a/pkg/sentry/kernel/table_test.go
+++ /dev/null
@@ -1,110 +0,0 @@
-// Copyright 2018 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 kernel
-
-import (
- "testing"
-
- "gvisor.dev/gvisor/pkg/abi"
- "gvisor.dev/gvisor/pkg/sentry/arch"
-)
-
-const (
- maxTestSyscall = 1000
-)
-
-func createSyscallTable() *SyscallTable {
- m := make(map[uintptr]Syscall)
- for i := uintptr(0); i <= maxTestSyscall; i++ {
- j := i
- m[i] = Syscall{
- Fn: func(*Task, arch.SyscallArguments) (uintptr, *SyscallControl, error) {
- return j, nil, nil
- },
- }
- }
-
- s := &SyscallTable{
- OS: abi.Linux,
- Arch: arch.AMD64,
- Table: m,
- }
-
- RegisterSyscallTable(s)
- return s
-}
-
-func TestTable(t *testing.T) {
- table := createSyscallTable()
- defer func() {
- // Cleanup registered tables to keep tests separate.
- allSyscallTables = []*SyscallTable{}
- }()
-
- // Go through all functions and check that they return the right value.
- for i := uintptr(0); i < maxTestSyscall; i++ {
- fn := table.Lookup(i)
- if fn == nil {
- t.Errorf("Syscall %v is set to nil", i)
- continue
- }
-
- v, _, _ := fn(nil, arch.SyscallArguments{})
- if v != i {
- t.Errorf("Wrong return value for syscall %v: expected %v, got %v", i, i, v)
- }
- }
-
- // Check that values outside the range return nil.
- for i := uintptr(maxTestSyscall + 1); i < maxTestSyscall+100; i++ {
- fn := table.Lookup(i)
- if fn != nil {
- t.Errorf("Syscall %v is not nil: %v", i, fn)
- continue
- }
- }
-}
-
-func BenchmarkTableLookup(b *testing.B) {
- table := createSyscallTable()
-
- b.ResetTimer()
-
- j := uintptr(0)
- for i := 0; i < b.N; i++ {
- table.Lookup(j)
- j = (j + 1) % 310
- }
-
- b.StopTimer()
- // Cleanup registered tables to keep tests separate.
- allSyscallTables = []*SyscallTable{}
-}
-
-func BenchmarkTableMapLookup(b *testing.B) {
- table := createSyscallTable()
-
- b.ResetTimer()
-
- j := uintptr(0)
- for i := 0; i < b.N; i++ {
- table.mapLookup(j)
- j = (j + 1) % 310
- }
-
- b.StopTimer()
- // Cleanup registered tables to keep tests separate.
- allSyscallTables = []*SyscallTable{}
-}
diff --git a/pkg/sentry/kernel/task_list.go b/pkg/sentry/kernel/task_list.go
new file mode 100755
index 000000000..57d3f098d
--- /dev/null
+++ b/pkg/sentry/kernel/task_list.go
@@ -0,0 +1,173 @@
+package kernel
+
+// ElementMapper provides an identity mapping by default.
+//
+// This can be replaced to provide a struct that maps elements to linker
+// objects, if they are not the same. An ElementMapper is not typically
+// required if: Linker is left as is, Element is left as is, or Linker and
+// Element are the same type.
+type taskElementMapper struct{}
+
+// linkerFor maps an Element to a Linker.
+//
+// This default implementation should be inlined.
+//
+//go:nosplit
+func (taskElementMapper) linkerFor(elem *Task) *Task { return elem }
+
+// List is an intrusive list. Entries can be added to or removed from the list
+// in O(1) time and with no additional memory allocations.
+//
+// The zero value for List is an empty list ready to use.
+//
+// To iterate over a list (where l is a List):
+// for e := l.Front(); e != nil; e = e.Next() {
+// // do something with e.
+// }
+//
+// +stateify savable
+type taskList struct {
+ head *Task
+ tail *Task
+}
+
+// Reset resets list l to the empty state.
+func (l *taskList) Reset() {
+ l.head = nil
+ l.tail = nil
+}
+
+// Empty returns true iff the list is empty.
+func (l *taskList) Empty() bool {
+ return l.head == nil
+}
+
+// Front returns the first element of list l or nil.
+func (l *taskList) Front() *Task {
+ return l.head
+}
+
+// Back returns the last element of list l or nil.
+func (l *taskList) Back() *Task {
+ return l.tail
+}
+
+// PushFront inserts the element e at the front of list l.
+func (l *taskList) PushFront(e *Task) {
+ taskElementMapper{}.linkerFor(e).SetNext(l.head)
+ taskElementMapper{}.linkerFor(e).SetPrev(nil)
+
+ if l.head != nil {
+ taskElementMapper{}.linkerFor(l.head).SetPrev(e)
+ } else {
+ l.tail = e
+ }
+
+ l.head = e
+}
+
+// PushBack inserts the element e at the back of list l.
+func (l *taskList) PushBack(e *Task) {
+ taskElementMapper{}.linkerFor(e).SetNext(nil)
+ taskElementMapper{}.linkerFor(e).SetPrev(l.tail)
+
+ if l.tail != nil {
+ taskElementMapper{}.linkerFor(l.tail).SetNext(e)
+ } else {
+ l.head = e
+ }
+
+ l.tail = e
+}
+
+// PushBackList inserts list m at the end of list l, emptying m.
+func (l *taskList) PushBackList(m *taskList) {
+ if l.head == nil {
+ l.head = m.head
+ l.tail = m.tail
+ } else if m.head != nil {
+ taskElementMapper{}.linkerFor(l.tail).SetNext(m.head)
+ taskElementMapper{}.linkerFor(m.head).SetPrev(l.tail)
+
+ l.tail = m.tail
+ }
+
+ m.head = nil
+ m.tail = nil
+}
+
+// InsertAfter inserts e after b.
+func (l *taskList) InsertAfter(b, e *Task) {
+ a := taskElementMapper{}.linkerFor(b).Next()
+ taskElementMapper{}.linkerFor(e).SetNext(a)
+ taskElementMapper{}.linkerFor(e).SetPrev(b)
+ taskElementMapper{}.linkerFor(b).SetNext(e)
+
+ if a != nil {
+ taskElementMapper{}.linkerFor(a).SetPrev(e)
+ } else {
+ l.tail = e
+ }
+}
+
+// InsertBefore inserts e before a.
+func (l *taskList) InsertBefore(a, e *Task) {
+ b := taskElementMapper{}.linkerFor(a).Prev()
+ taskElementMapper{}.linkerFor(e).SetNext(a)
+ taskElementMapper{}.linkerFor(e).SetPrev(b)
+ taskElementMapper{}.linkerFor(a).SetPrev(e)
+
+ if b != nil {
+ taskElementMapper{}.linkerFor(b).SetNext(e)
+ } else {
+ l.head = e
+ }
+}
+
+// Remove removes e from l.
+func (l *taskList) Remove(e *Task) {
+ prev := taskElementMapper{}.linkerFor(e).Prev()
+ next := taskElementMapper{}.linkerFor(e).Next()
+
+ if prev != nil {
+ taskElementMapper{}.linkerFor(prev).SetNext(next)
+ } else {
+ l.head = next
+ }
+
+ if next != nil {
+ taskElementMapper{}.linkerFor(next).SetPrev(prev)
+ } else {
+ l.tail = prev
+ }
+}
+
+// Entry is a default implementation of Linker. Users can add anonymous fields
+// of this type to their structs to make them automatically implement the
+// methods needed by List.
+//
+// +stateify savable
+type taskEntry struct {
+ next *Task
+ prev *Task
+}
+
+// Next returns the entry that follows e in the list.
+func (e *taskEntry) Next() *Task {
+ return e.next
+}
+
+// Prev returns the entry that precedes e in the list.
+func (e *taskEntry) Prev() *Task {
+ return e.prev
+}
+
+// SetNext assigns 'entry' as the entry that follows e in the list.
+func (e *taskEntry) SetNext(elem *Task) {
+ e.next = elem
+}
+
+// SetPrev assigns 'entry' as the entry that precedes e in the list.
+func (e *taskEntry) SetPrev(elem *Task) {
+ e.prev = elem
+}
diff --git a/pkg/sentry/kernel/task_test.go b/pkg/sentry/kernel/task_test.go
deleted file mode 100644
index cfcde9a7a..000000000
--- a/pkg/sentry/kernel/task_test.go
+++ /dev/null
@@ -1,69 +0,0 @@
-// Copyright 2018 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 kernel
-
-import (
- "testing"
-
- "gvisor.dev/gvisor/pkg/sentry/kernel/sched"
-)
-
-func TestTaskCPU(t *testing.T) {
- for _, test := range []struct {
- mask sched.CPUSet
- tid ThreadID
- cpu int32
- }{
- {
- mask: []byte{0xff},
- tid: 1,
- cpu: 0,
- },
- {
- mask: []byte{0xff},
- tid: 10,
- cpu: 1,
- },
- {
- // more than 8 cpus.
- mask: []byte{0xff, 0xff},
- tid: 10,
- cpu: 9,
- },
- {
- // missing the first cpu.
- mask: []byte{0xfe},
- tid: 1,
- cpu: 1,
- },
- {
- mask: []byte{0xfe},
- tid: 10,
- cpu: 3,
- },
- {
- // missing the fifth cpu.
- mask: []byte{0xef},
- tid: 10,
- cpu: 2,
- },
- } {
- assigned := assignCPU(test.mask, test.tid)
- if test.cpu != assigned {
- t.Errorf("assignCPU(%v, %v) got %v, want %v", test.mask, test.tid, assigned, test.cpu)
- }
- }
-
-}
diff --git a/pkg/sentry/kernel/time/BUILD b/pkg/sentry/kernel/time/BUILD
deleted file mode 100644
index 4e4de0512..000000000
--- a/pkg/sentry/kernel/time/BUILD
+++ /dev/null
@@ -1,20 +0,0 @@
-load("//tools/go_stateify:defs.bzl", "go_library")
-
-package(licenses = ["notice"])
-
-go_library(
- name = "time",
- srcs = [
- "context.go",
- "time.go",
- ],
- importpath = "gvisor.dev/gvisor/pkg/sentry/kernel/time",
- visibility = ["//pkg/sentry:internal"],
- deps = [
- "//pkg/abi/linux",
- "//pkg/sentry/context",
- "//pkg/sync",
- "//pkg/syserror",
- "//pkg/waiter",
- ],
-)
diff --git a/pkg/sentry/kernel/time/time_state_autogen.go b/pkg/sentry/kernel/time/time_state_autogen.go
new file mode 100755
index 000000000..77b676cec
--- /dev/null
+++ b/pkg/sentry/kernel/time/time_state_autogen.go
@@ -0,0 +1,56 @@
+// automatically generated by stateify.
+
+package time
+
+import (
+ "gvisor.dev/gvisor/pkg/state"
+)
+
+func (x *Time) beforeSave() {}
+func (x *Time) save(m state.Map) {
+ x.beforeSave()
+ m.Save("ns", &x.ns)
+}
+
+func (x *Time) afterLoad() {}
+func (x *Time) load(m state.Map) {
+ m.Load("ns", &x.ns)
+}
+
+func (x *Setting) beforeSave() {}
+func (x *Setting) save(m state.Map) {
+ x.beforeSave()
+ m.Save("Enabled", &x.Enabled)
+ m.Save("Next", &x.Next)
+ m.Save("Period", &x.Period)
+}
+
+func (x *Setting) afterLoad() {}
+func (x *Setting) load(m state.Map) {
+ m.Load("Enabled", &x.Enabled)
+ m.Load("Next", &x.Next)
+ m.Load("Period", &x.Period)
+}
+
+func (x *Timer) beforeSave() {}
+func (x *Timer) save(m state.Map) {
+ x.beforeSave()
+ m.Save("clock", &x.clock)
+ m.Save("listener", &x.listener)
+ m.Save("setting", &x.setting)
+ m.Save("paused", &x.paused)
+}
+
+func (x *Timer) afterLoad() {}
+func (x *Timer) load(m state.Map) {
+ m.Load("clock", &x.clock)
+ m.Load("listener", &x.listener)
+ m.Load("setting", &x.setting)
+ m.Load("paused", &x.paused)
+}
+
+func init() {
+ state.Register("time.Time", (*Time)(nil), state.Fns{Save: (*Time).save, Load: (*Time).load})
+ state.Register("time.Setting", (*Setting)(nil), state.Fns{Save: (*Setting).save, Load: (*Setting).load})
+ state.Register("time.Timer", (*Timer)(nil), state.Fns{Save: (*Timer).save, Load: (*Timer).load})
+}
diff --git a/pkg/sentry/kernel/timekeeper_test.go b/pkg/sentry/kernel/timekeeper_test.go
deleted file mode 100644
index 849c5b646..000000000
--- a/pkg/sentry/kernel/timekeeper_test.go
+++ /dev/null
@@ -1,156 +0,0 @@
-// Copyright 2018 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 kernel
-
-import (
- "testing"
-
- "gvisor.dev/gvisor/pkg/sentry/context/contexttest"
- "gvisor.dev/gvisor/pkg/sentry/pgalloc"
- sentrytime "gvisor.dev/gvisor/pkg/sentry/time"
- "gvisor.dev/gvisor/pkg/sentry/usage"
- "gvisor.dev/gvisor/pkg/sentry/usermem"
- "gvisor.dev/gvisor/pkg/syserror"
-)
-
-// mockClocks is a sentrytime.Clocks that simply returns the times in the
-// struct.
-type mockClocks struct {
- monotonic int64
- realtime int64
-}
-
-// Update implements sentrytime.Clocks.Update. It does nothing.
-func (*mockClocks) Update() (monotonicParams sentrytime.Parameters, monotonicOk bool, realtimeParam sentrytime.Parameters, realtimeOk bool) {
- return
-}
-
-// Update implements sentrytime.Clocks.GetTime.
-func (c *mockClocks) GetTime(id sentrytime.ClockID) (int64, error) {
- switch id {
- case sentrytime.Monotonic:
- return c.monotonic, nil
- case sentrytime.Realtime:
- return c.realtime, nil
- default:
- return 0, syserror.EINVAL
- }
-}
-
-// stateTestClocklessTimekeeper returns a test Timekeeper which has not had
-// SetClocks called.
-func stateTestClocklessTimekeeper(tb testing.TB) *Timekeeper {
- ctx := contexttest.Context(tb)
- mfp := pgalloc.MemoryFileProviderFromContext(ctx)
- fr, err := mfp.MemoryFile().Allocate(usermem.PageSize, usage.Anonymous)
- if err != nil {
- tb.Fatalf("failed to allocate memory: %v", err)
- }
- return &Timekeeper{
- params: NewVDSOParamPage(mfp, fr),
- }
-}
-
-func stateTestTimekeeper(tb testing.TB) *Timekeeper {
- t := stateTestClocklessTimekeeper(tb)
- t.SetClocks(sentrytime.NewCalibratedClocks())
- return t
-}
-
-// TestTimekeeperMonotonicZero tests that monotonic time starts at zero.
-func TestTimekeeperMonotonicZero(t *testing.T) {
- c := &mockClocks{
- monotonic: 100000,
- }
-
- tk := stateTestClocklessTimekeeper(t)
- tk.SetClocks(c)
- defer tk.Destroy()
-
- now, err := tk.GetTime(sentrytime.Monotonic)
- if err != nil {
- t.Errorf("GetTime err got %v want nil", err)
- }
- if now != 0 {
- t.Errorf("GetTime got %d want 0", now)
- }
-
- c.monotonic += 10
-
- now, err = tk.GetTime(sentrytime.Monotonic)
- if err != nil {
- t.Errorf("GetTime err got %v want nil", err)
- }
- if now != 10 {
- t.Errorf("GetTime got %d want 10", now)
- }
-}
-
-// TestTimekeeperMonotonicJumpForward tests that monotonic time jumps forward
-// after restore.
-func TestTimekeeperMonotonicForward(t *testing.T) {
- c := &mockClocks{
- monotonic: 900000,
- realtime: 600000,
- }
-
- tk := stateTestClocklessTimekeeper(t)
- tk.restored = make(chan struct{})
- tk.saveMonotonic = 100000
- tk.saveRealtime = 400000
- tk.SetClocks(c)
- defer tk.Destroy()
-
- // The monotonic clock should jump ahead by 200000 to 300000.
- //
- // The new system monotonic time (900000) is irrelevant to what the app
- // sees.
- now, err := tk.GetTime(sentrytime.Monotonic)
- if err != nil {
- t.Errorf("GetTime err got %v want nil", err)
- }
- if now != 300000 {
- t.Errorf("GetTime got %d want 300000", now)
- }
-}
-
-// TestTimekeeperMonotonicJumpBackwards tests that monotonic time does not jump
-// backwards when realtime goes backwards.
-func TestTimekeeperMonotonicJumpBackwards(t *testing.T) {
- c := &mockClocks{
- monotonic: 900000,
- realtime: 400000,
- }
-
- tk := stateTestClocklessTimekeeper(t)
- tk.restored = make(chan struct{})
- tk.saveMonotonic = 100000
- tk.saveRealtime = 600000
- tk.SetClocks(c)
- defer tk.Destroy()
-
- // The monotonic clock should remain at 100000.
- //
- // The new system monotonic time (900000) is irrelevant to what the app
- // sees and we don't want to jump the monotonic clock backwards like
- // realtime did.
- now, err := tk.GetTime(sentrytime.Monotonic)
- if err != nil {
- t.Errorf("GetTime err got %v want nil", err)
- }
- if now != 100000 {
- t.Errorf("GetTime got %d want 100000", now)
- }
-}
diff --git a/pkg/sentry/kernel/tty.go b/pkg/sentry/kernel/tty.go
index 464d2306a..464d2306a 100644..100755
--- a/pkg/sentry/kernel/tty.go
+++ b/pkg/sentry/kernel/tty.go
diff --git a/pkg/sentry/kernel/uncaught_signal.proto b/pkg/sentry/kernel/uncaught_signal.proto
deleted file mode 100644
index 0bdb062cb..000000000
--- a/pkg/sentry/kernel/uncaught_signal.proto
+++ /dev/null
@@ -1,37 +0,0 @@
-// Copyright 2018 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.
-
-syntax = "proto3";
-
-package gvisor;
-
-import "pkg/sentry/arch/registers.proto";
-
-message UncaughtSignal {
- // Thread ID.
- int32 tid = 1;
-
- // Process ID.
- int32 pid = 2;
-
- // Registers at the time of the fault or signal.
- Registers registers = 3;
-
- // Signal number.
- int32 signal_number = 4;
-
- // The memory location which caused the fault (set if applicable, 0
- // otherwise). This will be set for SIGILL, SIGFPE, SIGSEGV, and SIGBUS.
- uint64 fault_addr = 5;
-}
diff --git a/pkg/sentry/kernel/uncaught_signal_go_proto/uncaught_signal.pb.go b/pkg/sentry/kernel/uncaught_signal_go_proto/uncaught_signal.pb.go
new file mode 100755
index 000000000..822e549ab
--- /dev/null
+++ b/pkg/sentry/kernel/uncaught_signal_go_proto/uncaught_signal.pb.go
@@ -0,0 +1,119 @@
+// Code generated by protoc-gen-go. DO NOT EDIT.
+// source: pkg/sentry/kernel/uncaught_signal.proto
+
+package gvisor
+
+import (
+ fmt "fmt"
+ proto "github.com/golang/protobuf/proto"
+ registers_go_proto "gvisor.dev/gvisor/pkg/sentry/arch/registers_go_proto"
+ math "math"
+)
+
+// Reference imports to suppress errors if they are not otherwise used.
+var _ = proto.Marshal
+var _ = fmt.Errorf
+var _ = math.Inf
+
+// This is a compile-time assertion to ensure that this generated file
+// is compatible with the proto package it is being compiled against.
+// A compilation error at this line likely means your copy of the
+// proto package needs to be updated.
+const _ = proto.ProtoPackageIsVersion3 // please upgrade the proto package
+
+type UncaughtSignal struct {
+ Tid int32 `protobuf:"varint,1,opt,name=tid,proto3" json:"tid,omitempty"`
+ Pid int32 `protobuf:"varint,2,opt,name=pid,proto3" json:"pid,omitempty"`
+ Registers *registers_go_proto.Registers `protobuf:"bytes,3,opt,name=registers,proto3" json:"registers,omitempty"`
+ SignalNumber int32 `protobuf:"varint,4,opt,name=signal_number,json=signalNumber,proto3" json:"signal_number,omitempty"`
+ FaultAddr uint64 `protobuf:"varint,5,opt,name=fault_addr,json=faultAddr,proto3" json:"fault_addr,omitempty"`
+ XXX_NoUnkeyedLiteral struct{} `json:"-"`
+ XXX_unrecognized []byte `json:"-"`
+ XXX_sizecache int32 `json:"-"`
+}
+
+func (m *UncaughtSignal) Reset() { *m = UncaughtSignal{} }
+func (m *UncaughtSignal) String() string { return proto.CompactTextString(m) }
+func (*UncaughtSignal) ProtoMessage() {}
+func (*UncaughtSignal) Descriptor() ([]byte, []int) {
+ return fileDescriptor_5ca9e03e13704688, []int{0}
+}
+
+func (m *UncaughtSignal) XXX_Unmarshal(b []byte) error {
+ return xxx_messageInfo_UncaughtSignal.Unmarshal(m, b)
+}
+func (m *UncaughtSignal) XXX_Marshal(b []byte, deterministic bool) ([]byte, error) {
+ return xxx_messageInfo_UncaughtSignal.Marshal(b, m, deterministic)
+}
+func (m *UncaughtSignal) XXX_Merge(src proto.Message) {
+ xxx_messageInfo_UncaughtSignal.Merge(m, src)
+}
+func (m *UncaughtSignal) XXX_Size() int {
+ return xxx_messageInfo_UncaughtSignal.Size(m)
+}
+func (m *UncaughtSignal) XXX_DiscardUnknown() {
+ xxx_messageInfo_UncaughtSignal.DiscardUnknown(m)
+}
+
+var xxx_messageInfo_UncaughtSignal proto.InternalMessageInfo
+
+func (m *UncaughtSignal) GetTid() int32 {
+ if m != nil {
+ return m.Tid
+ }
+ return 0
+}
+
+func (m *UncaughtSignal) GetPid() int32 {
+ if m != nil {
+ return m.Pid
+ }
+ return 0
+}
+
+func (m *UncaughtSignal) GetRegisters() *registers_go_proto.Registers {
+ if m != nil {
+ return m.Registers
+ }
+ return nil
+}
+
+func (m *UncaughtSignal) GetSignalNumber() int32 {
+ if m != nil {
+ return m.SignalNumber
+ }
+ return 0
+}
+
+func (m *UncaughtSignal) GetFaultAddr() uint64 {
+ if m != nil {
+ return m.FaultAddr
+ }
+ return 0
+}
+
+func init() {
+ proto.RegisterType((*UncaughtSignal)(nil), "gvisor.UncaughtSignal")
+}
+
+func init() {
+ proto.RegisterFile("pkg/sentry/kernel/uncaught_signal.proto", fileDescriptor_5ca9e03e13704688)
+}
+
+var fileDescriptor_5ca9e03e13704688 = []byte{
+ // 210 bytes of a gzipped FileDescriptorProto
+ 0x1f, 0x8b, 0x08, 0x00, 0x00, 0x00, 0x00, 0x00, 0x02, 0xff, 0x4c, 0x8e, 0x4d, 0x4a, 0xc6, 0x30,
+ 0x10, 0x86, 0x89, 0xfd, 0x81, 0xc6, 0x1f, 0x34, 0xab, 0x20, 0x88, 0x45, 0x17, 0x76, 0xd5, 0x80,
+ 0x9e, 0xc0, 0x0b, 0xb8, 0x88, 0xb8, 0x2e, 0x69, 0x13, 0xd3, 0xd0, 0x9a, 0x86, 0x49, 0x22, 0x78,
+ 0x24, 0x6f, 0x29, 0x4d, 0xd4, 0xef, 0xdb, 0x0d, 0xcf, 0xbc, 0xf3, 0xcc, 0x8b, 0x1f, 0xdc, 0xa2,
+ 0x99, 0x57, 0x36, 0xc0, 0x17, 0x5b, 0x14, 0x58, 0xb5, 0xb2, 0x68, 0x27, 0x11, 0xf5, 0x1c, 0x06,
+ 0x6f, 0xb4, 0x15, 0x6b, 0xef, 0x60, 0x0b, 0x1b, 0xa9, 0xf5, 0xa7, 0xf1, 0x1b, 0x5c, 0xdf, 0x1e,
+ 0x1d, 0x08, 0x98, 0x66, 0x06, 0x4a, 0x1b, 0x1f, 0x14, 0xf8, 0x1c, 0xbc, 0xfb, 0x46, 0xf8, 0xe2,
+ 0xed, 0x57, 0xf1, 0x9a, 0x0c, 0xe4, 0x12, 0x17, 0xc1, 0x48, 0x8a, 0x5a, 0xd4, 0x55, 0x7c, 0x1f,
+ 0x77, 0xe2, 0x8c, 0xa4, 0x27, 0x99, 0x38, 0x23, 0x09, 0xc3, 0xcd, 0xbf, 0x89, 0x16, 0x2d, 0xea,
+ 0x4e, 0x1f, 0xaf, 0xfa, 0xfc, 0xb3, 0xe7, 0x7f, 0x0b, 0x7e, 0xc8, 0x90, 0x7b, 0x7c, 0x9e, 0x0b,
+ 0x0e, 0x36, 0x7e, 0x8c, 0x0a, 0x68, 0x99, 0x64, 0x67, 0x19, 0xbe, 0x24, 0x46, 0x6e, 0x30, 0x7e,
+ 0x17, 0x71, 0x0d, 0x83, 0x90, 0x12, 0x68, 0xd5, 0xa2, 0xae, 0xe4, 0x4d, 0x22, 0xcf, 0x52, 0xc2,
+ 0x58, 0xa7, 0xca, 0x4f, 0x3f, 0x01, 0x00, 0x00, 0xff, 0xff, 0xfd, 0x62, 0x54, 0xdf, 0x06, 0x01,
+ 0x00, 0x00,
+}