From 810748f5c9c72f713d81d14bcc89a8eb4ca49eb6 Mon Sep 17 00:00:00 2001
From: Nicolas Lacasse <nlacasse@google.com>
Date: Tue, 16 Jun 2020 08:47:04 -0700
Subject: Port aio to VFS2.

In order to make sure all aio goroutines have stopped during S/R, a new
WaitGroup was added to TaskSet, analagous to runningGoroutines. This WaitGroup
is incremented with each aio goroutine, and waited on during kernel.Pause.

The old VFS1 aio code was changed to use this new WaitGroup, rather than
fs.Async. The only uses of fs.Async are now inode and mount Release operations,
which do not call fs.Async recursively. This fixes a lock-ordering violation
that can cause deadlocks.

Updates #1035.

PiperOrigin-RevId: 316689380
---
 pkg/abi/linux/aio.go | 60 ++++++++++++++++++++++++++++++++++++++++++++++++++--
 1 file changed, 58 insertions(+), 2 deletions(-)

(limited to 'pkg/abi/linux/aio.go')

diff --git a/pkg/abi/linux/aio.go b/pkg/abi/linux/aio.go
index 3c6e0079d..86ee3f8b5 100644
--- a/pkg/abi/linux/aio.go
+++ b/pkg/abi/linux/aio.go
@@ -14,7 +14,63 @@
 
 package linux
 
+import "encoding/binary"
+
+// AIORingSize is sizeof(struct aio_ring).
+const AIORingSize = 32
+
+// I/O commands.
 const (
-	// AIORingSize is sizeof(struct aio_ring).
-	AIORingSize = 32
+	IOCB_CMD_PREAD  = 0
+	IOCB_CMD_PWRITE = 1
+	IOCB_CMD_FSYNC  = 2
+	IOCB_CMD_FDSYNC = 3
+	// 4 was the experimental IOCB_CMD_PREADX.
+	IOCB_CMD_POLL    = 5
+	IOCB_CMD_NOOP    = 6
+	IOCB_CMD_PREADV  = 7
+	IOCB_CMD_PWRITEV = 8
 )
+
+// I/O flags.
+const (
+	IOCB_FLAG_RESFD  = 1
+	IOCB_FLAG_IOPRIO = 2
+)
+
+// IOCallback describes an I/O request.
+//
+// The priority field is currently ignored in the implementation below. Also
+// note that the IOCB_FLAG_RESFD feature is not supported.
+type IOCallback struct {
+	Data uint64
+	Key  uint32
+	_    uint32
+
+	OpCode  uint16
+	ReqPrio int16
+	FD      int32
+
+	Buf    uint64
+	Bytes  uint64
+	Offset int64
+
+	Reserved2 uint64
+	Flags     uint32
+
+	// eventfd to signal if IOCB_FLAG_RESFD is set in flags.
+	ResFD int32
+}
+
+// IOEvent describes an I/O result.
+//
+// +stateify savable
+type IOEvent struct {
+	Data    uint64
+	Obj     uint64
+	Result  int64
+	Result2 int64
+}
+
+// IOEventSize is the size of an ioEvent encoded.
+var IOEventSize = binary.Size(IOEvent{})
-- 
cgit v1.2.3