This series of patches is a followup submission for review of another chunk of my large series supporting TLS across chardevs, VNC and migration, previously shown here:
RFC: https://lists.gnu.org/archive/html/qemu-devel/2015-09/msg00829.html v1: https://lists.gnu.org/archive/html/qemu-devel/2015-09/msg04853.html In this series, I have provided only the general I/O channels framework, which will ultimately be used by VNC, chardev and migration code, whicih currently work as follows: - VNC: uses raw sockets APIs directly, layering in TLS, SASL and websockets where needed. This has resulted in what should be fairly generic code for TLS and websockets being entangled with the rest of the VNC server code. - Chardevs: uses GLib's GIOChannel framework. This only provides implementations for sockets and files. Its API lacks support for using struct iovec for read/writes, file descriptor passing, misc sockets ioctls/fcntls. While you can work around these problems by accessing the underling file descriptor directly, this breaks the encapsulation, requiring callers to know about specific implementations. It also does not have integration with QEMU Error reporting framework. So while the GIOChannel is extensible, extending it would result in throwing away pretty much the entire of the existing implementations - Migration: uses QEMUFile framework. The provides an abstract interface for I/O channels used during migration. It has impls for sockets, files, memory buffers & commands. While it has struct iovec support for writes, it does not have the same for reads. It also doesn't support file descriptor passing or control of the sockets ioctls/fcntls. It also does not have any explicit event loop integration, expecting the callers to directly access the underling file desctriptor and setup events on that. This limits its utility in cases where you need channels which are not backed by file descriptors. It has no integration with QEMU Error object for error reporting, has fixed semantics wrt writes ie must always do a full write, no partial writes, and most strangely forces either input or output mode, but refuses to allow both, so no bi-directional channels! Out of all of these, the migration code is probably closest to what is needed, but is still a good way off from being a generic framework that be can reused outside of the migration code. There is also the GLib GIO library which provides a generic framework, but we can't depend on that due to the minimum GLib requirement. It also has various missing pieces such as file descriptor passing, and no support for struct iovec either. Hence, this series was born, which tries to take the best of the designs for the GIOChannel, QIO and QEMUFile code to provide QIOChannel. Right from the start this new framework is explicitly isolated from any other QEMU subsystem, so its implementation will not get mixed up with specific use cases. The QIOChannel abstract base class defines the overall API contract - qio_channel_{write,writev,write_full} for writing data. The underling impl always uses struct iovec, but non-iov APIs are provided as simple wrappers for convenience - qio_channel_{read,readv,read_full} for reading data. The underling impl always uses struct iovec, but non-iov APIs are provided as simple wrappers for convenience - qio_channel_has_feature to determine which optional features the channel supports - eg file descriptor passing, nagle, etc - qio_channel_set_{blocking,delay,cork} for various fcntl and ioctl controls - qio_channel_{close,shutdown} for closing the I/O stream or individual channels - qio_channel_seek for random access channels - qio_channel_{add,create}_watch for integration into the main loop for event notifications - qio_channel_wait for blocking of execution pending an event notification - qio_channel_yield for switching coroutine until an event notification All the APIs support Error ** object arguments where needed. The object is built using QOM, in order to get reference counting and sub-classing with type checking. They are *not* user creatable/visible objects though - these are internal infrastructure, so we will be free to adapt APIs/objects at will over time. In this series there are a variety of implementations. Some of them provide an actual base layer data transport, while others provide a data translation layer: In this series there are a variety of implementations. Some of them provide an actual base layer data transport, while others provide a data translation layer: - QIOChannelSocket - for socket based I/O, IPv4, IPV6 & UNIX, both stream and datagram. - QIOChannelFile - any non-socket file, plain file, char dev, fifo, pipe, etc - QIOChannelTLS - data translation layer to apply TLS protocol - QIOChannelWebsock - data translation layer to apply the websockets server protocol - QIOChannelCommand - spawn & communicate with a command via pipes - QIOChannelBuffer - a simple in-memory buffer These 6 implementations are sufficient to support the current needs of the VNC server, chardev network backends, and all the various migration protocols, except RDMA (which is part of the larger RFC series I've posted previously). The sockets implementation in particular solves a long standing limitation of the qemu-sockets.c API by providing a truely asynchronous API for establishing listener sockets / connections. The current code is only async for the socket establishment, but still blocks the caller for DNS resolution. I have not attempted to make the socket implementation support multiple listener sockets in a single object. I looked at this but it would significantly complicate the QIOChannelSocket implementation. I intend to suggest an alternative approach for that problem at a later date, whereby we define a QIONetworkService object, which provides the infrastructure for managing multiple QIOChannelSocket listeners, and clients, which was resuable for any QEMU network service. There are unit tests for the socket, file, tls, and command channel implementations. I didn't create unit test for the websock impl, as it needs a websock client to usefully test it which doesn't exist in QEMU yet. The memory buffer impl also lacks unit tests for now, simply due to not getting around to it yet. This series merely introduces all the new framework. To keep it an acceptable length for review, it doesn't include any of the patches which actually use the new APIs. If reviewers want to see real world usage of these APIs, they can refer to my previous RFC series. https://lists.gnu.org/archive/html/qemu-devel/2015-09/msg00829.html In particular - VNC conversion to QIOChannelSocket https://lists.gnu.org/archive/html/qemu-devel/2015-09/msg00851.html - VNC conversion to QIOChannelTLS https://lists.gnu.org/archive/html/qemu-devel/2015-09/msg00845.html - VNC conversion to QIOChannelWebsock https://lists.gnu.org/archive/html/qemu-devel/2015-09/msg00842.html - Chardev conversion to QIOChannelSocket https://lists.gnu.org/archive/html/qemu-devel/2015-09/msg00849.html - Chardev addition of TLS support https://lists.gnu.org/archive/html/qemu-devel/2015-09/msg00848.html - All the migration ones, but most interesting addition of TLS https://lists.gnu.org/archive/html/qemu-devel/2015-09/msg00871.html Finally I've listed myself as maintainer for the new io/ and include/io/ directory prefixes. Changed in v2: - Put the Buffer APIs extracted from VNC code into util/ instead of io/ directories (Paolo) - Added Kevin & Stefan to MAINTAINERS for coroutine code (Paolo) - Remove feature checks for TCP_NODELAY and TCP_CORK and treat them as hints which cannot fail (Paolo) - Push feature checking for FD passing up into QIOChannel base class (Paolo) - Use more constants in websockets code instead of magic values (David Gilbert) Daniel P. Berrange (16): sockets: add helpers for creating SocketAddress from a socket sockets: move qapi_copy_SocketAddress into qemu-sockets.c sockets: allow port to be NULL when listening on IP address ui: convert VNC startup code to use SocketAddress osdep: add qemu_fork() wrapper for safely handling signals coroutine: move into libqemuutil.a library util: pull Buffer code out of VNC module io: add abstract QIOChannel classes io: add helper module for creating watches on FDs io: add QIOTask class for async operations io: add QIOChannelSocket class io: add QIOChannelFile class io: add QIOChannelTLS class io: add QIOChannelWebsock class io: add QIOChannelCommand class io: add QIOChannelBuffer class MAINTAINERS | 20 + Makefile | 2 + Makefile.objs | 9 +- Makefile.target | 2 + block.c | 2 +- block/qcow2.h | 2 +- block/vdi.c | 2 +- block/write-threshold.c | 2 +- blockjob.c | 2 +- configure | 11 + hw/9pfs/codir.c | 2 +- hw/9pfs/cofile.c | 2 +- hw/9pfs/cofs.c | 2 +- hw/9pfs/coxattr.c | 2 +- hw/9pfs/virtio-9p-coth.c | 2 +- hw/9pfs/virtio-9p-coth.h | 2 +- hw/9pfs/virtio-9p.h | 2 +- include/block/block.h | 2 +- include/block/block_int.h | 2 +- include/io/channel-buffer.h | 59 ++ include/io/channel-command.h | 91 ++ include/io/channel-file.h | 93 ++ include/io/channel-socket.h | 244 ++++++ include/io/channel-tls.h | 142 +++ include/io/channel-watch.h | 72 ++ include/io/channel-websock.h | 108 +++ include/io/channel.h | 506 +++++++++++ include/io/task.h | 256 ++++++ include/qemu/buffer.h | 118 +++ include/{block => qemu}/coroutine.h | 0 include/{block => qemu}/coroutine_int.h | 2 +- include/qemu/osdep.h | 16 + include/qemu/sockets.h | 34 + io/Makefile.objs | 9 + io/channel-buffer.c | 255 ++++++ io/channel-command.c | 369 ++++++++ io/channel-file.c | 237 +++++ io/channel-socket.c | 737 ++++++++++++++++ io/channel-tls.c | 405 +++++++++ io/channel-watch.c | 200 +++++ io/channel-websock.c | 975 +++++++++++++++++++++ io/channel.c | 283 ++++++ io/task.c | 159 ++++ migration/qemu-file-buf.c | 2 +- migration/qemu-file-stdio.c | 2 +- migration/qemu-file-unix.c | 2 +- migration/qemu-file.c | 2 +- migration/rdma.c | 2 +- nbd.c | 2 +- qemu-char.c | 25 - scripts/create_config | 9 + tests/.gitignore | 7 + tests/Makefile | 16 + tests/io-channel-helpers.c | 247 ++++++ tests/io-channel-helpers.h | 33 + tests/test-coroutine.c | 4 +- tests/test-io-channel-command.c | 122 +++ tests/test-io-channel-file.c | 91 ++ tests/test-io-channel-socket.c | 367 ++++++++ tests/test-io-channel-tls.c | 335 +++++++ tests/test-io-task.c | 276 ++++++ tests/test-vmstate.c | 2 +- thread-pool.c | 2 +- trace-events | 56 ++ ui/vnc.c | 204 ++--- ui/vnc.h | 16 +- util/Makefile.objs | 4 + util/buffer.c | 65 ++ coroutine-gthread.c => util/coroutine-gthread.c | 2 +- .../coroutine-sigaltstack.c | 2 +- coroutine-ucontext.c => util/coroutine-ucontext.c | 2 +- coroutine-win32.c => util/coroutine-win32.c | 2 +- util/oslib-posix.c | 71 ++ util/oslib-win32.c | 9 + qemu-coroutine-io.c => util/qemu-coroutine-io.c | 2 +- .../qemu-coroutine-lock.c | 4 +- .../qemu-coroutine-sleep.c | 2 +- qemu-coroutine.c => util/qemu-coroutine.c | 4 +- util/qemu-sockets.c | 158 +++- 79 files changed, 7396 insertions(+), 197 deletions(-) create mode 100644 include/io/channel-buffer.h create mode 100644 include/io/channel-command.h create mode 100644 include/io/channel-file.h create mode 100644 include/io/channel-socket.h create mode 100644 include/io/channel-tls.h create mode 100644 include/io/channel-watch.h create mode 100644 include/io/channel-websock.h create mode 100644 include/io/channel.h create mode 100644 include/io/task.h create mode 100644 include/qemu/buffer.h rename include/{block => qemu}/coroutine.h (100%) rename include/{block => qemu}/coroutine_int.h (98%) create mode 100644 io/Makefile.objs create mode 100644 io/channel-buffer.c create mode 100644 io/channel-command.c create mode 100644 io/channel-file.c create mode 100644 io/channel-socket.c create mode 100644 io/channel-tls.c create mode 100644 io/channel-watch.c create mode 100644 io/channel-websock.c create mode 100644 io/channel.c create mode 100644 io/task.c create mode 100644 tests/io-channel-helpers.c create mode 100644 tests/io-channel-helpers.h create mode 100644 tests/test-io-channel-command.c create mode 100644 tests/test-io-channel-file.c create mode 100644 tests/test-io-channel-socket.c create mode 100644 tests/test-io-channel-tls.c create mode 100644 tests/test-io-task.c create mode 100644 util/buffer.c rename coroutine-gthread.c => util/coroutine-gthread.c (99%) rename coroutine-sigaltstack.c => util/coroutine-sigaltstack.c (99%) rename coroutine-ucontext.c => util/coroutine-ucontext.c (99%) rename coroutine-win32.c => util/coroutine-win32.c (98%) rename qemu-coroutine-io.c => util/qemu-coroutine-io.c (99%) rename qemu-coroutine-lock.c => util/qemu-coroutine-lock.c (98%) rename qemu-coroutine-sleep.c => util/qemu-coroutine-sleep.c (96%) rename qemu-coroutine.c => util/qemu-coroutine.c (98%) -- 2.4.3
