For CONFIG_LINUX, implement the new zero copy flag and the optional callback
io_flush on QIOChannelSocket, but enables it only when MSG_ZEROCOPY
feature is available in the host kernel, which is checked on
qio_channel_socket_connect_sync()
qio_channel_socket_flush() was implemented by counting how many times
sendmsg(...,MSG_ZEROCOPY) was successfully called, and then reading the
socket's error queue, in order to find how many of them finished sending.
Flush will loop until those counters are the same, or until some error occurs.
Notes on using writev() with QIO_CHANNEL_WRITE_FLAG_ZERO_COPY:
1: Buffer
- As MSG_ZEROCOPY tells the kernel to use the same user buffer to avoid copying,
some caution is necessary to avoid overwriting any buffer before it's sent.
If something like this happen, a newer version of the buffer may be sent instead.
- If this is a problem, it's recommended to call qio_channel_flush() before freeing
or re-using the buffer.
2: Locked memory
- When using MSG_ZERCOCOPY, the buffer memory will be locked after queued, and
unlocked after it's sent.
- Depending on the size of each buffer, and how often it's sent, it may require
a larger amount of locked memory than usually available to non-root user.
- If the required amount of locked memory is not available, writev_zero_copy
will return an error, which can abort an operation like migration,
- Because of this, when an user code wants to add zero copy as a feature, it
requires a mechanism to disable it, so it can still be accessible to less
privileged users.
Signed-off-by: Leonardo Bras <leobras@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20220513062836.965425-4-leobras@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Add flags to io_writev and introduce io_flush as optional callback to
QIOChannelClass, allowing the implementation of zero copy writes by
subclasses.
How to use them:
- Write data using qio_channel_writev*(...,QIO_CHANNEL_WRITE_FLAG_ZERO_COPY),
- Wait write completion with qio_channel_flush().
Notes:
As some zero copy write implementations work asynchronously, it's
recommended to keep the write buffer untouched until the return of
qio_channel_flush(), to avoid the risk of sending an updated buffer
instead of the buffer state during write.
As io_flush callback is optional, if a subclass does not implement it, then:
- io_flush will return 0 without changing anything.
Also, some functions like qio_channel_writev_full_all() were adapted to
receive a flag parameter. That allows shared code between zero copy and
non-zero copy writev, and also an easier implementation on new flags.
Signed-off-by: Leonardo Bras <leobras@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20220513062836.965425-3-leobras@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
The qemu_*block() functions are meant to be be used with sockets (the
win32 implementation expects SOCKET)
Over time, those functions where used with Win32 SOCKET or
file-descriptors interchangeably. But for portability, they must only be
used with socket-like file-descriptors. FDs can use
g_unix_set_fd_nonblocking() instead.
Rename the functions with "socket" in the name to prevent bad usages.
This is effectively reverting commit f9e8cacc55 ("oslib-posix:
rename socket_set_nonblock() to qemu_set_nonblock()").
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
The socket API wrappers were initially introduced in commit
00aa0040 ("Wrap recv to avoid warnings"), but made redundant with
commit a2d96af4 ("osdep: add wrappers for socket functions") which fixes
the win32 declarations and thus removed the earlier warnings.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Adaptive polling measures the execution time of the polling check plus
handlers called when a polled event becomes ready. Handlers can take a
significant amount of time, making it look like polling was running for
a long time when in fact the event handler was running for a long time.
For example, on Linux the io_submit(2) syscall invoked when a virtio-blk
device's virtqueue becomes ready can take 10s of microseconds. This
can exceed the default polling interval (32 microseconds) and cause
adaptive polling to stop polling.
By excluding the handler's execution time from the polling check we make
the adaptive polling calculation more accurate. As a result, the event
loop now stays in polling mode where previously it would have fallen
back to file descriptor monitoring.
The following data was collected with virtio-blk num-queues=2
event_idx=off using an IOThread. Before:
168k IOPS, IOThread syscalls:
9837.115 ( 0.020 ms): IO iothread1/620155 io_submit(ctx_id: 140512552468480, nr: 16, iocbpp: 0x7fcb9f937db0) = 16
9837.158 ( 0.002 ms): IO iothread1/620155 write(fd: 103, buf: 0x556a2ef71b88, count: 8) = 8
9837.161 ( 0.001 ms): IO iothread1/620155 write(fd: 104, buf: 0x556a2ef71b88, count: 8) = 8
9837.163 ( 0.001 ms): IO iothread1/620155 ppoll(ufds: 0x7fcb90002800, nfds: 4, tsp: 0x7fcb9f1342d0, sigsetsize: 8) = 3
9837.164 ( 0.001 ms): IO iothread1/620155 read(fd: 107, buf: 0x7fcb9f939cc0, count: 512) = 8
9837.174 ( 0.001 ms): IO iothread1/620155 read(fd: 105, buf: 0x7fcb9f939cc0, count: 512) = 8
9837.176 ( 0.001 ms): IO iothread1/620155 read(fd: 106, buf: 0x7fcb9f939cc0, count: 512) = 8
9837.209 ( 0.035 ms): IO iothread1/620155 io_submit(ctx_id: 140512552468480, nr: 32, iocbpp: 0x7fca7d0cebe0) = 32
174k IOPS (+3.6%), IOThread syscalls:
9809.566 ( 0.036 ms): IO iothread1/623061 io_submit(ctx_id: 140539805028352, nr: 32, iocbpp: 0x7fd0cdd62be0) = 32
9809.625 ( 0.001 ms): IO iothread1/623061 write(fd: 103, buf: 0x5647cfba5f58, count: 8) = 8
9809.627 ( 0.002 ms): IO iothread1/623061 write(fd: 104, buf: 0x5647cfba5f58, count: 8) = 8
9809.663 ( 0.036 ms): IO iothread1/623061 io_submit(ctx_id: 140539805028352, nr: 32, iocbpp: 0x7fd0d0388b50) = 32
Notice that ppoll(2) and eventfd read(2) syscalls are eliminated because
the IOThread stays in polling mode instead of falling back to file
descriptor monitoring.
As usual, polling is not implemented on Windows so this patch ignores
the new io_poll_read() callback in aio-win32.c.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
Message-id: 20211207132336.36627-2-stefanha@redhat.com
[Fixed up aio_set_event_notifier() calls in
tests/unit/test-fdmon-epoll.c added after this series was queued.
--Stefan]
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
MSG_CMSG_CLOEXEC cleans up received fd's; it's really only for Unix
sockets, but currently we enable it for everything; some socket types
(IP_MPTCP) don't like this.
Only enable it when we're giving the recvmsg room to receive fd's
anyway.
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-Id: <20210421112834.107651-2-dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
There is no "version 2" of the "Lesser" General Public License.
It is either "GPL version 2.0" or "Lesser GPL version 2.1".
This patch replaces all occurrences of "Lesser GPL version 2" with
"Lesser GPL version 2.1" in comment section.
Signed-off-by: Chetan Pant <chetan4windows@gmail.com>
Acked-by: Daniel P. Berrangé <berrange@redhat.com>
Message-Id: <20201014134033.14095-1-chetan4windows@gmail.com>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
The Error ** argument must be NULL, &error_abort, &error_fatal, or a
pointer to a variable containing NULL. Passing an argument of the
latter kind twice without clearing it in between is wrong: if the
first call sets an error, it no longer points to NULL for the second
call.
qio_channel_socket_close() passes @errp first to
socket_listen_cleanup(), and then, if closesocket() fails, to
error_setg_errno(). If socket_listen_cleanup() failed, this will trip
the assertion in error_setv().
Fix by ignoring a second error.
Fixes: 73564c407c
Cc: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-Id: <20200422130719.28225-11-armbru@redhat.com>
Current parameter was always one. We continue with that value for now
in all callers.
Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
---
Moved trace to socket_listen
No header includes qemu-common.h after this commit, as prescribed by
qemu-common.h's file comment.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20190523143508.25387-5-armbru@redhat.com>
[Rebased with conflicts resolved automatically, except for
include/hw/arm/xlnx-zynqmp.h hw/arm/nrf51_soc.c hw/arm/msf2-soc.c
block/qcow2-refcount.c block/qcow2-cluster.c block/qcow2-cache.c
target/arm/cpu.h target/lm32/cpu.h target/m68k/cpu.h target/mips/cpu.h
target/moxie/cpu.h target/nios2/cpu.h target/openrisc/cpu.h
target/riscv/cpu.h target/tilegx/cpu.h target/tricore/cpu.h
target/unicore32/cpu.h target/xtensa/cpu.h; bsd-user/main.c and
net/tap-bsd.c fixed up]
The qio_channel_socket_close method for was mistakenly unlinking the
UNIX server socket, even if the channel was a client connection. This
was not noticed with chardevs, since they never call close, but with the
VNC server, this caused the VNC server socket to be deleted after the
first client quit.
The qio_channel_socket_close method also needlessly reimplemented the
logic that already exists in socket_listen_cleanup(). Just call that
method directly, for listen sockets only.
This fixes a regression introduced in QEMU 3.0.0 with
commit d66f78e1ea
Author: Pavel Balaev <mail@void.so>
Date: Mon May 21 19:17:35 2018 +0300
Delete AF_UNIX socket after close
Fixes launchpad #1795100
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Since version 2.12.0 AF_UNIX socket created for QMP exchange is not
deleted on instance shutdown.
This is due to the fact that function qio_channel_socket_finalize() is
called after qio_channel_socket_close().
Signed-off-by: Pavel Balaev <mail@void.so>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
We have worked on qio_task_run_in_thread() already. Further, let
all the qio channel APIs use that context.
Signed-off-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
qio_task_run_in_thread() allows main thread to run blocking operations
in the background. However it has an assumption on that it's always
working with the default context. This patch tries to allow the threaded
QIO task framework to run with non-default gcontext.
Currently no functional change so far, so the QIOTasks are still always
running on main context.
Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
Signed-off-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
In my "build everything" tree, a change to the types in
qapi-schema.json triggers a recompile of about 4800 out of 5100
objects.
The previous commit split up qmp-commands.h, qmp-event.h, qmp-visit.h,
qapi-types.h. Each of these headers still includes all its shards.
Reduce compile time by including just the shards we actually need.
To illustrate the benefits: adding a type to qapi/migration.json now
recompiles some 2300 instead of 4800 objects. The next commit will
improve it further.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20180211093607.27351-24-armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
[eblake: rebase to master]
Signed-off-by: Eric Blake <eblake@redhat.com>
The non-blocking connect mechanism is obsolete, and it doesn't
work well in inet connection, because it will call getaddrinfo
first and getaddrinfo will blocks on DNS lookups. Since commit
e65c67e4 & d984464e, the non-blocking connect of migration goes
through QIOChannel in a different manner(using a thread), and
nobody use this old non-blocking connect anymore.
Any newly written code which needs a non-blocking connect should
use the QIOChannel code, so we can drop NonBlockingConnectHandler
as a concept entirely.
Suggested-by: Daniel P. Berrange <berrange@redhat.com>
Signed-off-by: Cao jin <caoj.fnst@cn.fujitsu.com>
Signed-off-by: Mao Zhongyi <maozy.fnst@cn.fujitsu.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
When accept failed, we should setup errp with the reason. More
importantly, the caller may assume errp be non-NULL when error happens,
and not setting the errp may crash QEMU.
At the same time, move the trace_qio_channel_socket_accept_fail() after
the if check on EINTR. Two reasons:
1. when EINTR happened, it's not really a fault (we should just try
again), so we should not log with an "accept failure".
2. trace_*() functions may overwrite errno, then the old errno will be
missing. We need to either check errno before trace_*() calls, or
reserve the errno.
Signed-off-by: Peter Xu <peterx@redhat.com>
Message-Id: <1501666880-10159-3-git-send-email-peterx@redhat.com>
Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
The channel socket was initialized manually, but forgot to set
QIO_CHANNEL_FEATURE_SHUTDOWN. Thus, the colo_process_incoming_thread
would hang at recvmsg. This patch just call qio_channel_socket_new to
get channel, Which set QIO_CHANNEL_FEATURE_SHUTDOWN already.
Signed-off-by: Wang Guang<wang.guang55@zte.com.cn>
Signed-off-by: zhanghailiang <zhang.zhanghailiang@huawei.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
This is in preparation for making qio_channel_yield work on
AioContexts other than the main one.
Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Message-id: 20170213135235.12274-6-pbonzini@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Now that task objects have a directly associated error,
there's no need for an an Error **errp parameter to
the QIOTask thread worker function. It already has a
QIOTask object, so can directly set the error on it.
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Incrementing the reference in qio_task_get_source is
not necessary, since we're not running concurrently
with any other code touching the QIOTask. This
minimizes chances of further memory leaks.
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
The SO_ACCEPTCONN ioctl is not portable across OS, with
some BSD versions and OS-X not supporting it. There is
no viable alternative to this, so instead just set the
feature explicitly when creating a listener socket.
The current users of qio_channel_socket_new_fd() won't
ever be given a listening socket, so there's no problem
with no auto-detecting it in this scenario
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Testing QIOChannel feature support can be done with a helper called
qio_channel_has_feature(). Setting feature support, however, was
done manually with a logical OR. This patch introduces a new helper
called qio_channel_set_feature() and makes use of it where applicable.
Signed-off-by: Felipe Franciosi <felipe@nutanix.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Parts of the code have been testing QIOChannel features directly with a
logical AND. This patch makes it all consistent by using the
qio_channel_has_feature() function to test if a feature is present.
Signed-off-by: Felipe Franciosi <felipe@nutanix.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
When QIOChannels were introduced in 666a3af9, the feature bits were
already defined shifted. However, when using them, the code was shifting
them again. The incorrect use was consistent until 74b6ce43, where
QIO_CHANNEL_FEATURE_LISTEN was defined shifted but tested unshifted.
This patch changes the definition to be unshifted and fixes the
incorrect usage introduced on 74b6ce43.
Signed-off-by: Felipe Franciosi <felipe@nutanix.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Rather than rolling our own clone via an expensive conversion
in and back out of QObject, use the new clone visitor.
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <1465490926-28625-15-git-send-email-eblake@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
qemu leaves unix socket files behind when removing a listening chardev
or leaving. qemu could clean that up, even if doing so isn't race-free.
Fixes:
https://bugzilla.redhat.com/show_bug.cgi?id=1347077
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <1466105332-10285-4-git-send-email-marcandre.lureau@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Commit 57cb38b included qapi/error.h into qemu/osdep.h to get the
Error typedef. Since then, we've moved to include qemu/osdep.h
everywhere. Its file comment explains: "To avoid getting into
possible circular include dependencies, this file should not include
any other QEMU headers, with the exceptions of config-host.h,
compiler.h, os-posix.h and os-win32.h, all of which are doing a
similar job to this file and are under similar constraints."
qapi/error.h doesn't do a similar job, and it doesn't adhere to
similar constraints: it includes qapi-types.h. That's in excess of
100KiB of crap most .c files don't actually need.
Add the typedef to qemu/typedefs.h, and include that instead of
qapi/error.h. Include qapi/error.h in .c files that need it and don't
get it now. Include qapi-types.h in qom/object.h for uint16List.
Update scripts/clean-includes accordingly. Update it further to match
reality: replace config.h by config-target.h, add sysemu/os-posix.h,
sysemu/os-win32.h. Update the list of includes in the qemu/osdep.h
comment quoted above similarly.
This reduces the number of objects depending on qapi/error.h from "all
of them" to less than a third. Unfortunately, the number depending on
qapi-types.h shrinks only a little. More work is needed for that one.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
[Fix compilation without the spice devel packages. - Paolo]
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Now that QEMU wraps the Win32 sockets methods to automatically
set errno upon failure, there is no reason for callers to use
the socket_error() method. They can rely on accessing errno
even on Win32. Remove all use of socket_error() from general
code, leaving it as a static method in oslib-win32.c only.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
On Win32 we cannot directly poll on socket handles. Instead we
create a Win32 event object and associate the socket handle with
the event. When the event signals readyness we then have to
use select to determine which events are ready. Creating Win32
events is moderately heavyweight, so we don't want todo it
every time we create a GSource, so this associates a single
event with a QIOChannel.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Since we now canonicalize WSAEWOULDBLOCK into EAGAIN there is
no longer any need to explicitly check EWOULDBLOCK for Win32.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
The QIOChannelSocket code mistakenly uses the bare accept()
function which does not set SOCK_CLOEXEC.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Sockets are not in the same namespace as file descriptors on Windows.
As an initial step, introduce separate APIs for file descriptor and
socket watches.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
s/write/read/ in the error message reported after
readmsg() fails
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Two wrongs make a right, but they should be fixed anyway.
Cc: Daniel P. Berrange <berrange@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <1455015557-15106-1-git-send-email-pbonzini@redhat.com>
Clean up includes so that osdep.h is included first and headers
which it implies are not included manually.
This commit was created with scripts/clean-includes.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1454089805-5470-14-git-send-email-peter.maydell@linaro.org
Some versions of GCC on OS-X complain about CMSG_SPACE
not being constant size, which prevents use of { 0 }
io/channel-socket.c: In function 'qio_channel_socket_writev':
io/channel-socket.c:497:18: error: variable-sized object may not be initialized
char control[CMSG_SPACE(sizeof(int) * SOCKET_MAX_FDS)] = { 0 };
The compiler is at fault here, but it is nicer to avoid
tickling this compiler bug by using memset instead.
Reviewed-By: John Arbuckle <programmingkidx@gmail.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
When reporting the number of FDs has been exceeded, pass
EINVAL to error_setg_errno, rather than -EINVAL.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
When sending file descriptors over a socket, we have to
allocate a data buffer to hold the FDs in the scmsghdr.
Unfortunately we allocated the buffer on the stack inside
an if () {} block, but called sendmsg() outside the block.
So the stack bytes holding the FDs were liable to be
overwritten with other data. By luck this was not a problem
when sending 1 FD, but if sending 2 or more then it would
fail.
The fix is to simply move the variables outside the nested
'if' block. To keep valgrind quiet we also zero-initialize
the 'control' buffer.
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
The QIO_CHANNEL_FEATURE_FD_PASS feature flag is set in the
qio_channel_socket_set_fd() method, however, this only deals
with client side connections.
To ensure server side connections also have the feature flag
set, we must set it in qio_channel_socket_accept() too. This
also highlighted a typo fix where the code updated the
sockaddr struct in the wrong object instance.
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Implement a QIOChannel subclass that supports sockets I/O.
The implementation is able to manage a single socket file
descriptor, whether a TCP/UNIX listener, TCP/UNIX connection,
or a UDP datagram. It provides APIs which can listen and
connect either asynchronously or synchronously. Since there
is no asynchronous DNS lookup API available, it uses the
QIOTask helper for spawning a background thread to ensure
non-blocking operation.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Leonardo Bras
Marc-André Lureau
Stefan Hajnoczi
Dr. David Alan Gilbert
Chetan Pant
Markus Armbruster
Juan Quintela
Daniel P. Berrangé
Pavel Balaev
Peter Xu
Cao jin
Fam Zheng
Wang guang
Paolo Bonzini
Felipe Franciosi
Eric Blake
Peter Maydell