On Sun, Jun 12, 2011 at 9:46 PM, Stefan Hajnoczi <stefa...@linux.vnet.ibm.com> wrote: > QEMU is event-driven and suffers when blocking operations are performed > because > VM execution may be stopped until the operation completes. Therefore many > operations that could block are performed asynchronously and a callback is > invoked when the operation has completed. This allows QEMU to continue > executing while the operation is pending. > > The downside to callbacks is that they split up code into many smaller > functions, each of which is a single step in a state machine that quickly > becomes complex and hard to understand. Callback functions also result in > lots > of noise as variables are packed and unpacked into temporary structs that pass > state to the callback function. > > This patch series introduces coroutines as a solution for writing asynchronous > code while still having a nice sequential control flow. The semantics are > explained in the second patch. The fourth patch adds automated tests. > > A nice feature of coroutines is that it is relatively easy to take synchronous > code and lift it into a coroutine to make it asynchronous. Work has been done > to move qcow2 request processing into coroutines and thereby make it > asynchronous (today qcow2 will perform synchronous metadata accesses). This > qcow2 work is still ongoing and not quite ready for mainline yet. > > Coroutines are also being used for virtfs (virtio-9p) so I have submitted this > patch now because virtfs patches that depend on coroutines are being > published. > > v5: > * GThread-based implementation for platforms without makecontext(3) (Aneesh > Kumar K.V <aneesh.ku...@linux.vnet.ibm.com>) > * Switch to gtester test framework > > v4: > * Windows Fibers support (Paolo Bonzini <pbonz...@redhat.com>) > * Return-after-setjmp() fix (Aneesh Kumar K.V > <aneesh.ku...@linux.vnet.ibm.com>) > * Re-entrancy for multi-threaded coroutines support > * qemu-coroutine.h cleanup and documentation > > v3: > * Updated LGPL v2 license header to use web link > * Removed atexit(3) pool freeing > * Removed thread-local current/leader > * Documented thread-safety limitation > * Disabled trace events > > v2: > * Added ./check-coroutine --lifecycle-benchmark for performance measurement > * Split pooling into a separate patch with performance justification > * Set maximum pool size to prevent holding onto too many free coroutines > * Added atexit(3) handler to free pool > * Coding style cleanups > > Aneesh Kumar K.V (1): > coroutine: implement coroutines using gthread > > Anthony Liguori (1): > Add hard build dependency on glib > > Kevin Wolf (1): > coroutine: introduce coroutines > > Stefan Hajnoczi (2): > coroutine: add check-coroutine automated tests > coroutine: add check-coroutine --benchmark-lifecycle > > .gitignore | 1 + > Makefile | 5 +- > Makefile.objs | 13 +++ > Makefile.target | 1 + > configure | 29 +++++++ > coroutine-gthread.c | 131 ++++++++++++++++++++++++++++ > coroutine-ucontext.c | 229 > ++++++++++++++++++++++++++++++++++++++++++++++++++ > coroutine-win32.c | 92 ++++++++++++++++++++ > qemu-coroutine-int.h | 48 +++++++++++ > qemu-coroutine.c | 75 ++++++++++++++++ > qemu-coroutine.h | 95 +++++++++++++++++++++ > test-coroutine.c | 192 ++++++++++++++++++++++++++++++++++++++++++ > trace-events | 5 + > 13 files changed, 915 insertions(+), 1 deletions(-) > create mode 100644 coroutine-gthread.c > create mode 100644 coroutine-ucontext.c > create mode 100644 coroutine-win32.c > create mode 100644 qemu-coroutine-int.h > create mode 100644 qemu-coroutine.c > create mode 100644 qemu-coroutine.h > create mode 100644 test-coroutine.c
Any objections to merging this? Stefan
