[Qemu-devel] Re: [PATCH 4/7] add win32 qemu-thread implementation

[Paolo Bonzini]

On 02/10/2011 08:46 PM, Stefan Weil wrote:
> Am 10.02.2011 18:37, schrieb Paolo Bonzini:
> > For now, qemu_cond_timedwait and qemu_mutex_timedlock are left as
> > POSIX-only functions. They can be removed later, once the patches
> > that remove their uses are in.
> >
> > Signed-off-by: Paolo Bonzini <address@hidden>
> > Cc: Stefan Weil <address@hidden>
> > Cc: Blue Swirl <address@hidden>
> > ---
> > Makefile.objs | 4 +-
> > qemu-thread.c => qemu-thread-posix.c | 0
> > qemu-thread-posix.h | 18 +++
> > qemu-thread-win32.c | 272 ++++++++++++++++++++++++++++++++++
> > qemu-thread-win32.h | 22 +++
> > qemu-thread.h | 27 ++--
> > 6 files changed, 326 insertions(+), 17 deletions(-)
> > rename qemu-thread.c => qemu-thread-posix.c (100%)
> > create mode 100644 qemu-thread-posix.h
> > create mode 100644 qemu-thread-win32.c
> > create mode 100644 qemu-thread-win32.h
>
> What about using MinGW's pthread support?
> http://sourceforge.net/projects/mingw/files/MinGW/pthreads-w32/
>
> I used it for the threaded vnc server in QEMU, but I don't
> have more practical experience with it.

1) For threads, we'd need to do the same in order to keep the HANDLE for 
iothread IPIs (SuspendThread/ResumeThread).  We could use nonportable 
functions from pthread-w32, but if for example in the future QEMU 
switched to glib these wouldn't be available.  So QEMU would need a 
separate Win32 implementation anyway sooner or later.

2) For condvars, we can make some more assumptions and make it 
simpler/faster.

3) It has some namespace pollution issues.  These wouldn't be a problem 
for QEMU at the moment, however.

Paolo