On Tue, Jun 05, 2012 at 07:49:07PM +0800, Liu Yuan wrote: > On 06/05/2012 07:39 PM, Christoph Hellwig wrote: > > > After running into the first coroutine problems I've not moved further > > with testing any coroutine based qemu versions. > > > Maybe I am not at any qualification to speak design choice of QEMU, but > at least to me, coroutine is really a disaster (Even seems that there > coroutine is nested), I can't even debug a simple bug if involved with > coroutine. This bug is very reproducible and looks to me just a simple > racy problem, but after a long time debugging with it, it just scares me > away without any fruit. In the long run, I think coroutine based > sheepdog block driver will risk maintainability of the code. I completely agree with your opinion there - if you look at old list archives you'll see that I argued against adding them to qemu but lost the fight. There has been some recent discussion where people advocated for moving away from it again, though. |