Anthony Liguori writes: > This patch implements posix-aio using pthreads. It immediately > eliminates the need for fd pooling.
Well in principle I think I approve. Having read the discussion I think emulation of preadv/pwritev with pread/pwrite is probably fine. However I did want to make one comment on this: I think it would be valuable to try to settle on a good and stable design for the aio functionality. I've spent the last several weeks tracking down a mysterious failure of qemu-dm in our testing systems which turned out to be due to a bug in the RHEL 4.3 glibc. Obviously my own workstation and test box are much newer and better than that. Last time I merged from qemu upstream I also spent quite a time discovering a different glibc bug. These kind of threading/aio/etc. features are often immature in older libcs, some of which advertise them but have subtle bugs. If we can at least pick an approach and an implementation, rather than constantly changing, we'll have a chance to find and work around those bugs. But having said that I think avoiding too much reliance on glibc-implemented primitives (which often seem to be buggy in some suble way) and doing it ourselves seems like a better approach. Ian. -- To unsubscribe from this list: send the line "unsubscribe kvm" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html
