On 09/20/2013 06:49 PM, Michael Tokarev wrote: > Can you please send a v4 with the above changes? Hopefully you agree > with them. Overall, I like this new version too.
Ok, I'll do that ASAP. I'll try to find a better wording for the error messages. At first I tried to modify the code as little as possible to be eligible for qemu-trivial, but clearly I am in favor of adding %m or perror when a syscall fails. BTW, there are some other issues in the ivshmem code, like the maximum size is not enforced (256M in 32b and 4G in 64b) but I am currently investigating a bug in the linux kernel when the size is > 256M (this leads to really weird behavior when coupled with VT-d), so I'll wait to be sure it is really in the kernel and not QEMU related before submitting this part. Thanks, -- Damien