virtio — archive

[Cornelia Huck <mst@redhat.com>]

On Mon, Feb 13 2023, "Michael S. Tsirkin" <mst@redhat.com> wrote: > On Mon, Feb 13, 2023 at 02:13:43PM +0100, Cornelia Huck wrote: >> On Mon, Feb 13 2023, Max Gurtovoy <mgurtovoy@nvidia.com> wrote: >> >> > On 13/02/2023 14:42, Cornelia Huck wrote: >> >> On Mon, Feb 13 2023, Max Gurtovoy <mgurtovoy@nvidia.com> wrote: >> >> >> >>> On 13/02/2023 10:16, Michael S. Tsirkin wrote: >> >>>> On Mon, Feb 13, 2023 at 02:54:47AM +0200, Max Gurtovoy wrote: >> >>>>>> For some system calls and library functions (e.g., >> >>>>>> getpriority(2)), -1 is a valid return on success. In such cases, >> >>>>>> a successful return can be distinguished from an error return by >> >>>>>> setting errno to zero before the call, and then, if the call >> >>>>>> returns a status that indicates that an error may have occurred, >> >>>>>> checking to see if errno has a nonzero value. >> >>>>>> >> >>>>>> >> >>>>>> >> >>>>>>> Description is already good enough to describe what they are. >> >>>>>>> Can we please drop Linux wording? >> >>>>>> >> >>>>>> But why should we? It's where 22 comes from so this way people are not >> >>>>>> wondering about the value, and it's somewhat helpful for Linux >> >>>>>> developers. >> >>>>>> >> >>>>> >> >>>>> I also think we should not mention Linux. I don't think it's mentioned >> >>>>> currently in the spec and no good reason to do so now. >> >>>> >> >>>> But we do: fuse, input at least both do. >> >>>> >> >>>>> Also value of 22 is not mandatory for this EINVAL status code. It can be >> >>>>> just 1 (the first number after the OK status). >> >>>> >> >>>> 22 makes it a tiny bit easier for kvm. So why not. >> >>> >> >>> Because people comment on that in the review :) and because a >> >>> specification should be OS independent. >> >>> 22 might be EINVAL in Linux but a success in MyOS is my lucky number is 22. >> >>> >> >>> not a big issue for me, but I prefer to have a cleaner spec than maybe >> >>> simplify something in a very specific OS. >> >> >> >> Well, my take is: >> >> - we have to pick *something* >> >> - using EINVAL == 22 is very convenient for one of the most common (the >> >> most common?) implementors >> >> - noting that we're trying to follow what Linux uses makes it clear what >> >> to pick for any new return codes >> >> >> > >> > Can we agree on 22 without mentioning Linux ? >> >> See my third point above: we want people to be aware why we chose 22, so >> any new values will match up as well. > > In fact I think it is a good idea to > 1. actually link to where one can see EINVAL=22 so people know where to find > more values > 2. ask driver to handle any other error same as EINVAL for now - we don't want > new feature bits if all we do down the road is add more error codes. > > I'll add this in the next version. Wfm.