On Wed, Jul 02 2025, Magnus Kulke <magnusku...@linux.microsoft.com> wrote:
> On Wed, Jul 02, 2025 at 11:11:41AM +0200, Cornelia Huck wrote: >> On Tue, Jul 01 2025, Magnus Kulke <magnusku...@linux.microsoft.com> wrote: >> >> > Introduce headers for the Microsoft Hypervisor (MSHV) userspace ABI, >> > including IOCTLs and structures used to interface with the hypervisor. >> > >> > These definitions are based on the upstream Linux MSHV interface and >> > will be used by the MSHV accelerator backend in later patches. >> > >> > Note that for the time being the header `linux-mshv.h` is also being >> > included to allow building on machines that do not ship the header yet. >> > The header will be available in kernel 6.15 (at the time of writing >> > we're at -rc6) we will probably drop it in later revisions of the >> > patch set. >> >> The right way to handle header updates is to split the linux-header >> updates into a "dummy" update (that just adds the header) and replace >> that patch with a proper header update once the changes hit Linux >> mainline. >> > > Ah, that's right, the commit message is not accurate any more, I'll change > it in the next revision. The driver has been released as part of 6.15. > > I was following Paolo's advice to include the headers in > `linux-headers/linux` and add an entry to > `scripts/update-linux-headers.sh` > > https://lists.gnu.org/archive/html/qemu-devel/2025-05/msg04791.html > > I might not have completely understood the process though, do you > suggest to run `update-linux-headers.sh` on a current kernel (the other > headers seem to be updated from 6.16-rc*, so I assume the latest rc?) > and include the resulting changes in a seperate commit? Yes, if you need to use headers imported from Linux, you should add a separate commit stating the release (rc) you used, and update *all* of the headers. As you import a new header, I would update the script to add the header in the same commit. > >> I have not looked at the contents of the series otherwise -- but if this >> is a system header (and not something that defines the kernel<->vmm >> interface), how do you make sure that your system has that installed? >> Maybe I'm misunderstanding. >> > > The kernel header does define the ioctls and ABI for the driver, albeit not > completely, some hyperv-specific types and defines are shipped as seperate > `include/hw/hyperv` header files. I would have hoped that the Linux header would be more complete, but we have to work with what we get.
