Re: [PATCH for-4.20] hvmloader: Rework hypercall infrastructure

2024-07-29 Thread Roger Pau Monné
On Thu, Jul 25, 2024 at 08:18:27AM +0200, Jan Beulich wrote: > On 24.07.2024 19:24, Andrew Cooper wrote: > > On 17/07/2024 1:37 pm, Jan Beulich wrote: > >> On 17.07.2024 13:12, Andrew Cooper wrote: > >>> static inline int > >>> hypercall_sched_op( > >>> int cmd, void *arg) > >>> { > >>> -

Re: [PATCH for-4.20] hvmloader: Rework hypercall infrastructure

2024-07-24 Thread Jan Beulich
On 24.07.2024 19:24, Andrew Cooper wrote: > On 17/07/2024 1:37 pm, Jan Beulich wrote: >> On 17.07.2024 13:12, Andrew Cooper wrote: >>> static inline int >>> hypercall_sched_op( >>> int cmd, void *arg) >>> { >>> -return _hypercall2(int, sched_op, cmd, arg); >>> +return _hypercall2(in

Re: [PATCH for-4.20] hvmloader: Rework hypercall infrastructure

2024-07-24 Thread Andrew Cooper
On 17/07/2024 1:37 pm, Jan Beulich wrote: > On 17.07.2024 13:12, Andrew Cooper wrote: >> Right now, the hypercall page is at a hardcoded physical address, and making >> hypercalls involves indirect calls to compile-time constant addresses. >> e.g. HYPERCALL_memory_op is: >> >> mov$0x80180,%ea

Re: [PATCH for-4.20] hvmloader: Rework hypercall infrastructure

2024-07-17 Thread Jan Beulich
On 17.07.2024 13:12, Andrew Cooper wrote: > Right now, the hypercall page is at a hardcoded physical address, and making > hypercalls involves indirect calls to compile-time constant addresses. > e.g. HYPERCALL_memory_op is: > > mov$0x80180,%eax > call *%eax > > Instead, import the hype