Re: [PATCH v1 7/7] xen/vm_event: consolidate CONFIG_VM_EVENT

[Jan Beulich <jbeulich@xxxxxxxx>]

On 13.11.2025 04:16, Penny Zheng wrote:
> File hvm/vm_event.c and x86/vm_event.c are the extend to vm_event handling
> routines, and its compilation shall be guarded by CONFIG_VM_EVENT too.
> 
> Although CONFIG_VM_EVENT is right now forcibly enabled on x86 via
> MEM_ACCESS_ALWAYS_ON, we could disable it through disabling
> CONFIG_MGMT_HYPERCALLS later. So we remove MEM_ACCESS_ALWAYS_ON and
> make VM_EVENT=y on default only on x86 to retain the same.
> 
> The following functions are developed on the basis of vm event framework, or
> only invoked by vm_event.c, so they all shall be wrapped with CONFIG_VM_EVENT
> (otherwise they will become unreachable and
> violate Misra rule 2.1 when VM_EVENT=n):
> - hvm_toggle_singlestep
> - hvm_fast_singlestep
> - hvm_emulate_one_vm_event
>     - 
> hvmemul_write{,cmpxchg,rep_ins,rep_outs,rep_movs,rep_stos,read_io,write_io}_discard
> And Function vm_event_check_ring() needs stub to pass compilation when
> VM_EVENT=n.
> 
> Signed-off-by: Penny Zheng <Penny.Zheng@xxxxxxx>
> ---
> v1 -> v2:
> - split out XSM changes

Isn't that split out patch also needed as a prereq to the one here? The one
here on its own:
Acked-by: Jan Beulich <jbeulich@xxxxxxxx>

But the possibly missing dependency question needs addressing before this may
go in.

> - remove unnecessary stubs
> - move "struct p2m_domain" declaration ahead of the #ifdef
> ---
> v2 -> v3:
> - move .enable_msr_interception and .set_descriptor_access_exiting together
> - with the introduction of "vm_event_is_enabled()", all hvm_monitor_xxx()
> stubs are no longer needed
> - change to use in-place stubs in do_altp2m_op()
> - no need to add stub for monitor_traps(), __vm_event_claim_slot(),
> vm_event_put_request() and vm_event_vcpu_pause()
> - remove MEM_ACCESS_ALWAYS_ON
> - return default p2m_access_rwx for xenmem_access_to_p2m_access() when
> VM_EVENT=n
> - add wrapping for hvm_emulate_one_vm_event/
> hvmemul_write{,cmpxchg,rep_ins,rep_outs,rep_movs,rep_stos,read_io,write_io}_discard
> ---

No changes at all in this version? The v3 patch was much larger, after all.

Jan