Re: [PATCH 3/3] x86/mem_sharing: make fork_reset more configurable

[Roger Pau Monné <roger.pau@xxxxxxxxxx>]

On Tue, Mar 22, 2022 at 01:41:39PM -0400, Tamas K Lengyel wrote:
> diff --git a/xen/arch/x86/mm/mem_sharing.c b/xen/arch/x86/mm/mem_sharing.c
> index a21c781452..bfa6082f13 100644
> --- a/xen/arch/x86/mm/mem_sharing.c
> +++ b/xen/arch/x86/mm/mem_sharing.c
> @@ -1892,15 +1892,19 @@ static int fork(struct domain *cd, struct domain *d, 
> uint16_t flags)
>   * footprints the hypercall continuation should be implemented (or if this
>   * feature needs to be become "stable").
>   */
> -static int mem_sharing_fork_reset(struct domain *d)
> +int mem_sharing_fork_reset(struct domain *d, bool reset_state,
> +                           bool reset_memory)
>  {
> -    int rc;
> +    int rc = 0;
>      struct domain *pd = d->parent;
>      struct p2m_domain *p2m = p2m_get_hostp2m(d);
>      struct page_info *page, *tmp;
>  
>      domain_pause(d);

I would assert that at least one of reset_sate or reset_memory is set
here, as callers already do the checks.

>  
> +    if ( !reset_memory )
> +        goto state;

I don't like using labels and goto like this as I think it makes the
code harder to follow, and so more likely to introduce bugs. I would
rather place the memory reset parts inside of an if ( reset_memory ) {
... }, but that's my taste.

> +
>      /* need recursive lock because we will free pages */
>      spin_lock_recursive(&d->page_alloc_lock);
>      page_list_for_each_safe(page, tmp, &d->page_list)
> @@ -1933,7 +1937,9 @@ static int mem_sharing_fork_reset(struct domain *d)
>      }
>      spin_unlock_recursive(&d->page_alloc_lock);
>  
> -    rc = copy_settings(d, pd, d->arch.hvm.mem_sharing.skip_special_pages);
> + state:
> +    if ( reset_state )
> +        rc = copy_settings(d, pd, 
> d->arch.hvm.mem_sharing.skip_special_pages);
>  
>      domain_unpause(d);
>  
> @@ -2239,15 +2245,21 @@ int 
> mem_sharing_memop(XEN_GUEST_HANDLE_PARAM(xen_mem_sharing_op_t) arg)
>  
>      case XENMEM_sharing_op_fork_reset:
>      {
> +        bool reset_state = mso.u.fork.flags & XENMEM_FORK_RESET_STATE;
> +        bool reset_memory = mso.u.fork.flags & XENMEM_FORK_RESET_MEMORY;
> +
>          rc = -EINVAL;
> -        if ( mso.u.fork.pad || mso.u.fork.flags )
> +        if ( mso.u.fork.pad || (!reset_state && !reset_memory) )
> +            goto out;
> +        if ( mso.u.fork.flags &
> +             ~(XENMEM_FORK_RESET_STATE | XENMEM_FORK_RESET_MEMORY) )
>              goto out;
>  
>          rc = -ENOSYS;
>          if ( !d->parent )
>              goto out;
>  
> -        rc = mem_sharing_fork_reset(d);
> +        rc = mem_sharing_fork_reset(d, reset_state, reset_memory);
>          break;
>      }
>  
> diff --git a/xen/common/vm_event.c b/xen/common/vm_event.c
> index 84cf52636b..a7b192be0d 100644
> --- a/xen/common/vm_event.c
> +++ b/xen/common/vm_event.c
> @@ -28,6 +28,11 @@
>  #include <asm/p2m.h>
>  #include <asm/monitor.h>
>  #include <asm/vm_event.h>
> +
> +#ifdef CONFIG_MEM_SHARING
> +#include <asm/mem_sharing.h>
> +#endif
> +
>  #include <xsm/xsm.h>
>  #include <public/hvm/params.h>
>  
> @@ -394,6 +399,15 @@ static int vm_event_resume(struct domain *d, struct 
> vm_event_domain *ved)
>              if ( rsp.reason == VM_EVENT_REASON_MEM_PAGING )
>                  p2m_mem_paging_resume(d, &rsp);
>  #endif
> +#ifdef CONFIG_MEM_SHARING
> +            do {
> +                bool reset_state = rsp.flags & 
> VM_EVENT_FLAG_RESET_FORK_STATE;
> +                bool reset_mem = rsp.flags & VM_EVENT_FLAG_RESET_FORK_MEMORY;
> +
> +                if ( reset_state || reset_mem )
> +                    mem_sharing_fork_reset(d, reset_state, reset_mem);

You seem to drop the error code returned by mem_sharing_fork_reset.

> +            } while(0);
> +#endif

I think you can avoid the do {} while(0); just using the braces will
allow you to define local variables in the inner block.

>              /*
>               * Check emulation flags in the arch-specific handler only, as it
> diff --git a/xen/include/public/memory.h b/xen/include/public/memory.h
> index 208d8dcbd9..30ce23c5a7 100644
> --- a/xen/include/public/memory.h
> +++ b/xen/include/public/memory.h
> @@ -541,12 +541,14 @@ struct xen_mem_sharing_op {
>                  uint32_t gref;     /* IN: gref to debug         */
>              } u;
>          } debug;
> -        struct mem_sharing_op_fork {      /* OP_FORK */
> +        struct mem_sharing_op_fork {      /* OP_FORK/_RESET */
>              domid_t parent_domain;        /* IN: parent's domain id */
>  /* These flags only makes sense for short-lived forks */
>  #define XENMEM_FORK_WITH_IOMMU_ALLOWED (1u << 0)
>  #define XENMEM_FORK_BLOCK_INTERRUPTS   (1u << 1)
>  #define XENMEM_FORK_SKIP_SPECIAL_PAGES (1u << 2)
> +#define XENMEM_FORK_RESET_STATE        (1u << 3)
> +#define XENMEM_FORK_RESET_MEMORY       (1u << 4)

For backward compatibility purposes should the flags be added
backwards, ie:

#define XENMEM_FORK_KEEP_STATE        (1u << 3)
#define XENMEM_FORK_KEEP_MEMORY       (1u << 4)

So that existing callers of XENMEM_sharing_op_fork_reset will continue
working as expected?

Or we don't care about that as the interface is protected with
__XEN_TOOLS__?

Thanks, Roger.