[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] [PATCH] mem_access: Use monitor_traps instead of mem_access_send_req



On Jul 29, 2016 01:27, "Razvan Cojocaru" <rcojocaru@xxxxxxxxxxxxxxx> wrote:
>
> On 07/28/2016 10:35 PM, Tamas K Lengyel wrote:
> > The two functions monitor_traps and mem_access_send_req duplicate
> > some of the same functionality. The mem_access_send_req however leaves a
> > lot of the standard vm_event fields to be filled by other functions.
> >
> > Since mem_access events go on the monitor ring in this patch we consolidate
> > all paths to use monitor_traps to place events on the ring and to fill in
> > the common parts of the requests.
> >
> > Signed-off-by: Tamas K Lengyel <tamas.lengyel@xxxxxxxxxxxx>
> > ---
> > Cc: Stefano Stabellini <sstabellini@xxxxxxxxxx>
> > Cc: Julien Grall <julien.grall@xxxxxxx>
> > Cc: Jan Beulich <jbeulich@xxxxxxxx>
> > Cc: Andrew Cooper <andrew.cooper3@xxxxxxxxxx>
> > Cc: Razvan Cojocaru <rcojocaru@xxxxxxxxxxxxxxx>
> > Cc: George Dunlap <george.dunlap@xxxxxxxxxxxxx>
> > ---
> >  xen/arch/arm/p2m.c                | 69 +++++++++++++++++++--------------------
> >  xen/arch/x86/hvm/hvm.c            | 16 ++++++---
> >  xen/arch/x86/hvm/monitor.c        |  6 ++++
> >  xen/arch/x86/mm/p2m.c             | 24 ++------------
> >  xen/common/mem_access.c           | 11 -------
> >  xen/include/asm-x86/hvm/monitor.h |  2 ++
> >  xen/include/asm-x86/p2m.h         | 13 +++++---
> >  xen/include/xen/mem_access.h      |  7 ----
> >  8 files changed, 63 insertions(+), 85 deletions(-)
> >
> > diff --git a/xen/arch/arm/p2m.c b/xen/arch/arm/p2m.c
> > index d82349c..df898a3 100644
> > --- a/xen/arch/arm/p2m.c
> > +++ b/xen/arch/arm/p2m.c
> > @@ -5,7 +5,7 @@
> >  #include <xen/domain_page.h>
> >  #include <xen/bitops.h>
> >  #include <xen/vm_event.h>
> > -#include <xen/mem_access.h>
> > +#include <xen/monitor.h>
> >  #include <xen/iocap.h>
> >  #include <public/vm_event.h>
> >  #include <asm/flushtlb.h>
> > @@ -1642,12 +1642,41 @@ void __init setup_virt_paging(void)
> >      smp_call_function(setup_virt_paging_one, (void *)val, 1);
> >  }
> >
> > +static int
> > +__p2m_mem_access_send_req(paddr_t gpa, vaddr_t gla, const struct npfec npfec,
> > +                          xenmem_access_t xma)
> > +{
> > +    struct vcpu *v = current;
> > +    vm_event_request_t req = {};
> > +    bool_t sync = (xma == XENMEM_access_n2rwx) ? 0 : 1;
> > +
> > +    req.reason = VM_EVENT_REASON_MEM_ACCESS;
> > +
> > +    /* Send request to mem access subscriber */
> > +    req.u.mem_access.gfn = gpa >> PAGE_SHIFT;
> > +    req.u.mem_access.offset = gpa & ((1 << PAGE_SHIFT) - 1);
> > +    if ( npfec.gla_valid )
> > +    {
> > +        req.u.mem_access.flags |= MEM_ACCESS_GLA_VALID;
> > +        req.u.mem_access.gla = gla;
> > +
> > +        if ( npfec.kind == npfec_kind_with_gla )
> > +            req.u.mem_access.flags |= MEM_ACCESS_FAULT_WITH_GLA;
> > +        else if ( npfec.kind == npfec_kind_in_gpt )
> > +            req.u.mem_access.flags |= MEM_ACCESS_FAULT_IN_GPT;
> > +    }
> > +    req.u.mem_access.flags |= npfec.read_access    ? MEM_ACCESS_R : 0;
> > +    req.u.mem_access.flags |= npfec.write_access   ? MEM_ACCESS_W : 0;
> > +    req.u.mem_access.flags |= npfec.insn_fetch     ? MEM_ACCESS_X : 0;
> > +
> > +    return monitor_traps(v, sync, &req);
> > +}
> > +
> >  bool_t p2m_mem_access_check(paddr_t gpa, vaddr_t gla, const struct npfec npfec)
> >  {
> >      int rc;
> >      bool_t violation;
> >      xenmem_access_t xma;
> > -    vm_event_request_t *req;
> >      struct vcpu *v = current;
> >      struct p2m_domain *p2m = p2m_get_hostp2m(v->domain);
> >
> > @@ -1734,40 +1763,8 @@ bool_t p2m_mem_access_check(paddr_t gpa, vaddr_t gla, const struct npfec npfec)
> >          return false;
> >      }
> >
> > -    req = xzalloc(vm_event_request_t);
> > -    if ( req )
> > -    {
> > -        req->reason = VM_EVENT_REASON_MEM_ACCESS;
> > -
> > -        /* Pause the current VCPU */
> > -        if ( xma != XENMEM_access_n2rwx )
> > -            req->flags |= VM_EVENT_FLAG_VCPU_PAUSED;
> > -
> > -        /* Send request to mem access subscriber */
> > -        req->u.mem_access.gfn = gpa >> PAGE_SHIFT;
> > -        req->u.mem_access.offset =  gpa & ((1 << PAGE_SHIFT) - 1);
> > -        if ( npfec.gla_valid )
> > -        {
> > -            req->u.mem_access.flags |= MEM_ACCESS_GLA_VALID;
> > -            req->u.mem_access.gla = gla;
> > -
> > -            if ( npfec.kind == npfec_kind_with_gla )
> > -                req->u.mem_access.flags |= MEM_ACCESS_FAULT_WITH_GLA;
> > -            else if ( npfec.kind == npfec_kind_in_gpt )
> > -                req->u.mem_access.flags |= MEM_ACCESS_FAULT_IN_GPT;
> > -        }
> > -        req->u.mem_access.flags |= npfec.read_access    ? MEM_ACCESS_R : 0;
> > -        req->u.mem_access.flags |= npfec.write_access   ? MEM_ACCESS_W : 0;
> > -        req->u.mem_access.flags |= npfec.insn_fetch     ? MEM_ACCESS_X : 0;
> > -        req->vcpu_id = v->vcpu_id;
>
> The line setting req->vcpu_id has been removed here ...
>
> > -
> > -        mem_access_send_req(v->domain, req);
> > -        xfree(req);
> > -    }
> > -
> > -    /* Pause the current VCPU */
> > -    if ( xma != XENMEM_access_n2rwx )
> > -        vm_event_vcpu_pause(v);
> > +    if ( __p2m_mem_access_send_req(gpa, gla, npfec, xma) < 0 )
> > +        domain_crash(v->domain);
> >
> >      return false;
> >  }
> > diff --git a/xen/arch/x86/hvm/hvm.c b/xen/arch/x86/hvm/hvm.c
> > index daaee1d..688370d 100644
> > --- a/xen/arch/x86/hvm/hvm.c
> > +++ b/xen/arch/x86/hvm/hvm.c
> > @@ -1707,7 +1707,7 @@ int hvm_hap_nested_page_fault(paddr_t gpa, unsigned long gla,
> >      int rc, fall_through = 0, paged = 0;
> >      int sharing_enomem = 0;
> >      vm_event_request_t *req_ptr = NULL;
> > -    bool_t ap2m_active;
> > +    bool_t ap2m_active, sync = 0;
> >
> >      /* On Nested Virtualization, walk the guest page table.
> >       * If this succeeds, all is fine.
> > @@ -1846,11 +1846,12 @@ int hvm_hap_nested_page_fault(paddr_t gpa, unsigned long gla,
> >                  }
> >              }
> >
> > -            if ( p2m_mem_access_check(gpa, gla, npfec, &req_ptr) )
> > -            {
> > +            sync = p2m_mem_access_check(gpa, gla, npfec, &req_ptr);
> > +
> > +            if ( !sync ) {
> >                  fall_through = 1;
> >              } else {
> > -                /* Rights not promoted, vcpu paused, work here is done */
> > +                /* Rights not promoted (aka. sync event), work here is done */
> >                  rc = 1;
> >                  goto out_put_gfn;
> >              }
> > @@ -1956,7 +1957,12 @@ out:
> >      }
> >      if ( req_ptr )
> >      {
> > -        mem_access_send_req(currd, req_ptr);
> > +        if ( hvm_monitor_mem_access(curr, sync, req_ptr) < 0 )
> > +        {
> > +            /* Crash the domain */
> > +            rc = 0;
> > +        }
> > +
> >          xfree(req_ptr);
> >      }
> >      return rc;
> > diff --git a/xen/arch/x86/hvm/monitor.c b/xen/arch/x86/hvm/monitor.c
> > index 7277c12..c7285c6 100644
> > --- a/xen/arch/x86/hvm/monitor.c
> > +++ b/xen/arch/x86/hvm/monitor.c
> > @@ -152,6 +152,12 @@ int hvm_monitor_cpuid(unsigned long insn_length)
> >      return monitor_traps(curr, 1, &req);
> >  }
> >
> > +int hvm_monitor_mem_access(struct vcpu* v, bool_t sync,
> > +                           vm_event_request_t *req)
> > +{
> > +    return monitor_traps(v, sync, req);
> > +}
> > +
> >  /*
> >   * Local variables:
> >   * mode: C
> > diff --git a/xen/arch/x86/mm/p2m.c b/xen/arch/x86/mm/p2m.c
> > index 812dbf6..27f9d26 100644
> > --- a/xen/arch/x86/mm/p2m.c
> > +++ b/xen/arch/x86/mm/p2m.c
> > @@ -1728,13 +1728,8 @@ bool_t p2m_mem_access_check(paddr_t gpa, unsigned long gla,
> >      if ( req )
> >      {
> >          *req_ptr = req;
> > -        req->reason = VM_EVENT_REASON_MEM_ACCESS;
> > -
> > -        /* Pause the current VCPU */
> > -        if ( p2ma != p2m_access_n2rwx )
> > -            req->flags |= VM_EVENT_FLAG_VCPU_PAUSED;
> >
> > -        /* Send request to mem event */
> > +        req->reason = VM_EVENT_REASON_MEM_ACCESS;
> >          req->u.mem_access.gfn = gfn;
> >          req->u.mem_access.offset = gpa & ((1 << PAGE_SHIFT) - 1);
> >          if ( npfec.gla_valid )
> > @@ -1750,23 +1745,10 @@ bool_t p2m_mem_access_check(paddr_t gpa, unsigned long gla,
> >          req->u.mem_access.flags |= npfec.read_access    ? MEM_ACCESS_R : 0;
> >          req->u.mem_access.flags |= npfec.write_access   ? MEM_ACCESS_W : 0;
> >          req->u.mem_access.flags |= npfec.insn_fetch     ? MEM_ACCESS_X : 0;
> > -        req->vcpu_id = v->vcpu_id;
>
> ... and here, and as such it doesn't seem to get set anywhere else now.
> Am I missing an code path outside of this patch where req->vcpu_id is
> being correctly set so this has become unnecessary?
>

Ah yes, I forgot to add that to monitor_traps. The idea is that common parts if all requests should be set there, including the vcpu_id so we can cut down on code duplication.

Tamas

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
https://lists.xen.org/xen-devel

 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.