[PATCH] Hypercall continuation cancelation in compat mode for XENMEM_get/set_pod_target.

[PATCH] Hypercall continuation cancelation in compat mode for XENMEM_get/set_pod_target.

[Jean Guyader]

[-- Attachment #1: Type: text/plain, Size: 472 bytes --]


If copy_to_guest failed in the compat code after a continuation as been
done in the native code we need to cancel it so we won't reexecute the
hypercall but return from the hypercall with the appropriate error.

Signed-off-by: Jean Guyader <jean.guyader@eu.citrix.com>
---
 xen/arch/x86/domain.c           |   18 ++++++++++++++++++
 xen/arch/x86/x86_64/compat/mm.c |    4 ++++
 xen/include/xen/sched.h         |    1 +
 3 files changed, 23 insertions(+), 0 deletions(-)


[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: 0001-Hypercall-continuation-cancelation-in-compat-mode-fo.patch --]
[-- Type: text/x-patch; name="0001-Hypercall-continuation-cancelation-in-compat-mode-fo.patch", Size: 1854 bytes --]

diff --git a/xen/arch/x86/domain.c b/xen/arch/x86/domain.c
index 52c7f37..acc3241 100644
--- a/xen/arch/x86/domain.c
+++ b/xen/arch/x86/domain.c
@@ -1723,6 +1723,24 @@ void sync_vcpu_execstate(struct vcpu *v)
     __arg;                                                                  \
 })
 
+void hypercall_cancel_continuation(void)
+{
+    struct cpu_user_regs *regs = guest_cpu_user_regs();
+    struct mc_state *mcs = &current->mc_state;
+
+    if ( test_bit(_MCSF_in_multicall, &mcs->flags) )
+    {
+        __clear_bit(_MCSF_call_preempted, &mcs->flags);
+    }
+    else
+    {
+        if ( !is_hvm_vcpu(current) )
+            regs->eip += 2; /* skip re-execute 'syscall' / 'int $xx' */
+        else
+            current->arch.hvm_vcpu.hcall_preempted = 0;
+    }
+}
+
 unsigned long hypercall_create_continuation(
     unsigned int op, const char *format, ...)
 {
diff --git a/xen/arch/x86/x86_64/compat/mm.c b/xen/arch/x86/x86_64/compat/mm.c
index 2c05099..3ef08a5 100644
--- a/xen/arch/x86/x86_64/compat/mm.c
+++ b/xen/arch/x86/x86_64/compat/mm.c
@@ -133,7 +133,11 @@ int compat_arch_memory_op(int op, XEN_GUEST_HANDLE(void) arg)
         XLAT_pod_target(&cmp, nat);
 
         if ( copy_to_guest(arg, &cmp, 1) )
+        {
+            if ( rc == __HYPERVISOR_memory_op )
+                hypercall_cancel_continuation();
             rc = -EFAULT;
+        }
 
         break;
     }
diff --git a/xen/include/xen/sched.h b/xen/include/xen/sched.h
index 3ba5495..80d5c4b 100644
--- a/xen/include/xen/sched.h
+++ b/xen/include/xen/sched.h
@@ -554,6 +554,7 @@ extern void (*dead_idle) (void);
  */
 unsigned long hypercall_create_continuation(
     unsigned int op, const char *format, ...);
+void hypercall_cancel_continuation(void);
 
 #define hypercall_preempt_check() (unlikely(    \
         softirq_pending(smp_processor_id()) |   \

[-- Attachment #3: Type: text/plain, Size: 138 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel

Re: [PATCH] Hypercall continuation cancelation in compat mode for XENMEM_get/set_pod_target.

[Jan Beulich]

>>> On 10.11.11 at 18:17, Jean Guyader <jean.guyader@eu.citrix.com> wrote:

> If copy_to_guest failed in the compat code after a continuation as been
> done in the native code we need to cancel it so we won't reexecute the
> hypercall but return from the hypercall with the appropriate error.
> 
> Signed-off-by: Jean Guyader <jean.guyader@eu.citrix.com>

Acked-by: Jan Beulich <jbeulich@suse.com>

(But I'd like to be certain that Keir also took a look at this before
committing, not the least because this also should go into 4.1-testing.)

> ---
>  xen/arch/x86/domain.c           |   18 ++++++++++++++++++
>  xen/arch/x86/x86_64/compat/mm.c |    4 ++++
>  xen/include/xen/sched.h         |    1 +
>  3 files changed, 23 insertions(+), 0 deletions(-)

Re: [PATCH] Hypercall continuation cancelation in compat mode for XENMEM_get/set_pod_target.

[Keir Fraser]

On 11/11/2011 08:38, "Jan Beulich" <JBeulich@suse.com> wrote:

>>>> On 10.11.11 at 18:17, Jean Guyader <jean.guyader@eu.citrix.com> wrote:
> 
>> If copy_to_guest failed in the compat code after a continuation as been
>> done in the native code we need to cancel it so we won't reexecute the
>> hypercall but return from the hypercall with the appropriate error.
>> 
>> Signed-off-by: Jean Guyader <jean.guyader@eu.citrix.com>
> 
> Acked-by: Jan Beulich <jbeulich@suse.com>
> 
> (But I'd like to be certain that Keir also took a look at this before
> committing, not the least because this also should go into 4.1-testing.)

Looks okay to me.

Acked-by: Keir Fraser <keir@xen.org>

>> ---
>>  xen/arch/x86/domain.c           |   18 ++++++++++++++++++
>>  xen/arch/x86/x86_64/compat/mm.c |    4 ++++
>>  xen/include/xen/sched.h         |    1 +
>>  3 files changed, 23 insertions(+), 0 deletions(-)
> 
> 
> 

Re: [PATCH] Hypercall continuation cancelation in compat mode for XENMEM_get/set_pod_target.

[Jean Guyader]

On 10/11 05:17, Jean Guyader wrote:
> 
> If copy_to_guest failed in the compat code after a continuation as been
> done in the native code we need to cancel it so we won't reexecute the
> hypercall but return from the hypercall with the appropriate error.
> 
> Signed-off-by: Jean Guyader <jean.guyader@eu.citrix.com>

Did that make it into 4.1?

Jean

Re: [PATCH] Hypercall continuation cancelation in compat mode for XENMEM_get/set_pod_target.

[Jan Beulich]

>>> On 17.11.11 at 07:01, Jean Guyader <jean.guyader@eu.citrix.com> wrote:
> On 10/11 05:17, Jean Guyader wrote:
>> 
>> If copy_to_guest failed in the compat code after a continuation as been
>> done in the native code we need to cancel it so we won't reexecute the
>> hypercall but return from the hypercall with the appropriate error.
>> 
>> Signed-off-by: Jean Guyader <jean.guyader@eu.citrix.com>
> 
> Did that make it into 4.1?
> 
> Jean

Not so far - Keir?

Jan

Re: Re: [PATCH] Hypercall continuation cancelation in compat mode for XENMEM_get/set_pod_target.

[Keir Fraser]

On 17/11/2011 06:01, "Jean Guyader" <jean.guyader@eu.citrix.com> wrote:

> On 10/11 05:17, Jean Guyader wrote:
>> 
>> If copy_to_guest failed in the compat code after a continuation as been
>> done in the native code we need to cancel it so we won't reexecute the
>> hypercall but return from the hypercall with the appropriate error.
>> 
>> Signed-off-by: Jean Guyader <jean.guyader@eu.citrix.com>
> 
> Did that make it into 4.1?

Done.

> Jean
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xensource.com
> http://lists.xensource.com/xen-devel