Skip to contentSkip to conversation

x86/mm: Clean up p2m_finish_type_change return value

PATCH In review Alexandru Stefan ISAILA 4 participants 6 messages Mar 29, 2019 – Apr 10, 2019

lore
Share & tools
lore
#0 lore raw

PATCH v5 1/1 x86/mm: Clean up p2m_finish_type_change return value

In the case of any errors, finish_type_change() passes values returned from p2m->recalc() up the stack (with some exceptions in the case where an error is expected); this eventually ends up being returned to the XEN_DOMOP_map_mem_type_to_ioreq_server hypercall.

However, on Intel processors (but not on AMD processor), p2m->recalc() can also return '1' as well as '0'. This case is handled very
inconsistently: finish_type_change() will return the value of the final entry it attempts, discarding results for other entries; p2m_finish_type_change() will attempt to accumulate '1's, so that it returns '1' if any of the calls to finish_type_change() returns '1'; and dm_op() will again return '1' only if the very last call to p2m_finish_type_change() returns '1'. The result is that the XEN_DMOP_map_mem_type_to_ioreq_server() hypercall will sometimes return 0 and sometimes return 1 on success, in an unpredictable manner.

The hypercall documentation doesn't mention return values; but it's not clear what the caller could do with the information about whether entries had been changed or not. At the moment it's always 0 on AMD boxes, and *usually* 1 on Intel boxes; so nothing can be relying on a '1' return value for correctness (or if it is, it's broken).

Make the return value on success consistently '0' by only returning 0/-ERROR from finish_type_change(). Also remove the accumulation code from p2m_finish_type_change().

---
Changes since V4:
- Undo changes in p2m->recalc() and have 1 turn into 0 in finish_type_change().

---
 xen/arch/x86/mm/p2m.c | 16 +++++-----------
 1 file changed, 5 insertions(+), 11 deletions(-)
xen/arch/x86/mm/p2m.c +5 -11
index b9bbb8f485..5508cedc55 100644
@@ -1158,7 +1158,7 @@ void p2m_change_type_range(struct domain *d,
  * Finish p2m type change for gfns which are marked as need_recalc in a range.
  * Uses the current p2m's max_mapped_pfn to further clip the invalidation
  * range for alternate p2ms.
- * Returns: 0/1 for success, negative for failure
+ * Returns: 0 for success, negative for failure
  */
 static int finish_type_change(struct p2m_domain *p2m,
                               gfn_t first_gfn, unsigned long max_nr)
@@ -1174,9 +1174,9 @@ static int finish_type_change(struct p2m_domain *p2m,
         /*
          * ept->recalc could return 0/1/-ENOMEM. pt->recalc could return
          * 0/-ENOMEM/-ENOENT, -ENOENT isn't an error as we are looping
-         * gfn here.
+         * gfn here. If rc is 1 we need to have it 0 for success.
          */
-        if ( rc == -ENOENT )
+        if ( rc == -ENOENT || rc == 1 )
             rc = 0;
         else if ( rc < 0 )
         {
@@ -1213,19 +1213,13 @@ int p2m_finish_type_change(struct domain *d,
             if ( d->arch.altp2m_eptp[i] != mfn_x(INVALID_MFN) )
             {
                 struct p2m_domain *altp2m = d->arch.altp2m_p2m[i];
-                int rc1;
 
                 p2m_lock(altp2m);
-                rc1 = finish_type_change(altp2m, first_gfn, max_nr);
+                rc = finish_type_change(altp2m, first_gfn, max_nr);
                 p2m_unlock(altp2m);
 
-                if ( rc1 < 0 )
-                {
-                    rc = rc1;
+                if ( rc < 0 )
                     goto out;
-                }
-
-                rc |= rc1;
             }
     }
 #endif
Signature · 7 lines
2.17.1


_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Reviewed-by on v5 1/1
#1 lore raw

xen/arch/x86/mm/p2m.c static int finish_type_change(struct p2m_domain *p2m, · line 1179

Quoted 10 lines
Quoted 1 line
Quoted 1 line · On 29.03.19 at 13:50, <[email protected]> wrote:
xen/arch/x86/mm/p2m.c +2 -2
@@ -1174,9 +1174,9 @@ static int finish_type_change(struct p2m_domain *p2m,
         /*
          * ept->recalc could return 0/1/-ENOMEM. pt->recalc could return
          * 0/-ENOMEM/-ENOENT, -ENOENT isn't an error as we are looping
-         * gfn here.
+         * gfn here. If rc is 1 we need to have it 0 for success.
          */
-        if ( rc == -ENOENT )
+        if ( rc == -ENOENT || rc == 1 )

I guess you mean "rc > 0" here? This and the comment are of course easy enough to adjust while committing, and with the adjustments

Quoted 5 lines
xen/arch/x86/mm/p2m.c +0 -1
@@ -1213,19 +1213,13 @@ int p2m_finish_type_change(struct domain *d,
             if ( d->arch.altp2m_eptp[i] != mfn_x(INVALID_MFN) )
             {
                 struct p2m_domain *altp2m = d->arch.altp2m_p2m[i];
-                int rc1;

xen/arch/x86/mm/p2m.c int p2m_finish_type_change(struct domain *d, · line 1221

Quoted 10 lines
xen/arch/x86/mm/p2m.c +2 -4
 
                 p2m_lock(altp2m);
-                rc1 = finish_type_change(altp2m, first_gfn, max_nr);
+                rc = finish_type_change(altp2m, first_gfn, max_nr);
                 p2m_unlock(altp2m);
 
-                if ( rc1 < 0 )
-                {
-                    rc = rc1;
+                if ( rc < 0 )

Along the lines of an earlier comment - I think it is generally better to not special case negative return values when positive ones don't have special meaning. But I know the code base is very inconsistent about this, so I don't really mind leaving it as is.

Jan

_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel

Reviewed-by
#2 lore raw
Quoted 13 lines · Jan Beulich

xen/arch/x86/mm/p2m.c static int finish_type_change(struct p2m_domain *p2m, · line 1179

Quoted 10 lines
Quoted 1 line
Quoted 1 line · On 29.03.19 at 13:50, <[email protected]> wrote:
xen/arch/x86/mm/p2m.c +2 -2
@@ -1174,9 +1174,9 @@ static int finish_type_change(struct p2m_domain *p2m,
         /*
          * ept->recalc could return 0/1/-ENOMEM. pt->recalc could return
          * 0/-ENOMEM/-ENOENT, -ENOENT isn't an error as we are looping
-         * gfn here.
+         * gfn here. If rc is 1 we need to have it 0 for success.
          */
-        if ( rc == -ENOENT )
+        if ( rc == -ENOENT || rc == 1 )

I guess you mean "rc > 0" here? This and the comment are of course easy enough to adjust while committing, and with the adjustments

+1

Quoted 1 line

_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel

#3 lore raw
In reply to Alexandru Stefan ISAILA

Roger/Paul are you ok with the latest changes? Can this go in?

Regards,
Alex

Quoted 75 lines · Alexandru Stefan ISAILA

In the case of any errors, finish_type_change() passes values returned from p2m->recalc() up the stack (with some exceptions in the case where an error is expected); this eventually ends up being returned to the XEN_DOMOP_map_mem_type_to_ioreq_server hypercall.

However, on Intel processors (but not on AMD processor), p2m->recalc() can also return '1' as well as '0'. This case is handled very
inconsistently: finish_type_change() will return the value of the final entry it attempts, discarding results for other entries; p2m_finish_type_change() will attempt to accumulate '1's, so that it returns '1' if any of the calls to finish_type_change() returns '1'; and dm_op() will again return '1' only if the very last call to p2m_finish_type_change() returns '1'. The result is that the XEN_DMOP_map_mem_type_to_ioreq_server() hypercall will sometimes return 0 and sometimes return 1 on success, in an unpredictable manner.

The hypercall documentation doesn't mention return values; but it's not clear what the caller could do with the information about whether entries had been changed or not. At the moment it's always 0 on AMD boxes, and *usually* 1 on Intel boxes; so nothing can be relying on a '1' return value for correctness (or if it is, it's broken).

Make the return value on success consistently '0' by only returning 0/-ERROR from finish_type_change(). Also remove the accumulation code from p2m_finish_type_change().

---
Changes since V4:
- Undo changes in p2m->recalc() and have 1 turn into 0 in finish_type_change().

---
  xen/arch/x86/mm/p2m.c | 16 +++++-----------
  1 file changed, 5 insertions(+), 11 deletions(-)
xen/arch/x86/mm/p2m.c +4 -6
index b9bbb8f485..5508cedc55 100644
@@ -1158,7 +1158,7 @@ void p2m_change_type_range(struct domain *d,
   * Finish p2m type change for gfns which are marked as need_recalc in a range.
   * Uses the current p2m's max_mapped_pfn to further clip the invalidation
   * range for alternate p2ms.
- * Returns: 0/1 for success, negative for failure
+ * Returns: 0 for success, negative for failure
   */
  static int finish_type_change(struct p2m_domain *p2m,
                                gfn_t first_gfn, unsigned long max_nr)
@@ -1174,9 +1174,9 @@ static int finish_type_change(struct p2m_domain *p2m,
          /*
           * ept->recalc could return 0/1/-ENOMEM. pt->recalc could return
           * 0/-ENOMEM/-ENOENT, -ENOENT isn't an error as we are looping
-         * gfn here.
+         * gfn here. If rc is 1 we need to have it 0 for success.
           */
-        if ( rc == -ENOENT )
+        if ( rc == -ENOENT || rc == 1 )
              rc = 0;
          else if ( rc < 0 )
          {
@@ -1213,19 +1213,13 @@ int p2m_finish_type_change(struct domain *d,
              if ( d->arch.altp2m_eptp[i] != mfn_x(INVALID_MFN) )
              {
                  struct p2m_domain *altp2m = d->arch.altp2m_p2m[i];
-                int rc1;
 
                  p2m_lock(altp2m);
-                rc1 = finish_type_change(altp2m, first_gfn, max_nr);
+                rc = finish_type_change(altp2m, first_gfn, max_nr);
                  p2m_unlock(altp2m);
 
-                if ( rc1 < 0 )

xen/arch/x86/mm/p2m.c int p2m_finish_type_change(struct domain *d, · line 1225

Quoted 10 lines
xen/arch/x86/mm/p2m.c +1 -5
-                {
-                    rc = rc1;
+                if ( rc < 0 )
                      goto out;
-                }
-
-                rc |= rc1;
              }
      }
  #endif

_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel

#4 lore raw
Quoted 11 lines

-----Original Message-----
From: Xen-devel [mailto:[email protected]] On Behalf Of Alexandru Stefan ISAILA
Sent: 10 April 2019 09:19
To: [email protected]; George Dunlap <[email protected]>; Paul Durrant <[email protected]>
Cc: Andrew Cooper <[email protected]>; Wei Liu <[email protected]>; [email protected]; Roger Pau Monne <[email protected]>
Subject: Re: [Xen-devel] [PATCH v5] x86/mm: Clean up p2m_finish_type_change return value

Roger/Paul are you ok with the latest changes? Can this go in?

Yes, I am happy for my R-b to be added if so desired.

Paul

Quoted 77 lines

Regards,
Alex

Quoted 73 lines · Alexandru Stefan ISAILA

In the case of any errors, finish_type_change() passes values returned from p2m->recalc() up the stack (with some exceptions in the case where an error is expected); this eventually ends up being returned to the XEN_DOMOP_map_mem_type_to_ioreq_server hypercall.

However, on Intel processors (but not on AMD processor), p2m->recalc() can also return '1' as well as '0'. This case is handled very
inconsistently: finish_type_change() will return the value of the final entry it attempts, discarding results for other entries; p2m_finish_type_change() will attempt to accumulate '1's, so that it returns '1' if any of the calls to finish_type_change() returns '1'; and dm_op() will again return '1' only if the very last call to p2m_finish_type_change() returns '1'. The result is that the XEN_DMOP_map_mem_type_to_ioreq_server() hypercall will sometimes return 0 and sometimes return 1 on success, in an unpredictable manner.

The hypercall documentation doesn't mention return values; but it's not clear what the caller could do with the information about whether entries had been changed or not. At the moment it's always 0 on AMD boxes, and *usually* 1 on Intel boxes; so nothing can be relying on a '1' return value for correctness (or if it is, it's broken).

Make the return value on success consistently '0' by only returning 0/-ERROR from finish_type_change(). Also remove the accumulation code from p2m_finish_type_change().

---
Changes since V4:
- Undo changes in p2m->recalc() and have 1 turn into 0 in finish_type_change().

---
  xen/arch/x86/mm/p2m.c | 16 +++++-----------
  1 file changed, 5 insertions(+), 11 deletions(-)
xen/arch/x86/mm/p2m.c +4 -5
index b9bbb8f485..5508cedc55 100644
@@ -1158,7 +1158,7 @@ void p2m_change_type_range(struct domain *d,
   * Finish p2m type change for gfns which are marked as need_recalc in a range.
   * Uses the current p2m's max_mapped_pfn to further clip the invalidation
   * range for alternate p2ms.
- * Returns: 0/1 for success, negative for failure
+ * Returns: 0 for success, negative for failure
   */
  static int finish_type_change(struct p2m_domain *p2m,
                                gfn_t first_gfn, unsigned long max_nr)
@@ -1174,9 +1174,9 @@ static int finish_type_change(struct p2m_domain *p2m,
          /*
           * ept->recalc could return 0/1/-ENOMEM. pt->recalc could return
           * 0/-ENOMEM/-ENOENT, -ENOENT isn't an error as we are looping
-         * gfn here.
+         * gfn here. If rc is 1 we need to have it 0 for success.
           */
-        if ( rc == -ENOENT )
+        if ( rc == -ENOENT || rc == 1 )
              rc = 0;
          else if ( rc < 0 )
          {
@@ -1213,19 +1213,13 @@ int p2m_finish_type_change(struct domain *d,
              if ( d->arch.altp2m_eptp[i] != mfn_x(INVALID_MFN) )
              {
                  struct p2m_domain *altp2m = d->arch.altp2m_p2m[i];
-                int rc1;
 
                  p2m_lock(altp2m);
-                rc1 = finish_type_change(altp2m, first_gfn, max_nr);
+                rc = finish_type_change(altp2m, first_gfn, max_nr);
                  p2m_unlock(altp2m);
 
Quoted 16 lines

xen/arch/x86/mm/p2m.c int p2m_finish_type_change(struct domain *d, · line 1225

Quoted 12 lines
xen/arch/x86/mm/p2m.c +1 -6
-                if ( rc1 < 0 )
-                {
-                    rc = rc1;
+                if ( rc < 0 )
                      goto out;
-                }
-
-                rc |= rc1;
              }
      }
  #endif

_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel

_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel

#5 lore raw
In reply to Alexandru Stefan ISAILA
Quoted 2 lines
Quoted 1 line
Quoted 1 line · On 10.04.19 at 10:19, <[email protected]> wrote:

Roger/Paul are you ok with the latest changes? Can this go in?

The thing keeping it from going in is the lack of George's ack. Judging from the To: list of your mail you may even have meant him instead of Roger.

Jan

_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel