xen/vm_event: fix rc check for uninitialized ring
PATCH 1/1 xen/vm_event: fix rc check for uninitialized ring
vm_event_claim_slot() returns -EOPNOTSUPP for an uninitialized ring since commit 15e4dd5e866b43bbc ("common/vm_event: Initialize vm_event lists on domain creation"), but the callers test for -ENOSYS.
Correct the callers.
--- xen/arch/x86/mm/p2m.c | 2 +- xen/common/monitor.c | 2 +- xen/common/vm_event.c | 2 +- 3 files changed, 3 insertions(+), 3 deletions(-)
xen/arch/x86/mm/p2m.c +1 -1
@@ -1705,7 +1705,7 @@ void p2m_mem_paging_populate(struct domain *d, unsigned long gfn_l)/* We're paging. There should be a ring */ int rc =vm_event_claim_slot (d, d->vm_event_paging); -if ( rc == -ENOSYS ) +if ( rc == -EOPNOTSUPP ) {gdprintk (XENLOG_ERR ,"Domain %hu paging gfn %lx yet no ring " "in place\n" , d->domain_id, gfn_l);
xen/common/monitor.c +1 -1
@@ -98,7 +98,7 @@ int monitor_traps(struct vcpu *v, bool sync, vm_event_request_t *req) {case 0 :break ; -case -ENOSYS : +case -EOPNOTSUPP :/* * If there was no ring to handle the event, then * simply continue executing normally.
xen/common/vm_event.c +1 -1
@@ -513,7 +513,7 @@ bool vm_event_check_ring(struct vm_event_domain *ved)* this function will always return 0 for a guest. For a non-guest, we check * for space and return -EBUSY if the ring is not available. * -* Return codes: -ENOSYS: the ring is not yet configured +* Return codes: -EOPNOTSUPP: the ring is not yet configured * -EBUSY: the ring is busy * 0: a spot has been reserved *
Signature · 7 lines
2.16.4 _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Quoted 45 lines · Juergen Gross
vm_event_claim_slot()returns-EOPNOTSUPPfor an uninitialized ring since commit15e4dd5e866b43bbc("common/vm_event: Initializevm_eventlists on domain creation"), but the callers test for-ENOSYS.Correct the callers.
Signed-off-by Juergen Gross <[email protected]>--- xen/arch/x86/mm/p2m.c | 2 +- xen/common/monitor.c | 2 +- xen/common/vm_event.c | 2 +- 3 files changed, 3 insertions(+), 3 deletions(-)xen/arch/x86/mm/p2m.c +1
-1@@ -1705,7 +1705,7 @@ void p2m_mem_paging_populate(struct domain *d, unsigned long gfn_l)/* We're paging. There should be a ring */ int rc =vm_event_claim_slot (d, d->vm_event_paging); -if ( rc == -ENOSYS ) +if ( rc == -EOPNOTSUPP ) {gdprintk (XENLOG_ERR ,"Domain %hu paging gfn %lx yet no ring " "in place\n" , d->domain_id, gfn_l);xen/common/monitor.c +1
-1@@ -98,7 +98,7 @@ int monitor_traps(struct vcpu *v, bool sync, vm_event_request_t *req) {case 0 :break ; -case -ENOSYS : +case -EOPNOTSUPP :/* * If there was no ring to handle the event, then * simply continue executing normally. xen/common/vm_event.c +0
-0
xen/common/vm_event.c bool vm_event_check_ring(struct vm_event_domain *ved) · line 519
Quoted 8 lines
xen/common/vm_event.c +1
-1@@ -513,7 +513,7 @@ bool vm_event_check_ring(struct vm_event_domain *ved)* this function will always return 0 for a guest. For a non-guest, we check * for space and return -EBUSY if the ring is not available. * -* Return codes: -ENOSYS: the ring is not yet configured +* Return codes: -EOPNOTSUPP: the ring is not yet configured * -EBUSY: the ring is busy * 0: a spot has been reserved *
But vm_event_grab_slot() (which ends up being what vm_event_wait_slot() returns), still returns -ENOSYS:
463 static int vm_event_grab_slot(struct vm_event_domain *ved, int foreign) 464 {
465 unsigned int avail_req;
466
467 if ( !ved->ring_page )
468 return -ENOSYS;
Although we can't get to that part if vm_event_check_ring() returns false, we should probably return -EOPNOTSUPP from there as well. In fact, I wonder if any of the -ENOSYS returns in vm_event.c should not be replaced with return -EOPNOTSUPPs.
Anyway, the change does clearly improve the code even without settling the above questions, so:
Thanks,
Razvan
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Quoted 67 lines · Razvan Cojocaru
Quoted 46 lines · Juergen Gross
vm_event_claim_slot()returns-EOPNOTSUPPfor an uninitialized ring since commit15e4dd5e866b43bbc("common/vm_event: Initializevm_eventlists on domain creation"), but the callers test for-ENOSYS.Correct the callers.
Signed-off-by Juergen Gross <[email protected]>---
xen/arch/x86/mm/p2m.c| 2 +-
xen/common/monitor.c | 2 +-
xen/common/vm_event.c| 2 +-3 files changed, 3 insertions(+), 3 deletions(-)xen/arch/x86/mm/p2m.c +1
-1@@ -1705,7 +1705,7 @@ void p2m_mem_paging_populate(struct domain *d, unsigned long gfn_l) /* We're paging. There should be a ring */ int rc = vm_event_claim_slot(d, d->vm_event_paging); - if ( rc == -ENOSYS ) + if ( rc == -EOPNOTSUPP ) {
gdprintk(XENLOG_ERR, "Domain %hu paging gfn %lx yet no ring " "in place\n", d->domain_id, gfn_l);xen/common/monitor.c +0
-0@@ -98,7 +98,7 @@ int monitor_traps(struct vcpu *v, bool sync, vm_event_request_t *req) {case 0:
break;
- case-ENOSYS:
+ case-EOPNOTSUPP:
/*
* If there was no ring to handle the event, then
* simply continue executing normally.xen/common/vm_event.c +0
-0@@ -513,7 +513,7 @@ bool vm_event_check_ring(struct vm_event_domain *ved) * this function will always return 0 for a guest. For a non-guest,Quoted 9 lines
we check
* for space and return-EBUSYif the ring is not available. *
- * Return codes:-ENOSYS: the ring is not yet configured
+ * Return codes:-EOPNOTSUPP: the ring is not yet configured *-EBUSY: the ring is busy
* 0: a spot has been reserved
*But
vm_event_grab_slot()(which ends up being whatvm_event_wait_slot()returns), still returns-ENOSYS:463 static int
vm_event_grab_slot(struct vm_event_domain *ved, int foreign)464 {
465 unsigned intavail_req;
466
467 if ( !ved->ring_page )
468 return-ENOSYS;
Quoted 2 lines
Although we can't get to that part if
vm_event_check_ring()returns false, we should probably return-EOPNOTSUPPfrom there as well. In
Hmm, in case the page is removed while a vcpu is waiting for a slot to become free ENOSYS should still be possible. There is, however, a race in vm_event_grab_slot(): ved->ring_page is tested outside the lock, so it could be zeroed just after the test occurred leading to a "use after free" when calling vm_event_ring_available().
Quoted 2 lines
fact, I wonder if any of the
-ENOSYSreturns invm_event.cshould not be replaced with return -EOPNOTSUPPs.
I believe the ones for the ring page not existing should be replaced, while the ones for wrong hypercall sub-commands should either remain or be replaced by EINVAL.
Quoted 4 lines
Anyway, the change does clearly improve the code even without settling the above questions, so:
Acked-by Razvan Cojocaru <[email protected]>
Juergen
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
xen/arch/x86/mm/p2m.c void p2m_mem_paging_populate(struct domain *d, unsigned long gfn_l) · line 1708
Quoted 9 lines
Quoted 1 line
Quoted 1 line · On 24.05.19 at 15:15, <[email protected]> wrote:
xen/arch/x86/mm/p2m.c +1
-1@@ -1705,7 +1705,7 @@ void p2m_mem_paging_populate(struct domain *d, unsigned long gfn_l)/* We're paging. There should be a ring */ int rc =vm_event_claim_slot (d, d->vm_event_paging); -if ( rc == -ENOSYS ) +if ( rc == -EOPNOTSUPP )
Perhaps while committing (or if a v2 should become necessary) the missing blank line could be added here at the same time.
Jan
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Quoted 5 lines · Jan Beulich
Quoted 5 lines
Quoted 1 line
Quoted 1 line · On 24.05.19 at 15:15, <[email protected]> wrote:
xen/arch/x86/mm/p2m.c +0
-0@@ -1705,7 +1705,7 @@ void p2m_mem_paging_populate(struct domain *d, unsigned long gfn_l)
Quoted 7 lines
xen/arch/x86/mm/p2m.c void p2m_mem_paging_populate(struct domain *d, unsigned long gfn_l) · line 1708
Quoted 4 lines
xen/arch/x86/mm/p2m.c +1
-1/* We're paging. There should be a ring */ int rc =vm_event_claim_slot (d, d->vm_event_paging); -if ( rc == -ENOSYS ) +if ( rc == -EOPNOTSUPP )Perhaps while committing (or if a v2 should become necessary) the missing blank line could be added here at the same time.
I'll send a V2 with the two missing ENOSYS replacements.
Juergen
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel