Current implementation of virDomainMemoryDefCheckConflict() does
only a one way comparison, i.e. if there's a memory device within
def->mems[] which address falls in [mem->address, mem->address +
mem->size] range (mem is basically an iterator within
def->mems[]). And for static XML this works just fine. Problem is
with hot/cold plugging of a memory device. Then mem points to
freshly parsed memory device and these half checks are
insufficient. Not only we must check whether an existing memory
device doesn't clash with freshly parsed memory device, but also
whether freshly parsed memory device does not fall into range of
already existing memory device.

Resolves: https://issues.redhat.com/browse/RHEL-4452
Signed-off-by: Michal Privoznik <mpriv...@redhat.com>
---
 src/conf/domain_validate.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/src/conf/domain_validate.c b/src/conf/domain_validate.c
index d485ec4fb1..14148a18d3 100644
--- a/src/conf/domain_validate.c
+++ b/src/conf/domain_validate.c
@@ -2264,6 +2264,7 @@ virDomainMemoryDefCheckConflict(const virDomainMemoryDef 
*mem,
     for (i = 0; i < def->nmems; i++) {
         const virDomainMemoryDef *other = def->mems[i];
         unsigned long long otherStart = 0;
+        unsigned long long otherEnd = 0;
 
         if (other == mem)
             continue;
@@ -2315,7 +2316,10 @@ virDomainMemoryDefCheckConflict(const virDomainMemoryDef 
*mem,
         if (thisStart == 0 || otherStart == 0)
             continue;
 
-        if (thisStart <= otherStart && thisEnd > otherStart) {
+        otherEnd = otherStart + other->size;
+
+        if ((thisStart <= otherStart && thisEnd > otherStart) ||
+            (otherStart <= thisStart && otherEnd > thisStart)) {
             virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                            _("memory device address [0x%1$llx:0x%2$llx] 
overlaps with other memory device (0x%3$llx)"),
                            thisStart, thisEnd, otherStart);
-- 
2.43.0
_______________________________________________
Devel mailing list -- devel@lists.libvirt.org
To unsubscribe send an email to devel-le...@lists.libvirt.org

Reply via email to