Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Fix pg_size_pretty() to avoid overflow for inputs close to INT64_MAX.
authorTom Lane <tgl@sss.pgh.pa.us>
Mon, 25 Apr 2011 20:22:20 +0000 (16:22 -0400)
committerTom Lane <tgl@sss.pgh.pa.us>
Mon, 25 Apr 2011 20:22:20 +0000 (16:22 -0400)
The expression that tried to round the value to the nearest TB could
overflow, leading to bogus output as reported in bug #5993 from Nicola
Cossu.  This isn't likely to ever happen in the intended usage of the
function (if it could, we'd be needing to use a wider datatype instead);
but it's not hard to give the expected output, so let's do so.

src/backend/utils/adt/dbsize.c

index 3a85292f057bf4dcb94ccb28f53a6980f54d5a28..ab50c06171ea1d78c5fa513ff34b340d55a4a754 100644 (file)
@@ -402,9 +402,15 @@ pg_size_pretty(PG_FUNCTION_ARGS)
                             (size + mult / 2) / mult);
                else
                {
+                   /* Here we have to worry about avoiding overflow */
+                   int64   val;
+
                    mult *= 1024;
+                   val = size / mult;
+                   if ((size % mult) >= (mult / 2))
+                       val++;
                    snprintf(buf, sizeof(buf), INT64_FORMAT " TB",
-                            (size + mult / 2) / mult);
+                            val);
                }
            }
        }