Does it help to wait before reattaching?

Revert the map/unmap dance I tried in commit 73042b8d1; that helps
not at all.

Instead, speculate that the unwanted allocation is being done on
another thread, and thus timing variations explain the apparent
unpredictability.  Temporarily add a 1-second sleep before the
VirtualFree call, in hopes that any such other threads will
quiesce and not jog our elbow.

This is obviously not a desirable long-term fix, but as a means of
investigation it seems useful.

Discussion: https://postgr.es/m/25495.1524517820@sss.pgh.pa.us

diff --git a/src/backend/port/win32_shmem.c b/src/backend/port/win32_shmem.c
index 0f140e7a36987754bae8d86ff287ce801a1b69c7..71b1bc3ec4c4590287f16f3a3ae270000159e622 100644 (file)
--- a/src/backend/port/win32_shmem.c
+++ b/src/backend/port/win32_shmem.c
@@ -456,22 +456,11 @@ PGSharedMemoryReAttach(void)
        /* Ensure buf is big enough that it won't grow mid-operation */
        initStringInfo(&buf);
        enlargeStringInfo(&buf, 128 * 1024);
+       /* ... and let's just be sure all that space is committed */
+       memset(buf.data, 0, buf.maxlen);
 
-       dumpmem(&buf, "beginning PGSharedMemoryReAttach");
-
-       hdr = (PGShmemHeader *) MapViewOfFileEx(UsedShmemSegID, FILE_MAP_READ | FILE_MAP_WRITE, 0, 0, 0, NULL);
-       if (hdr)
-       {
-               if (!UnmapViewOfFile(hdr))
-                       elog(LOG, "could not unmap temporary view of shared memory: error code %lu",
-                                GetLastError());
-       }
-       else
-       {
-               /* This isn't fatal, just unpromising ... */
-               elog(LOG, "could not attach to shared memory (key=%p) at any address: error code %lu",
-                        UsedShmemSegID, GetLastError());
-       }
+       /* Test: see if this lets the process address space quiesce */
+       pg_usleep(1000000L);
 
        dumpmem(&buf, "before VirtualFree");