Re: Return pg_control from pg_backup_stop().

On Fri, Oct 23, 2026 at 06:47:24PM +0300, Heikki Linnakangas wrote:
> Looks good to me.

Apologies for not following closely what you have been doing lately in
this area, David. I should have followed, but well..

+void
+backup_control_file(uint8 *controlFile)
+{
+ ControlFileData controlData;
+
+ LWLockAcquire(ControlFileLock, LW_SHARED);
+ memcpy(&controlData, ControlFile, sizeof(ControlFileData));
+ LWLockRelease(ControlFileLock);
[...]
/* ... and pg_control after everything else. */
- if (lstat(XLOG_CONTROL_FILE, &statbuf) != 0)
- ereport(ERROR,
- (errcode_for_file_access(),
- errmsg("could not stat file \"%s\": %m",
- XLOG_CONTROL_FILE)));
- sendFile(sink, XLOG_CONTROL_FILE, XLOG_CONTROL_FILE, &statbuf,
- false, InvalidOid, InvalidOid,
- InvalidRelFileNumber, 0, &manifest, 0, NULL, 0);
+ backup_control_file(controlFile);
+ sendFileWithContent(sink, XLOG_CONTROL_FILE,
+ (char *) controlFile, PG_CONTROL_FILE_SIZE,
+ &manifest);

Yep, that's clearly something that should be backpatched, at quick
glance. Heikki, are you planning to take care of this part?
--
Michael