Fiji and Samoa. Historical corrections for Hong Kong.
The original coding in tuplestore_trim() was only meant to work efficiently in cases where each trim call deleted most of the tuples in the store. Which, in fact, was the pattern of the original usage with a Material node supporting mark/restore operations underneath a MergeJoin. However, WindowAgg now uses tuplestores and it has considerably less friendly trimming behavior. In particular it can attempt to trim one tuple at a time off a large tuplestore. tuplestore_trim() had O(N^2) runtime in this situation because of repeatedly shifting its tuple pointer array. Fix by avoiding shifting the array until a reasonably large number of tuples have been deleted. This can waste some pointer space, but we do still reclaim the tuples themselves, so the percentage wastage should be pretty small. Per Jie Li's report of slow percent_rank() evaluation. cume_dist() and ntile() would certainly be affected as well, along with any other window function that has a moving frame start and requires reading substantially ahead of the current row. Back-patch to 8.4, where window functions were introduced. There's no need to tweak it before that.
Recent versions of the Linux system header files cause xlogdefs.h to believe that open_datasync should be the default sync method, whereas formerly fdatasync was the default on Linux. open_datasync is a bad choice, first because it doesn't actually outperform fdatasync (in fact the reverse), and second because we try to use O_DIRECT with it, causing failures on certain filesystems (e.g., ext4 with data=journal option). This part of the patch is largely per a proposal from Marti Raudsepp. More extensive changes are likely to follow in HEAD, but this is as much change as we want to back-patch. Also clean up confusing code and incorrect documentation surrounding the fsync_writethrough option. Those changes shouldn't result in any actual behavioral change, but I chose to back-patch them anyway to keep the branches looking similar in this area. In 9.0 and HEAD, also do some copy-editing on the WAL Reliability documentation section. Back-patch to all supported branches, since any of them might get used on modern Linux versions.
There are some code paths, such as SPI_execute(), where we invoke copyObject() on raw parse trees before doing parse analysis on them. Since the bison grammar is capable of building heavily nested parsetrees while itself using only minimal stack depth, this means that copyObject() can be the front-line function that hits stack overflow before anything else does. Accordingly, it had better have a check_stack_depth() call. I did a bit of performance testing and found that this slows down copyObject() by only a few percent, so the hit ought to be negligible in the context of complete processing of a query. Per off-list report from Toshihide Katayama. Back-patch to all supported branches.
There were corner cases in which the planner would attempt to inline such a function, which would result in a failure at runtime due to loss of information about exactly what the result record type is. Fix by disabling inlining when the function's recorded result type is RECORD. There might be some sub-cases where inlining could still be allowed, but this is a simple and backpatchable fix, so leave refinements for another day. Per bug #5777 from Nate Carson. Back-patch to all supported branches. 8.1 happens to avoid a core-dump here, but it still does the wrong thing.
Most of the functions that execute XPath queries leaked the data structures created by libxml2. This memory would not be recovered until end of session, so it mounts up pretty quickly in any serious use of the feature. Per report from Pavel Stehule, though this isn't his patch. Back-patch to all supported branches.
except creating new connections.