Commit4ce424d

committed

aio: Experimental heuristics to increase batching in read_stream.c

Author:Reviewed-by:Discussion:https://postgr.es/m/Backpatch:

1 parent11f0ca7 commit4ce424dCopy full SHA for 4ce424d

File tree

+24

-0

lines changed

+24

-0

lines changed

Lines changed: 24 additions & 0 deletions

Original file line number	Diff line number	Diff line change
`@@ -404,6 +404,30 @@ read_stream_start_pending_read(ReadStream *stream)`
`404`	`404`	`staticvoid`
`405`	`405`	`read_stream_look_ahead(ReadStream*stream)`
`406`	`406`	`{`
	`407`	`+/*`
	`408`	`+ * Batch-submitting multiple IOs is more efficient than doing so`
	`409`	`+ * one-by-one. If we just ramp up to the max, we'll only be allowed to`
	`410`	`+ * submit one io_combine_limit sized IO. Defer submitting IO in that case.`
	`411`	`+ *`
	`412`	`+ * FIXME: This needs better heuristics.`
	`413`	`+ */`
	`414`	`+#if1`
	`415`	`+if (!stream->sync_mode&&stream->distance> (io_combine_limit*8))`
	`416`	`+{`
	`417`	`+if (stream->pinned_buffers+stream->pending_read_nblocks> ((stream->distance*3) /4))`
	`418`	`+{`
	`419`	`+#if0`
	`420`	`+ereport(LOG,`
	`421`	`+errmsg("reduce reduce reduce: pinned: %d, pending: %d, distance: %d",`
	`422`	`+stream->pinned_buffers,`
	`423`	`+stream->pending_read_nblocks,`
	`424`	`+stream->distance));`
	`425`	`+#endif`
	`426`	`+return;`
	`427`	`+}`
	`428`	`+}`
	`429`	`+#endif`
	`430`	`+`
`407`	`431`	`/*`
`408`	`432`	`* Allow amortizing the cost of submitting IO over multiple IOs. This`
`409`	`433`	`* requires that we don't do any operations that could lead to a deadlock`

Comments

(0)