Overview

Every Oracle execution plan is logically a tree of row source operators (row sources)

A rowsource operator (sometimes called rowsource operation, execution plan operation or just an execution plan line) is:

1. Physically just a C function
2. ...which calls its child rowsource functions in the execution plan tree
   
3. ...fetches rows returned from the child rowsources (functions)
   
4. ...processes the rows (depending on what kind of rowsource it is)
5. ...passes the rows up the tree to parent rowsources
6. ...once the rows reach the root in the execution plan tree, they are returned back to fetch call which sends them back to the client (in case of a SELECT statement). In case of DML statements, the root rowsource function is the one which performs the required changes on the rows returned to the tree (update, delete).
   

Note that a few rowsources in Oracle are "hidden", these are not listed here (nor in explain plan / V$SQL_PLAN output).
An example of a hidden rowsource is the Statistics rowsource, which measures time and passes rows through from its child operator. The usage of hidden rowsources can be detected using OS level tools like Tanel's os_explain command. Another hidden rowsource is a plain filter rowsource, which throws away rows not matching the filter condition. This filter rowsource does not show up as FILTER operation (which is a different thing), but is rather just shown as "filter -" in the predicate section of execution plan output.

Different operations do different things and they only know how to do "their thing". For example, a TABLE ACCESS operation does not know how walk an index and find ROWID's from there, instead the TABLE ACCESS will call INDEX rowsource which walks the index structure and returns a list of ROWID's (physical row addresses) to the TABLE ACCESS. Then the TABLE ACCESS uses these ROWIDs to fetch their corresponding rows from their physical locations.

Many operators also have options which affect how exactly the operator works. For example INDEX operator has options such as UNIQUE, RANGE SCAN, SKIP SCAN etc, which determine the way the INDEX operator internally works.

Starting from Oracle 11.1 it is possible to list all (non-hidden) rowsources available by querying X$XPLTON (eXPLain rowsource Operator Names).
The rowsource operator options are listed in X$XPLTOO (eXPLain rowsource Operator Options). Starting from Oracle 11.2, the same data can be queried from AWR views DBA_HIST_PLAN_OPERATION_NAME and DBA_HIST_PLAN_OPTION_NAME.

Oracle Rowsource Operator Name list

1. BITMAP KEY ITERATION
2. NESTED LOOPS
3. MERGE JOIN
4. HASH JOIN
5. AND-EQUAL
6. BITMAP AND
7. INTERSECTION
8. MINUS
9. BITMAP MINUS
10. GENERATE
11. BITMAP CONVERSION
12. BITMAP INDEX
13. CONNECT BY
14. CONNECT BY PUMP
15. COUNT
16. DELETE
17. UPDATE
18. FIRST ROW
19. FILTER
20. FOR UPDATE
21. FIXED TABLE
22. INLIST ITERATOR
23. INDEX
24. INDEX BUILD
25. LOAD AS SELECT
26. MULTI-TABLE INSERT
27. INTO
28. DIRECT LOAD INTO
29. COLLECTION ITERATOR
30. DOMAIN INDEX
31. PARTITION RANGE
32. PARTITION HASH
33. PARTITION LIST
34. PARTITION SYSTEM
35. PARTITION COMBINED
36. GRANULE ITERATOR
37. SEQUENCE
38. TABLE ACCESS
39. TABLE QUEUE
40. INDEX MAINTENANCE
41. FIFO BUFFER
42. WINDOW
43. TEMP TABLE TRANSFORMATION
44. SORT
45. BUFFER
46. BITMAP COMPACTION
47. BITMAP CONSTRUCTION
48. BITMAP OR
49. BITMAP MERGE
50. CONCATENATION
51. UNION-ALL
52. VIEW PUSHED PREDICATE
53. VIEW
54. REMOTE
55. SELECT STATEMENT
56. INSERT STATEMENT
57. UPDATE STATEMENT
58. MERGE STATEMENT
59. DELETE STATEMENT
60. CREATE TABLE STATEMENT
61. ALTER INDEX STATEMENT
62. CREATE INDEX STATEMENT
63. DDL STATEMENT
64. TEMP TABLE GENERATION
65. TRUNCATE TEMP TABLE
66. RECURSIVE EXECUTION
67. MERGE
68. EXTERNAL TABLE ACCESS
69. BITMAP JOIN INDEX UPDATE
70. BITMAP JOIN INDEX UPDATE STATEMENT
71. VIEW HIERARCHY
72. SQL MODEL
73. MAT_VIEW ACCESS
74. MAT_VIEW REWRITE ACCESS
75. FREQUENT ITEMSET COUNTING
76. FIC ENUMERATE FEED
77. FIC LOAD BITMAPS
78. FIC LOAD ITEMSETS
79. FIC DETECT END
80. FIC RECURSIVE ITERATION
81. DOMAIN INDEX BUILD
82. DOMAIN INDEX PARTITION BUILD
83. FAST DUAL
84. REFERENCE MODEL
85. PX RECEIVE
86. QC SEND
87. PX SEND
88. PX COORDINATOR
89. PX BLOCK
90. PX PARTITION RANGE
91. PX PARTITION HASH
92. PX PARTITION LIST
93. WINDOW (IN SQL MODEL)
94. UNION ALL PUSHED PREDICATE
95. DST ENUMERATE FEED
96. DST DETECT END
97. DST RECURSIVE ITERATION
98. DST SPLIT
99. DECISION TREE CLASSIFICATION
100. DST PRUNE
101. JOIN FILTER
102. HASH
103. ERROR LOGGING
104. LOAD TABLE CONVENTIONAL
105. PART JOIN FILTER
106. AW SCAN
107. PARTITION REFERENCE
108. PX PARTITION REFERENCE
109. XPATH EVALUATION
110. RESULT CACHE
111. DM FEED PREDICTOR-TARGET PAIRS
112. DM SUPERVISED BINNING
113. DM HASH-BROADCAST COUNTS
114. BULK BINDS GET
115. APPROXIMATE NDV
116. UNPIVOT
117. TRANSPOSE
118. MONITORING
119. CUBE SCAN
120. JOINED CUBE SCAN
121. CUBE ACCESS
122. MAT_VIEW CUBE ACCESS
123. MAT_VIEW REWRITE CUBE ACCESS
124. CUBE TABLE SCAN
125. JOINED CUBE TABLE SCAN
126. CUBE TABLE ACCESS
127. UNION ALL (RECURSIVE WITH)
128. RECURSIVE WITH PUMP

Oracle Rowsource Operator Options list

1. OUTER
2. ANTI
3. SEMI
4. CARTESIAN
5. TO ROWIDS
6. FROM ROWIDS
7. COUNT
8. SINGLE VALUE
9. SAMPLE FAST FULL SCAN
10. FAST FULL SCAN
11. FULL SCAN
12. RANGE SCAN
13. UNIQUE SCAN
14. SKIP SCAN
15. FULL SCAN (MIN/MAX)
16. RANGE SCAN (MIN/MAX)
17. FULL SCAN DESCENDING
18. RANGE SCAN DESCENDING
19. SKIP SCAN DESCENDING
20. WITH FILTERING
21. WITHOUT FILTERING
22. STOPKEY
23. FIXED INDEX
24. FULL
25. HASH
26. INDEX BUILD
27. UNIQUE (LOCAL)
28. NON UNIQUE (LOCAL)
29. UNIQUE
30. NON UNIQUE
31. PICKLER FETCH
32. CONSTRUCTOR FETCH
33. SUBQUERY FETCH
34. REMOTE
35. SAMPLE
36. SAMPLE BY ROWID RANGE
37. BY ROWID RANGE
38. BY LOCAL INDEX ROWID
39. BY GLOBAL INDEX ROWID
40. BY INDEX ROWID
41. BY USER ROWID
42. CLUSTER
43. NOSORT
44. BUFFER PUSHED RANK
45. BUFFER
46. CHILD PUSHED RANK
47. CHILD
48. SORT PUSHED RANK
49. SINGLE
50. EMPTY
51. INLIST
52. ALL
53. ITERATOR
54. PARTITION
55. AGGREGATE
56. GROUP BY STOPKEY
57. GROUP BY
58. UNIQUE NOSORT
59. GROUP BY NOSORT
60. GROUP BY ROLLUP
61. CUBE
62. ROLLUP
63. SORT
64. CREATE INDEX
65. UNIQUE STOPKEY
66. ORDER BY STOPKEY
67. ORDER BY
68. JOIN
69. GROUP BY NOSORT ROLLUP
70. CUSTOM OPTION
71. ROWID RANGE
72. EXTERNAL CHUNK
73. ACYCLIC
74. CYCLIC
75. ACYCLIC FAST
76. ORDERED
77. ORDERED FAST
78. RIGHT OUTER
79. RIGHT ANTI
80. RIGHT SEMI
81. PARTITION OUTER
82. (LOCAL)
83. PARTITION (ROWID)
84. PARTITION (KEY)
85. RANGE
86. ROUND-ROBIN
87. BROADCAST
88. QC (ORDER)
89. QC (RANDOM)
90. BROADCAST LOCAL
91. HASH LOCAL
92. HYBRID (ROWID PKEY)
93. RANDOM LOCAL
94. HASH (BLOCK ADDRESS)
95. MIXED HASH BCAST LOCAL
96. MIXED HASH BCAST
97. OUTER BUFFERED
98. RIGHT OUTER BUFFERED
99. ANTI BUFFERED
100. RIGHT ANTI BUFFERED
101. SEMI BUFFERED
102. RIGHT SEMI BUFFERED
103. CARTESIAN BUFFERED
104. PARTITION OUTER BUFFERED
105. HASH ALL
106. RANGE ALL
107. LIST
108. LIST ALL
109. SYSTEM
110. FORCED SERIAL
111. PARTITION JOIN
112. AW HASH
113. CREATE
114. USE
115. SUBQUERY
116. OR
117. MULTI-COLUMN
118. NOSORT STOPKEY
119. BUFFERED
120. FULL OUTER
121. FULL OUTER BUFFERED
122. NO FILTERING WITH START-WITH
123. JOIN-FILTER
124. ANTI NA
125. ANTI SNA
126. ANTI NA BUFFERED
127. ANTI SNA BUFFERED
128. RIGHT ANTI NA
129. RIGHT ANTI SNA
130. RIGHT ANTI NA BUFFERED
131. RIGHT ANTI SNA BUFFERED
132. GROUP BY PIVOT
133. GROUP BY NOSORT PIVOT
134. DUMP
135. VERIFY
136. STRIP
137. STORAGE FULL
138. STORAGE SAMPLE
139. STORAGE SAMPLE BY ROWID RANGE
140. STORAGE BY ROWID RANGE
141. PARTIAL OUTER
142. OPTIMIZED
143. SPECIFIED
144. STORAGE SAMPLE FAST FULL SCAN
145. STORAGE FAST FULL SCAN
146. STORAGE FULL SCAN
147. STORAGE RANGE SCAN
148. STORAGE FULL SCAN (MIN/MAX)
149. WITH FILTERING (UNIQUE)
150. WITHOUT FILTERING (UNIQUE)
151. NO FILTERING WITH SW (UNIQUE)
152. BREADTH FIRST
153. DEPTH FIRST
154. AND

I will (slowly) start adding additional documentation about these operators and options here... Feel free to request which ones I should document first at the Oracle Living Book entry page