1. CS 36 – February 2 Control structures –Sequence √ –Loops √ –Selection If-statement Case-statement Finding loops

2. Variants of “if” Fortran: if goto More “structured” version: if then else –the “else” arm is optional –How do we recognize the end of the ? –body is single stmt or a block of statements

3. Ambiguous “else” What should this mean: if c1 then if c2 then s1 else s2 ??? Using { } blocks makes meaning clear: if c1 then { if c2 then s1 } else s2 OR if c1 then { if c2 then s1 else s2 } Language should provide meaning in case we omit the { }.

4. Case/switch statement Purpose: for multiple choices, less tedious than a long string of if-else’s First appeared in Algol Typical rules –Case values are constant –Case values are unique –Cases are integral type

5. Pascal example case month of 1,3,5,7,8,10,12: days := 31; 4,6,9,11: days := 30; 2 : if year mod 4 = 0 then days := 29 else days := 28 end;

6. Case variants Some languages allow a default case Example from Ada: case ch is when ‘0’.. ‘9’ => put_line(“digit”); when ‘a’.. ‘z’ => put_line(“letter”); when others => put_line(“special character”); end case; –Avoids a runtime error or mysterious bug Ada’s case statement allows range!

7. C, C++, Java Case statement called “switch” Break statement is required, or else we fall into the next case! switch(month) { case 4: case 6: case 9: case 11: days = 30; break; case 2: days = (y%4==0)? 29 : 28; break; default: days = 31; } Having to say “case” for each value is annoying 

8. Dijkstra’s “Guarded” stmt Functions like a non-deterministic case statement: if x >= y  max := x [] y >= x  max := y fi Evaluate all the conditions. If more than 1 is true, pick any and do corresponding statement. Very few languages use non-determinism (e.g. Haskell). Hard to verify correctness.

9. Finding loops For many optimizations, compiler needs to know where loops are. Can’t rely on “for,” “while” because they are gone by the time we generate code. Procedure –Represent control-flow as a graph. –Focus on blocks of code instead of tiny instructions so not bogged down by unnecessary detail.

10. Basic block In assembly code, a basic block is a sequence of instructions either beginning with a label or ending with a branch/jump. Let’s assume the compiler has already figured out where the blocks are (CS 25). A typical function may have 50 instructions, but only 10 blocks. Easier to analyze.

11. Where are the loops? Not hard for us But control-flow data just has sequence of blocks Need to gather info about transitions between blocks. 1 2 3 4 5 6 7

12. Finding loops For each block, determine –Successors Where can I go immediately after this block? –Predecessors Where could I have just come from? –Dominators Where must I have been, to reach here?

13. Example 1 2 3 4 5 6 7 blockpredsuccdom 1-2 21,63 324,6 43,55 544,6 63,52,7 76-

14. Example 1 2 3 4 5 6 7 blockpredsuccdom 1-2 1 21,63 1,2 324,6 1,2,3 43,55 1,2,3,4 544,6 1,2,3,4,5 63,52,7 1,2,3,6 76- 1,2,3,6,7

15. Aha! A loop We have a loop whenever a block can say: “One of my successors is also one of my dominators.” In other words, I’m going to a place I’ve already been. Hence a back edge, and a loop.