Why `do { ...; exit(...); } while (0)` in C? [duplicate]

Question

This question already has answers here:

Why use apparently meaningless do-while and if-else statements in macros? (9 answers)

how does do{} while(0) work in macro? (3 answers)

Closed 12 years ago.

As a C newbie I'm having trouble understanding the following code:

#define errExit(msg) do { perror(msg); exit(EXIT_FAILURE); \
 } while (0)

I gathered that the reason this function is #defined is to override an existing function, but what is the point of a do ... while(0) loop with an unconditional exit() statement? Is it not possible to write this without the loop construct?

• Kiril Kirov: Nice question, I've wondered the same! +1
• l0b0: Too bad the title duplicate detector didn't pick up this...

Answer 1

Many duplicates here I think.

The do...while(0) trick enables you to use errExit in various contexts without breaking anything:

if(x) errExit(msg);
else return 1;

is translated to:

if(x) do { ...; ...; } while(0);
else return 1;

If you omit the do...while(0) part, then you could not reliably add a semicolon for example.

• Joe: Why would plain old braces not do the job?
• Kiril Kirov: Nice! +1. That's really clever and tricky.
• l0b0: Ah, so the whole idea is to keep the resulting code to a single statement as seen from the surrounding code? Neat!
• asheeshr: @Joe Read the complete answer here stackoverflow.com/a/154138/1488917
• Benoit: @Joe: No. Because of semicolon. As said by AshRj the answer provided elsewhere is thorough.

Answer 2

Assume the macro didn't have the do { ... } while(0) loop, just the 2 statements inside. Now, what if I were to write

if( foo() )
 errExit("foo!" );

My conditional exit has become a non-conditional exit.

Answer 3

The do { ... } while(0) construct is common, and usually considered best practice, for macro functions of multiple statements, such as this. It allows use as a single statement, so there are no surprises.