I ran into a bug in sparse when I added a label for a goto. I changed
the code I was planning to add to avoid it (by adding a continue at the
bottom of the loop) but it shows that there is a bug in parsing goto
labels. The sparse version is current (at least matches the one that
was in sparse.bkbits.net). The sparse tool generated the following:
test.c:15:2: warning: Expected ; at end of statement
test.c:15:2: warning: got }
builtin:0:0: warning: Expected } at end of compound statement
builtin:0:0: warning: got end-of-input
builtin:0:0: warning: Expected } at end of function
builtin:0:0: warning: got end-of-input
You can see it from the following tests which differ in only one line
(the 1st generates the warning, the second does not).
void some_junk_function_with_sparse_error(void)
{
int i;
int j = 0;
while(j!=2) {
for(i=0;i<10;i++) {
if(i==9) {
j = 1;
goto out;
}
}
j = 2;
out:
}
}
The error can be avoided by adding a superflous continue statement e.g.
void some_junk_function_without_sparse_error(void)
{
int i;
int j = 0;
while(j!=2) {
for(i=0;i<10;i++) {
if(i==9) {
j = 1;
goto out;
}
}
j = 2;
out:
continue;
}
}
which is ugly. As an aside, I hate gotos, but the kind of case I was
dealing with was someting C did not provide - ie a way to use break or
continue as if we were in the outer while loop while inside the inner
for loop. It can be coded differently, but it ends up with longer code
- and while I was debating which was less ugly (to use goto or not), I
hit the sparse bug which the junk code above demonstrates.
-
To unsubscribe from this list: send the line "unsubscribe linux-sparse" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
