Re: [lwip-users] possible bug in sys_timeout()

[Kieran Mansley]

On Mon, 16 Feb 2004, John Taylor wrote:
> There appears to be bug in the sys_timeout() code at line 40 below.  Is
> it pilot error on my part or a real bug?
>
> To exit the for-loop at line 32, the if statement at line 34 must
> evaluate to true.  When the IF statement is true then "t->next == NULL"
> or "t->next->time > timeout->time" is true. BUT if "t->next == NULL"
> then I should crash at line 40.

Looks (tentatively) OK to me.  Are you seeing a crash, or just think it
should be?

If t->next == NULL then it will only do lines 39-41:

39:             timeout->next = t->next;
40:             t->next = timeout;
41:             break;

Which will become:

39:             timeout->next = NULL
40:             t->next = timeout;
41:             break;

The only things we dereference there are timeout and t, which we know are
not NULL.  We don't dereference t->next, so the fact that it points to
NULL shouldn't matter.

So, I guess I'm saying that although the code is a bit cryptic, I think
it's OK!

Kieran