This is the mail archive of the
mailing list for the GCC project.
Re: [Bug c/18602] segfault on a huge switch statement.
- From: DJ Delorie <dj at redhat dot com>
- To: gcc-bugzilla at gcc dot gnu dot org
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Mon, 22 Nov 2004 16:36:17 -0500
- Subject: Re: [Bug c/18602] segfault on a huge switch statement.
- References: <firstname.lastname@example.org> <email@example.com>
> Confirmed, the problem is because of stack overflow. Either
> splay_tree_delete_helper needs a little help or the C/C++ front-end
> needs to stop using splay trees.
How about this?
/* Deallocate NODE (a member of SP), and all its sub-trees. */
splay_tree_delete_helper (sp, node)
splay_tree_node pending = 0;
splay_tree_node active = 0;
#define KDEL(x) if (sp->delete_key) (*sp->delete_key)(x);
#define VDEL(x) if (sp->delete_value) (*sp->delete_value)(x);
/* We use the "key" field to hold the "next" pointer. */
node->key = (splay_tree_key)pending;
pending = (splay_tree_node)node;
/* Now, keep processing the pending list until there aren't any
more. This is a little more complicated than just recursing, but
it doesn't toast the stack for large trees. */
active = pending;
pending = 0;
/* active points to a node which has its key and value
deallocated, we just need to process left and right. */
active->left->key = (splay_tree_key)pending;
pending = (splay_tree_node)(active->left);
active->right->key = (splay_tree_key)pending;
pending = (splay_tree_node)(active->right);
temp = active;
active = (splay_tree_node)(temp->key);
(*sp->deallocate) ((char*) temp, sp->allocate_data);