Regardless of anything else, won't this break booting with initcall_debug on PPC64/IA64 machines? (see the definition of print_fn_descriptor_symbol() in kallsyms.h)
Correct, thanks for pointing that out. The code below is probably better:
static void __init do_initcalls(void)
{
initcall_t *call;
@@ -547,6 +558,9 @@ static void __init do_initcalls(void)
for (call = __initcall_start; call < __initcall_end; call++) {
char *msg;
+ if (!*call) + continue; + if (initcall_debug) { printk(KERN_DEBUG "Calling initcall 0x%p", *call); print_fn_descriptor_symbol(": %s()", (unsigned long) *call);
Yes, that looks more sensible. It hides the fact that the initcall ever existed, rather than explicitly telling you that it's been skipped, but I don't imagine that that's ever going to cause a problem in practice (i.e., I don't think anyone would ever enable "force_ohci1394=off" by mistake and also without noticing).
And I guess the idea of replacing the initcall pointer with NULL will work both with and without function descriptors, right? So we should be safe on IA64 and PPC64.
I think so, though I don't really know a great deal about this area.
An IA64 descriptor is of the form { &code, &data_context }, and a function pointer is a pointer to such a descriptor. Presumably, setting a function pointer to NULL will either end up setting the pointer-to-descriptor to NULL or the code pointer to NULL, but either way, I would expect the 'if (!*call)' comparison to work as intended.
Best thing would be to get someone on IA64 and/or PPC64 to check this for you. Also might be worth checking that the patch works as intended with CONFIG_MODULES=n (assuming you haven't already).
Regards, Malcolm - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
