MOD_INC_USE_COUNT usage

With the high risk of being flamed, I send this to LKML. Feel free to tell me not to if this is a huge no-no, and it won't happen again.

I was studying some of the warnings that appears when compiling 2.6.1-mm3, and one common warning is the MOD_INC_USE_COUNT macro being used. Apparently it's deprecated, but if that's the case, why is it still being used in the kernel?

I googled a bit and found a few discussions about the issue (mostly concerning the /mm-subsystem) where some people found it okay to just remove the call to MOD_INC_USE_COUNT totally, since it is totally unnecessary under 2.5 (and under 2.6 too, I presume). Since the discussions provided a lot of contradictions and no solution, I thought asking a question to someone who has an idea of what this is about might be a good thing.

I'm not asking you to remove this call, but I do want to know why it shouldn't be removed and what is being done to replace a function (or macro... whatever it is) that is deprecated. Any explanations would be gratefully accepted. Thank you.

Since I'm no kernel-hacker (and not much of a coder at all) don't expect me to understand a lot of jargon, but this question is being asked because I'm really interested in understanding the Linux kernel.

So, why shouldn't this patch be applied?:

@@ -121,6 +121,7 @@ static int __devinit generic_init_one(st

		return 1;
	}
	ide_setup_pci_device(dev, d);
+	MOD_INC_USE_COUNT;
	return 0;

}

Tim Cambrant

Hello,

As I can see, this is really redundant to have a MOD_INC_USE_COUNT for almost all the modules. But this should be checked carefully before removing ;)

umka