Re: [PATCH 01/10] IOCHK interface for I/O error handling/detecting

[Hidetoshi Seto <seto.hidetoshi@jp.fujitsu.com>]

Hi Greg,

Thank you for giving me many useful advices!

Greg KH wrote:
> On Thu, Jun 09, 2005 at 09:48:15PM +0900, Hidetoshi Seto wrote:
> 
>>+void iochk_init(void) { ; }
>>+
>>+void iochk_clear(iocookie *cookie, struct pci_dev *dev)
>>+{
>>+	/* no-ops */
>>+}
> 
> A bit of a coding style difference between the two functions, yet they
> do the same thing :)

I intended to emphasize the pair. I'll unify them if not needed.

>>+int iochk_read(iocookie *cookie)
>>+{
>>+	/* no-ops */
>>+	return 0;
>>+}
> 
> Why not just return the cookie?  Can this ever fail?

In this time, no one initializes the cookie, so I just ignored it.

> Shouldn't these go into a .h file and be made "static inline" so they
> just compile away to nothing?

I'm not used to inlining...
In case of generic definition above, absolutely it should be inlined.
OK, I'll try.

>>+EXPORT_SYMBOL(iochk_clear);
>>+EXPORT_SYMBOL(iochk_read);
> 
> EXPORT_SYMBOL_GPL() perhaps?

Yea.

>>+#ifndef HAVE_ARCH_IOMAP_CHECK
>>+typedef unsigned long iocookie;
>>+#endif
> 
> Why typedef this if it isn't specified?

Because I stuck to have short name alias, and wanted to hide even
whether it is struct or not.

Thanks,
H.Seto