> On Tue, Mar 8, 2011 at 1:47 PM, KOSAKI Motohiro > <kosaki.motohiro@xxxxxxxxxxxxxx> wrote: > >> > > +#ifdef CONFIG_SMP > >> > > +static DEFINE_PER_CPU(struct pagevec, activate_page_pvecs); > >> > > >> > Why do we have to handle SMP and !SMP? > >> > We have been not separated in case of pagevec using in swap.c. > >> > If you have a special reason, please write it down. > >> this is to reduce memory footprint as suggested by akpm. > >> > >> Thanks, > >> Shaohua > > > > Hi Shaouhua, > > > > I agree with you. But, please please avoid full quote. I don't think > > it is so much difficult work. ;-) > > I didn't want to add new comment in the code but want to know why we > have to care of activate_page_pvecs specially. I think it's not a > matter of difficult work or easy work. If new thing is different with > existing things, at least some comment in description makes review > easy. > > If it's memory footprint issue, should we care of other pagevec to > reduce memory footprint in non-smp? If it is, it would be a TODO list > for consistency and memory footprint. Yeah. indeed. Shaoua, If my remember is correct, your previous version has code size comparision result. could you resurrect it? -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@xxxxxxxxxx For more info on Linux MM, see: http://www.linux-mm.org/ . Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/ Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>