Messages in this thread | | | Date | Sat, 24 Jan 2015 21:56:45 +0000 | From | Bryan O'Donoghue <> | Subject | Re: [PATCH v2 1/1] x86: Add Isolated Memory Regions for Quark X1000 |
| |
On 24/01/15 11:02, Andy Shevchenko wrote: > On Sat, Jan 24, 2015 at 3:48 AM, Ong, Boon Leong > <boon.leong.ong@intel.com> wrote: > >>> +static int imr_enabled(struct imr_regs *imr) >> Do we want to make it inline perhaps since it is 1 liner? > > Since it is declared static I would even suggest the new name is_imr_enabled().
I think imr_is_enabled() is a better name. Every other function is imr_ prefixed.
>>> + >>> + if ((imr_to_phys(imr.addr_lo) == base) && >>> + (imr_to_phys(imr.addr_hi) == max)) { >> I think we need to take care of the size that has been fix-up here ... >> >>> + found = 1; >>> + reg = i; > > According to your comment this line becomes redundant. >
Ah but I still think though.
We eventually do an imr_write(reg, &imr); which I think reads better than imr_write(i, &imr);
-- BOD
| |