Messages in this thread | | | Date | Fri, 16 Oct 2009 00:04:14 +0100 (BST) | From | Hugh Dickins <> | Subject | Re: [PATCH 2/9] swap_info: change to array of pointers |
| |
On Thu, 15 Oct 2009, Hugh Dickins wrote: > On Thu, 15 Oct 2009, KAMEZAWA Hiroyuki wrote: > > On Thu, 15 Oct 2009 01:48:01 +0100 (BST) > > Hugh Dickins <hugh.dickins@tiscali.co.uk> wrote: > > > @@ -1675,11 +1674,13 @@ static void *swap_start(struct seq_file > > > if (!l) > > > return SEQ_START_TOKEN; > > > > > > - for (i = 0; i < nr_swapfiles; i++, ptr++) { > > > - if (!(ptr->flags & SWP_USED) || !ptr->swap_map) > > > + for (type = 0; type < nr_swapfiles; type++) { > > > + smp_rmb(); /* read nr_swapfiles before swap_info[type] */ > > > + si = swap_info[type]; > > > > if (!si) ?
Re-reading, I see that I missed your interjection there.
Precisely because we read swap_info[type] after reading nr_swapfiles, with smp_rmb() here to enforce that, and smp_wmb() where they're set in swapon, there is no way for si to be seen as NULL here. Is there?
Or are you asking for a further comment here on why that's so? I think I'd rather just switch to taking swap_lock in swap_start() and swap_next(), than be adding comments on why we don't need it.
Hugh
| |