Quoting Matt Helsley (matthltc@xxxxxxxxxx):
> This completes the work necessary to make checkpoint/restart of
> thousands of epoll items more reliable when higher order kmallocs
> would fail. We grab a piece of memory suitable to store a "chunk"
> of items for input. Read the input one chunk at a time and add
> epoll items for each item in the chunk.
>
> Signed-off-by: Matt Helsley <matthltc@xxxxxxxxxx>
Acked-by: Serge Hallyn <serue@xxxxxxxxxx>
> ---
> fs/eventpoll.c | 58 +++++++++++++++++++++++++++++++++++--------------------
> 1 files changed, 37 insertions(+), 21 deletions(-)
>
> diff --git a/fs/eventpoll.c b/fs/eventpoll.c
> index 2506b40..c261263 100644
> --- a/fs/eventpoll.c
> +++ b/fs/eventpoll.c
> @@ -1594,7 +1594,7 @@ static int ep_items_restore(void *data)
> struct ckpt_eventpoll_item *items = NULL;
> struct eventpoll *ep;
> struct file *epfile = NULL;
> - int ret, num_items, i = 0;
> + int ret, num_items, nchunk;
>
> h = ckpt_read_obj_type(ctx, sizeof(*h), CKPT_HDR_EPOLL_ITEMS);
> if (IS_ERR(h))
> @@ -1611,34 +1611,50 @@ static int ep_items_restore(void *data)
> if (!num_items)
> return 0;
>
> - ret = ckpt_read_payload(ctx, (void**)&items, num_items*sizeof(*items),
> - CKPT_HDR_BUFFER);
> + ret = _ckpt_read_obj_type(ctx, NULL, 0, CKPT_HDR_BUFFER);
> + if (ret < 0)
> + return ret;
> + /* Make sure the items match the size we expect */
> + if (num_items != (ret / sizeof(*items)))
> + return -EINVAL;
> +
> + nchunk = num_items;
> + do {
> + items = kzalloc(sizeof(*items)*nchunk, GFP_KERNEL);
> + if (items)
> + break;
> + nchunk = nchunk/2;
> + } while (nchunk > 0);
> if (!items)
> return -ENOMEM;
>
> - /* Make sure the items match the size we expect */
> - if (num_items != (ret / sizeof(*items))) {
> - ret = -EINVAL;
> - goto out;
> - }
> -
> ep = epfile->private_data;
>
> - /* Restore the epoll items/watches */
> - for (ret = 0, i = 0; !ret && i < num_items; i++) {
> - struct epoll_event epev;
> - struct file *tfile;
> + while (num_items > 0) {
> + int n = min(num_items, nchunk);
> + int j;
>
> - tfile = ckpt_obj_fetch(ctx, items[i].file_objref,
> - CKPT_OBJ_FILE);
> - if (IS_ERR(tfile)) {
> - ret = PTR_ERR(tfile);
> + ret = ckpt_kread(ctx, items, n*sizeof(*items));
> + if (ret < 0)
> break;
> +
> + /* Restore the epoll items/watches */
> + for (j = 0; !ret && j < n; j++) {
> + struct epoll_event epev;
> + struct file *tfile;
> +
> + tfile = ckpt_obj_fetch(ctx, items[j].file_objref,
> + CKPT_OBJ_FILE);
> + if (IS_ERR(tfile)) {
> + ret = PTR_ERR(tfile);
> + goto out;
> + }
> + epev.events = items[j].events;
> + epev.data = items[j].data;
> + ret = do_epoll_ctl(EPOLL_CTL_ADD, items[j].fd,
> + epfile, tfile, &epev);
> }
> - epev.events = items[i].events;
> - epev.data = items[i].data;
> - ret = do_epoll_ctl(EPOLL_CTL_ADD, items[i].fd,
> - epfile, tfile, &epev);
> + num_items -= n;
> }
> out:
> kfree(items);
> --
> 1.5.6.3
>
> _______________________________________________
> Containers mailing list
> Containers@xxxxxxxxxxxxxxxxxxxxxxxxxx
> https://lists.linux-foundation.org/mailman/listinfo/containers
_______________________________________________
Containers mailing list
Containers@xxxxxxxxxxxxxxxxxxxxxxxxxx
https://lists.linux-foundation.org/mailman/listinfo/containers
