Quoting Nathan Lynch (ntl@xxxxxxxxx):
> Similar to our handling of fds that have been subject to F_SETOWN,
> detect when an fd has had its f_owner->signum changed from the
> default.
>
> Signed-off-by: Nathan Lynch <ntl@xxxxxxxxx>
Hey Nathan,
Can you give more motivation for this? Do you just feel that it
isn't worth the risk of mis-coding the check at restart?
For safety check, what about forcing such a task to be restarted
in a private pidns?
I'm not nacking it, don't mind it going in temporarily, but this
commit message makes it sound like using F_SETSIG is an application
error.
thanks,
-serge
> ---
> fs/checkpoint.c | 8 ++++++++
> 1 files changed, 8 insertions(+), 0 deletions(-)
>
> diff --git a/fs/checkpoint.c b/fs/checkpoint.c
> index fd539c5..bf4d2d4 100644
> --- a/fs/checkpoint.c
> +++ b/fs/checkpoint.c
> @@ -265,6 +265,7 @@ static int checkpoint_file_desc(struct ckpt_ctx *ctx,
> struct file *file = NULL;
> struct fdtable *fdt;
> int objref, ret;
> + int signum;
> int coe = 0; /* avoid gcc warning */
> pid_t pid;
>
> @@ -311,6 +312,13 @@ static int checkpoint_file_desc(struct ckpt_ctx *ctx,
> goto out;
> }
>
> + signum = ACCESS_ONCE(file->f_owner.signum);
> + if (signum != 0) {
> + ret = -EBUSY;
> + ckpt_err(ctx, ret, "%(T)fd %d has a signal set (%d)\n", fd, signum);
> + goto out;
> + }
> +
> /*
> * if seen first time, this will add 'file' to the objhash, keep
> * a reference to it, dump its state while at it.
> --
> 1.7.4.4
>
> _______________________________________________
> 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
