Re: [PATCH] IB/core: Print error when umem fails due to locked memory limit.

[Doug Ledford]

On 10/15/2015 10:21 AM, j.glisse@xxxxxxxxx wrote:
> From: JÃrÃme Glisse <jglisse@xxxxxxxxxx>
> 
> It can be rather tedious to find why userspace is failing when only
> thing kernel report is -ENOMEM. This add an error message so that
> user can figure out why they are getting -ENOMEM.
> 
> Signed-off-by: JÃrÃme Glisse <jglisse@xxxxxxxxxx>
> cc: <linux-rdma@xxxxxxxxxxxxxxx>
> Cc: Haggai Eran <haggaie@xxxxxxxxxxxx>
> Cc: Sagi Grimberg <sagig@xxxxxxxxxxxx>
> Cc: Shachar Raindel <raindel@xxxxxxxxxxxx>
> Cc: Doug Ledford <dledford@xxxxxxxxxx>
> ---
>  drivers/infiniband/core/umem.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/drivers/infiniband/core/umem.c b/drivers/infiniband/core/umem.c
> index 38acb3c..a66929e 100644
> --- a/drivers/infiniband/core/umem.c
> +++ b/drivers/infiniband/core/umem.c
> @@ -169,6 +169,7 @@ struct ib_umem *ib_umem_get(struct ib_ucontext *context, unsigned long addr,
>  	lock_limit = rlimit(RLIMIT_MEMLOCK) >> PAGE_SHIFT;
>  
>  	if ((locked > lock_limit) && !capable(CAP_IPC_LOCK)) {
> +		pr_err("locked memory quota exhausted (see ulimit -l)\n");
>  		ret = -ENOMEM;
>  		goto out;
>  	}
> 

This looks like it could easily cause a flood of messages (possibly even
enough to DoS the machine).  Please convert this to some sort of rate
limited output (preferably, you would issue this warning only once per
task, and the error would include the command name).

-- 
Doug Ledford <dledford@xxxxxxxxxx>
              GPG KeyID: 0E572FDD

Attachment: signature.asc
Description: OpenPGP digital signature