diff mbox series

add kernel log when allocate a new virtual mapping failed because of overcommit

Message ID 20210630074433.41992-1-xuhuijie2@huawei.com (mailing list archive)
State New
Headers show
Series add kernel log when allocate a new virtual mapping failed because of overcommit | expand

Commit Message

Xu Huijie June 30, 2021, 7:44 a.m. UTC
From: xuhuijie <xuhuijie2@huawei.com>

Comments

Matthew Wilcox June 30, 2021, 1:29 p.m. UTC | #1
On Wed, Jun 30, 2021 at 03:44:33PM +0800, Xu Huijie wrote:
> From: xuhuijie <xuhuijie2@huawei.com>
> 
> diff --git a/mm/mmap.c b/mm/mmap.c
> index aa9de981b..f019fc6b2 100644
> --- a/mm/mmap.c
> +++ b/mm/mmap.c
> @@ -1758,8 +1758,10 @@ unsigned long mmap_region(struct file *file, unsigned long addr,
>  	 */
>  	if (accountable_mapping(file, vm_flags)) {
>  		charged = len >> PAGE_SHIFT;
> -		if (security_vm_enough_memory_mm(mm, charged))
> +		if (security_vm_enough_memory_mm(mm, charged)) {
> +			pr_err("Unable to allocate a new virtual mapping because of the overcommit\n");

No, we don't let unprivileged processes flood the kernel logs.  If this
is genuinely useful, something like pr_info_ratelimited() might be
appropriate.  It certainly doesn't rise to the level of "error".
diff mbox series

Patch

diff --git a/mm/mmap.c b/mm/mmap.c
index aa9de981b..f019fc6b2 100644
--- a/mm/mmap.c
+++ b/mm/mmap.c
@@ -1758,8 +1758,10 @@  unsigned long mmap_region(struct file *file, unsigned long addr,
 	 */
 	if (accountable_mapping(file, vm_flags)) {
 		charged = len >> PAGE_SHIFT;
-		if (security_vm_enough_memory_mm(mm, charged))
+		if (security_vm_enough_memory_mm(mm, charged)) {
+			pr_err("Unable to allocate a new virtual mapping because of the overcommit\n");
 			return -ENOMEM;
+		}
 		vm_flags |= VM_ACCOUNT;
 	}