[PATCH] mm: kfence: fix handling discontiguous page

[Muchun Song]

The struct pages could be discontiguous when the kfence pool is allocated
via alloc_contig_pages() with CONFIG_SPARSEMEM and !CONFIG_SPARSEMEM_VMEMMAP.
So, the iteration should use nth_page().

Fixes: 0ce20dd84089 ("mm: add Kernel Electric-Fence infrastructure")
Signed-off-by: Muchun Song <songm...@bytedance.com>
---
mm/kfence/core.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/mm/kfence/core.c b/mm/kfence/core.c
index d66092dd187c..1065e0568d05 100644
--- a/mm/kfence/core.c
+++ b/mm/kfence/core.c
@@ -556,7 +556,7 @@ static unsigned long kfence_init_pool(void)
* enters __slab_free() slow-path.
*/
for (i = 0; i < KFENCE_POOL_SIZE / PAGE_SIZE; i++) {
- struct slab *slab = page_slab(&pages[i]);
+ struct slab *slab = page_slab(nth_page(pages, i));

if (!i || (i % 2))
continue;
@@ -602,7 +602,7 @@ static unsigned long kfence_init_pool(void)

reset_slab:
for (i = 0; i < KFENCE_POOL_SIZE / PAGE_SIZE; i++) {
- struct slab *slab = page_slab(&pages[i]);
+ struct slab *slab = page_slab(nth_page(pages, i));

if (!i || (i % 2))
continue;
--
2.11.0

[Marco Elver]

On Thu, 23 Mar 2023 at 03:50, 'Muchun Song' via kasan-dev
<kasa...@googlegroups.com> wrote:
>
> The struct pages could be discontiguous when the kfence pool is allocated
> via alloc_contig_pages() with CONFIG_SPARSEMEM and !CONFIG_SPARSEMEM_VMEMMAP.
> So, the iteration should use nth_page().
>
> Fixes: 0ce20dd84089 ("mm: add Kernel Electric-Fence infrastructure")
> Signed-off-by: Muchun Song <songm...@bytedance.com>

Reviewed-by: Marco Elver <el...@google.com>

> ---
> mm/kfence/core.c | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/mm/kfence/core.c b/mm/kfence/core.c
> index d66092dd187c..1065e0568d05 100644
> --- a/mm/kfence/core.c
> +++ b/mm/kfence/core.c
> @@ -556,7 +556,7 @@ static unsigned long kfence_init_pool(void)
> * enters __slab_free() slow-path.
> */
> for (i = 0; i < KFENCE_POOL_SIZE / PAGE_SIZE; i++) {
> - struct slab *slab = page_slab(&pages[i]);
> + struct slab *slab = page_slab(nth_page(pages, i));
>
> if (!i || (i % 2))
> continue;
> @@ -602,7 +602,7 @@ static unsigned long kfence_init_pool(void)
>
> reset_slab:
> for (i = 0; i < KFENCE_POOL_SIZE / PAGE_SIZE; i++) {
> - struct slab *slab = page_slab(&pages[i]);
> + struct slab *slab = page_slab(nth_page(pages, i));
>
> if (!i || (i % 2))
> continue;
> --
> 2.11.0
>

> --
> You received this message because you are subscribed to the Google Groups "kasan-dev" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to kasan-dev+...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/kasan-dev/20230323025003.94447-1-songmuchun%40bytedance.com.

[Andrew Morton]

On Thu, 23 Mar 2023 10:50:03 +0800 Muchun Song <songm...@bytedance.com> wrote:

> The struct pages could be discontiguous when the kfence pool is allocated
> via alloc_contig_pages() with CONFIG_SPARSEMEM and !CONFIG_SPARSEMEM_VMEMMAP.
> So, the iteration should use nth_page().

What are the user-visible runtime effects of this flaw?

Thanks.

[Muchun Song]

Set the PG_slab and memcg_data to a arbitrary address (may be not used as a struct
page), so the worst case may corrupt the kernel.

Thanks.

>
> Thanks.

[Kefeng Wang]

On 2023/3/23 10:50, Muchun Song wrote:
> The struct pages could be discontiguous when the kfence pool is allocated
> via alloc_contig_pages() with CONFIG_SPARSEMEM and !CONFIG_SPARSEMEM_VMEMMAP.
> So, the iteration should use nth_page().
>

Reviewed-by: Kefeng Wang <wangkef...@huawei.com>