Re: [PATCH v5] mm/hugetlb_cma: Fix null nodemask dereference in hugetlb_cma_alloc_frozen_folio

[Muchun Song <muchun.song@linux.dev>]

> On Aug 10, 2026, at 04:11, Sourav Panda <souravpanda@google.com> wrote:
> 
> On Sat, Aug 8, 2026 at 11:58 PM Muchun Song <muchun.song@linux.dev> wrote:
>> 
>> 
>> 
>>> On Aug 9, 2026, at 12:32, Sourav Panda <souravpanda@google.com> wrote:
>>> 
>>> alloc_buddy_hugetlb_folio_with_mpol() can pass a NULL nodemask to
>>> alloc_fresh_hugetlb_folio() as a fallback to allocate from all
>>> nodes. If order is gigantic, alloc_fresh_hugetlb_folio() propagates
>>> the NULL nodemask down to hugetlb_cma_alloc_frozen_folio() via
>>> alloc_gigantic_frozen_folio().
>>> 
>>> Additionally, hugetlb_cma_alloc_frozen_folio() previously attempted
>>> allocation on hugetlb_cma[nid] without verifying if nid is included in
>>> the caller's nodemask. Adding a node_isset(nid, *nodemask) check ensures
>>> the initial preferred node allocation honors the memory policy / nodemask.
>>> 
>>> However, hugetlb_cma_alloc_frozen_folio() dereferences the nodemask in
>>> node_isset(nid, *nodemask) and for_each_node_mask(node, *nodemask),
>>> leading to a null pointer dereference kernel panic when nodemask is NULL.
>>> 
>>> Fix this by checking if nodemask is NULL in
>>> hugetlb_cma_alloc_frozen_folio() and defaulting it to
>>> cpuset_current_mems_allowed (safely read using a seqcount retry loop).
>>> This ensures that the initial node check and fallback loop safely honor
>>> the task's cpuset without violating cpuset constraints or causing NULL
>>> pointer dereferences.
>>> 
>>> From a userspace perspective, this bug allows an unprivileged user to
>>> crash the kernel (trigger a panic) by requesting a gigantic hugepage
>>> allocation with MPOL_PREFERRED_MANY on a system where CMA is only
>>> configured on a subset of NUMA nodes.
>>> 
>>> This can be reproduced by booting a VM with two NUMA nodes, restricting
>>> CMA to Node 1 (e.g., hugetlb_cma=1:1G default_hugepagesz=1G
>>> hugepagesz=1G hugepages=0), and running a program that allocates a
>>> 1GB hugepage area without reserving, restricts allocation to Node 0
>>> using mbind() with MPOL_PREFERRED_MANY, and triggers a page fault:
>>> 
>>> void *ptr = mmap(NULL, 1UL << 30, PROT_READ | PROT_WRITE,
>>>                  MAP_PRIVATE | MAP_ANONYMOUS | MAP_HUGETLB |
>>>                  MAP_HUGE_1GB | MAP_NORESERVE, -1, 0);
>>> unsigned long nodemask = 1; /* Node 0 */
>>> mbind(ptr, 1UL << 30, MPOL_PREFERRED_MANY, &nodemask,
>>>       sizeof(nodemask) * 8, 0);
>>> memset(ptr, 0, 1UL << 30); /* Trigger fault */
>>> 
>>> This results in a NULL pointer dereference:
>>> 
>>> BUG: kernel NULL pointer dereference, address: 0000000000000000
>>> #PF: supervisor read access in kernel mode
>>> #PF: error_code(0x0000) - not-present page
>>> Oops: Oops: 0000 [#1] SMP NOPTI
>>> RIP: 0010:hugetlb_cma_alloc_frozen_folio+0x75/0x120
>>> Call Trace:
>>>  <TASK>
>>>  only_alloc_fresh_hugetlb_folio.isra.0+0x2c/0x160
>>>  alloc_surplus_hugetlb_folio+0x6d/0x100
>>>  alloc_hugetlb_folio+0x3c5/0x660
>>>  hugetlb_no_page+0x3d9/0x650
>>> 
>>> Fixes: eb02f14c4a2b ("mm/hugetlb: allow overcommitting gigantic hugepages")
>>> Cc: stable@vger.kernel.org
>>> Signed-off-by: Sourav Panda <souravpanda@google.com>
>>> ---
>>> Changes in v5:
>>> - Replaced defaulting nodemask to &node_states[N_MEMORY] with safely reading
>>> cpuset_current_mems_allowed using a seqcount retry loop, ensuring fallback
>>> allocations comply with task hardwall cpusets as suggested by Usama Arif.
>>> - v4: https://lore.kernel.org/linux-mm/20260726072935.3513996-1-souravpanda@google.com/
>>> - v3: https://lore.kernel.org/linux-mm/20260705175119.440599-1-souravpanda@google.com/
>>> - v2: https://lore.kernel.org/linux-mm/20260704174930.2885785-1-souravpanda@google.com/
>>> - v1: https://lore.kernel.org/linux-mm/20260702215713.627941-1-souravpanda@google.com/
>>> 
>>> mm/hugetlb_cma.c | 15 ++++++++++++++-
>>> 1 file changed, 14 insertions(+), 1 deletion(-)
>>> 
>>> diff --git a/mm/hugetlb_cma.c b/mm/hugetlb_cma.c
>>> index 39344d6c78d8..3ae9347078e9 100644
>>> --- a/mm/hugetlb_cma.c
>>> +++ b/mm/hugetlb_cma.c
>>> @@ -3,6 +3,7 @@
>>> #include <linux/mm.h>
>>> #include <linux/cma.h>
>>> #include <linux/compiler.h>
>>> +#include <linux/cpuset.h>
>>> #include <linux/mm_inline.h>
>>> 
>>> #include <asm/page.h>
>>> @@ -30,11 +31,23 @@ struct folio *hugetlb_cma_alloc_frozen_folio(int order, gfp_t gfp_mask,
>>>      int node;
>>>      struct folio *folio;
>>>      struct page *page = NULL;
>>> +     nodemask_t local_node_mask;
>>> 
>>>      if (!hugetlb_cma_size)
>>>              return NULL;
>>> 
>>> -     if (hugetlb_cma[nid])
>>> +     if (!nodemask) {
>>> +             unsigned int cpuset_mems_cookie;
>>> +
>>> +             do {
>>> +                     cpuset_mems_cookie = read_mems_allowed_begin();
>>> +                     local_node_mask = cpuset_current_mems_allowed;
>> 
>> I think it only makes sense to move cma_alloc_frozen_compound() inside the
>> loop. Otherwise, during subsequent memory allocations, we could still face
>> unexpected allocation failures.
>> 
>> Muchun,
>> Thanks.
> 
> Hi Muchun!

Hi,

> 
> What do you think of the following diff? This way
> cma_alloc_frozen_compound is also protected from stale nodemasks.
> 
>         if (!hugetlb_cma_size)
>             return NULL;
> 
>    -    if (hugetlb_cma[nid])
>    +retry_cpuset:
>    +    if (!nodemask) {
>    +        cpuset_mems_cookie = read_mems_allowed_begin();
>    +        local_node_mask = cpuset_current_mems_allowed;
>    +        nmask = &local_node_mask;
>    +    } else {
>    +        nmask = nodemask;
>    +    }
>    +
>    +    if (hugetlb_cma[nid] && node_isset(nid, *nmask))
>             page = cma_alloc_frozen_compound(hugetlb_cma[nid], order);
> 
>         if (!page && !(gfp_mask & __GFP_THISNODE)) {
>    -        for_each_node_mask(node, *nodemask) {
>    +        for_each_node_mask(node, *nmask) {
>                 if (node == nid || !hugetlb_cma[node])
>                     continue;
> 
>    @@ -48,6 +61,9 @@ struct folio *hugetlb_cma_alloc_frozen_folio(int
> order, gfp_t gfp_mask,
>             }
>         }
> 
>    +    if (!page && !nodemask &&
> unlikely(read_mems_allowed_retry(cpuset_mems_cookie)))
>    +        goto retry_cpuset;
>    +
>         if (!page)
>             return NULL;

The overall looks good to me. But can we change it to this here?

	if (!page) {
		if (!nodemask &&
		    unlikely(read_mems_allowed_retry(cpuset_mems_cookie)))
			goto retry_cpuset;
		return NULL;
	}

That way, we can remove the 'if (!page)' check here.

Muchun,
Thanks.

> 
> If this looks good to you, I'll send out v6 with this change.
> 
> With regards,
> Sourav Panda
> 
>> 
>>> +             } while (read_mems_allowed_retry(cpuset_mems_cookie));
>>> +
>>> +             nodemask = &local_node_mask;
>>> +     }
>>> +
>>> +     if (hugetlb_cma[nid] && node_isset(nid, *nodemask))
>>>              page = cma_alloc_frozen_compound(hugetlb_cma[nid], order);
>>> 
>>>      if (!page && !(gfp_mask & __GFP_THISNODE)) {
>>> --
>>> 2.55.0