Add a sanity check to madvise_dontneed_free() to address a corner case in madvise where a race condition causes the current vma being processed to be backed by a different page size.
During a madvise(MADV_DONTNEED) call on a memory region registered with a userfaultfd, there's a period of time where the process mm lock is temporarily released in order to send a UFFD_EVENT_REMOVE and let userspace handle the event. During this time, the vma covering the current address range may change due to an explicit mmap done concurrently by another thread.
If, after that change, the memory region, which was originally backed by 4KB pages, is now backed by hugepages, the end address is rounded down to a hugepage boundary to avoid data loss (see "Fixes" below). This rounding may cause the end address to be truncated to the same address as the start.
Make this corner case follow the same semantics as in other similar cases where the requested region has zero length (ie. return 0).
This will make madvise_walk_vmas() continue to the next vma in the range (this time holding the process mm lock) which, due to the prev pointer becoming stale because of the vma change, will be the same hugepage-backed vma that was just checked before. The next time madvise_dontneed_free() runs for this vma, if the start address isn't aligned to a hugepage boundary, it'll return -EINVAL, which is also in line with the madvise api.
From userspace perspective, madvise() will return EINVAL because the start address isn't aligned according to the new vma alignment requirements (hugepage), even though it was correctly page-aligned when the call was issued.
Fixes: 8ebe0a5eaaeb ("mm,madvise,hugetlb: fix unexpected data loss with MADV_DONTNEED on hugetlbfs") Cc: stable@vger.kernel.org Signed-off-by: Ricardo Cañuelo Navarro rcn@igalia.com --- mm/madvise.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/mm/madvise.c b/mm/madvise.c index 49f3a75046f6..c8e28d51978a 100644 --- a/mm/madvise.c +++ b/mm/madvise.c @@ -933,7 +933,9 @@ static long madvise_dontneed_free(struct vm_area_struct *vma, */ end = vma->vm_end; } - VM_WARN_ON(start >= end); + if (start == end) + return 0; + VM_WARN_ON(start > end); }
if (behavior == MADV_DONTNEED || behavior == MADV_DONTNEED_LOCKED)
Hi all,
Some more context about the patch. The issue (WARNING in madvise_vma_behavior) was found by a private syzbot instance, so I can't share the link, but it can be triggered by an unprivileged user with this reproducer:
--8<------------------------------------------------------------------
#define _GNU_SOURCE
#include <endian.h> #include <errno.h> #include <fcntl.h> #include <pthread.h> #include <stdint.h> #include <stdio.h> #include <stdlib.h> #include <string.h> #include <sys/mount.h> #include <sys/stat.h> #include <sys/syscall.h> #include <sys/types.h> #include <time.h> #include <unistd.h>
#include <linux/futex.h>
#ifndef __NR_userfaultfd #define __NR_userfaultfd 323 #endif
static void sleep_ms(uint64_t ms) { usleep(ms * 1000); }
static uint64_t current_time_ms(void) { struct timespec ts; if (clock_gettime(CLOCK_MONOTONIC, &ts)) exit(1); return (uint64_t)ts.tv_sec * 1000 + (uint64_t)ts.tv_nsec / 1000000; }
static void thread_start(void* (*fn)(void*), void* arg) { pthread_t th; pthread_attr_t attr; pthread_attr_init(&attr); pthread_attr_setstacksize(&attr, 128 << 10); int i = 0; for (; i < 100; i++) { if (pthread_create(&th, &attr, fn, arg) == 0) { pthread_attr_destroy(&attr); return; } if (errno == EAGAIN) { usleep(50); continue; } break; } exit(1); }
typedef struct { int state; } event_t;
static void event_init(event_t* ev) { ev->state = 0; }
static void event_reset(event_t* ev) { ev->state = 0; }
static void event_set(event_t* ev) { if (ev->state) exit(1); __atomic_store_n(&ev->state, 1, __ATOMIC_RELEASE); syscall(SYS_futex, &ev->state, FUTEX_WAKE | FUTEX_PRIVATE_FLAG, 1000000); }
static void event_wait(event_t* ev) { while (!__atomic_load_n(&ev->state, __ATOMIC_ACQUIRE)) syscall(SYS_futex, &ev->state, FUTEX_WAIT | FUTEX_PRIVATE_FLAG, 0, 0); }
static int event_isset(event_t* ev) { return __atomic_load_n(&ev->state, __ATOMIC_ACQUIRE); }
static int event_timedwait(event_t* ev, uint64_t timeout) { uint64_t start = current_time_ms(); uint64_t now = start; for (;;) { uint64_t remain = timeout - (now - start); struct timespec ts; ts.tv_sec = remain / 1000; ts.tv_nsec = (remain % 1000) * 1000 * 1000; syscall(SYS_futex, &ev->state, FUTEX_WAIT | FUTEX_PRIVATE_FLAG, 0, &ts); if (__atomic_load_n(&ev->state, __ATOMIC_ACQUIRE)) return 1; now = current_time_ms(); if (now - start > timeout) return 0; } }
struct thread_t { int created, call; event_t ready, done; };
static struct thread_t threads[16]; static void execute_call(int call); static int running;
static void* thr(void* arg) { struct thread_t* th = (struct thread_t*)arg; for (;;) { event_wait(&th->ready); event_reset(&th->ready); execute_call(th->call); __atomic_fetch_sub(&running, 1, __ATOMIC_RELAXED); event_set(&th->done); } return 0; }
static void loop(void) { if (write(1, "executing program\n", sizeof("executing program\n") - 1)) { } int i, call, thread; for (call = 0; call < 8; call++) { for (thread = 0; thread < (int)(sizeof(threads) / sizeof(threads[0])); thread++) { struct thread_t* th = &threads[thread]; if (!th->created) { th->created = 1; event_init(&th->ready); event_init(&th->done); event_set(&th->done); thread_start(thr, th); } if (!event_isset(&th->done)) continue; event_reset(&th->done); th->call = call; __atomic_fetch_add(&running, 1, __ATOMIC_RELAXED); event_set(&th->ready); event_timedwait(&th->done, 50); break; } } for (i = 0; i < 100 && __atomic_load_n(&running, __ATOMIC_RELAXED); i++) sleep_ms(1); }
uint64_t r[1] = {0xffffffffffffffff};
void execute_call(int call) { intptr_t res = 0; switch (call) { case 0: *(uint64_t*)0x20000040 = 0x20000004; *(uint32_t*)0x20000048 = 4; *(uint32_t*)0x2000004c = 2; *(uint32_t*)0x20000050 = 0; syscall(__NR_mq_notify, /*mqd=*/-1, /*notif=*/0x20000040ul); break; case 1: syscall(__NR_mremap, /*addr=*/0x20002000ul, /*len=*/0x3000ul, /*newlen=*/0x4000ul, /*flags=MREMAP_FIXED|MREMAP_MAYMOVE*/ 3ul, /*newaddr=*/0x20422000ul); break; case 2: res = syscall(__NR_userfaultfd, /*flags=UFFD_USER_MODE_ONLY|O_CLOEXEC*/ 0x80001ul); if (res != -1) r[0] = res; break; case 3: *(uint64_t*)0x200000c0 = 0xaa; *(uint64_t*)0x200000c8 = 0xc; *(uint64_t*)0x200000d0 = 0; syscall(__NR_ioctl, /*fd=*/r[0], /*cmd=*/0xc018aa3f, /*arg=*/0x200000c0ul); break; case 4: *(uint64_t*)0x20000140 = 0x200e2000; *(uint64_t*)0x20000148 = 0xc00000; *(uint64_t*)0x20000150 = 1; *(uint64_t*)0x20000158 = 0; syscall(__NR_ioctl, /*fd=*/r[0], /*cmd=*/0xc020aa00, /*arg=*/0x20000140ul); break; case 5: syscall( __NR_mmap, /*addr=*/0x20000000ul, /*len=*/0x400000ul, /*prot=PROT_GROWSUP|PROT_SEM|PROT_WRITE|PROT_READ|PROT_EXEC*/ 0x200000ful, /*flags=MAP_SYNC|MAP_NONBLOCK|MAP_HUGETLB|MAP_FIXED|MAP_ANONYMOUS|0x2*/ 0xd0032ul, /*fd=*/-1, /*offset=*/0ul); break; case 6: syscall(__NR_madvise, /*addr=*/0x20000000ul, /*len=*/0x600003ul, /*advice=MADV_DONTNEED*/ 4ul); break; case 7: syscall( __NR_mmap, /*addr=*/0x20000000ul, /*len=*/0xff5000ul, /*prot=*/0ul, /*flags=MAP_POPULATE|MAP_NORESERVE|MAP_NONBLOCK|MAP_HUGETLB|MAP_FIXED|0x2000000000821*/ 0x200000005c831ul, /*fd=*/-1, /*offset=*/0ul); break; } } int main(void) { syscall(__NR_mmap, /*addr=*/0x1ffff000ul, /*len=*/0x1000ul, /*prot=*/0ul, /*flags=MAP_FIXED|MAP_ANONYMOUS|MAP_PRIVATE*/ 0x32ul, /*fd=*/-1, /*offset=*/0ul); syscall(__NR_mmap, /*addr=*/0x20000000ul, /*len=*/0x1000000ul, /*prot=PROT_WRITE|PROT_READ|PROT_EXEC*/ 7ul, /*flags=MAP_FIXED|MAP_ANONYMOUS|MAP_PRIVATE*/ 0x32ul, /*fd=*/-1, /*offset=*/0ul); syscall(__NR_mmap, /*addr=*/0x21000000ul, /*len=*/0x1000ul, /*prot=*/0ul, /*flags=MAP_FIXED|MAP_ANONYMOUS|MAP_PRIVATE*/ 0x32ul, /*fd=*/-1, /*offset=*/0ul); loop(); return 0; }
--8<------------------------------------------------------------------
Cheers, Ricardo
On Fri, 31 Jan 2025 15:37:49 +0100 Ricardo Cañuelo Navarro rcn@igalia.com wrote:
Add a sanity check to madvise_dontneed_free() to address a corner case in madvise where a race condition causes the current vma being processed to be backed by a different page size.
During a madvise(MADV_DONTNEED) call on a memory region registered with a userfaultfd, there's a period of time where the process mm lock is temporarily released in order to send a UFFD_EVENT_REMOVE and let userspace handle the event. During this time, the vma covering the current address range may change due to an explicit mmap done concurrently by another thread.
If, after that change, the memory region, which was originally backed by 4KB pages, is now backed by hugepages, the end address is rounded down to a hugepage boundary to avoid data loss (see "Fixes" below). This rounding may cause the end address to be truncated to the same address as the start.
Make this corner case follow the same semantics as in other similar cases where the requested region has zero length (ie. return 0).
This will make madvise_walk_vmas() continue to the next vma in the range (this time holding the process mm lock) which, due to the prev pointer becoming stale because of the vma change, will be the same hugepage-backed vma that was just checked before. The next time madvise_dontneed_free() runs for this vma, if the start address isn't aligned to a hugepage boundary, it'll return -EINVAL, which is also in line with the madvise api.
From userspace perspective, madvise() will return EINVAL because the
start address isn't aligned according to the new vma alignment requirements (hugepage), even though it was correctly page-aligned when the call was issued.
...
--- a/mm/madvise.c +++ b/mm/madvise.c @@ -933,7 +933,9 @@ static long madvise_dontneed_free(struct vm_area_struct *vma, */ end = vma->vm_end; }
VM_WARN_ON(start >= end);
if (start == end)return 0;VM_WARN_ON(start > end);
Perhaps add a comment telling the user how this situation can come about?
Hi Andrew,
On Fri, Jan 31 2025 at 14:23:21, Andrew Morton akpm@linux-foundation.org wrote:
Perhaps add a comment telling the user how this situation can come about?
Sure, I'll add the description to v2. Thanks for reviewing it!
Ricardo
linux-stable-mirror@lists.linaro.org
-
Andrew Morton -
Ricardo Cañuelo Navarro