This patch addresses a bug in the RT variant of the kernel where a
"sleeping function called from invalid context" warning may occur in
kmemleak_seq_show under specific conditions:
- CONFIG_PREEMPT_RT=y
- SELinux is the LSM for the system
- `kptr_restrict` is set to 1.
- The kmemleak buffer contains at least one item.
Commit 8c96f1bc6fc49c724c4cdd22d3e99260263b7384 ("mm/kmemleak: turn
kmemleak_lock and object->lock to raw_spinlock_t") introduced a change
where kmemleak_seq_show is executed in atomic context within the RT kernel.
However, the SELinux capability check within this function flow still
relies on regular spinlocks, leading to potential race conditions that
trigger the error when printing the kmemleak backtrace.
To resolve this, the backtrace printing has been moved out of the critical
section.
Signed-off-by: Alessandro Carminati <acarmina@xxxxxxxxxx>
---
Please read previous conversation in the RFC
https://lore.kernel.org/all/20241115145410.114376-1-acarmina@xxxxxxxxxx/
Splash triggering this patch:
```
[ 159.247069] BUG: sleeping function called from invalid context at kernel/locking/spinlock_rt.c:48
[ 159.247193] in_atomic(): 1, irqs_disabled(): 1, non_block: 0, pid: 136, name: cat
[ 159.247241] preempt_count: 1, expected: 0
[ 159.247277] RCU nest depth: 2, expected: 2
[ 159.247388] 6 locks held by cat/136:
[ 159.247438] #0: ffff32e64bcbf950 (&p->lock){+.+.}-{3:3}, at: seq_read_iter+0xb8/0xe30
[ 159.248835] #1: ffffafe6aaa9dea0 (scan_mutex){+.+.}-{3:3}, at: kmemleak_seq_start+0x34/0x128
[ 159.249053] #3: ffff32e6546b1cd0 (&object->lock){....}-{2:2}, at: kmemleak_seq_show+0x3c/0x1e0
[ 159.249127] #4: ffffafe6aa8d8560 (rcu_read_lock){....}-{1:2}, at: has_ns_capability_noaudit+0x8/0x1b0
[ 159.249205] #5: ffffafe6aabbc0f8 (notif_lock){+.+.}-{2:2}, at: avc_compute_av+0xc4/0x3d0
[ 159.249364] irq event stamp: 136660
[ 159.249407] hardirqs last enabled at (136659): [<ffffafe6a80fd7a0>] _raw_spin_unlock_irqrestore+0xa8/0xd8
[ 159.249465] hardirqs last disabled at (136660): [<ffffafe6a80fd85c>] _raw_spin_lock_irqsave+0x8c/0xb0
[ 159.249518] softirqs last enabled at (0): [<ffffafe6a5d50b28>] copy_process+0x11d8/0x3df8
[ 159.249571] softirqs last disabled at (0): [<0000000000000000>] 0x0
[ 159.249970] Preemption disabled at:
[ 159.249988] [<ffffafe6a6598a4c>] kmemleak_seq_show+0x3c/0x1e0
[ 159.250609] CPU: 1 UID: 0 PID: 136 Comm: cat Tainted: G E 6.11.0-rt7+ #34
[ 159.250797] Tainted: [E]=UNSIGNED_MODULE
[ 159.250822] Hardware name: linux,dummy-virt (DT)
[ 159.251050] Call trace:
[ 159.251079] dump_backtrace+0xa0/0x128
[ 159.251132] show_stack+0x1c/0x30
[ 159.251156] dump_stack_lvl+0xe8/0x198
[ 159.251180] dump_stack+0x18/0x20
[ 159.251227] rt_spin_lock+0x8c/0x1a8
[ 159.251273] avc_perm_nonode+0xa0/0x150
[ 159.251316] cred_has_capability.isra.0+0x118/0x218
[ 159.251340] selinux_capable+0x50/0x80
[ 159.251363] security_capable+0x7c/0xd0
[ 159.251388] has_ns_capability_noaudit+0x94/0x1b0
[ 159.251412] has_capability_noaudit+0x20/0x30
[ 159.251437] restricted_pointer+0x21c/0x4b0
[ 159.251461] pointer+0x298/0x760
[ 159.251482] vsnprintf+0x330/0xf70
[ 159.251504] seq_printf+0x178/0x218
[ 159.251526] print_unreferenced+0x1a4/0x2d0
[ 159.251551] kmemleak_seq_show+0xd0/0x1e0
[ 159.251576] seq_read_iter+0x354/0xe30
[ 159.251599] seq_read+0x250/0x378
[ 159.251622] full_proxy_read+0xd8/0x148
[ 159.251649] vfs_read+0x190/0x918
[ 159.251672] ksys_read+0xf0/0x1e0
[ 159.251693] __arm64_sys_read+0x70/0xa8
[ 159.251716] invoke_syscall.constprop.0+0xd4/0x1d8
[ 159.251767] el0_svc+0x50/0x158
[ 159.251813] el0t_64_sync+0x17c/0x180
```
mm/kmemleak.c | 46 ++++++++++++++++++++++++++++++++++++----------
1 file changed, 36 insertions(+), 10 deletions(-)
diff --git a/mm/kmemleak.c b/mm/kmemleak.c
index 0400f5e8ac60..c77899af3e9e 100644
--- a/mm/kmemleak.c
+++ b/mm/kmemleak.c
@@ -356,14 +356,9 @@ static bool unreferenced_object(struct kmemleak_object *object)
* Printing of the unreferenced objects information to the seq file. The
* print_unreferenced function must be called with the object->lock held.
*/
-static void print_unreferenced(struct seq_file *seq,
+static depot_stack_handle_t print_unreferenced(struct seq_file *seq,
struct kmemleak_object *object)
{
- int i;
- unsigned long *entries;
- unsigned int nr_entries;
-
- nr_entries = stack_depot_fetch(object->trace_handle, &entries);
warn_or_seq_printf(seq, "unreferenced object 0x%08lx (size %zu):\n",
object->pointer, object->size);
warn_or_seq_printf(seq, " comm \"%s\", pid %d, jiffies %lu\n",
@@ -371,6 +366,23 @@ static void print_unreferenced(struct seq_file *seq,
hex_dump_object(seq, object);
warn_or_seq_printf(seq, " backtrace (crc %x):\n", object->checksum);
+ return object->trace_handle;
+}
+
+/*
+ * Prints stack traces of unreferenced objects outside of the lock context.
+ * This avoids potential issues with printing pointers that might require
+ * additional locking.
+ */
+static void print_stack_trace(struct seq_file *seq,
+ depot_stack_handle_t h)
+{
+ int i;
+ unsigned long *entries;
+ unsigned int nr_entries;
+
+ nr_entries = stack_depot_fetch(h, &entries);
+
for (i = 0; i < nr_entries; i++) {
void *ptr = (void *)entries[i];
warn_or_seq_printf(seq, " [<%pK>] %pS\n", ptr, ptr);
@@ -1621,7 +1633,9 @@ static void kmemleak_cond_resched(struct kmemleak_object *object)
*/
static void kmemleak_scan(void)
{
+ depot_stack_handle_t stackdepot_handle;
struct kmemleak_object *object;
+ bool do_print = false;
struct zone *zone;
int __maybe_unused i;
int new_leaks = 0;
@@ -1783,12 +1797,17 @@ static void kmemleak_scan(void)
!(object->flags & OBJECT_REPORTED)) {
object->flags |= OBJECT_REPORTED;
- if (kmemleak_verbose)
- print_unreferenced(NULL, object);
+ if (kmemleak_verbose) {
+ stackdepot_handle = print_unreferenced(NULL, object);
+ do_print = true;
+ }
new_leaks++;
}
raw_spin_unlock_irq(&object->lock);
+ if (kmemleak_verbose && do_print)
+ print_stack_trace(NULL, stackdepot_handle);
+
}
rcu_read_unlock();
@@ -1937,13 +1956,20 @@ static void kmemleak_seq_stop(struct seq_file *seq, void *v)
*/
static int kmemleak_seq_show(struct seq_file *seq, void *v)
{
+ depot_stack_handle_t stackdepot_handle;
struct kmemleak_object *object = v;
+ bool do_print = false;
unsigned long flags;
raw_spin_lock_irqsave(&object->lock, flags);
- if ((object->flags & OBJECT_REPORTED) && unreferenced_object(object))
- print_unreferenced(seq, object);
+ if ((object->flags & OBJECT_REPORTED) && unreferenced_object(object)) {
+ stackdepot_handle = print_unreferenced(seq, object);
+ do_print = true;
+ }
raw_spin_unlock_irqrestore(&object->lock, flags);
+ if (do_print)
+ print_stack_trace(seq, stackdepot_handle);
+
return 0;
}
--
2.34.1
