From: huangshaobo <huangshaobo6@xxxxxxxxxx> Out-of-bounds accesses that aren't caught by a guard page will result in corruption of canary memory. In pathological cases, where an object has certain alignment requirements, an out-of-bounds access might never be caught by the guard page. Such corruptions, however, are only detected on kfree() normally. If the bug causes the kernel to panic before kfree(), KFENCE has no opportunity to report the issue. Such corruptions may also indicate failing memory or other faults. To provide some more information in such cases, add the option to check canary bytes on panic. This might help narrow the search for the panic cause; but, due to only having the allocation stack trace, such reports are difficult to use to diagnose an issue alone. In most cases, such reports are inactionable, and is therefore an opt-in feature (disabled by default). Suggested-by: chenzefeng <chenzefeng2@xxxxxxxxxx> Signed-off-by: huangshaobo <huangshaobo6@xxxxxxxxxx> --- v3: - use Marco's description replace the commit message - keep these includes sorted alphabetically - "in panic" replaced with "on panic" in title and comments - Blank line between /* === ... */ and function. v2: - it is only detected in panic. - it is disabled by default. - can only be enabled via boot parameter. - the code is moved to the specified partition. https://lore.kernel.org/all/20220424105949.50016-1-huangshaobo6@xxxxxxxxxx/ v1: https://lore.kernel.org/all/20220420104927.59056-1-huangshaobo6@xxxxxxxxxx/ Thanks again Marco for the suggestion. --- mm/kfence/core.c | 34 ++++++++++++++++++++++++++++++++++ 1 file changed, 34 insertions(+) diff --git a/mm/kfence/core.c b/mm/kfence/core.c index 9b2b5f56f4ae..06232d51e021 100644 --- a/mm/kfence/core.c +++ b/mm/kfence/core.c @@ -21,6 +21,8 @@ #include <linux/log2.h> #include <linux/memblock.h> #include <linux/moduleparam.h> +#include <linux/notifier.h> +#include <linux/panic_notifier.h> #include <linux/random.h> #include <linux/rcupdate.h> #include <linux/sched/clock.h> @@ -99,6 +101,10 @@ module_param_named(skip_covered_thresh, kfence_skip_covered_thresh, ulong, 0644) static bool kfence_deferrable __read_mostly = IS_ENABLED(CONFIG_KFENCE_DEFERRABLE); module_param_named(deferrable, kfence_deferrable, bool, 0444); +/* If true, check all canary bytes on panic. */ +static bool kfence_check_on_panic; +module_param_named(check_on_panic, kfence_check_on_panic, bool, 0444); + /* The pool of pages used for guard pages and objects. */ char *__kfence_pool __read_mostly; EXPORT_SYMBOL(__kfence_pool); /* Export for test modules. */ @@ -727,6 +733,31 @@ static int __init kfence_debugfs_init(void) late_initcall(kfence_debugfs_init); +/* === Panic Notifier ====================================================== */ + +static void kfence_check_all_canary(void) +{ + int i; + + for (i = 0; i < CONFIG_KFENCE_NUM_OBJECTS; i++) { + struct kfence_metadata *meta = &kfence_metadata[i]; + + if (meta->state == KFENCE_OBJECT_ALLOCATED) + for_each_canary(meta, check_canary_byte); + } +} + +static int kfence_check_canary_callback(struct notifier_block *nb, + unsigned long reason, void *arg) +{ + kfence_check_all_canary(); + return NOTIFY_OK; +} + +static struct notifier_block kfence_check_canary_notifier = { + .notifier_call = kfence_check_canary_callback, +}; + /* === Allocation Gate Timer ================================================ */ static struct delayed_work kfence_timer; @@ -804,6 +835,9 @@ static void kfence_init_enable(void) else INIT_DELAYED_WORK(&kfence_timer, toggle_allocation_gate); + if (kfence_check_on_panic) + atomic_notifier_chain_register(&panic_notifier_list, &kfence_check_canary_notifier); + WRITE_ONCE(kfence_enabled, true); queue_delayed_work(system_unbound_wq, &kfence_timer, 0); -- 2.12.3
