debugobject: Make it work with deferred page initialization - again

debugobjects uses __GFP_HIGH for allocations as it might be invoked within locked regions. That worked perfectly fine until v6.18. It still works correctly when deferred page initialization is disabled and works by chance when no page allocation is required before deferred page initialization has completed.

Since v6.18 allocations w/o a reclaim flag cause new_slab() to end up in alloc_frozen_pages_nolock_noprof(), which returns early when deferred page initialization has not yet completed. As the deferred page initialization takes quite a while the debugobject pool is depleted and debugobjects are disabled.

This can be worked around when PREEMPT_COUNT is enabled as that allows debugobjects to add __GFP_KSWAPD_RECLAIM to the GFP flags when the context is preemtible. When PREEMPT_COUNT is disabled the context is unknown and the reclaim bit can't be set because the caller might hold locks which might deadlock in the allocator.

In preemptible context the reclaim bit is harmless and not a performance issue as that's usually invoked from slow path initialization context.

That makes debugobjects depend on PREEMPT_COUNT || !DEFERRED_STRUCT_PAGE_INIT.

Fixes af92793e52c3 ("slab: Introduce kmalloc_nolock() and kfree_nolock().")

Signed-off-by Thomas Gleixner <[email protected]>

debugobjects uses __GFP_HIGH for allocations as it might be invoked within locked regions. That worked perfectly fine until v6.18. It still works correctly when deferred page initialization is disabled and works by chance when no page allocation is required before deferred page initialization has completed.

Since v6.18 allocations w/o a reclaim flag cause new_slab() to end up in alloc_frozen_pages_nolock_noprof(), which returns early when deferred page initialization has not yet completed. As the deferred page initialization takes quite a while the debugobject pool is depleted and debugobjects are disabled.

This can be worked around when PREEMPT_COUNT is enabled as that allows debugobjects to add __GFP_KSWAPD_RECLAIM to the GFP flags when the context is preemtible. When PREEMPT_COUNT is disabled the context is unknown and the reclaim bit can't be set because the caller might hold locks which might deadlock in the allocator.

In preemptible context the reclaim bit is harmless and not a performance issue as that's usually invoked from slow path initialization context.

That makes debugobjects depend on PREEMPT_COUNT || !DEFERRED_STRUCT_PAGE_INIT.

Fixes af92793e52c3 ("slab: Introduce kmalloc_nolock() and kfree_nolock().")

Signed-off-by Thomas Gleixner <[email protected]>

---
 lib/Kconfig.debug  |    1 +
 lib/debugobjects.c |   19 ++++++++++++++++++-
 2 files changed, 19 insertions(+), 1 deletion(-)

lib/Kconfig.debug +1 -0

@@ -753,6 +753,7 @@ source "mm/Kconfig.debug"
 
 config DEBUG_OBJECTS
 	bool "Debug object operations"
+	depends on PREEMPT_COUNT || !DEFERRED_STRUCT_PAGE_INIT
 	depends on DEBUG_KERNEL
 	help
 	  If you say Y here, additional code will be inserted into the

lib/debugobjects.c +18 -1

@@ -398,9 +398,26 @@ static void fill_pool(void)
 
 	atomic_inc(&cpus_allocating);
 	while (pool_should_refill(&pool_global)) {
+		gfp_t gfp = __GFP_HIGH | __GFP_NOWARN;
 		HLIST_HEAD(head);
 
-		if (!kmem_alloc_batch(&head, obj_cache, __GFP_HIGH | __GFP_NOWARN))
+		/*
+		 * Allow reclaim only in preemptible context and during
+		 * early boot. If not preemptible, the caller might hold
+		 * locks causing a deadlock in the allocator.
+		 *
+		 * If the reclaim flag is not set during early boot then
+		 * allocations, which happen before deferred page
+		 * initialization has completed, will fail.
+		 *
+		 * In preemptible context the flag is harmless and not a
+		 * performance issue as that's usually invoked from slow
+		 * path initialization context.
+		 */
+		if (preemptible() || system_state < SYSTEM_SCHEDULING)
+			gfp |= __GFP_KSWAPD_RECLAIM;
+
+		if (!kmem_alloc_batch(&head, obj_cache, gfp))
 			break;
 
 		guard(raw_spinlock_irqsave)(&pool_lock);

That makes debugobjects depend on PREEMPT_COUNT || !DEFERRED_STRUCT_PAGE_INIT.

Fixes af92793e52c3 ("slab: Introduce kmalloc_nolock() and kfree_nolock().")

Signed-off-by Thomas Gleixner <[email protected]>

debugobjects uses __GFP_HIGH for allocations as it might be invoked within locked regions. That worked perfectly fine until v6.18. It still works correctly when deferred page initialization is disabled and works by chance when no page allocation is required before deferred page initialization has completed.

Since v6.18 allocations w/o a reclaim flag cause new_slab() to end up in alloc_frozen_pages_nolock_noprof(), which returns early when deferred page initialization has not yet completed. As the deferred page initialization takes quite a while the debugobject pool is depleted and debugobjects are disabled.

This can be worked around when PREEMPT_COUNT is enabled as that allows debugobjects to add __GFP_KSWAPD_RECLAIM to the GFP flags when the context is preemtible. When PREEMPT_COUNT is disabled the context is unknown and the reclaim bit can't be set because the caller might hold locks which might deadlock in the allocator.

In preemptible context the reclaim bit is harmless and not a performance issue as that's usually invoked from slow path initialization context.

That makes debugobjects depend on PREEMPT_COUNT || !DEFERRED_STRUCT_PAGE_INIT.

Fixes af92793e52c3 ("slab: Introduce kmalloc_nolock() and kfree_nolock().")

Signed-off-by Thomas Gleixner <[email protected]>

---
 lib/Kconfig.debug  |    1 +
 lib/debugobjects.c |   19 ++++++++++++++++++-
 2 files changed, 19 insertions(+), 1 deletion(-)

lib/Kconfig.debug +1 -0

@@ -753,6 +753,7 @@ source "mm/Kconfig.debug"
 
 config DEBUG_OBJECTS
 	bool "Debug object operations"
+	depends on PREEMPT_COUNT || !DEFERRED_STRUCT_PAGE_INIT
 	depends on DEBUG_KERNEL
 	help
 	  If you say Y here, additional code will be inserted into the

lib/debugobjects.c +18 -1

@@ -398,9 +398,26 @@ static void fill_pool(void)
 
 	atomic_inc(&cpus_allocating);
 	while (pool_should_refill(&pool_global)) {
+		gfp_t gfp = __GFP_HIGH | __GFP_NOWARN;
 		HLIST_HEAD(head);
 
-		if (!kmem_alloc_batch(&head, obj_cache, __GFP_HIGH | __GFP_NOWARN))
+		/*
+		 * Allow reclaim only in preemptible context and during
+		 * early boot. If not preemptible, the caller might hold
+		 * locks causing a deadlock in the allocator.
+		 *
+		 * If the reclaim flag is not set during early boot then
+		 * allocations, which happen before deferred page
+		 * initialization has completed, will fail.
+		 *
+		 * In preemptible context the flag is harmless and not a
+		 * performance issue as that's usually invoked from slow
+		 * path initialization context.
+		 */
+		if (preemptible() || system_state < SYSTEM_SCHEDULING)
+			gfp |= __GFP_KSWAPD_RECLAIM;
+
+		if (!kmem_alloc_batch(&head, obj_cache, gfp))
 			break;
 
 		guard(raw_spinlock_irqsave)(&pool_lock);

Quoted 7 lines · Thomas Gleixner

In preemptible context the reclaim bit is harmless and not a performance issue as that's usually invoked from slow path initialization context.

That makes debugobjects depend on PREEMPT_COUNT || !DEFERRED_STRUCT_PAGE_INIT.

Fixes af92793e52c3 ("slab: Introduce kmalloc_nolock() and kfree_nolock().")

Signed-off-by Thomas Gleixner <[email protected]>

FYI, it should now be possible to revert this workaround:
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=335c347686e76df9d2c7d7f61b5ea627a4c5cb4c