From nobody Wed Dec 17 06:27:14 2025 Received: from out-182.mta1.migadu.com (out-182.mta1.migadu.com [95.215.58.182]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 151F83D0AB for ; Tue, 19 Dec 2023 21:19:58 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=linux.dev Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=linux.dev Authentication-Results: smtp.subspace.kernel.org; dkim=pass (1024-bit key) header.d=linux.dev header.i=@linux.dev header.b="elb7wJjm" X-Report-Abuse: Please report any abuse attempt to abuse@migadu.com and include these headers. DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=linux.dev; s=key1; t=1703020797; h=from:from:reply-to:subject:subject:date:date:message-id:message-id: to:to:cc:cc:mime-version:mime-version: content-transfer-encoding:content-transfer-encoding: in-reply-to:in-reply-to:references:references; bh=nejB0rN/5EGuA3SQQpabp67CuxIikS5y1NrnTFuMISI=; b=elb7wJjmHVmMnqrahneo3nmWE9hHEyTRVGuiSBGymfQVeOW15J6gqNykSrIgFdw9PIMbZF ta1kf9P/xFptBLk6yUizJYveQGUO6hzXz1/PIciaF5j55mJ3tEib32Ug4OJ35XYc+zlmf1 5OklhsBMYdcTGxBkNXOkqzloDUn2vJw= From: andrey.konovalov@linux.dev To: Andrew Morton Cc: Andrey Konovalov , Marco Elver , Alexander Potapenko , Dmitry Vyukov , Vlastimil Babka , kasan-dev@googlegroups.com, Evgenii Stepanov , Tetsuo Handa , linux-mm@kvack.org, linux-kernel@vger.kernel.org, Andrey Konovalov Subject: [PATCH v3 mm 1/4] lib/stackdepot: add printk_deferred_enter/exit guards Date: Tue, 19 Dec 2023 22:19:50 +0100 Message-Id: <82092f9040d075a161d1264377d51e0bac847e8a.1703020707.git.andreyknvl@google.com> In-Reply-To: References: Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable X-Migadu-Flow: FLOW_OUT Content-Type: text/plain; charset="utf-8" From: Andrey Konovalov Stack depot functions can be called from various contexts that do allocations, including with console locks taken. At the same time, stack depot functions might print WARNING's or refcount-related failures. This can cause a deadlock on console locks. Add printk_deferred_enter/exit guards to stack depot to avoid this. Reported-by: Tetsuo Handa Closes: https://lore.kernel.org/all/000000000000f56750060b9ad216@google.com/ Fixes: 108be8def46e ("lib/stackdepot: allow users to evict stack traces") Fixes: cd11016e5f52 ("mm, kasan: stackdepot implementation. Enable stackdep= ot for SLAB") Reviewed-by: Marco Elver Signed-off-by: Andrey Konovalov --- lib/stackdepot.c | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/lib/stackdepot.c b/lib/stackdepot.c index 870cce2f4cbd..a0be5d05c7f0 100644 --- a/lib/stackdepot.c +++ b/lib/stackdepot.c @@ -506,12 +506,14 @@ depot_stack_handle_t stack_depot_save_flags(unsigned = long *entries, bucket =3D &stack_table[hash & stack_hash_mask]; =20 read_lock_irqsave(&pool_rwlock, flags); + printk_deferred_enter(); =20 /* Fast path: look the stack trace up without full locking. */ found =3D find_stack(bucket, entries, nr_entries, hash); if (found) { if (depot_flags & STACK_DEPOT_FLAG_GET) refcount_inc(&found->count); + printk_deferred_exit(); read_unlock_irqrestore(&pool_rwlock, flags); goto exit; } @@ -520,6 +522,7 @@ depot_stack_handle_t stack_depot_save_flags(unsigned lo= ng *entries, if (new_pool_required) need_alloc =3D true; =20 + printk_deferred_exit(); read_unlock_irqrestore(&pool_rwlock, flags); =20 /* @@ -541,6 +544,7 @@ depot_stack_handle_t stack_depot_save_flags(unsigned lo= ng *entries, } =20 write_lock_irqsave(&pool_rwlock, flags); + printk_deferred_enter(); =20 found =3D find_stack(bucket, entries, nr_entries, hash); if (!found) { @@ -562,6 +566,7 @@ depot_stack_handle_t stack_depot_save_flags(unsigned lo= ng *entries, depot_keep_new_pool(&prealloc); } =20 + printk_deferred_exit(); write_unlock_irqrestore(&pool_rwlock, flags); exit: if (prealloc) { @@ -600,9 +605,11 @@ unsigned int stack_depot_fetch(depot_stack_handle_t ha= ndle, return 0; =20 read_lock_irqsave(&pool_rwlock, flags); + printk_deferred_enter(); =20 stack =3D depot_fetch_stack(handle); =20 + printk_deferred_exit(); read_unlock_irqrestore(&pool_rwlock, flags); =20 *entries =3D stack->entries; @@ -619,6 +626,7 @@ void stack_depot_put(depot_stack_handle_t handle) return; =20 write_lock_irqsave(&pool_rwlock, flags); + printk_deferred_enter(); =20 stack =3D depot_fetch_stack(handle); if (WARN_ON(!stack)) @@ -633,6 +641,7 @@ void stack_depot_put(depot_stack_handle_t handle) } =20 out: + printk_deferred_exit(); write_unlock_irqrestore(&pool_rwlock, flags); } EXPORT_SYMBOL_GPL(stack_depot_put); --=20 2.25.1 From nobody Wed Dec 17 06:27:14 2025 Received: from out-180.mta1.migadu.com (out-180.mta1.migadu.com [95.215.58.180]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id E3B0B3D0A2 for ; Tue, 19 Dec 2023 21:19:59 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=linux.dev Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=linux.dev Authentication-Results: smtp.subspace.kernel.org; dkim=pass (1024-bit key) header.d=linux.dev header.i=@linux.dev header.b="IR+Whe6+" X-Report-Abuse: Please report any abuse attempt to abuse@migadu.com and include these headers. DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=linux.dev; s=key1; t=1703020798; h=from:from:reply-to:subject:subject:date:date:message-id:message-id: to:to:cc:cc:mime-version:mime-version: content-transfer-encoding:content-transfer-encoding: in-reply-to:in-reply-to:references:references; bh=IugD6msAA76PLOkulY8RL+cRtUx1nL2a5VcYM7ZPFvs=; b=IR+Whe6+qpxv+At+vArbXY01tYtvfspW3PKw8UAncGDGnEEps0UIMWhP+wPzEx5NxY9M8h m1TDVtQZm9HrpXOj1mV5z/Lc+HwiQjOjMkmc/bfj2auMjP1/9llRH/NH6L89Gq+qP7zEz3 M7vUVsknaJSSv98tDJGKvpfbqCVOHN0= From: andrey.konovalov@linux.dev To: Andrew Morton Cc: Andrey Konovalov , Marco Elver , Alexander Potapenko , Dmitry Vyukov , Vlastimil Babka , kasan-dev@googlegroups.com, Evgenii Stepanov , Tetsuo Handa , linux-mm@kvack.org, linux-kernel@vger.kernel.org, Andrey Konovalov , syzbot+186b55175d8360728234@syzkaller.appspotmail.com Subject: [PATCH v3 mm 2/4] kasan: handle concurrent kasan_record_aux_stack calls Date: Tue, 19 Dec 2023 22:19:51 +0100 Message-Id: <1606b960e2f746862d1f459515972f9695bf448a.1703020707.git.andreyknvl@google.com> In-Reply-To: References: Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable X-Migadu-Flow: FLOW_OUT Content-Type: text/plain; charset="utf-8" From: Andrey Konovalov kasan_record_aux_stack can be called concurrently on the same object. This might lead to a race condition when rotating the saved aux stack trace handles, which in turns leads to incorrect accounting of stack depot handles and refcount underflows in the stack depot code. Fix by introducing a raw spinlock to protect the aux stack trace handles in kasan_record_aux_stack. Reported-by: Tetsuo Handa Reported-by: syzbot+186b55175d8360728234@syzkaller.appspotmail.com Closes: https://lore.kernel.org/all/000000000000784b1c060b0074a2@google.com/ Fixes: 773688a6cb24 ("kasan: use stack_depot_put for Generic mode") Signed-off-by: Andrey Konovalov Reviewed-by: Marco Elver --- Changes v2->v3: - Use raw spinlock to avoid lockdep complaints on RT kernels. Changes v1->v2: - Use per-object spinlock instead of a global one. --- mm/kasan/generic.c | 32 +++++++++++++++++++++++++++++--- mm/kasan/kasan.h | 8 ++++++++ 2 files changed, 37 insertions(+), 3 deletions(-) diff --git a/mm/kasan/generic.c b/mm/kasan/generic.c index 54e20b2bc3e1..55e6b5db2cae 100644 --- a/mm/kasan/generic.c +++ b/mm/kasan/generic.c @@ -25,6 +25,7 @@ #include #include #include +#include #include #include #include @@ -471,8 +472,18 @@ void kasan_init_object_meta(struct kmem_cache *cache, = const void *object) struct kasan_free_meta *free_meta; =20 alloc_meta =3D kasan_get_alloc_meta(cache, object); - if (alloc_meta) + if (alloc_meta) { __memset(alloc_meta, 0, sizeof(*alloc_meta)); + + /* + * Temporarily disable KASAN bug reporting to allow instrumented + * raw_spin_lock_init to access aux_lock, which resides inside + * of a redzone. + */ + kasan_disable_current(); + raw_spin_lock_init(&alloc_meta->aux_lock); + kasan_enable_current(); + } free_meta =3D kasan_get_free_meta(cache, object); if (free_meta) __memset(free_meta, 0, sizeof(*free_meta)); @@ -502,6 +513,8 @@ static void __kasan_record_aux_stack(void *addr, depot_= flags_t depot_flags) struct kmem_cache *cache; struct kasan_alloc_meta *alloc_meta; void *object; + depot_stack_handle_t new_handle, old_handle; + unsigned long flags; =20 if (is_kfence_address(addr) || !slab) return; @@ -512,9 +525,22 @@ static void __kasan_record_aux_stack(void *addr, depot= _flags_t depot_flags) if (!alloc_meta) return; =20 - stack_depot_put(alloc_meta->aux_stack[1]); + new_handle =3D kasan_save_stack(0, depot_flags); + + /* + * Temporarily disable KASAN bug reporting to allow instrumented + * spinlock functions to access aux_lock, which resides inside of a + * redzone. + */ + kasan_disable_current(); + raw_spin_lock_irqsave(&alloc_meta->aux_lock, flags); + old_handle =3D alloc_meta->aux_stack[1]; alloc_meta->aux_stack[1] =3D alloc_meta->aux_stack[0]; - alloc_meta->aux_stack[0] =3D kasan_save_stack(0, depot_flags); + alloc_meta->aux_stack[0] =3D new_handle; + raw_spin_unlock_irqrestore(&alloc_meta->aux_lock, flags); + kasan_enable_current(); + + stack_depot_put(old_handle); } =20 void kasan_record_aux_stack(void *addr) diff --git a/mm/kasan/kasan.h b/mm/kasan/kasan.h index 5e298e3ac909..69e4f5e58e33 100644 --- a/mm/kasan/kasan.h +++ b/mm/kasan/kasan.h @@ -6,6 +6,7 @@ #include #include #include +#include #include =20 #if defined(CONFIG_KASAN_SW_TAGS) || defined(CONFIG_KASAN_HW_TAGS) @@ -249,6 +250,13 @@ struct kasan_global { struct kasan_alloc_meta { struct kasan_track alloc_track; /* Free track is stored in kasan_free_meta. */ + /* + * aux_lock protects aux_stack from accesses from concurrent + * kasan_record_aux_stack calls. It is a raw spinlock to avoid sleeping + * on RT kernels, as kasan_record_aux_stack_noalloc can be called from + * non-sleepable contexts. + */ + raw_spinlock_t aux_lock; depot_stack_handle_t aux_stack[2]; }; =20 --=20 2.25.1 From nobody Wed Dec 17 06:27:14 2025 Received: from out-175.mta1.migadu.com (out-175.mta1.migadu.com [95.215.58.175]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 770AD3D0CE for ; Tue, 19 Dec 2023 21:20:00 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=linux.dev Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=linux.dev Authentication-Results: smtp.subspace.kernel.org; dkim=pass (1024-bit key) header.d=linux.dev header.i=@linux.dev header.b="qSi6bCK8" X-Report-Abuse: Please report any abuse attempt to abuse@migadu.com and include these headers. DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=linux.dev; s=key1; t=1703020798; h=from:from:reply-to:subject:subject:date:date:message-id:message-id: to:to:cc:cc:mime-version:mime-version: content-transfer-encoding:content-transfer-encoding: in-reply-to:in-reply-to:references:references; bh=fOHzqh8IZvvUQEcAVUngjlYsg+ArhHiPZJoDS72WUGY=; b=qSi6bCK82bZzb5eDpiDbrLAgv900bdtXCroZwQk+HMxNuCcAH6RDEUBSEuZXglh3sQ52dp y4hliOr/Z2u8q2lAKSTk3GSjbIMKCqYtE46R+/r0JsXgn8CkNa8Ohm/TliCh9Gv/qOmJ9H dqjaS5WqDKLwR54ujuGm3mhTRxi7Yx0= From: andrey.konovalov@linux.dev To: Andrew Morton Cc: Andrey Konovalov , Marco Elver , Alexander Potapenko , Dmitry Vyukov , Vlastimil Babka , kasan-dev@googlegroups.com, Evgenii Stepanov , Tetsuo Handa , linux-mm@kvack.org, linux-kernel@vger.kernel.org, Andrey Konovalov Subject: [PATCH v3 mm 3/4] kasan: memset free track in qlink_free Date: Tue, 19 Dec 2023 22:19:52 +0100 Message-Id: In-Reply-To: References: Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable X-Migadu-Flow: FLOW_OUT Content-Type: text/plain; charset="utf-8" From: Andrey Konovalov Instead of only zeroing out the stack depot handle when evicting the free stack trace in qlink_free, zero out the whole track. Do this just to produce a similar effect for alloc and free meta. The other fields of the free track besides the stack trace handle are considered invalid at this point anyway, so no harm in zeroing them out. Fixes: 773688a6cb24 ("kasan: use stack_depot_put for Generic mode") Reviewed-by: Marco Elver Signed-off-by: Andrey Konovalov --- mm/kasan/quarantine.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/mm/kasan/quarantine.c b/mm/kasan/quarantine.c index 265ca2bbe2dd..782e045da911 100644 --- a/mm/kasan/quarantine.c +++ b/mm/kasan/quarantine.c @@ -157,7 +157,7 @@ static void qlink_free(struct qlist_node *qlink, struct= kmem_cache *cache) if (free_meta && *(u8 *)kasan_mem_to_shadow(object) =3D=3D KASAN_SLAB_FREETRACK) { stack_depot_put(free_meta->free_track.stack); - free_meta->free_track.stack =3D 0; + __memset(&free_meta->free_track, 0, sizeof(free_meta->free_track)); } =20 /* --=20 2.25.1 From nobody Wed Dec 17 06:27:14 2025 Received: from out-180.mta1.migadu.com (out-180.mta1.migadu.com [95.215.58.180]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 1569C3D0D5 for ; Tue, 19 Dec 2023 21:20:00 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=linux.dev Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=linux.dev Authentication-Results: smtp.subspace.kernel.org; dkim=pass (1024-bit key) header.d=linux.dev header.i=@linux.dev header.b="vWTr3qhn" X-Report-Abuse: Please report any abuse attempt to abuse@migadu.com and include these headers. DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=linux.dev; s=key1; t=1703020799; h=from:from:reply-to:subject:subject:date:date:message-id:message-id: to:to:cc:cc:mime-version:mime-version: content-transfer-encoding:content-transfer-encoding: in-reply-to:in-reply-to:references:references; bh=nCFM+nFcyoFd/gCQWZCb7EKvVZ+qSBSDeoPEA8Jfh0A=; b=vWTr3qhn77AyDPItr7Jb1Va+xZtAYsJMZbCeKfs5+hQd3T8gMEEYCJMliC2zyzhGEytQiS pdedlVWK/90vwulDxVVraP5pN7YgbdG3MBwinwjLSYYIyJMPDTtGDljZA7elKcrlFiXdm6 5vpKeJm4v97c+Ecf3o21M8gf1GjoJH0= From: andrey.konovalov@linux.dev To: Andrew Morton Cc: Andrey Konovalov , Marco Elver , Alexander Potapenko , Dmitry Vyukov , Vlastimil Babka , kasan-dev@googlegroups.com, Evgenii Stepanov , Tetsuo Handa , linux-mm@kvack.org, linux-kernel@vger.kernel.org, Andrey Konovalov Subject: [PATCH v3 mm 4/4] lib/stackdepot: fix comment in include/linux/stackdepot.h Date: Tue, 19 Dec 2023 22:19:53 +0100 Message-Id: <0ebe712d91f8d302a8947d3c9e9123bc2b1b8440.1703020707.git.andreyknvl@google.com> In-Reply-To: References: Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable X-Migadu-Flow: FLOW_OUT Content-Type: text/plain; charset="utf-8" From: Andrey Konovalov As stack traces can now be evicted from the stack depot, remove the comment saying that they are never removed. Fixes: 108be8def46e ("lib/stackdepot: allow users to evict stack traces") Reviewed-by: Marco Elver Signed-off-by: Andrey Konovalov --- include/linux/stackdepot.h | 2 -- 1 file changed, 2 deletions(-) diff --git a/include/linux/stackdepot.h b/include/linux/stackdepot.h index a6796f178913..adcbb8f23600 100644 --- a/include/linux/stackdepot.h +++ b/include/linux/stackdepot.h @@ -11,8 +11,6 @@ * SLUB_DEBUG needs 256 bytes per object for that). Since allocation and f= ree * stack traces often repeat, using stack depot allows to save about 100x = space. * - * Stack traces are never removed from the stack depot. - * * Author: Alexander Potapenko * Copyright (C) 2016 Google, Inc. * --=20 2.25.1