From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from out-178.mta1.migadu.com (out-178.mta1.migadu.com [95.215.58.178]) (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 6EBDE31619B for ; Tue, 14 Oct 2025 13:20:03 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=95.215.58.178 ARC-Seal:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1760448005; cv=none; b=nTCri/WMIGG71S0YowE53aNWyAh+kUF7a8RFeyg8Jp85Vg02J7JA/Ym/n6macZ2sTX97a3Pj4f9+xYQU/h4636Xt1m4x4ly88v17pt8Jta/946EkIu4WSd8PCz5mSPgHMf3OrKv3K7u02TZFpDEr3hTotxKesaAOUUUanRTnWWE= ARC-Message-Signature:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1760448005; c=relaxed/simple; bh=Wim9o18MKObwnzJcT2quzHlf2IzqkHTysWZ4YPTqIKE=; h=Message-ID:Date:MIME-Version:Subject:To:Cc:References:From: In-Reply-To:Content-Type; b=m5QLdXMJZLosbXGhDTj6u/fciDs3WUltGRLdhpgvM79UUdZl1lE987+8Y57434r4oXk8BJdV5xowvAnoQpNtM5vRK8zLcFSfgAb3YaFyILDw9AICJKtmdi4cBq5MyjKY7v06XN/ylM+HSlrlhrTaJbNZtUbRBnT2UMROG1LXjOA= ARC-Authentication-Results:i=1; smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=linux.dev; spf=pass smtp.mailfrom=linux.dev; dkim=pass (1024-bit key) header.d=linux.dev header.i=@linux.dev header.b=RnVJgjVA; arc=none smtp.client-ip=95.215.58.178 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="RnVJgjVA" Message-ID: <8507df5e-0170-4316-b732-9ebb11bab4e2@linux.dev> DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=linux.dev; s=key1; t=1760447991; h=from:from:reply-to:subject:subject:date:date:message-id:message-id: to:to:cc:cc:mime-version:mime-version:content-type:content-type: content-transfer-encoding:content-transfer-encoding: in-reply-to:in-reply-to:references:references; bh=JE+Kn2HjCSTg/qS1EjXYbWfpt2UDhrJQ2s6zULpht+4=; b=RnVJgjVA3Fn7aQ0zhP00Tom2evqdyeKrpAcE/pTudMQPYc9RoBK4RHqzpY37mrC+w67AXk rn+BogiWQxs/UF4rw+OzxNCZa7SxN+d61b1bH2QzsoenKDMi+MPeBe+LucJomb0095Mh3A PY2NfGRNCUii2yRKYTml87SztdkZ9Ns= Date: Tue, 14 Oct 2025 21:18:50 +0800 Precedence: bulk X-Mailing-List: cgroups@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Subject: Re: [PATCH] slab: Introduce __SECOND_OBJEXT_FLAG for objext_flags To: Vlastimil Babka , Harry Yoo Cc: Alexei Starovoitov , Andrew Morton , Johannes Weiner , Shakeel Butt , Michal Hocko , Roman Gushchin , Muchun Song , Suren Baghdasaryan , cgroups@vger.kernel.org, linux-mm@kvack.org, linux-kernel@vger.kernel.org, Hao Ge References: <20251014093124.300012-1-hao.ge@linux.dev> Content-Language: en-US X-Report-Abuse: Please report any abuse attempt to abuse@migadu.com and include these headers. From: Hao Ge In-Reply-To: Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 7bit X-Migadu-Flow: FLOW_OUT On 2025/10/14 20:49, Vlastimil Babka wrote: > On 10/14/25 12:27, Harry Yoo wrote: >> On Tue, Oct 14, 2025 at 05:31:24PM +0800, Hao Ge wrote: >>> From: Hao Ge >>> >>> We should not reuse the first bit for OBJEXTS_ALLOC_FAIL. >>> This is because the following scenarios may be encountered: >>> >>> Under heavy system load, certain sequences of events can trigger the >> Hi Hao, thanks for catching it! >> >> It's late at night and my brain is tired so I may be missing something, >> but let me leave comment anyway... >> >>> VM_BUG_ON_FOLIO(folio->memcg_data & MEMCG_DATA_OBJEXTS, folio) check: >> Should we check (folio->memcg_data != OBJEXTS_ALLOC_FAIL) && >> (folio->memcg_data & MEMCG_DATA_OBJEXTS) instead then? > Yes, we already went that direction, but seems we need to expand to more > places due to 7612833192d5 ("slab: Reuse first bit for OBJEXTS_ALLOC_FAIL"): > > https://lore.kernel.org/all/20250915200918.3855580-2-surenb@google.com/ >> Not clearing a valid folio->memcg_data is considered an error, but freeing a >> folio that is marked OBJEXTS_ALLOC_FAIL isn't. >> >>> 1. High system pressure may cause objext allocation failure for a slab. >>> 2. When objext allocation fails, slab->obj_exts is set to >>> OBJEXTS_ALLOC_FAIL (value 1). >>> 3. Later, this slab may enter the release process. >>> 4. During release of the associated folio, the existing >>> VM_BUG_ON_FOLIO check validates folio->memcg_data. >>> If the MEMCG_DATA_OBJEXTS bit is unexpectedly >>> set here, the bug check gets triggered. >>> >>> We have obtained the following logs: >>> [ 7108.343437] page: refcount:0 mapcount:0 mapping:0000000000000000 index:0xffff0002deb97600 pfn:0x31eb96 >>> [ 7108.343482] head: order:1 mapcount:0 entire_mapcount:0 nr_pages_mapped:0 pincount:0 >>> [ 7108.343500] memcg:1 >>> [ 7108.343507] flags: 0x17ffff800000040(head|node=0|zone=2|lastcpupid=0xfffff) >>> [ 7108.343523] raw: 017ffff800000040 ffff0000c000cac0 dead000000000100 0000000000000000 >>> [ 7108.343528] raw: ffff0002deb97600 0000000000240000 00000000ffffffff 0000000000000001 >>> [ 7108.343534] head: 017ffff800000040 ffff0000c000cac0 dead000000000100 0000000000000000 >>> [ 7108.343539] head: ffff0002deb97600 0000000000240000 00000000ffffffff 0000000000000001 >>> [ 7108.343562] head: 017ffff800000001 fffffdffcb7ae581 00000000ffffffff 00000000ffffffff >>> [ 7108.343569] head: ffffffffffffffff 0000000000000000 00000000ffffffff 0000000000000002 >>> [ 7108.343574] page dumped because: VM_BUG_ON_FOLIO(folio->memcg_data & MEMCG_DATA_OBJEXTS) >>> [ 7108.343601] ------------[ cut here ]------------ >>> [ 7108.343607] kernel BUG at ./include/linux/memcontrol.h:537! >>> [ 7108.343617] Internal error: Oops - BUG: 00000000f2000800 [#1] SMP >>> [ 7108.345751] Modules linked in: squashfs isofs vhost_vsock vhost_net vmw_vsock_virtio_transport_common vfio_iommu_type1 vhost vfio vsock vhost_iotlb iommufd tap binfmt_misc nfsv3 nfs_acl nfs lockd grace netfs tls rds dns_resolver tun brd overlay ntfs3 exfat btrfs blake2b_generic xor xor_neon raid6_pq loop sctp ip6_udp_tunnel udp_tunnel nft_fib_inet nft_fib_ipv4 nft_fib_ipv6 nft_fib nft_reject_inet nf_reject_ipv4 nf_reject_ipv6 nft_reject nft_ct nft_chain_nat nf_nat nf_conntrack nf_defrag_ipv6 nf_defrag_ipv4 nf_tables rfkill ip_set sunrpc vfat fat joydev sg sch_fq_codel nfnetlink virtio_gpu drm_client_lib virtio_dma_buf drm_shmem_helper sr_mod drm_kms_helper cdrom drm ghash_ce virtio_net virtio_scsi backlight virtio_console virtio_blk net_failover failover virtio_mmio dm_mirror dm_region_hash dm_log dm_multipath dm_mod fuse i2c_dev virtio_pci virtio_pci_legacy_dev virtio_pci_modern_dev virtio virtio_ring autofs4 aes_neon_bs aes_ce_blk [last unloaded: hwpoison_inject] >>> [ 7108.355662] CPU: 7 UID: 0 PID: 4470 Comm: kylin-process-m Kdump: loaded Not tainted 6.18.0-rc1-dirty #54 PREEMPT(voluntary) >>> [ 7108.356864] Hardware name: QEMU KVM Virtual Machine, BIOS unknown 2/2/2022 >>> [ 7108.357621] pstate: 60400005 (nZCv daif +PAN -UAO -TCO -DIT -SSBS BTYPE=--) >>> [ 7108.358981] pc : __free_frozen_pages+0xf18/0x18e8 >>> [ 7108.359834] lr : __free_frozen_pages+0xf18/0x18e8 >>> [ 7108.360379] sp : ffff8000a2bb7580 >>> [ 7108.360786] x29: ffff8000a2bb7580 x28: fffffdffcb7ae580 x27: fffffdffcb7ae580 >>> [ 7108.362013] x26: fffffdffcb7ae588 x25: 1fffffbff96f5cb1 x24: 1fffffbff96f5cb0 >>> [ 7108.362804] x23: ffff8000839d6ba0 x22: ffff8000839d6000 x21: 0000000000000000 >>> [ 7108.363596] x20: 0000000000000000 x19: 0000000000000001 x18: 0000000000000000 >>> [ 7108.364393] x17: 445f47434d454d20 x16: 2620617461645f67 x15: 636d656d3e2d6f69 >>> [ 7108.365498] x14: 6c6f66284f494c4f x13: 0000000000000001 x12: ffff600063fece93 >>> [ 7108.366317] x11: 1fffe00063fece92 x10: ffff600063fece92 x9 : dfff800000000000 >>> [ 7108.367610] x8 : 00009fff9c01316e x7 : ffff00031ff67493 x6 : 0000000000000001 >>> [ 7108.368455] x5 : ffff00031ff67490 x4 : ffff600063fece93 x3 : 0000000000000000 >>> [ 7108.369276] x2 : 0000000000000000 x1 : ffff000103fe5d40 x0 : 000000000000004c >>> [ 7108.370140] Call trace: >>> [ 7108.370463] __free_frozen_pages+0xf18/0x18e8 (P) >>> [ 7108.371011] free_frozen_pages+0x1c/0x30 >>> [ 7108.372040] __free_slab+0xd0/0x250 >>> [ 7108.372471] free_slab+0x38/0x118 >>> [ 7108.372882] free_to_partial_list+0x1d4/0x340 >>> [ 7108.373813] __slab_free+0x24c/0x348 >>> [ 7108.374253] ___cache_free+0xf0/0x110 >>> [ 7108.374699] qlist_free_all+0x78/0x130 >>> [ 7108.375156] kasan_quarantine_reduce+0x114/0x148 >>> [ 7108.375695] __kasan_slab_alloc+0x7c/0xb0 >>> [ 7108.376668] kmem_cache_alloc_noprof+0x164/0x5c8 >>> [ 7108.377206] __alloc_object+0x44/0x1f8 >>> [ 7108.377659] __create_object+0x34/0xc8 >>> [ 7108.378196] kmemleak_alloc+0xb8/0xd8 >>> [ 7108.378644] kmem_cache_alloc_noprof+0x368/0x5c8 >>> [ 7108.379224] getname_flags.part.0+0xa4/0x610 >>> [ 7108.379733] getname_flags+0x80/0xd8 >>> [ 7108.380169] do_sys_openat2+0xb4/0x178 >>> [ 7108.380921] __arm64_sys_openat+0x134/0x1d0 >>> [ 7108.381952] invoke_syscall+0xd4/0x258 >>> [ 7108.382408] el0_svc_common.constprop.0+0xb4/0x240 >>> [ 7108.382965] do_el0_svc+0x48/0x68 >>> [ 7108.383375] el0_svc+0x40/0xe0 >>> [ 7108.383757] el0t_64_sync_handler+0xa0/0xe8 >>> [ 7108.384465] el0t_64_sync+0x1ac/0x1b0 >>> [ 7108.385284] Code: 91398021 aa1b03e0 91138021 97fd35e3 (d4210000) >>> [ 7108.386553] SMP: stopping secondary CPUs >>> [ 7108.389714] Starting crashdump kernel... >>> [ 7108.390190] Bye! >>> >>> So, introduce __SECOND_OBJEXT_FLAG for objext_flags, adjust >>> the corresponding order accordingly, and ensure that OBJEXTS_ALLOC_FAIL >>> is no longer reused. >>> >>> Fixes: 7612833192d5 ("slab: Reuse first bit for OBJEXTS_ALLOC_FAIL") >> Hmm using a new bit was suggested at that time, but that would >> require bumping up the alignment when allocating slabobj_ext array? >> (see alloc_slab_obj_exts()) Hi Vlastimil and Harry Now I understand the alignment you're referring to, and indeed, my solution does require bumping up the alignment when allocating the slabobj_ext array. And syzbot ci has reported a bug: https://lore.kernel.org/all/68ee41cb.050a0220.91a22.020b.GAE@google.com/ This is because the data of slabobj_ext has been corrupted. >> And we can still distinguish two cases where >> >> 1) MEMCG_DATA_OBJEXTS is set, but upper bits are not set, >> so it should mean obj_exts allocation failed (OBJEXTS_ALLOC_FAIL), >> thus do not report error, or >> >> 2) MEMCG_DATA_OBJEXTS is set, and upper bits are also set, so someone >> did not clear a valid folio->memcg_data before freeing the folio >> (report error). >> >> without introducing a new bit, right? > Agreed. Okay, now I understand what you mean. I will send out the V2 version soon. Thank you for your guidance. Thanks Best Regards Hao > >>> Signed-off-by: Hao Ge >>> --- >>> include/linux/memcontrol.h | 16 ++++++---------- >>> 1 file changed, 6 insertions(+), 10 deletions(-) >>> >>> diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h >>> index 873e510d6f8d..8ea023944fac 100644 >>> --- a/include/linux/memcontrol.h >>> +++ b/include/linux/memcontrol.h >>> @@ -341,27 +341,23 @@ enum page_memcg_data_flags { >>> __NR_MEMCG_DATA_FLAGS = (1UL << 2), >>> }; >>> >>> -#define __OBJEXTS_ALLOC_FAIL MEMCG_DATA_OBJEXTS >>> #define __FIRST_OBJEXT_FLAG __NR_MEMCG_DATA_FLAGS >>> +#define __SECOND_OBJEXT_FLAG (__FIRST_OBJEXT_FLAG << 1) >>> >>> #else /* CONFIG_MEMCG */ >>> >>> -#define __OBJEXTS_ALLOC_FAIL (1UL << 0) >>> #define __FIRST_OBJEXT_FLAG (1UL << 0) >>> +#define __SECOND_OBJEXT_FLAG (1UL << 0) >>> >>> #endif /* CONFIG_MEMCG */ >>> >>> enum objext_flags { >>> - /* >>> - * Use bit 0 with zero other bits to signal that slabobj_ext vector >>> - * failed to allocate. The same bit 0 with valid upper bits means >>> - * MEMCG_DATA_OBJEXTS. >>> - */ >>> - OBJEXTS_ALLOC_FAIL = __OBJEXTS_ALLOC_FAIL, >>> + /* slabobj_ext vector failed to allocate */ >>> + OBJEXTS_ALLOC_FAIL = __FIRST_OBJEXT_FLAG, >>> /* slabobj_ext vector allocated with kmalloc_nolock() */ >>> - OBJEXTS_NOSPIN_ALLOC = __FIRST_OBJEXT_FLAG, >>> + OBJEXTS_NOSPIN_ALLOC = __SECOND_OBJEXT_FLAG, >>> /* the next bit after the last actual flag */ >>> - __NR_OBJEXTS_FLAGS = (__FIRST_OBJEXT_FLAG << 1), >>> + __NR_OBJEXTS_FLAGS = (__SECOND_OBJEXT_FLAG << 1), >>> }; >>> >>> #define OBJEXTS_FLAGS_MASK (__NR_OBJEXTS_FLAGS - 1) >>> -- >>> 2.25.1 >>>