Message-ID: <CAJuCfpH2FrQ2OCB7sjXNGf72mLJJPEefXzOu=Ywi=MO92p0HzQ@mail.gmail.com>
Date: Tue, 28 Oct 2025 15:52:46 -0700
From: Suren Baghdasaryan <surenb@...gle.com>
To: Hao Ge <hao.ge@...ux.dev>
Cc: Vlastimil Babka <vbabka@...e.cz>, Andrew Morton <akpm@...ux-foundation.org>, 
	Christoph Lameter <cl@...two.org>, David Rientjes <rientjes@...gle.com>, 
	Roman Gushchin <roman.gushchin@...ux.dev>, Harry Yoo <harry.yoo@...cle.com>, 
	Shakeel Butt <shakeel.butt@...ux.dev>, linux-mm@...ck.org, linux-kernel@...r.kernel.org, 
	Hao Ge <gehao@...inos.cn>
Subject: Re: [PATCH] codetag: debug: Handle existing CODETAG_EMPTY in
 mark_objexts_empty for slabobj_ext

On Tue, Oct 28, 2025 at 9:58 AM Hao Ge <hao.ge@...ux.dev> wrote:
>
> HI Suren
>
>
> On 2025/10/29 00:19, Suren Baghdasaryan wrote:
> > On Mon, Oct 27, 2025 at 6:31 PM Hao Ge <hao.ge@...ux.dev> wrote:
> >> Hi Suren
> >>
> >>
> >> On 2025/10/28 03:44, Suren Baghdasaryan wrote:
> >>> On Mon, Oct 27, 2025 at 1:53 AM Hao Ge <hao.ge@...ux.dev> wrote:
> >>>> From: Hao Ge <gehao@...inos.cn>
> >>>>
> >>>> Even though obj_exts was created with the __GFP_NO_OBJ_EXT flag,
> >>>> objects in the same slab may have their extensions allocated via
> >>>> alloc_slab_obj_exts, and handle_failed_objexts_alloc may be called
> >>>> within alloc_slab_obj_exts to set their codetag to CODETAG_EMPTY.
> >>>>
> >>>> Therefore, both NULL and CODETAG_EMPTY are valid for the codetag of
> >>>> slabobj_ext, as we do not need to re-set it to CODETAG_EMPTY if it
> >>>> is already CODETAG_EMPTY. It also resolves the warning triggered when
> >>>> the codetag is CODETAG_EMPTY during slab freeing.
> >>> I'm not sure what scenario leads to handle_failed_objexts_alloc() and
> >>> mark_objexts_empty() being used against the same codetag reference.
> >>> Could you please explain the exact scenario you hit?
> >>>
> >>> handle_failed_objexts_alloc() assigns CODETAG_EMPTY to the elements of
> >>> the obj_exts vector while mark_objexts_empty() assigns CODETAG_EMPTY
> >>> to the obj_ext of the obj_exts vector itself. In what case do these
> >>> two calls operate on the same reference?
> >> This issue also occurred during our memory stress testing.
> >>
> >> I apologize for the incorrect description in my commit message:
> >>
> >> The possibility of its occurrence should be as follows:
> >>
> >> When a slab allocates a slabobj_ext, the slab to which this slabobj_ext
> >> belongs may have already allocated its own slabobj_ext
> >>
> >> and called handle_failed_objexts_alloc. That is to say, the codetag of
> >> this slabobj_ext has been set to CODETAG_EMPTY.
> > Ah, ok. Let's see if I understood this correctly. The slab from which
> > slabobj_ext objects are allocated (let's call it slabA) tries to
> > allocate its own slabobj_ext, fails, then succeeds and calls
> > handle_failed_objexts_alloc() which sets all elements of its
> > slabA->obj_exts[] to CODETAG_EMPTY. Then slabobj_ext object is
> > allocated from slabA with __GFP_NO_OBJ_EXT, therefore it leaves
> > slabA->obj_exts[obj_idx] intact and equal to CODETAG_EMPTY instead of
> > NULL. At this point, slabB->obj_exts points to an object in slabA.
> > When slabB is freed and free_slab_obj_exts() gets called to free
> > slabB->obj_exts, it detects that slabA->obj_exts[obj_idx] is not NULL
> > and generates this warning. In your callstack stack the freeing seems
> > to happen inside kmem_cache_alloc_noprof() and I think that's because
> > we lost the race and freeing the vector we allocated here:
> > https://elixir.bootlin.com/linux/v6.18-rc3/source/mm/slub.c#L2165 but
> > the scenario is pretty much the same as if we didn't lose the race.
> > The warning would happen later on I think, when we would be freeing
> > slabB->obj_exts. The key for reproducing this is to fail allocation
> > for a slab that is used to allocate slabobj_ext objects. Is my
> > understanding correct?
>
>
> Yes, your explanation is extremely detailed and thorough.

Ok, let's work on the changelog them. How about something like this:
```
When alloc_slab_obj_exts() fails and then later succeeds in allocating
a slab extension vector, it calls handle_failed_objexts_alloc() to
mark all objects in the vector as empty. As a result all objects in
this slab (slabA) will have their extensions set to CODETAG_EMPTY.
Later on if this slabA is used to allocate a slabobj_ext vector for
another slab (slabB), we end up with the slabB->obj_exts pointing to a
slabobj_ext vector that itself has a non-NULL slabobj_ext equal to
CODETAG_EMPTY. When slabB gets freed, free_slab_obj_exts() is called
to free slabB->obj_exts vector. free_slab_obj_exts() calls
mark_objexts_empty(slabB->obj_exts) which will generate a warning
because it expects slabobj_ext vectors to have a NULL obj_ext, not
CODETAG_EMPTY.

Modify mark_objexts_empty() to skip the warning and setting the
obj_ext value if it's already set to CODETAG_EMPTY.
```
I know it's still hard to read but I could not come up with a better one :/

>
>
> > BTW, your fix seems fine to me, I just want to understand the scenario
> > so that we can explain it correctly.
> >
> >> To quickly detect this WARN, I modified the code
> >> from:WARN_ON(slab_exts[offs].ref.ct) to WARN_ON(slab_exts[offs].ref.ct
> >> == 1);
> >>
> >> We then obtained this message:
> >>
> >> [21630.898561] ------------[ cut here ]------------
> >> [21630.898596] kernel BUG at mm/slub.c:2050!
> >> [21630.898611] Internal error: Oops - BUG: 00000000f2000800 [#1] SMP
> >> [21630.900372] Modules linked in: squashfs isofs vfio_iommu_type1
> >> vhost_vsock vfio vhost_net vmw_vsock_virtio_transport_common vhost tap
> >> vhost_iotlb iommufd vsock 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 sr_mod cdrom drm_client_lib virtio_dma_buf drm_shmem_helper
> >> drm_kms_helper drm ghash_ce backlight virtio_net virtio_blk virtio_scsi
> >> net_failover virtio_console 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]
> >> [21630.909177] CPU: 3 UID: 0 PID: 3787 Comm: kylin-process-m Kdump:
> >> loaded Tainted: G        W           6.18.0-rc1+ #74 PREEMPT(voluntary)
> >> [21630.910495] Tainted: [W]=WARN
> >> [21630.910867] Hardware name: QEMU KVM Virtual Machine, BIOS unknown
> >> 2/2/2022
> >> [21630.911625] pstate: 80400005 (Nzcv daif +PAN -UAO -TCO -DIT -SSBS
> >> BTYPE=--)
> >> [21630.912392] pc : __free_slab+0x228/0x250
> >> [21630.912868] lr : __free_slab+0x18c/0x250[21630.913334] sp :
> >> ffff8000a02f73e0
> >> [21630.913830] x29: ffff8000a02f73e0 x28: fffffdffc43fc800 x27:
> >> ffff0000c0011c40
> >> [21630.914677] x26: ffff0000c000cac0 x25: ffff00010fe5e5f0 x24:
> >> ffff000102199b40
> >> [21630.915469] x23: 0000000000000003 x22: 0000000000000003 x21:
> >> ffff0000c0011c40
> >> [21630.916259] x20: fffffdffc4086600 x19: fffffdffc43fc800 x18:
> >> 0000000000000000
> >> [21630.917048] x17: 0000000000000000 x16: 0000000000000000 x15:
> >> 0000000000000000
> >> [21630.917837] x14: 0000000000000000 x13: 0000000000000000 x12:
> >> ffff70001405ee66
> >> [21630.918640] x11: 1ffff0001405ee65 x10: ffff70001405ee65 x9 :
> >> ffff800080a295dc
> >> [21630.919442] x8 : ffff8000a02f7330 x7 : 0000000000000000 x6 :
> >> 0000000000003000
> >> [21630.920232] x5 : 0000000024924925 x4 : 0000000000000001 x3 :
> >> 0000000000000007
> >> [21630.921021] x2 : 0000000000001b40 x1 : 000000000000001f x0 :
> >> 0000000000000001
> >> [21630.921810] Call trace:
> >> [21630.922130]  __free_slab+0x228/0x250 (P)
> >> [21630.922669]  free_slab+0x38/0x118
> >> [21630.923079]  free_to_partial_list+0x1d4/0x340
> >> [21630.923591]  __slab_free+0x24c/0x348
> >> [21630.924024]  ___cache_free+0xf0/0x110
> >> [21630.924468]  qlist_free_all+0x78/0x130
> >> [21630.924922]  kasan_quarantine_reduce+0x114/0x148
> >> [21630.925525]  __kasan_slab_alloc+0x7c/0xb0
> >> [21630.926006]  kmem_cache_alloc_noprof+0x164/0x5c8
> >> [21630.926699]  __alloc_object+0x44/0x1f8
> >> [21630.927153]  __create_object+0x34/0xc8
> >> [21630.927604]  kmemleak_alloc+0xb8/0xd8
> >> [21630.928052]  kmem_cache_alloc_noprof+0x368/0x5c8
> >> [21630.928606]  getname_flags.part.0+0xa4/0x610
> >> [21630.929112]  getname_flags+0x80/0xd8
> >> [21630.929557]  vfs_fstatat+0xc8/0xe0
> >> [21630.929975]  __do_sys_newfstatat+0xa0/0x100
> >> [21630.930469]  __arm64_sys_newfstatat+0x90/0xd8
> >> [21630.931046]  invoke_syscall+0xd4/0x258
> >> [21630.931685]  el0_svc_common.constprop.0+0xb4/0x240
> >> [21630.932467]  do_el0_svc+0x48/0x68
> >> [21630.932972]  el0_svc+0x40/0xe0
> >> [21630.933472]  el0t_64_sync_handler+0xa0/0xe8
> >> [21630.934151]  el0t_64_sync+0x1ac/0x1b0
> >> [21630.934923] Code: aa1803e0 97ffef2b a9446bf9 17ffff9c (d4210000)
> >> [21630.936461] SMP: stopping secondary CPUs
> >> [21630.939550] Starting crashdump kernel...
> >> [21630.940108] Bye!
> >>
> >>>> Fixes: 09c46563ff6d ("codetag: debug: introduce OBJEXTS_ALLOC_FAIL to mark failed slab_ext allocations")
> >>>> Signed-off-by: Hao Ge <gehao@...inos.cn>
> >>>> ---
> >>>>    mm/slub.c | 12 +++++++++++-
> >>>>    1 file changed, 11 insertions(+), 1 deletion(-)
> >>>>
> >>>> diff --git a/mm/slub.c b/mm/slub.c
> >>>> index d4367f25b20d..cda8f75b72e7 100644
> >>>> --- a/mm/slub.c
> >>>> +++ b/mm/slub.c
> >>>> @@ -2046,7 +2046,17 @@ static inline void mark_objexts_empty(struct slabobj_ext *obj_exts)
> >>>>           if (slab_exts) {
> >>>>                   unsigned int offs = obj_to_index(obj_exts_slab->slab_cache,
> >>>>                                                    obj_exts_slab, obj_exts);
> >>>> -               /* codetag should be NULL */
> >>>> +
> >>>> +               /*
> >>>> +                * codetag should be either NULL or CODETAG_EMPTY.
> >>>> +                * When the same slab calls handle_failed_objexts_alloc,
> >>>> +                * it will set us to CODETAG_EMPTY.
> >>>> +                *
> >>>> +                * If codetag is already CODETAG_EMPTY, no action is needed here.
> >>>> +                */

This comment is more confusing than helpful IMO. I would leave the
current comment intact and just add this check before it like this:

+               if (unlikely(is_codetag_empty(&slab_exts[offs].ref)))
+                       return;
+
                /* codetag should be NULL */
                WARN_ON(slab_exts[offs].ref.ct);
                ...

> >>>> +               if (unlikely(is_codetag_empty(&slab_exts[offs].ref)))
> >>>> +                       return;
> >>>> +
> >>>>                   WARN_ON(slab_exts[offs].ref.ct);
> >>>>                   set_codetag_empty(&slab_exts[offs].ref);
> >>>>           }
> >>>> --
> >>>> 2.25.1
> >>>>

Powered by blists - more mailing lists