Kernel KVM virtualization development
 help / color / mirror / Atom feed
From: sashiko-bot@kernel.org
To: "Peng Hao" <flyingpenghao@gmail.com>
Cc: kvm@vger.kernel.org
Subject: Re: [RFC PATCH 2/3] KVM: Add a module param to disable the per-VM debugfs directories
Date: Mon, 16 Nov 2026 03:43:26 +0000	[thread overview]
Message-ID: <sashiko-outbox-216141@kernel.org> (raw)
In-Reply-To: <20261116032748.54582-3-flyingpeng@tencent.com>

Thank you for your contribution! Sashiko AI review found 1 potential
issue to consider:
- [Medium] KVM: debugfs leak and UAF in
  kvm_nested_s2_ptdump_create_debugfs()
--

commit 6b8cad730d46b448b05e32a4f22f6d9fe9cf399b
Author: Peng Hao <flyingpenghao@gmail.com>

KVM: Add a module param to disable the per-VM debugfs directories

This commit adds the kvm.per_vm_debugfs module parameter to allow skipping the
creation of per-VM and per-vCPU debugfs directories and files.

> diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
> index fff19e9239d2b..54a2924f9b928 100644
> --- a/virt/kvm/kvm_main.c
> +++ b/virt/kvm/kvm_main.c

[ ... ]

> @@ -1024,7 +1032,7 @@ static int kvm_create_vm_debugfs(struct kvm *kvm, const char *fdname)
>  	int kvm_debugfs_num_entries = kvm_vm_stats_header.num_desc +
>  				      kvm_vcpu_stats_header.num_desc;
>  
> -	if (!debugfs_initialized())
> +	if (!debugfs_initialized() || !per_vm_debugfs)
>  		return 0;

[Severity: Medium]
On arm64 with CONFIG_PTDUMP_STAGE2_DEBUGFS=y, does returning early here when
per_vm_debugfs is false cause kvm_nested_s2_ptdump_create_debugfs() to leak
nested_mmu* files in the root of debugfs and leave dangling pointers?

kvm_create_vm() initializes kvm->debugfs_dentry to ERR_PTR(-ENOENT), so
callers passing it as a parent to debugfs_create_file() or
debugfs_create_dir() fail as expected.  However, kvm->arch.debugfs_nv_dentry
is zero-allocated (NULL) and is only populated when kvm_create_vm_debugfs()
reaches kvm_arch_create_vm_debugfs() -> kvm_s2_ptdump_create_debugfs().

When per_vm_debugfs is false while debugfs_initialized() is true,
kvm->arch.debugfs_nv_dentry remains NULL.  Initializing a nested vCPU in
kvm_vcpu_init_nested() then calls kvm_nested_s2_ptdump_create_debugfs(),
passing NULL as the parent dentry:

arch/arm64/kvm/ptdump.c:kvm_nested_s2_ptdump_create_debugfs() {
...
	debugfs_create_file(file_name, 0400,
			    mmu->arch->debugfs_nv_dentry, mmu,
			    &kvm_ptdump_guest_fops);
}

debugfs interprets a NULL parent as the root of the debugfs mount and creates
/sys/kernel/debug/nested_mmu* with inode->i_private pointing to mmu.  When
the VM is destroyed, these root files are never removed while
kvm_destroy_nested() frees the nested_mmus arrays and kvm_destroy_vm() frees
struct kvm, so opening /sys/kernel/debug/nested_mmu* afterwards triggers a
use-after-free in kvm_ptdump_guest_open().

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261116032748.54582-1-flyingpeng@tencent.com?part=2

  reply	other threads:[~2026-11-16  3:43 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-11-16  3:27 [RFC PATCH 0/3] KVM: Rework how halt-polling stats and per-VM debugfs use memory Peng Hao
2026-11-16  3:27 ` [RFC PATCH 1/3] KVM: Add the halt-polling time to the kvm_vcpu_wakeup tracepoint Peng Hao
2026-11-16  3:27 ` [RFC PATCH 2/3] KVM: Add a module param to disable the per-VM debugfs directories Peng Hao
2026-11-16  3:43   ` sashiko-bot [this message]
2026-11-16  3:27 ` [RFC PATCH 3/3] KVM: Move the halt-polling histograms out of struct kvm_vcpu_stat Peng Hao

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=sashiko-outbox-216141@kernel.org \
    --to=sashiko-bot@kernel.org \
    --cc=flyingpenghao@gmail.com \
    --cc=kvm@vger.kernel.org \
    --cc=sashiko-reviews@lists.linux.dev \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox