| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <4c083431-bdba-15f8-bcbc-f80192cb02c8@linux.ibm.com>
Date: Fri, 9 Apr 2021 09:26:56 -0400
From: Tony Krowiak <akrowiak@...ux.ibm.com>
To: Halil Pasic <pasic@...ux.ibm.com>
Cc: linux-s390@...r.kernel.org, linux-kernel@...r.kernel.org,
kvm@...r.kernel.org, jjherne@...ux.ibm.com, freude@...ux.ibm.com,
borntraeger@...ibm.com, cohuck@...hat.com, mjrosato@...ux.ibm.com,
alex.williamson@...hat.com, kwankhede@...dia.com,
fiuczy@...ux.ibm.com, frankja@...ux.ibm.com, david@...hat.com,
hca@...ux.ibm.com, gor@...ux.ibm.com,
"Jason J . Herne" <jjherne@...ux.ibm.com>
Subject: Re: [PATCH v15 00/13] s390/vfio-ap: dynamic configuration support
On 4/8/21 4:38 PM, Halil Pasic wrote:
> On Tue, 6 Apr 2021 11:31:09 -0400
> Tony Krowiak <akrowiak@...ux.ibm.com> wrote:
>
>> Tony Krowiak (13):
>> s390/vfio-ap: fix circular lockdep when setting/clearing crypto masks
> The subsequent patches, re introduce this circular locking dependency
> problem. See my kernel messages for the details. The link we severe
> in the above patch is re-introduced at several places. One of them is
> assign_adapter_store().
Like in the patch referenced above, the lockdep splat occurs when
the APCB masks are set which requires acquisition of the kvm lock.
Patch 08/13, allow hot plug/unplug of AP resources using mdev,
introduces code that updates the APCB masks whenever an
adapter, domain or control domain is assigned or unassigned
as well as when a queue device is probed or removed.
I think the solution from the patch above can be implemented
here to resolve this problem.
>
> Regards,
> Halil
>
> [ +0.000236] vfio_ap matrix: MDEV: Registered
> [ +0.037919] vfio_mdev 4f77ad87-1e62-4959-8b7a-c677c98d2194: Adding to iommu group 1
> [ +0.000092] vfio_mdev 4f77ad87-1e62-4959-8b7a-c677c98d2194: MDEV: group_id = 1
>
> [Apr 8 22:31] ======================================================
> [ +0.000002] WARNING: possible circular locking dependency detected
> [ +0.000002] 5.12.0-rc6-00016-g5bea90816c56 #57 Not tainted
> [ +0.000002] ------------------------------------------------------
> [ +0.000002] CPU 1/KVM/6651 is trying to acquire lock:
> [ +0.000002] 00000000cef9d508 (&matrix_dev->lock){+.+.}-{3:3}, at: handle_pqap+0x56/0x1c8 [vfio_ap]
> [ +0.000011]
> but task is already holding lock:
> [ +0.000001] 00000000d41f4308 (&vcpu->mutex){+.+.}-{3:3}, at: kvm_vcpu_ioctl+0x90/0x898 [kvm]
> [ +0.000038]
> which lock already depends on the new lock.
>
> [ +0.000002]
> the existing dependency chain (in reverse order) is:
> [ +0.000001]
> -> #2 (&vcpu->mutex){+.+.}-{3:3}:
> [ +0.000004] validate_chain+0x796/0xa20
> [ +0.000006] __lock_acquire+0x420/0x7c8
> [ +0.000003] lock_acquire.part.0+0xec/0x1e8
> [ +0.000002] lock_acquire+0xb8/0x208
> [ +0.000002] __mutex_lock+0xa2/0x928
> [ +0.000005] mutex_lock_nested+0x32/0x40
> [ +0.000002] kvm_s390_cpus_to_pv+0x4e/0xf8 [kvm]
> [ +0.000019] kvm_s390_handle_pv+0x1ce/0x6b0 [kvm]
> [ +0.000018] kvm_arch_vm_ioctl+0x3ec/0x550 [kvm]
> [ +0.000019] kvm_vm_ioctl+0x40e/0x4a8 [kvm]
> [ +0.000018] __s390x_sys_ioctl+0xc0/0x100
> [ +0.000004] do_syscall+0x7e/0xd0
> [ +0.000043] __do_syscall+0xc0/0xd8
> [ +0.000004] system_call+0x72/0x98
> [ +0.000004]
> -> #1 (&kvm->lock){+.+.}-{3:3}:
> [ +0.000004] validate_chain+0x796/0xa20
> [ +0.000002] __lock_acquire+0x420/0x7c8
> [ +0.000002] lock_acquire.part.0+0xec/0x1e8
> [ +0.000002] lock_acquire+0xb8/0x208
> [ +0.000003] __mutex_lock+0xa2/0x928
> [ +0.000002] mutex_lock_nested+0x32/0x40
> [ +0.000002] kvm_arch_crypto_set_masks+0x4a/0x2b8 [kvm]
> [ +0.000018] vfio_ap_mdev_refresh_apcb+0xd0/0xe0 [vfio_ap]
> [ +0.000003] assign_adapter_store+0x1f2/0x240 [vfio_ap]
> [ +0.000003] kernfs_fop_write_iter+0x13e/0x1e0
> [ +0.000003] new_sync_write+0x10a/0x198
> [ +0.000003] vfs_write.part.0+0x196/0x290
> [ +0.000002] ksys_write+0x6c/0xf8
> [ +0.000003] do_syscall+0x7e/0xd0
> [ +0.000002] __do_syscall+0xc0/0xd8
> [ +0.000003] system_call+0x72/0x98
> [ +0.000002]
> -> #0 (&matrix_dev->lock){+.+.}-{3:3}:
> [ +0.000004] check_noncircular+0x16e/0x190
> [ +0.000002] check_prev_add+0xec/0xf38
> [ +0.000002] validate_chain+0x796/0xa20
> [ +0.000002] __lock_acquire+0x420/0x7c8
> [ +0.000002] lock_acquire.part.0+0xec/0x1e8
> [ +0.000002] lock_acquire+0xb8/0x208
> [ +0.000002] __mutex_lock+0xa2/0x928
> [ +0.000002] mutex_lock_nested+0x32/0x40
> [ +0.000003] handle_pqap+0x56/0x1c8 [vfio_ap]
> [ +0.000002] handle_pqap+0xe2/0x1d8 [kvm]
> [ +0.000019] kvm_handle_sie_intercept+0x134/0x248 [kvm]
> [ +0.000019] vcpu_post_run+0x2b6/0x580 [kvm]
> [ +0.000018] __vcpu_run+0x27e/0x388 [kvm]
> [ +0.000019] kvm_arch_vcpu_ioctl_run+0x10a/0x278 [kvm]
> [ +0.000018] kvm_vcpu_ioctl+0x2cc/0x898 [kvm]
> [ +0.000018] __s390x_sys_ioctl+0xc0/0x100
> [ +0.000003] do_syscall+0x7e/0xd0
> [ +0.000002] __do_syscall+0xc0/0xd8
> [ +0.000002] system_call+0x72/0x98
> [ +0.000003]
> other info that might help us debug this:
>
> [ +0.000001] Chain exists of:
> &matrix_dev->lock --> &kvm->lock --> &vcpu->mutex
>
> [ +0.000005] Possible unsafe locking scenario:
>
> [ +0.000001] CPU0 CPU1
> [ +0.000001] ---- ----
> [ +0.000002] lock(&vcpu->mutex);
> [ +0.000002] lock(&kvm->lock);
> [ +0.000002] lock(&vcpu->mutex);
> [ +0.000002] lock(&matrix_dev->lock);
> [ +0.000002]
> *** DEADLOCK ***
>
> [ +0.000002] 2 locks held by CPU 1/KVM/6651:
> [ +0.000002] #0: 00000000d41f4308 (&vcpu->mutex){+.+.}-{3:3}, at: kvm_vcpu_ioctl+0x90/0x898 [kvm]
> [ +0.000023] #1: 00000000da2fc508 (&kvm->srcu){....}-{0:0}, at: __vcpu_run+0x1ec/0x388 [kvm]
> [ +0.000021]
> stack backtrace:
> [ +0.000002] CPU: 6 PID: 6651 Comm: CPU 1/KVM Not tainted 5.12.0-rc6-00016-g5bea90816c56 #57
> [ +0.000004] Hardware name: IBM 8561 T01 701 (LPAR)
> [ +0.000001] Call Trace:
> [ +0.000002] [<00000002010e7ef0>] show_stack+0x90/0xf8
> [ +0.000007] [<00000002010fb5b2>] dump_stack+0xba/0x108
> [ +0.000002] [<000000020053feb6>] check_noncircular+0x16e/0x190
> [ +0.000003] [<0000000200541424>] check_prev_add+0xec/0xf38
> [ +0.000002] [<0000000200542a06>] validate_chain+0x796/0xa20
> [ +0.000003] [<0000000200545430>] __lock_acquire+0x420/0x7c8
> [ +0.000002] [<00000002005441a4>] lock_acquire.part.0+0xec/0x1e8
> [ +0.000002] [<0000000200544358>] lock_acquire+0xb8/0x208
> [ +0.000003] [<000000020110aeea>] __mutex_lock+0xa2/0x928
> [ +0.000002] [<000000020110b7a2>] mutex_lock_nested+0x32/0x40
> [ +0.000003] [<000003ff8060fb5e>] handle_pqap+0x56/0x1c8 [vfio_ap]
> [ +0.000003] [<000003ff80597412>] handle_pqap+0xe2/0x1d8 [kvm]
> [ +0.000018] [<000003ff8058c924>] kvm_handle_sie_intercept+0x134/0x248 [kvm]
> [ +0.000020] [<000003ff80588e96>] vcpu_post_run+0x2b6/0x580 [kvm]
> [ +0.000019] [<000003ff805893de>] __vcpu_run+0x27e/0x388 [kvm]
> [ +0.000018] [<000003ff80589d0a>] kvm_arch_vcpu_ioctl_run+0x10a/0x278 [kvm]
> [ +0.000019] [<000003ff805704d4>] kvm_vcpu_ioctl+0x2cc/0x898 [kvm]
> [ +0.000019] [<0000000200801ee8>] __s390x_sys_ioctl+0xc0/0x100
> [ +0.000003] [<000000020046e7ae>] do_syscall+0x7e/0xd0
> [ +0.000003] [<00000002010ffc20>] __do_syscall+0xc0/0xd8
> [ +0.000002] [<0000000201110c42>] system_call+0x72/0x98
> [ +0.000003] INFO: lockdep is turned off.
> [ +6.846296] vfio_mdev 4f77ad87-1e62-4959-8b7a-c677c98d2194: Removing from iommu group 1
> [ +0.000028] vfio_mdev 4f77ad87-1e62-4959-8b7a-c677c98d2194: MDEV: detaching iommu
> [ +0.007677] vfio_ap matrix: MDEV: Unregistering
>
>
>> s390/vfio-ap: use new AP bus interface to search for queue devices
>> s390/vfio-ap: move probe and remove callbacks to vfio_ap_ops.c
>> s390/vfio-ap: manage link between queue struct and matrix mdev
>> s390/vfio-ap: introduce shadow APCB
>> s390/vfio-ap: refresh guest's APCB by filtering APQNs assigned to mdev
>> s390/vfio-ap: allow assignment of unavailable AP queues to mdev device
>> s390/vfio-ap: allow hot plug/unplug of AP resources using mdev device
>> s390/zcrypt: driver callback to indicate resource in use
>> s390/vfio-ap: implement in-use callback for vfio_ap driver
>> s390/vfio-ap: sysfs attribute to display the guest's matrix
>> s390/zcrypt: notify drivers on config changed and scan complete
>> callbacks
>> s390/vfio-ap: update docs to include dynamic config support
Powered by blists - more mailing lists