Am 09.02.22 um 16:22 schrieb Michael Mueller:
This patch enables the ultravisor adapter interruption vitualization
support indicated by UV feature BIT_UV_FEAT_AIV. This allows ISC
interruption injection directly into the GISA IPM for PV kvm guests.
Hardware that does not support this feature will continue to use the
UV interruption interception method to deliver ISC interruptions to
PV kvm guests. For this purpose, the ECA_AIV bit for all guest cpus
will be cleared and the GISA will be disabled during PV CPU setup.
In addition a check in __inject_io() has been removed. That reduces the
required instructions for interruption handling for PV and traditional
kvm guests.
Signed-off-by: Michael Mueller <mimu@xxxxxxxxxxxxx>
The CI said the following with gisa_disable in the calltrace.
Will drop from next for now.
LOCKDEP_CIRCULAR (suite: kvm-unit-tests-kvm, case: -)
WARNING: possible circular locking dependency detected
5.17.0-20220221.rc5.git1.b8f0356a093a.300.fc35.s390x+debug #1 Not tainted
------------------------------------------------------
qemu-system-s39/161139 is trying to acquire lock:
0000000280dc0b98 (&kvm->lock){+.+.}-{3:3}, at: kvm_s390_set_tod_clock+0x36/0x220 [kvm]
but task is already holding lock:
0000000280f4e4b8 (&vcpu->mutex){+.+.}-{3:3}, at: kvm_vcpu_ioctl+0x9a/0xa40 [kvm]
which lock already depends on the new lock.
the existing dependency chain (in reverse order) is:
-> #1 (&vcpu->mutex){+.+.}-{3:3}:
__lock_acquire+0x604/0xbd8
lock_acquire.part.0+0xe2/0x250
lock_acquire+0xb0/0x200
__mutex_lock+0x9e/0x8a0
mutex_lock_nested+0x32/0x40
kvm_s390_gisa_disable+0xa4/0x130 [kvm]
kvm_s390_handle_pv+0x718/0x778 [kvm]
kvm_arch_vm_ioctl+0x4ac/0x5f8 [kvm]
kvm_vm_ioctl+0x336/0x530 [kvm]
__s390x_sys_ioctl+0xbe/0x100
__do_syscall+0x1da/0x208
system_call+0x82/0xb0
-> #0 (&kvm->lock){+.+.}-{3:3}:
check_prev_add+0xe0/0xed8
validate_chain+0x736/0xb20
__lock_acquire+0x604/0xbd8
lock_acquire.part.0+0xe2/0x250
lock_acquire+0xb0/0x200
__mutex_lock+0x9e/0x8a0
mutex_lock_nested+0x32/0x40
kvm_s390_set_tod_clock+0x36/0x220 [kvm]
kvm_s390_handle_b2+0x378/0x728 [kvm]
kvm_handle_sie_intercept+0x13a/0x448 [kvm]
vcpu_post_run+0x28e/0x560 [kvm]
__vcpu_run+0x266/0x388 [kvm]
kvm_arch_vcpu_ioctl_run+0x10a/0x270 [kvm]
kvm_vcpu_ioctl+0x27c/0xa40 [kvm]
__s390x_sys_ioctl+0xbe/0x100
__do_syscall+0x1da/0x208
system_call+0x82/0xb0
other info that might help us debug this:
Possible unsafe locking scenario:
CPU0 CPU1
---- ----
lock(&vcpu->mutex);
lock(&kvm->lock);
lock(&vcpu->mutex);
lock(&kvm->lock);
*** DEADLOCK ***
2 locks held by qemu-system-s39/161139:
#0: 0000000280f4e4b8 (&vcpu->mutex){+.+.}-{3:3}, at: kvm_vcpu_ioctl+0x9a/0xa40 [kvm]
#1: 0000000280dc47c8 (&kvm->srcu){....}-{0:0}, at: __vcpu_run+0x1d4/0x388 [kvm]
stack backtrace:
CPU: 10 PID: 161139 Comm: qemu-system-s39 Not tainted 5.17.0-20220221.rc5.git1.b8f0356a093a.300.fc35.s390x+debug #1
Hardware name: IBM 8561 T01 701 (LPAR)
Call Trace:
[<00000001da4e89de>] dump_stack_lvl+0x8e/0xc8
[<00000001d9876c56>] check_noncircular+0x136/0x158
[<00000001d9877c70>] check_prev_add+0xe0/0xed8
[<00000001d987919e>] validate_chain+0x736/0xb20
[<00000001d987b23c>] __lock_acquire+0x604/0xbd8
[<00000001d987c432>] lock_acquire.part.0+0xe2/0x250
[<00000001d987c650>] lock_acquire+0xb0/0x200
[<00000001da4f72ae>] __mutex_lock+0x9e/0x8a0
[<00000001da4f7ae2>] mutex_lock_nested+0x32/0x40
[<000003ff8070cd6e>] kvm_s390_set_tod_clock+0x36/0x220 [kvm]
[<000003ff8071dd68>] kvm_s390_handle_b2+0x378/0x728 [kvm]
[<000003ff8071146a>] kvm_handle_sie_intercept+0x13a/0x448 [kvm]
[<000003ff8070dd46>] vcpu_post_run+0x28e/0x560 [kvm]
[<000003ff8070e27e>] __vcpu_run+0x266/0x388 [kvm]
[<000003ff8070eba2>] kvm_arch_vcpu_ioctl_run+0x10a/0x270 [kvm]
[<000003ff806f4044>] kvm_vcpu_ioctl+0x27c/0xa40 [kvm]
[<00000001d9b47ac6>] __s390x_sys_ioctl+0xbe/0x100
[<00000001da4ec152>] __do_syscall+0x1da/0x208
[<00000001da4fec42>] system_call+0x82/0xb0
INFO: lockdep is turned off.
