Re: [PATCH] x86/rcutorture move CONFIG_HYPERVISOR_GUEST to kvm-test-1-run.sh

2024-04-29 Thread Zhouyi Zhou
On Tue, Apr 30, 2024 at 7:22 AM Paul E. McKenney wrote: > > On Sat, Apr 27, 2024 at 12:56:26AM +, Zhouyi Zhou wrote: > > CONFIG_HYPERVISOR_GUEST is a x86 specific kernel option, move to > > kvm-test-1-run.sh to avoid ConfigFragment.diags in non-x86 platforms. > > > > Tested in both PPC VM of

Re: [PATCH] x86/rcutorture move CONFIG_HYPERVISOR_GUEST to kvm-test-1-run.sh

2024-04-29 Thread Paul E. McKenney
On Sat, Apr 27, 2024 at 12:56:26AM +, Zhouyi Zhou wrote: > CONFIG_HYPERVISOR_GUEST is a x86 specific kernel option, move to > kvm-test-1-run.sh to avoid ConfigFragment.diags in non-x86 platforms. > > Tested in both PPC VM of Open Source lab of Oregon State University and > local x86_64

Re: [PATCH 0/4] rcu/nocb cleanups

2024-04-29 Thread Paul E. McKenney
On Thu, Apr 25, 2024 at 04:18:31PM +0200, Frederic Weisbecker wrote: > Hi, > > Some cleanups before converting (de-)offloading to only work on offline > CPUs. Seeing no objections, I have queued this series for testing and review. Thanx,

Re: [PATCH] rcu: Fix suspicious RCU usage in __do_softirq()

2024-04-29 Thread Paul E. McKenney
On Sat, Apr 27, 2024 at 06:28:08PM +0800, Zqiang wrote: > Currently, the condition "__this_cpu_read(ksoftirqd) == current" is > checked to ensure the rcu_softirq_qs() is invoked in ksoftirqd tasks > context for non-RT kernels. however, in some scenarios, this condition > will be broken. > >

Re: [PATCH] Revert "rcu-tasks: Fix synchronize_rcu_tasks() VS zap_pid_ns_processes()"

2024-04-29 Thread Paul E. McKenney
On Mon, Apr 29, 2024 at 01:03:28PM +0200, Oleg Nesterov wrote: > On 04/25, Frederic Weisbecker wrote: > > > > This reverts commit 28319d6dc5e2ffefa452c2377dd0f71621b5bff0. The race > > it fixed was subject to conditions that don't exist anymore since: > > > > 1612160b9127 ("rcu-tasks:

Re: [PATCH] rcu: Remove unreachable logic

2024-04-29 Thread Mathieu Desnoyers
On 2024-04-29 13:34, Alan Huang wrote: call_rcu_core is only called from __call_rcu_common with interrupt disabled. This patch thus removes the unreachable logic and the would-be unused 'flags' parameter. Nack. call_rcu_core() receives a @flags parameter which are the _saved_ flags as they

[PATCH] rcu: Remove unreachable logic

2024-04-29 Thread Alan Huang
call_rcu_core is only called from __call_rcu_common with interrupt disabled. This patch thus removes the unreachable logic and the would-be unused 'flags' parameter. Signed-off-by: Alan Huang --- kernel/rcu/tree.c | 35 ++- 1 file changed, 2 insertions(+), 33

Re: [PATCH] Revert "rcu-tasks: Fix synchronize_rcu_tasks() VS zap_pid_ns_processes()"

2024-04-29 Thread Oleg Nesterov
On 04/25, Frederic Weisbecker wrote: > > This reverts commit 28319d6dc5e2ffefa452c2377dd0f71621b5bff0. The race > it fixed was subject to conditions that don't exist anymore since: > > 1612160b9127 ("rcu-tasks: Eliminate deadlocks involving do_exit() and > RCU tasks") > > This latter commit