| 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
| ||
|
Date: Tue, 14 May 2024 07:54:40 -0700
From: "Paul E. McKenney" <paulmck@...nel.org>
To: Zqiang <qiang.zhang1211@...il.com>
Cc: frederic@...nel.org, neeraj.upadhyay@...nel.org, joel@...lfernandes.org,
rcu@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] rcu/nocb: Fix using smp_processor_id() in preemptible
warning
On Thu, May 09, 2024 at 03:40:46PM +0800, Zqiang wrote:
> Currently, the this_cpu_ptr(&rcu_data) in rcu_rdp_is_offloaded() is called
> before the condition "!(IS_ENABLED(CONFIG_PREEMPT_COUNT) && preemptible())"
> is checked, and occurs in preemptible task context, this will trigger the
> following warning.
>
> [ 4.106221][ T18] BUG: using smp_processor_id() in preemptible [00000000] code: rcuop/0/18
> [ 4.107796][ T18] caller is debug_smp_processor_id (lib/smp_processor_id.c:61)
> [ 4.108547][ T18] CPU: 0 PID: 18 Comm: rcuop/0 Not tainted 6.9.0-rc2-00079-g4c66bc7cacc0 #1
> [ 4.109667][ T18] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.16.2-debian-1.16.2-1 04/01/2014
> [ 4.111064][ T18] Call Trace:
> [ 4.111064][ T18] <TASK>
> [ 4.111064][ T18] dump_stack_lvl (lib/dump_stack.c:116)
> [ 4.111064][ T18] dump_stack (lib/dump_stack.c:124)
> [ 4.111064][ T18] check_preemption_disabled (arch/x86/include/asm/preempt.h:84 (discriminator 15) lib/smp_processor_id.c:53 (discriminator 15))
> [ 4.111064][ T18] debug_smp_processor_id (lib/smp_processor_id.c:61)
> [ 4.111064][ T18] rcu_rdp_is_offloaded (kernel/rcu/tree_plugin.h:27 (discriminator 1))
> [ 4.111064][ T18] nocb_cb_wait (kernel/rcu/tree_nocb.h:936 (discriminator 2))
> [ 4.111064][ T18] rcu_nocb_cb_kthread (kernel/rcu/tree_nocb.h:983 (discriminator 1))
> [ 4.111064][ T18] ? nocb_cb_wait (kernel/rcu/tree_nocb.h:976)
> [ 4.111064][ T18] kthread (kernel/kthread.c:388)
> [ 4.111064][ T18] ? kthread (kernel/kthread.c:373 (discriminator 2))
> [ 4.111064][ T18] ? kthread_complete_and_exit (kernel/kthread.c:341)
> [ 4.111064][ T18] ret_from_fork (arch/x86/kernel/process.c:153)
> [ 4.111064][ T18] ? kthread_complete_and_exit (kernel/kthread.c:341)
> [ 4.111064][ T18] ret_from_fork_asm (arch/x86/entry/entry_64.S:256)
> [ 4.111064][ T18] </TASK>
>
> This commit fix this warning by priority check the condition
> "!(IS_ENABLED(CONFIG_PREEMPT_COUNT) && preemptible())" , to
> ensure whether the this_cpu_ptr(&rcu_data) can be executed in
> rcu_rdp_is_offloaded().
>
> Fixes: 8feeeba60711 ("rcu/nocb: Use kthread parking instead of ad-hoc implementation")
> Tested-by: kernel test robot <oliver.sang@...el.com>
> Signed-off-by: Zqiang <qiang.zhang1211@...il.com>
Hearing no objections, I have queued this wordsmithed version. As always,
please let me know if I have messed anything up.
Thanx, Paul
------------------------------------------------------------------------
commit 5271ad1de0fbcf0bd9caebcf721670c164e5fa9c
Author: Zqiang <qiang.zhang1211@...il.com>
Date: Thu May 9 15:40:46 2024 +0800
rcu/nocb: Don't use smp_processor_id() in preemptible code
Currently, rcu_rdp_is_offloaded() invokes this_cpu_ptr(&rcu_data) before
the condition "!(IS_ENABLED(CONFIG_PREEMPT_COUNT) && preemptible())"
is checked. When invoked in preemptible context in preemptible kernels,
this will trigger the following warning:
[ 4.106221][ T18] BUG: using smp_processor_id() in preemptible [00000000] code: rcuop/0/18
[ 4.107796][ T18] caller is debug_smp_processor_id (lib/smp_processor_id.c:61)
[ 4.108547][ T18] CPU: 0 PID: 18 Comm: rcuop/0 Not tainted 6.9.0-rc2-00079-g4c66bc7cacc0 #1
[ 4.109667][ T18] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.16.2-debian-1.16.2-1 04/01/2014
[ 4.111064][ T18] Call Trace:
[ 4.111064][ T18] <TASK>
[ 4.111064][ T18] dump_stack_lvl (lib/dump_stack.c:116)
[ 4.111064][ T18] dump_stack (lib/dump_stack.c:124)
[ 4.111064][ T18] check_preemption_disabled (arch/x86/include/asm/preempt.h:84 (discriminator 15) lib/smp_processor_id.c:53 (discriminator 15))
[ 4.111064][ T18] debug_smp_processor_id (lib/smp_processor_id.c:61)
[ 4.111064][ T18] rcu_rdp_is_offloaded (kernel/rcu/tree_plugin.h:27 (discriminator 1))
[ 4.111064][ T18] nocb_cb_wait (kernel/rcu/tree_nocb.h:936 (discriminator 2))
[ 4.111064][ T18] rcu_nocb_cb_kthread (kernel/rcu/tree_nocb.h:983 (discriminator 1))
[ 4.111064][ T18] ? nocb_cb_wait (kernel/rcu/tree_nocb.h:976)
[ 4.111064][ T18] kthread (kernel/kthread.c:388)
[ 4.111064][ T18] ? kthread (kernel/kthread.c:373 (discriminator 2))
[ 4.111064][ T18] ? kthread_complete_and_exit (kernel/kthread.c:341)
[ 4.111064][ T18] ret_from_fork (arch/x86/kernel/process.c:153)
[ 4.111064][ T18] ? kthread_complete_and_exit (kernel/kthread.c:341)
[ 4.111064][ T18] ret_from_fork_asm (arch/x86/entry/entry_64.S:256)
[ 4.111064][ T18] </TASK>
This commit therefore fixes this warning by checking the condition
"!(IS_ENABLED(CONFIG_PREEMPT_COUNT) && preemptible())" before invoking
this_cpu_ptr(), thus avoiding preemptible invocations.
Fixes: 8feeeba60711 ("rcu/nocb: Use kthread parking instead of ad-hoc implementation")
Tested-by: kernel test robot <oliver.sang@...el.com>
Signed-off-by: Zqiang <qiang.zhang1211@...il.com>
Signed-off-by: Paul E. McKenney <paulmck@...nel.org>
diff --git a/kernel/rcu/tree_plugin.h b/kernel/rcu/tree_plugin.h
index 31c539f09c150..475e20ebec5a5 100644
--- a/kernel/rcu/tree_plugin.h
+++ b/kernel/rcu/tree_plugin.h
@@ -28,8 +28,8 @@ static bool rcu_rdp_is_offloaded(struct rcu_data *rdp)
!(lockdep_is_held(&rcu_state.barrier_mutex) ||
(IS_ENABLED(CONFIG_HOTPLUG_CPU) && lockdep_is_cpus_held()) ||
rcu_lockdep_is_held_nocb(rdp) ||
- (rdp == this_cpu_ptr(&rcu_data) &&
- !(IS_ENABLED(CONFIG_PREEMPT_COUNT) && preemptible())) ||
+ (!(IS_ENABLED(CONFIG_PREEMPT_COUNT) && preemptible()) &&
+ rdp == this_cpu_ptr(&rcu_data)) ||
rcu_current_is_nocb_kthread(rdp)),
"Unsafe read of RCU_NOCB offloaded state"
);
Powered by blists - more mailing lists