| 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: <082fb8ba-91b8-448e-a472-195eb7b282fd@paulmck-laptop>
Date: Sun, 2 Nov 2025 13:44:29 -0800
From: "Paul E. McKenney" <paulmck@...nel.org>
To: rcu@...r.kernel.org
Cc: linux-kernel@...r.kernel.org, kernel-team@...a.com, rostedt@...dmis.org
Subject: [PATCH 0/18] SRCU updates for v6.19
Hello!
This series creates an srcu_expedite_current() function that allows
after-the-fact expediting of SRCU grace periods, adds a DEFINE_SRCU_FAST()
that further speeds things up by removing a conditional branch from
srcu_read_lock_fast*(), updates documentation, uses SRCU-fast to guard
event traces in PREEMPT_RT kernel (thus making such kernels safe for
event tracing), adds srcu_read_{,un}lock_fast_updown() functions that are
compatible with srcu_down_read() and srcu_up_read(), but do not permit
use in NMI handlers (to permit further optimization of SRCU-fast readers
by relieving them of the need to deal with irq/softirq/NMI handlers with
unbalanced lock/unlock calls), and optimizes SRCU-fast-updown for large
ARM servers that use LSE. It is expected that this optimization will be
obsoleted by some arm64 architecture-specific work:
https://lore.kernel.org/all/aQU7l-qMKJTx4znJ@arm.com/
The patches are as follows:
1. Permit Tiny SRCU srcu_read_unlock() with interrupts disabled.
2. Create an srcu_expedite_current() function.
3. Test srcu_expedite_current().
4. Create a DEFINE_SRCU_FAST().
5. Make grace-period determination use ssp->srcu_reader_flavor.
6. Exercise DEFINE_STATIC_SRCU_FAST() and init_srcu_struct_fast().
7. Require special srcu_struct define/init for SRCU-fast readers.
8. Make SRCU-fast readers enforce use of SRCU-fast definition/init.
9. Update for SRCU-fast definitions and initialization.
10. Guard __DECLARE_TRACE() use of __DO_TRACE_CALL() with SRCU-fast.
11. Mark diagnostic functions as notrace.
12. Permit kvm-again.sh to re-use the build directory.
13. Add SRCU_READ_FLAVOR_FAST_UPDOWN CPP macro.
14. Permit negative kvm.sh --kconfig numberic arguments.
15. Create an SRCU-fast-updown API.
16. Test SRCU-fast separately from SRCU-fast-updown.
17. Optimize SRCU-fast-updown for arm64.
18. Make srcu{,d}_torture_init() announce the SRCU type.
19. Remove redundant rcutorture_one_extend() from
rcu_torture_one_read().
Many of these patches were previously associated with another series
that re-implemented RCU tasks trace in terms of SRCU-fast. This work
is being deferred pending resolution of the ARM LSE situation on the one
hand or full debugging of the all-too-clever workaround optimization on
the other. ;-)
https://lore.kernel.org/all/7fa58961-2dce-4e08-8174-1d1cc592210f@paulmck-laptop/
Thanx, Paul
------------------------------------------------------------------------
b/Documentation/RCU/Design/Requirements/Requirements.rst | 33 +--
b/Documentation/RCU/checklist.rst | 12 -
b/Documentation/RCU/whatisRCU.rst | 3
b/include/linux/notifier.h | 2
b/include/linux/srcu.h | 16 +
b/include/linux/srcutiny.h | 1
b/include/linux/srcutree.h | 8
b/include/linux/tracepoint.h | 45 ++--
b/include/trace/perf.h | 4
b/include/trace/trace_events.h | 4
b/kernel/rcu/rcutorture.c | 12 +
b/kernel/rcu/srcutiny.c | 13 -
b/kernel/rcu/srcutree.c | 58 +++++
b/kernel/rcu/tree.c | 2
b/kernel/rcu/update.c | 8
b/kernel/tracepoint.c | 21 +-
b/tools/testing/selftests/rcutorture/bin/kvm-again.sh | 56 +++--
b/tools/testing/selftests/rcutorture/bin/kvm.sh | 2
include/linux/srcu.h | 133 ++++++++++--
include/linux/srcutiny.h | 30 ++
include/linux/srcutree.h | 157 +++++++++++----
kernel/rcu/rcutorture.c | 93 ++++++--
kernel/rcu/srcutree.c | 78 ++++++-
23 files changed, 624 insertions(+), 167 deletions(-)
Powered by blists - more mailing lists