Re: [PATCH] Remove GP_REPLAY state from rcu_sync

From: kbuild test robot
Date: Fri Oct 04 2019 - 18:04:14 EST

Next message:
Previous
In reply
Messages

Hi "Joel,

I love your patch!
[auto build test WARNING [cannot apply to v5.4-rc1 [if your patch is applied improve the system. base tree in git format-patch,
url: # save the make ARCH=x86_64

If you fix the issue, Reported-by: kbuild
All warnings (new ones
In file included kernel//rcu/sync.c: kernel//rcu/sync.c:187:23: if (rsp->gp_state ^
include/linux/compiler.h:58:52: #define __trace_if_var(cond) >> kernel//rcu/sync.c:187:2: if (rsp->gp_state ^~
kernel//rcu/sync.c:187:23: if (rsp->gp_state ^
include/linux/compiler.h:58:52: #define __trace_if_var(cond) >> kernel//rcu/sync.c:187:2: if (rsp->gp_state ^~

vim +/if +187 kernel//rcu/sync.c

cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov 6d1a4c2dfe7bb0 Joel Fernandes cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov 95bf33b55ff446 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov 3942a9bd7b5842 Peter Zijlstra 27fdb35fe99011 Paul 27fdb35fe99011 Paul 27fdb35fe99011 Paul 3942a9bd7b5842 Peter Zijlstra 3942a9bd7b5842 Peter Zijlstra 3942a9bd7b5842 Peter Zijlstra 3942a9bd7b5842 Peter Zijlstra 3942a9bd7b5842 Peter Zijlstra 3942a9bd7b5842 Peter Zijlstra 3942a9bd7b5842 Peter Zijlstra 3942a9bd7b5842 Peter Zijlstra 3942a9bd7b5842 Peter Zijlstra 3942a9bd7b5842 Peter Zijlstra 3942a9bd7b5842 Peter Zijlstra cc44ca848f5e51 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov 27fdb35fe99011 Paul cc44ca848f5e51 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov 27fdb35fe99011 Paul cc44ca848f5e51 Oleg Nesterov 27fdb35fe99011 Paul cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov 6d1a4c2dfe7bb0 Joel Fernandes 89da3b94bb9741 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov cc44ca848f5e51 Oleg Nesterov 07899a6e5f5613 Oleg Nesterov 07899a6e5f5613 Oleg Nesterov 07899a6e5f5613 Oleg Nesterov 07899a6e5f5613 Oleg Nesterov 07899a6e5f5613 Oleg Nesterov 07899a6e5f5613 Oleg Nesterov 07899a6e5f5613 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 07899a6e5f5613 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov 07899a6e5f5613 Oleg Nesterov 07899a6e5f5613 Oleg Nesterov 89da3b94bb9741 Oleg Nesterov
:::::: The code at :::::: 89da3b94bb97417ca2c5b
:::::: TO: Oleg Nesterov :::::: CC: Paul E.
---
0-DAY kernel test infrastructure Attachment: Description: application/gzip Michael Kelley: "RE: [PATCH 1/2] x86/hyperv: Allow guests to enable InvariantTSC" message: Linus Walleij: "Re: [PATCH 3/3] gpio: max77620: Fix interrupt handling" to: kbuild test robot: "Re: [PATCH] Remove GP_REPLAY state from rcu_sync" sorted by: [ date ] [ thread ] [ subject ] [ author ] Perhaps something to improve:
on rcu/dev]
next-20191004]
to the wrong git tree, please drop us a note to help
BTW, we also suggest to use '--base' option to specify the
please see https://stackoverflow.com/a/37406982]
href="https://github.com/0day-ci/linux/commits/Joel-Fernandes-Google/Remove-GP_REPLAY-state-from-rcu_sync/20191005-024257">https://github.com/0day-ci/linux/commits/Joel-Fernandes-Google/Remove-GP_REPLAY-state-from-rcu_sync/20191005-024257
href="https://git.kernel.org/pub/scm/linux/kernel/git/paulmck/linux-rcu.git">https://git.kernel.org/pub/scm/linux/kernel/git/paulmck/linux-rcu.git dev
(attached as .config)
7.4.0-13) 7.4.0
attached .config to linux build tree
kindly add following tag
test robot <lkp@xxxxxxxxx>
prefixed by >>):
from include/linux/export.h:44:0,
from include/linux/linkage.h:7,
from include/linux/kernel.h:8,
from include/linux/list.h:9,
from include/linux/wait.h:7,
from include/linux/rcu_sync.h:13,
from kernel//rcu/sync.c:10:
In function 'rcu_sync_dtor':
error: 'GP_REPLAY' undeclared (first use in this function)
== GP_REPLAY)
note: in definition of macro '__trace_if_var'
(__builtin_constant_p(cond) ? (cond) : __trace_if_value(cond))
^~~~
note: in expansion of macro 'if'
== GP_REPLAY)
note: each undeclared identifier is reported only once for each function it appears in
== GP_REPLAY)
note: in definition of macro '__trace_if_var'
(__builtin_constant_p(cond) ? (cond) : __trace_if_value(cond))
^~~~
note: in expansion of macro 'if'
== GP_REPLAY)
2015-08-21 @10 #include <linux/rcu_sync.h>
2015-08-21 11 #include <linux/sched.h>
2015-08-21 12
2019-10-04 13 enum { GP_IDLE = 0, GP_ENTER, GP_PASSED, GP_EXIT };
2015-08-21 14
2015-08-21 15 #define rss_lock gp_wait.lock
2015-08-21 16
2015-08-21 17 /**
2015-08-21 18 * rcu_sync_init() - Initialize an rcu_sync structure
2015-08-21 19 * @rsp: Pointer to rcu_sync structure to be initialized
2015-08-21 20 */
2019-04-23 21 void rcu_sync_init(struct rcu_sync *rsp)
2015-08-21 22 {
2015-08-21 23 memset(rsp, 0, sizeof(*rsp));
2015-08-21 24 init_waitqueue_head(&rsp->gp_wait);
2015-08-21 25 }
2015-08-21 26
2016-08-11 27 /**
E. McKenney 2017-10-19 28 * rcu_sync_enter_start - Force readers onto slow path for multiple updates
E. McKenney 2017-10-19 29 * @rsp: Pointer to rcu_sync structure to use for synchronization
E. McKenney 2017-10-19 30 *
2016-08-11 31 * Must be called after rcu_sync_init() and before first use.
2016-08-11 32 *
2016-08-11 33 * Ensures rcu_sync_is_idle() returns false and rcu_sync_{enter,exit}()
2016-08-11 34 * pairs turn into NO-OPs.
2016-08-11 35 */
2016-08-11 36 void rcu_sync_enter_start(struct rcu_sync *rsp)
2016-08-11 37 {
2016-08-11 38 rsp->gp_count++;
2016-08-11 39 rsp->gp_state = GP_PASSED;
2016-08-11 40 }
2016-08-11 41
2015-08-21 42
2019-04-25 43 static void rcu_sync_func(struct rcu_head *rhp);
2015-08-21 44
2019-04-25 45 static void rcu_sync_call(struct rcu_sync *rsp)
2019-04-25 46 {
2019-04-25 47 call_rcu(&rsp->cb_head, rcu_sync_func);
2015-08-21 48 }
2015-08-21 49
2015-08-21 50 /**
2015-08-21 51 * rcu_sync_func() - Callback function managing reader access to fastpath
E. McKenney 2017-10-19 52 * @rhp: Pointer to rcu_head in rcu_sync structure to use for synchronization
2015-08-21 53 *
2019-04-25 54 * This function is passed to call_rcu() function by rcu_sync_enter() and
2015-08-21 55 * rcu_sync_exit(), so that it is invoked after a grace period following the
2019-04-25 56 * that invocation of enter/exit.
2019-04-25 57 *
2019-04-25 58 * If it is called by rcu_sync_enter() it signals that all the readers were
2019-04-25 59 * switched onto slow path.
2019-04-25 60 *
2019-04-25 61 * If it is called by rcu_sync_exit() it takes action based on events that
2015-08-21 62 * have taken place in the meantime, so that closely spaced rcu_sync_enter()
2015-08-21 63 * and rcu_sync_exit() pairs need not wait for a grace period.
2015-08-21 64 *
2015-08-21 65 * If another rcu_sync_enter() is invoked before the grace period
2015-08-21 66 * ended, reset state to allow the next rcu_sync_exit() to let the
2015-08-21 67 * readers back onto their fastpaths (after a grace period). If both
2015-08-21 68 * another rcu_sync_enter() and its matching rcu_sync_exit() are invoked
2015-08-21 69 * before the grace period ended, re-invoke call_rcu() on behalf of that
2015-08-21 70 * rcu_sync_exit(). Otherwise, set all state back to idle so that readers
2015-08-21 71 * can again use their fastpaths.
2015-08-21 72 */
E. McKenney 2017-10-19 73 static void rcu_sync_func(struct rcu_head *rhp)
2015-08-21 74 {
E. McKenney 2017-10-19 75 struct rcu_sync *rsp = container_of(rhp, struct rcu_sync, cb_head);
2015-08-21 76 unsigned long flags;
2015-08-21 77
2019-04-25 78 WARN_ON_ONCE(READ_ONCE(rsp->gp_state) == GP_IDLE);
2019-04-25 79 WARN_ON_ONCE(READ_ONCE(rsp->gp_state) == GP_PASSED);
2015-08-21 80
2015-08-21 81 spin_lock_irqsave(&rsp->rss_lock, flags);
2015-08-21 82 if (rsp->gp_count) {
2015-08-21 83 /*
2019-04-25 84 * We're at least a GP after the GP_IDLE->GP_ENTER transition.
2015-08-21 85 */
2019-04-25 86 WRITE_ONCE(rsp->gp_state, GP_PASSED);
2019-04-25 87 wake_up_locked(&rsp->gp_wait);
2015-08-21 88 } else {
2015-08-21 89 /*
2019-04-25 90 * We're at least a GP after the last rcu_sync_exit(); eveybody
2019-04-25 91 * will now have observed the write side critical section.
2019-04-25 92 * Let 'em rip!.
2015-08-21 93 */
2019-04-25 94 WRITE_ONCE(rsp->gp_state, GP_IDLE);
2015-08-21 95 }
2015-08-21 96 spin_unlock_irqrestore(&rsp->rss_lock, flags);
2015-08-21 97 }
2015-08-21 98
2015-08-21 99 /**
2019-04-25 100 * rcu_sync_enter() - Force readers onto slowpath
2019-04-25 101 * @rsp: Pointer to rcu_sync structure to use for synchronization
2019-04-25 102 *
2019-04-25 103 * This function is used by updaters who need readers to make use of
2019-04-25 104 * a slowpath during the update. After this function returns, all
2019-04-25 105 * subsequent calls to rcu_sync_is_idle() will return false, which
2019-04-25 106 * tells readers to stay off their fastpaths. A later call to
2019-04-25 107 * rcu_sync_exit() re-enables reader slowpaths.
2019-04-25 108 *
2019-04-25 109 * When called in isolation, rcu_sync_enter() must wait for a grace
2019-04-25 110 * period, however, closely spaced calls to rcu_sync_enter() can
2019-04-25 111 * optimize away the grace-period wait via a state machine implemented
2019-04-25 112 * by rcu_sync_enter(), rcu_sync_exit(), and rcu_sync_func().
2019-04-25 113 */
2019-04-25 114 void rcu_sync_enter(struct rcu_sync *rsp)
2019-04-25 115 {
2019-04-25 116 int gp_state;
2019-04-25 117
2019-04-25 118 spin_lock_irq(&rsp->rss_lock);
2019-04-25 119 gp_state = rsp->gp_state;
2019-04-25 120 if (gp_state == GP_IDLE) {
2019-04-25 121 WRITE_ONCE(rsp->gp_state, GP_ENTER);
2019-04-25 122 WARN_ON_ONCE(rsp->gp_count);
2019-04-25 123 /*
2019-04-25 124 * Note that we could simply do rcu_sync_call(rsp) here and
2019-04-25 125 * avoid the "if (gp_state == GP_IDLE)" block below.
2019-04-25 126 *
2019-04-25 127 * However, synchronize_rcu() can be faster if rcu_expedited
2019-04-25 128 * or rcu_blocking_is_gp() is true.
2019-04-25 129 *
2019-04-25 130 * Another reason is that we can't wait for rcu callback if
2019-04-25 131 * we are called at early boot time but this shouldn't happen.
2019-04-25 132 */
2019-04-25 133 }
2019-04-25 134 rsp->gp_count++;
2019-04-25 135 spin_unlock_irq(&rsp->rss_lock);
2019-04-25 136
2019-04-25 137 if (gp_state == GP_IDLE) {
2019-04-25 138 /*
2019-04-25 139 * See the comment above, this simply does the "synchronous"
2019-04-25 140 * call_rcu(rcu_sync_func) which does GP_ENTER -> GP_PASSED.
2019-04-25 141 */
2019-04-25 142 synchronize_rcu();
2019-04-25 143 rcu_sync_func(&rsp->cb_head);
2019-04-25 144 /* Not really needed, wait_event() would see GP_PASSED. */
2019-04-25 145 return;
2019-04-25 146 }
2019-04-25 147
2019-04-25 148 wait_event(rsp->gp_wait, READ_ONCE(rsp->gp_state) >= GP_PASSED);
2019-04-25 149 }
2019-04-25 150
2019-04-25 151 /**
2019-04-25 152 * rcu_sync_exit() - Allow readers back onto fast path after grace period
2015-08-21 153 * @rsp: Pointer to rcu_sync structure to use for synchronization
2015-08-21 154 *
2015-08-21 155 * This function is used by updaters who have completed, and can therefore
2015-08-21 156 * now allow readers to make use of their fastpaths after a grace period
2015-08-21 157 * has elapsed. After this grace period has completed, all subsequent
2015-08-21 158 * calls to rcu_sync_is_idle() will return true, which tells readers that
2015-08-21 159 * they can once again use their fastpaths.
2015-08-21 160 */
2015-08-21 161 void rcu_sync_exit(struct rcu_sync *rsp)
2015-08-21 162 {
2019-10-04 163 WARN_ON_ONCE(READ_ONCE(rsp->gp_state) < GP_PASSED);
2019-04-25 164
2015-08-21 165 spin_lock_irq(&rsp->rss_lock);
2015-08-21 166 if (!--rsp->gp_count) {
2019-04-25 167 if (rsp->gp_state == GP_PASSED) {
2019-04-25 168 WRITE_ONCE(rsp->gp_state, GP_EXIT);
2019-04-25 169 rcu_sync_call(rsp);
2015-08-21 170 }
2015-08-21 171 }
2015-08-21 172 spin_unlock_irq(&rsp->rss_lock);
2015-08-21 173 }
2015-08-21 174
2015-08-21 175 /**
2015-08-21 176 * rcu_sync_dtor() - Clean up an rcu_sync structure
2015-08-21 177 * @rsp: Pointer to rcu_sync structure to be cleaned up
2015-08-21 178 */
2015-08-21 179 void rcu_sync_dtor(struct rcu_sync *rsp)
2015-08-21 180 {
2019-04-25 181 int gp_state;
2015-08-21 182
2019-04-25 183 WARN_ON_ONCE(READ_ONCE(rsp->gp_count));
2019-04-25 184 WARN_ON_ONCE(READ_ONCE(rsp->gp_state) == GP_PASSED);
2015-08-21 185
2015-08-21 186 spin_lock_irq(&rsp->rss_lock);
2019-04-25 @187 if (rsp->gp_state == GP_REPLAY)
line 187 was first introduced by commit
0ce3a28643819030247 rcu/sync: Simplify the state machine
<oleg@xxxxxxxxxx>
McKenney <paulmck@xxxxxxxxxxxxx>
Open Source Technology Center
href="https://lists.01.org/pipermail/kbuild-all">https://lists.01.org/pipermail/kbuild-all Intel Corporation
>.config.gz

Next message: Michael Kelley: "RE: [PATCH 1/2] x86/hyperv: Allow guests to enable InvariantTSC"
Previous message: Linus Walleij: "Re: [PATCH 3/3] gpio: max77620: Fix interrupt handling"
In reply to: kbuild test robot: "Re: [PATCH] Remove GP_REPLAY state from rcu_sync"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]