[incubator-nuttx] branch master updated: sched/irq_section.c: irq_waitlock is a private function; remove its prototype from irq.h and define it as static.

[ma...@apache.org]

This is an automated email from the ASF dual-hosted git repository.

masayuki pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/incubator-nuttx.git


The following commit(s) were added to refs/heads/master by this push:
     new 20d34cc  sched/irq_section.c: irq_waitlock is a private function; remove its prototype from irq.h and define it as static.
20d34cc is described below

commit 20d34cce98139a75f2657e904b509ae64f9d8912
Author: Abdelatif Guettouche <ab...@espressif.com>
AuthorDate: Tue Sep 14 18:31:07 2021 +0200

    sched/irq_section.c: irq_waitlock is a private function; remove its
    prototype from irq.h and define it as static.
    
    Signed-off-by: Abdelatif Guettouche <ab...@espressif.com>
---
 include/nuttx/irq.h      | 47 -----------------------------------------------
 sched/irq/irq_csection.c |  2 +-
 2 files changed, 1 insertion(+), 48 deletions(-)

diff --git a/include/nuttx/irq.h b/include/nuttx/irq.h
index 9500954..60ab8bd 100644
--- a/include/nuttx/irq.h
+++ b/include/nuttx/irq.h
@@ -155,53 +155,6 @@ int irqchain_detach(int irq, xcpt_t isr, FAR void *arg);
 #endif
 
 /****************************************************************************
- * Name: irq_waitlock
- *
- * Description:
- *   Spin to get g_irq_waitlock, handling a known deadlock condition:
- *
- *   A deadlock may occur if enter_critical_section is called from an
- *   interrupt handler.  Suppose:
- *
- *   - CPUn is in a critical section and has the g_cpu_irqlock spinlock.
- *   - CPUm takes an interrupt and attempts to enter the critical section.
- *   - It spins waiting on g_cpu_irqlock with interrupts disabled.
- *   - CPUn calls up_cpu_pause() to pause operation on CPUm.  This will
- *     issue an inter-CPU interrupt to CPUm
- *   - But interrupts are disabled on CPUm so the up_cpu_pause() is never
- *     handled, causing the deadlock.
- *
- *   This same deadlock can occur in the normal tasking case:
- *
- *   - A task on CPUn enters a critical section and has the g_cpu_irqlock
- *     spinlock.
- *   - Another task on CPUm attempts to enter the critical section but has
- *     to wait, spinning to get g_cpu_irqlock with interrupts disabled.
- *   - The task on CPUn causes a new task to become ready-to-run and the
- *     scheduler selects CPUm.  CPUm is requested to pause via a pause
- *     interrupt.
- *   - But the task on CPUm is also attempting to enter the critical
- *     section.  Since it is spinning with interrupts disabled, CPUm cannot
- *     process the pending pause interrupt, causing the deadlock.
- *
- *   This function detects this deadlock condition while spinning with \
- *   interrupts disabled.
- *
- * Input Parameters:
- *   cpu - The index of CPU that is trying to enter the critical section.
- *
- * Returned Value:
- *   True:  The g_cpu_irqlock spinlock has been taken.
- *   False: The g_cpu_irqlock spinlock has not been taken yet, but there is
- *          a pending pause interrupt request.
- *
- ****************************************************************************/
-
-#ifdef CONFIG_SMP
-bool irq_waitlock(int cpu);
-#endif
-
-/****************************************************************************
  * Name: enter_critical_section
  *
  * Description:
diff --git a/sched/irq/irq_csection.c b/sched/irq/irq_csection.c
index 798a222..4c76221 100644
--- a/sched/irq/irq_csection.c
+++ b/sched/irq/irq_csection.c
@@ -106,7 +106,7 @@ volatile uint8_t g_cpu_nestcount[CONFIG_SMP_NCPUS];
  ****************************************************************************/
 
 #ifdef CONFIG_SMP
-bool irq_waitlock(int cpu)
+static bool irq_waitlock(int cpu)
 {
 #ifdef CONFIG_SCHED_INSTRUMENTATION_SPINLOCKS
   FAR struct tcb_s *tcb = current_task(cpu);