Add new event priority for trace enable events so
that tracing gets turned on as the very first thing in the selected cycle (tick). --HG-- extra : convert_revision : c08f749ca42782af1b48e5aa5f0860bf7076bd3c
This commit is contained in:
parent
75ecd3be60
commit
e78684fc6c
1 changed files with 14 additions and 8 deletions
|
@ -120,10 +120,22 @@ class Event : public Serializable, public FastAlloc
|
||||||
/// priority; these values are used to control events that need to
|
/// priority; these values are used to control events that need to
|
||||||
/// be ordered within a cycle.
|
/// be ordered within a cycle.
|
||||||
enum Priority {
|
enum Priority {
|
||||||
/// Breakpoints should happen before anything else, so we
|
/// If we enable tracing on a particular cycle, do that as the
|
||||||
/// don't miss any action when debugging.
|
/// very first thing so we don't miss any of the events on
|
||||||
|
/// that cycle (even if we enter the debugger).
|
||||||
|
Trace_Enable_Pri = -101,
|
||||||
|
|
||||||
|
/// Breakpoints should happen before anything else (except
|
||||||
|
/// enabling trace output), so we don't miss any action when
|
||||||
|
/// debugging.
|
||||||
Debug_Break_Pri = -100,
|
Debug_Break_Pri = -100,
|
||||||
|
|
||||||
|
/// CPU switches schedule the new CPU's tick event for the
|
||||||
|
/// same cycle (after unscheduling the old CPU's tick event).
|
||||||
|
/// The switch needs to come before any tick events to make
|
||||||
|
/// sure we don't tick both CPUs in the same cycle.
|
||||||
|
CPU_Switch_Pri = -31,
|
||||||
|
|
||||||
/// For some reason "delayed" inter-cluster writebacks are
|
/// For some reason "delayed" inter-cluster writebacks are
|
||||||
/// scheduled before regular writebacks (which have default
|
/// scheduled before regular writebacks (which have default
|
||||||
/// priority). Steve?
|
/// priority). Steve?
|
||||||
|
@ -132,12 +144,6 @@ class Event : public Serializable, public FastAlloc
|
||||||
/// Default is zero for historical reasons.
|
/// Default is zero for historical reasons.
|
||||||
Default_Pri = 0,
|
Default_Pri = 0,
|
||||||
|
|
||||||
/// CPU switches schedule the new CPU's tick event for the
|
|
||||||
/// same cycle (after unscheduling the old CPU's tick event).
|
|
||||||
/// The switch needs to come before any tick events to make
|
|
||||||
/// sure we don't tick both CPUs in the same cycle.
|
|
||||||
CPU_Switch_Pri = -31,
|
|
||||||
|
|
||||||
/// Serailization needs to occur before tick events also, so
|
/// Serailization needs to occur before tick events also, so
|
||||||
/// that a serialize/unserialize is identical to an on-line
|
/// that a serialize/unserialize is identical to an on-line
|
||||||
/// CPU switch.
|
/// CPU switch.
|
||||||
|
|
Loading…
Reference in a new issue