src/share/vm/utilities/vmError.hpp

rev 12487 : 8166944: Hanging Error Reporting steps may lead to torn error logs.
Reviewed-by: cjplummer, dholmes
Summary: Interupt error reporting if reporting steps hang to enable subsequent reporting steps to run.
rev 12488 : [mq]: 8166944-Hanging-Error-Reporting-2

*** 1,7 ****
  /*
!  * Copyright (c) 2003, 2016, Oracle and/or its affiliates. All rights reserved.
   * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
   *
   * This code is free software; you can redistribute it and/or modify it
   * under the terms of the GNU General Public License version 2 only, as
   * published by the Free Software Foundation.
--- 1,7 ----
  /*
!  * Copyright (c) 2003, 2017, Oracle and/or its affiliates. All rights reserved.
   * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
   *
   * This code is free software; you can redistribute it and/or modify it
   * under the terms of the GNU General Public License version 2 only, as
   * published by the Free Software Foundation.
*** 72,81 ****
--- 72,91 ----
    // When coredump_status is set to true this will contain the name/path to the core/minidump,
    // if coredump_status if false, this will (hopefully) contain a useful error explaining why
    // no core/minidump has been written to disk
    static char coredump_message[O_BUFLEN];
  
+   // Timeout handling:
+   // Timestamp at which error reporting started; -1 if no error reporting in progress.
+   static volatile jlong _reporting_start_time;
+   // Whether or not error reporting did timeout.
+   static volatile bool _reporting_did_timeout;
+   // Timestamp at which the last error reporting step started; -1 if no error reporting
+   //   in progress.
+   static volatile jlong _step_start_time;
+   // Whether or not the last error reporting step did timeout.
+   static volatile bool _step_did_timeout;
  
    // set signal handlers on Solaris/Linux or the default exception filter
    // on Windows, to handle recursive crashes.
    static void reset_signal_handlers();
  
*** 102,115 ****
    static void report_and_die(const char* message, const char* detail_fmt, ...) ATTRIBUTE_PRINTF(2, 3);
  
    static fdStream out;
    static fdStream log; // error log used by VMError::report_and_die()
  
!   // Platform dependend helper functions used for timeout handling.
    static void reporting_started();
    static void interrupt_reporting_thread();
  
  public:
  
    // return a string to describe the error
    static char* error_string(char* buf, int buflen);
  
--- 112,135 ----
    static void report_and_die(const char* message, const char* detail_fmt, ...) ATTRIBUTE_PRINTF(2, 3);
  
    static fdStream out;
    static fdStream log; // error log used by VMError::report_and_die()
  
!   // Timeout handling.
!   // Hook functions for platform dependend functionality:
    static void reporting_started();
    static void interrupt_reporting_thread();
  
+   // Helper function to get the current timestamp.
+   static jlong get_current_timestamp();
+ 
+   // Accessors to get/set the start times for step and total timeout.
+   static void record_reporting_start_time();
+   static jlong get_reporting_start_time();
+   static void record_step_start_time();
+   static jlong get_step_start_time();
+ 
  public:
  
    // return a string to describe the error
    static char* error_string(char* buf, int buflen);
  
*** 150,161 ****
    // check to see if fatal error reporting is in progress
    static bool fatal_error_in_progress() { return first_error_tid != -1; }
  
    static intptr_t get_first_error_tid() { return first_error_tid; }
  
!   // Called by WatcherThread to check if the currently running error reporting did timeout.
!   //  Returns true if reporting did hit the global ErrorLogTimeout.
    static bool check_timeout();
  
  };
  
  #endif // SHARE_VM_UTILITIES_VMERROR_HPP
--- 170,181 ----
    // check to see if fatal error reporting is in progress
    static bool fatal_error_in_progress() { return first_error_tid != -1; }
  
    static intptr_t get_first_error_tid() { return first_error_tid; }
  
!   // Called by the WatcherThread to check if error reporting has timed-out.
!   //  Returns true if error reporting has not completed within the ErrorLogTimeout limit.
    static bool check_timeout();
  
  };
  
  #endif // SHARE_VM_UTILITIES_VMERROR_HPP