meta data doesn't get cleaned up on failed tests

[stevieb9]

We will somehow need a test to ensure that failed tests clean up properly...

                         '8e2c28516e3eda62fe1e1b847f1df430' => {
                                                                 'proc' => 19166,
                                                                 'label' => 't/335-shift_reg_adc.t'
                                                               },
                         '9666aba591fc178ab2112854ab2f4c11' => {
                                                                 'proc' => 18801,
                                                                 'label' => 't/150-cleanup.t'
                                                               },
                         'f693bbb439f879ee3fcf6344c16bc612' => {
                                                                 'proc' => 23802,
                                                                 'label' => 't/325-servo.t'
                                                               },
                         '46f2db2a844c876913207b4f4ccffe05' => {
                                                                 'proc' => 357,
                                                                 'label' => 't/105-pin.t'
                                                               },
                         '612f0ff8ee7b67a1e75c11c0b3b84edb' => {
                                                                 'proc' => 1470,
                                                                 'label' => 't/140-pwm_spi_adc.t'
                                                               },
                         '331bbd1a7bd10126666a0607ac8f956c' => {
                                                                 'label' => 't/150-cleanup.t',
                                                                 'proc' => 4548
                                                               },
                         'bf53328673cd2c4a39542a88da746c0b' => {
                                                                 'label' => 'multi_die',
                                                                 'proc' => 9634
                                                               }
                       },
          'pwm' => {
                     'in_use' => 0
                   },
          'pins' => {
                      '26' => {
                                'state' => 0,
                                'alt' => 0,
                                'comment' => '26:t/110-register.t',
                                'mode' => 0,
                                'users' => {
                                             '70e3b75d9fbcaff7be64c520f795766c' => 1
                                           }
                              },
                      '18' => {
                                'mode' => 0,
                                'users' => {
                                             '67badf1adb76d2d0f6167f234ac2684f' => 1
                                           },
                                'comment' => undef,
                                'alt' => 0,
                                'state' => 0
                              },
                      '12' => {
                                'comment' => '12:t/110-register.t',
                                'state' => 0,
                                'alt' => 0,
                                'users' => {
                                             '70e3b75d9fbcaff7be64c520f795766c' => 1
                                           },
                                'mode' => 0
                              }
                    },

[stevieb9]

The good thing here is that there's only a single shared memory segment and semaphgre at all times.

[stevieb9]

It's looking a whole lot better now. Going to run a dispatch until tomorrow and see the results before closing this.

[stevieb9]

Although I've added the ability to separate shared mem spaces between prod and test, this is still on the radar.

It appears as though it's related to PWM and multi tests. The tester seems to hang mid-flight.

Overall though, ~70 runs kept things relatively very clean.

[stevieb9]

This is becoming a bigger problem. t/310-dac.t fails, we have three pins remaining and can't be used again.

pi@pi-test:~/repos/rpi-wiringpi $ pimeta rpit
$VAR1 = {
          'objects' => {
                         '717545823cf2906f35e7ebd3dfc16ec0' => {
                                                                 'proc' => 872,
                                                                 'label' => 't/310-dac.t'
                                                               },
                         'd8250d07b27ee37c8fb97b7bff54c2a7' => {
                                                                 'label' => 'serial_arduino_display',
                                                                 'proc' => 374
                                                               },
                         'e369f185a7832fc05bda4ec12eb1c7c3' => {
                                                                 'proc' => 377,
                                                                 'label' => 'oled display date/time'
                                                               }
                       },
          'pwm' => {},
          'pins' => {
                      '12' => {
                                'mode' => 0,
                                'state' => 0,
                                'comment' => 'MCP4922 DAC CS',
                                'users' => {
                                             '717545823cf2906f35e7ebd3dfc16ec0' => 1
                                           },
                                'alt' => 0
                              },
                      '26' => {
                                'alt' => 0,
                                'users' => {
                                             '717545823cf2906f35e7ebd3dfc16ec0' => 1
                                           },
                                'comment' => 'MCP3008 ADC CS',
                                'state' => 0,
                                'mode' => 0
                              }
                    },
          'object_count' => 3,
          'testing' => {
                         'test_num' => '310',
                         'test_name' => 'dac'
                       }
        };

[stevieb9]

The latter comment refers to #160, but things should still clean up properly if a program or test file crashes

[stevieb9]

Unit test suite now cleans up properly