描述

I have a script that I run to create a project, run synthesis, and implementation runs. The script contains the following loop to clean up existing runs:

foreach run_name $previous_runs {
  delete_run -quiet ${run_name}
  file delete -force ${run_name}
}

When sourced, the script fails and reports the following:

Abnormal program termination (EXCEPTION_ACCESS_VIOLATION)

How can I work around the issue?

解决方案

In this use case, the Tcl script is deleting the previous run, thus causing a null pointer which is causing the tool to crash.

The second iteration of:

delete_run
foreach run_name $previous_runs {
    delete_run -quiet ${run_name}
    file delete -force ${run_name}
}

In the script, previous_runs contains the objects synth_1 and impl_1. So, the first delete_run deletes synth_1 which has a side effect of deleting impl_1. On the second iteration, run_name is a bad pointer which causes the crash.

The desired behavior can be achieved with the following:

set previous_syn_runs [get_runs -filter {IS_SYNTHESIS==1}] foreach syn_run_name $previous_syn_runs {
    #Check that the run still exists and was not deleted as a by product
    #of deleting another run.
    if {[lsearch [get_runs] ${syn_run_name}] != -1} {
        delete_run -quiet ${syn_run_name}
        file delete -force ${syn_run_name}
    }
}

You can also perform a reset_run, which will clean up the run directory, but the run names will remain in the GUI.

Note: In Vivado Design Suite 2013.4, the delete_run command was changed to delete_runs. In Vivado 2013.4 and later tool versions, the recommended methodology to delete all runs is to run detele_runs $all_runs.