#!/usr/bin/env expect
############################################################################
# Purpose: Test of Slurm functionality
#          Test --gpus-per-task with implicit task count.
############################################################################
# Copyright (C) 2019 SchedMD LLC
# Written by Morris Jette
#
# This file is part of Slurm, a resource management program.
# For details, see <https://slurm.schedmd.com/>.
# Please also read the included file: DISCLAIMER.
#
# Slurm is free software; you can redistribute it and/or modify it under
# the terms of the GNU General Public License as published by the Free
# Software Foundation; either version 2 of the License, or (at your option)
# any later version.
#
# Slurm is distributed in the hope that it will be useful, but WITHOUT ANY
# WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
# FOR A PARTICULAR PURPOSE.  See the GNU General Public License for more
# details.
#
# You should have received a copy of the GNU General Public License along
# with Slurm; if not, write to the Free Software Foundation, Inc.,
# 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA.
############################################################################
source ./globals

set exit_code      0
set file_in1       "test$test_id.input1"
set file_in2       "test$test_id.input2"
set number_commas  "\[0-9_,\]+"

if {[test_cons_tres]} {
	log_debug "Valid configuration, using select/cons_tres"
} else {
	skip "This test is only compatible with select/cons_tres"
}
if {[test_front_end]} {
	skip "This test is incompatible with front-end systems"
}

set nb_nodes [get_node_cnt_in_part]
set gpu_cnt [get_highest_gres_count $nb_nodes "gpu"]
if {$gpu_cnt < 0} {
	fail "Error getting GPU count"
}
if {$gpu_cnt < 1} {
	skip "This test requires 1 or more GPUs in the default partition"
}
log_debug "GPU count is $gpu_cnt"
log_debug "Default partition node count $nb_nodes"

if {$nb_nodes > 3} {
	set nb_nodes 3
}

#
# Build input script files
#
make_bash_script $file_in1 "
if \[ \$SLURM_LOCALID -eq 0 \]; then
echo HOST:\$SLURMD_NODENAME NODE_CNT:\$SLURM_NNODES CUDA_VISIBLE_DEVICES:\$CUDA_VISIBLE_DEVICES
fi
if \[ \$SLURM_PROCID -eq 0 \]; then
$scontrol -dd show job \$SLURM_JOB_ID | grep \"GRES=\"
fi
exit 0"
make_bash_script $file_in2 "echo HOST:\$SLURMD_NODENAME NODE_CNT:\$SLURM_NNODES CUDA_VISIBLE_DEVICES:\$CUDA_VISIBLE_DEVICES
if \[ \$SLURM_PROCID -eq 0 \]; then
$scontrol -dd show job \$SLURM_JOB_ID | grep \"NumCPUs=\"
$scontrol -dd show job \$SLURM_JOB_ID | grep \"GRES=\"
fi
exit 0"

#
# One GPU per task with node count (without range)
#
log_info "TEST: One GPU per task with node count (without range)"
set timeout $max_job_delay
set match 0
set node_cnt -1
set srun_pid [spawn $srun --nodes=$nb_nodes --gpus-per-task=1 -t1 -J "test$test_id" -l ./$file_in1]
expect {
	-re "NODE_CNT:($number_commas) CUDA_VISIBLE_DEVICES:($number_commas)" {
		set node_cnt $expect_out(1,string)
		incr match [cuda_count $expect_out(2,string)]
		exp_continue
	}
	timeout {
		log_error "srun not responding"
		slow_kill $srun_pid
		set exit_code 1
	}
	eof {
		wait
	}
}
if {$match != $node_cnt} {
	log_error "srun --gpus-per-task failure ($match != $node_cnt)"
	set exit_code 1
}

#
# Two GPUs per task with node count (without range)
#
if {$gpu_cnt > 1} {
	log_info "TEST: Two GPUs per task with node count (without range)"
	set match 0
	if {$nb_nodes > 1} {
		set min_nodes 2
	} else {
		set min_nodes 1
	}
	set node_cnt -1
	set srun_pid [spawn $srun --nodes=$nb_nodes --gpus-per-task=2 -t1 -J "test$test_id" -l ./$file_in1]
	expect {
		-re "NODE_CNT:($number_commas) CUDA_VISIBLE_DEVICES:($number_commas)" {
			set node_cnt $expect_out(1,string)
			incr match [cuda_count $expect_out(2,string)]
			exp_continue
		}
		timeout {
			log_error "srun not responding"
			slow_kill $srun_pid
			set exit_code 1
		}
		eof {
			wait
		}
	}
	set exp_gpu_cnt [expr $node_cnt * 2]
	if {$match != $exp_gpu_cnt} {
		log_error "srun --gpus-per-task failure ($match != $exp_gpu_cnt)"
		set exit_code 1
	}
}

#
# One GPU per task with node count range and task count resulting in uneven task distribution
#
if {$gpu_cnt > 1 && $nb_nodes > 1} {
	log_info "TEST: Node count range and uneven task distribution"
	set match 0
	set node_cnt -1
	set task_cnt [expr $nb_nodes + 1]
	set srun_pid [spawn $srun --nodes=2-$nb_nodes --ntasks=$task_cnt --gpus-per-task=1 -t1 -J "test$test_id" -l ./$file_in1]
	expect {
		-re "NODE_CNT:($number_commas) CUDA_VISIBLE_DEVICES:($number_commas)" {
			set node_cnt $expect_out(1,string)
			incr match [cuda_count $expect_out(2,string)]
			exp_continue
		}
		timeout {
			log_error "srun not responding"
			slow_kill $srun_pid
			set exit_code 1
		}
		eof {
			wait
		}
	}
	if {$match != $task_cnt} {
		log_error "srun --gpus-per-task failure ($match != $task_cnt)"
		set exit_code 1
	}
}

#
# One task and GPU for each GPU available at step level
#
if {$gpu_cnt > 1} {
	log_info "TEST: One task and GPU for each GPU available at step level"
	set match 0
	set task_cnt 0
# FIXME: RANGE CHECK CPU/GPU COUNT
	set salloc_pid [spawn $salloc -N1 --gpus=$gpu_cnt -t1 -J "test$test_id" $srun -n $gpu_cnt -O --gpus-per-task=1 ./$file_in2]
	expect {
		-re "CUDA_VISIBLE_DEVICES:($number_commas)" {
			incr task_cnt
			incr match [cuda_count $expect_out(1,string)]
			exp_continue
		}
		timeout {
			log_error "srun not responding"
			slow_kill $srun_pid
			set exit_code 1
		}
		eof {
			wait
		}
	}
	set exp_gpu_cnt [expr $task_cnt * $task_cnt]
	if {$task_cnt == 0} {
		log_error "srun --gpus-per-task test failed to run"
		set exit_code 1
	} elseif {$match != $exp_gpu_cnt} {
		log_error "srun --gpus-per-task failure ($match != $exp_gpu_cnt)"
		set exit_code 1
	}
}

#
# One task and two GPUs as resources available at step level
#
if {$gpu_cnt > 1} {
	log_info "TEST: One task and two GPUs as resources available at step level"
	set hostname "unknown"
	set match 0
	set num_cpus 0
	set task_cnt 0
	set tasks_spawned [expr $gpu_cnt / 2]
	set salloc_pid [spawn $salloc -N1 --exclusive --gpus=$gpu_cnt -t1 -J "test$test_id" $srun -n $tasks_spawned -O --gpus-per-task=2 ./$file_in2]
	expect {
		-re "CUDA_VISIBLE_DEVICES:($number_commas)" {
			incr task_cnt
			incr match [cuda_count $expect_out(1,string)]
			exp_continue
		}
		-re "NumCPUs=($number)" {
			set num_cpus $expect_out(1,string)
			exp_continue
		}
		-re " Nodes=($re_word_str)" {
			set hostname $expect_out(1,string)
			exp_continue
		}
		timeout {
			log_error "srun not responding"
			slow_kill $srun_pid
			set exit_code 1
		}
		eof {
			wait
		}
	}
	set exp_gpu_cnt [expr $task_cnt * $task_cnt * 2]
	if {$task_cnt == 0} {
		log_error "srun --gpus-per-task test failed to run"
		set exit_code 1
	} elseif {$match != $exp_gpu_cnt} {
		log_error "srun --gpus-per-task failure ($match != $exp_gpu_cnt)"
		set exit_code 1
	}
}

#
# Step allocation of GPUs based upon CPUs per task
#
if {$gpu_cnt > 1 && $num_cpus > 0} {
	log_info "TEST: Step allocation of GPUs based upon CPUs per task"
	set match 0
	set task_cnt 0
	if {$gpu_cnt > $num_cpus} {
		set cpus_per_task 1
		set gpus_per_task [expr $gpu_cnt / $num_cpus]
	} else {
		set cpus_per_task $num_cpus
		set gpus_per_task 1
	}
	set salloc_pid [spawn $salloc -N1 --exclusive -w $hostname --gpus=$gpu_cnt -t1 -J "test$test_id" $srun -c $cpus_per_task --gpus-per-task=$gpus_per_task ./$file_in2]
	expect {
		-re "CUDA_VISIBLE_DEVICES:($number_commas)" {
			incr task_cnt
			incr match [cuda_count $expect_out(1,string)]
			exp_continue
		}
		timeout {
			log_error "srun not responding"
			slow_kill $srun_pid
			set exit_code 1
		}
		eof {
			wait
		}
	}
	set exp_gpu_cnt [expr $task_cnt *$task_cnt * $gpus_per_task]
	if {$task_cnt == 0} {
		fail "srun --gpus-per-task test failed to run"
	} elseif {$match != $exp_gpu_cnt} {
		fail "srun --gpus-per-task failure ($match != $exp_gpu_cnt)"
	}
}

if {$exit_code == 0} {
	exec $bin_rm -f $file_in1 $file_in2
} else {
	fail "Test failed due to previous errors (\$exit_code = $exit_code)"
}