Re: SYSTEMROOT, cygpath, and /proc funniness

[Linda Walsh <cygwin at tlinx dot org>]

Nellis, Kenneth wrote:
> In .bash_profile I use cygpath to convert several environment 
> variables that contain Windows paths to instead contain Unix 
> paths. One in particular, SYSTEMROOT, has caused several 
> problems, one of which is causing ClearCase's cleartool to be 
> non-functional. Another is strictly within Cygwin itself 
> where there seems to be some funny relationship between 
> SYSTEMROOT, having the working directory be /proc, and the 
> cygpath command.
>
> The script, below, demonstrates the problem. It takes the
> following options:
> 0: run with defaults
> 1: run with cd /proc
> 2: run with SYSTEMROOT cygpath'd
> 3: run with both options
-----
	I tried your script on my system and couldn't reproduce
the results.

I altered the script to run all cases in succession.

My results on all 4 cases (calling script with "test.sh all 3")
case=all, count=3
5572: old priority 0, new priority -19
 case = 0, PWD=/tmp, SYSTEMROOT=C:\Windows,
Iter 0 @ 10:51:30: 0.07sec 0.01usr 0.00sys (19.73% cpu)
Iter 0 @ 10:51:31: 0.07sec 0.01usr 0.00sys (19.23% cpu)
Iter 0 @ 10:51:32: 0.07sec 0.01usr 0.00sys (19.23% cpu)
 case = 1, PWD=/proc, SYSTEMROOT=C:\Windows,
Iter 1 @ 10:51:32: 0.08sec 0.00usr 0.03sys (37.80% cpu)
Iter 1 @ 10:51:33: 0.07sec 0.03usr 0.00sys (39.24% cpu)
Iter 1 @ 10:51:33: 0.08sec 0.00usr 0.01sys (18.75% cpu)
 case = 2, PWD=/tmp, SYSTEMROOT=/Windows,
Iter 2 @ 10:51:34: 0.08sec 0.00usr 0.00sys (0.00% cpu)
Iter 2 @ 10:51:35: 0.07sec 0.00usr 0.01sys (19.23% cpu)
Iter 2 @ 10:51:35: 0.07sec 0.01usr 0.01sys (38.46% cpu)
 case = 3, PWD=/proc, SYSTEMROOT=/Windows,
Iter 3 @ 10:51:36: 0.07sec 0.00usr 0.01sys (19.48% cpu)
Iter 3 @ 10:51:36: 0.07sec 0.00usr 0.03sys (41.33% cpu)
Iter 3 @ 10:51:37: 0.07sec 0.01usr 0.01sys (40.25% cpu)

Script:

#!/bin/bash
TIMEFORMAT="%2Rsec %2Uusr %2Ssys (%P%% cpu)"
case=-1
count=20

uname -a

function usage {
   echo "Usage: $0 {0|1|2|3|all}"
   exit 1
}

function _setup {
 local case=$1
 count=${2:-$count}
 #echo "_setup1: case=$case, count=$count"
 case $case in
     0)  ;;
     1)  cd /proc
   ;;
     2)  SYSTEMROOT=$(cygpath "$SYSTEMROOT")
   ;;
     3)  cd /proc
   SYSTEMROOT=$(cygpath "$SYSTEMROOT")
   ;;
    all)
   ;;
     *)  echo "$0: invalid parameter: $1"
   usage
   exit 1
   ;;
 esac
 echo "$case $count"
}

function do_test {
 case=$1
 use_count=$2
 echo "  case = $case, PWD=$PWD, SYSTEMROOT=$SYSTEMROOT, "
 for ((n=1; n<=use_count; n++)); do
     echo -n "Iter $1 @ $(date +'%H:%M:%S'): "
     time cygpath /proc &>/dev/null
     sleep 0.3
 done
}

#get case & count from initial setup
read case count<<< "$(_setup "$@")"
echo "case=$case, count=$count"
# try for highest priority to minimize other procs 'stealing' cpu cycles
renice -19 $$
if [[ $case != all ]] ;then
 do_test $case $count
elif [[ $case == all ]]; then
 for ((case=0; case<=3; ++case)); do
   # call setup before each 'do_test' to simulate original script and do
   #   each in a subshell to isolate any unanticipated side-effects
   (
     _setup $case $count &>/dev/null
     do_test $case $count
   )
 done
fi



--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple