Ytukyg

I am writing a bin bash shell script to collect the information from multiple devices at once. However I notice if there is any adb commands executed in the while loop will break it. Not sure why as the loop is clearly not complete. I know there are multiple feeds discussing this topic but I am unable to find similar issue like my one. Any advise will be helpful. Thank you.

#!/bin/bash

echo "# Collecting attached devices under adb and record the project name..."

adb devices | while read line

do



   if  [ "$line" != "" ] && [ `echo $line | awk '{print $2}'` = "device" ]

then



device_sn=`echo $line | awk '{print $1}'`



project_name=`adb -s $device_sn shell getprop ro.product.device` #This line will break the while loop. If I remove it I can have all connected devices reboot to bootloader



echo "$project_name: $device_sn is rebooting to bootloader"



adb -s $device_sn reboot bootloader



   fi

done

I'm guessing here, but it might be that the adb command is reading the input that's intended for while read. This is a fairly common problem with the while read type loop, and the symptoms you describe match. One possibility is to redirect input on the internal command so it can't read from the unintended source, like this:

project_name=`adb -s $device_sn shell getprop ro.product.device` </dev/null

But my favorite solution to this is to pass the input to the loop over something other than standard input by changing the while read loop like this:

while read line <&3

do

    # normal loop contents here...

done 3< <(adb devices)

Those redirects pass input over file descriptor #3 (#0 is standard input, #1 is standard output, #2 is standard error, #3 is normally unused). With this form, you don't have to know which commands inside the loop might read from standard input, so no surprises (and if they do need to read from the Terminal, they can!). Also, unlike a standard pipe, this runs the loop in the main shell process rather than a subshell, avoiding a number of oddities (like variables set in the loop not being available after the loop).

There's one disadvantage, though: process substitution, <( ), is a bash-only feature. It's not even available in bash when it's run under the name sh! So you must use a bash shebang in the script (#!/bin/bash or #!/usr/bin/env bash), and don't override that by running the script with sh!