Solaris

sed -n "/2009 05:/,/2009 17:/p" trace.log | grep -f patterns.txt > temp.log
while read string ;do
count=$(grep -c "$string" temp.log)
echo "$count\t$string" >> count.log
done <patterns.txt

sed -n "/2009 05:/,/2009 17:/p" trace.log | egrep "PATTERN1|PATTERN2|PATTERN3|PATTERN4|PATTERN5|PATTERN6" > temp.log
while read string ;do
count=$(grep -c "$string" temp.log)
echo "$count\t$string" >> count.log
done <patterns.txt

/usr/xpg4/bin/grep

awk 'NR==FNR {P[$1]=0;next} /2009 05:/,/2009 17:/{for (i in P) if ($0~i) P[i]++}
     END {for (i in P) if (P[i]>0) print P[i]"\t"i}' patterns.txt trace.log > count.log

/usr/xpg4/bin/awk: line 0 (NR=2139569): Record too long (LIMIT: 19999 bytes)

sed -n '/PATTERN/ {N;p;}' trace.log