By setting the number of rows a file is before it's read, R speeds up tremendously and doesn't leak memory. Here's a quick trick to get a file's row count within R.

readRowCount.R

f.name <- "file.csv"
f.command <- paste("wc -l",
                   f.name,
                   "| cut -d \" \" -f 2")
f.rows <- as.numeric(system(f.command, intern = TRUE))
f <- read.csv(f.name,
              nrows=f.rows,
              col.names=c("user", "phone", "total", "percent"),
              colClasses=c("numeric","numeric","numeric","numeric"),
              header=FALSE)