Bug #18154
closedString#initialize leaks memory for STR_NOFREE strings
Description
GitHub PR: https://github.com/ruby/ruby/pull/4814
There is a memory leak in calling the constructor on a string that is marked STR_NOFREE
(e.g. a string created from a C string literal). The script below reproduces the memory leak. This is reproducible on all maintained Rubies (2.6.8, 2.7.4, 3.0.2, master) on Ubuntu 20.04.
We create a string marked STR_NOFREE
with 0.to_s
. to_s
for Fixnum has a special optimization for the value 0
(it directly converts it to a C string literal). When we call String#initialize
with a capacity it creates a buffer using malloc
but does not unset the STR_NOFREE
flag. This causes the buffer to be permanently leaked.
100.times do
1000.times do
# 0.to_s is a special case that creates a string from a C string literal.
# https://github.com/ruby/ruby/blob/26153667f91f0c883f6af6b61fac2c0df5312b45/numeric.c#L3393
# C string literals are always marked STR_NOFREE.
str = 0.to_s
# Call String#initialize again to create a buffer with a capacity of 10000
# characters.
str.send(:initialize, capacity: 10000)
end
# Output the Resident Set Size (memory usage, in KB) of the current Ruby process.
puts `ps -o rss= -p #{$$}`
end
We can see the leak through the following graph of the Resident Set Size (RSS) comparing the branch vs. master (at commit 26153667f91f0c883f6af6b61fac2c0df5312b45).