Fix memory leak in bz_reader_read #12
Conversation
Fix memory leak in bz_reader_read
|
@brianmario, you may need to re-re-revert this! 😁 It's causing "double free" segfaults for me on both Fedora 20 and CentOS 6.3. However, it only happens when I'm piping it through ruby-gpgme. |
|
The code in this gem is pretty bad. I'd recommend looking for alternatives. Unfortunately I haven't had any time to work on it. I wrote https://gist.github.com/brianmario/5833373 which only does a simple one-pass compress/decompress but you could use it as a starting point to build out an IO-like API. Sorry for the trouble :\ |
|
Thanks for that, it looks like a good start. However bad it may be though, this gem does work for me. I did try switching to rbzip2 because I was looking at JRuby but it doesn't implement gets, which I need. I've shelved JRuby for the moment. |
This reverts commit e58f154 because the "fix" is causing double frees when piped through ruby-gpgme. See issue brianmario#12.
|
Oh, this is showing up in #27 builds on Travis. @brianmario is it unsalvageable? |
4 participants
bz_reader_read never frees any of the buffers it allocates for uncompressed data.
I found this behavior to be sub-optimal so I made it free the buffers before returning the ruby string. Probably a better way to do this, but at least Bzip2.uncompress doesn't leak anymore...