_decompress consumes all memory

[NMigo]

Hi, performing a fastlz_decompresss on a non valid compressed string, the function consumes al memory:

php -a
Interactive shell
php > fastlz_decompress('£RES');

Fatal error: Allowed memory size of 1291845632 bytes exhausted (tried to allocate 3265483334 bytes) in php shell code on line 1

Call Stack:
13.5809 633472 1. {main}() php shell code:0
13.5810 633688 2. fastlz_decompress() php shell code:1

This is tested under Debian Linux 64bits, PHP CLI and the latest version from the repo. Please let us know if you need to re-test. Thanks,

[t3rmin4t0r]

I stored the size of the buffer as the first few bytes of the string.

I probably need to write a retry based reallocation to decompress data, similar to what the gzip case does.

Probably will fix it in a while.

[t3rmin4t0r]

Fixing this would break some of the existing compressed files.

I'm guessing that is fine? :)

[NMigo]

Well, for us is fine as we haven't put the code into production. ; )
But if anyone has compressed strings already in a DB, then it will be a problem...

[t3rmin4t0r]

Yeah, would somebody be crazy enough to use code I wrote on an afternoon out of boredom?

I'll add an "fastlz.format=old" ini option or something.

[NMigo]

Well your code has been featured in some talks...

On Fri, Feb 3, 2012 at 5:24 PM, Gopal V <
[email protected]

wrote:

Yeah, would somebody be crazy enough to use code I wrote on an afternoon
out of boredom?

I'll add an "fastlz.format=old" ini option or something.

---

Reply to this email directly or view it on GitHub:
#2 (comment)

[NMigo]

maybe this is fixed here at iliaal's repo? iliaal@f4ec15c

[NMigo]

just verified, it is solved indeed at iliaal's repo. However the README file still points to this repo.