GithubHelp home page GithubHelp logo

Stack buffer overflow in v043 about zstd HOT 5 CLOSED

facebook avatar facebook commented on April 20, 2024
Stack buffer overflow in v043

from zstd.

Comments (5)

Cyan4973 avatar Cyan4973 commented on April 20, 2024

Thanks for reporting @drach. I'll look into it.

There is also a chance that the problem lies into fullbench rather than zstd.
Since you provide the sample to reproduce the issue, investigation should find it.

from zstd.

Cyan4973 avatar Cyan4973 commented on April 20, 2024

I believe to have fixed this issue within latest update of "dev" branch 695a6cb.

The issue was within fullbench, zstd library looks unaffected (by this bug).
fullbench improperly allocated space for the ZSTD_decodeSeqHeaders test, which was extended by one cell in recent versions of FSE.

Since the issue is not specific to your sample, I'm surprised it has not show up earlier within Continuous Integration tests, as fullbench is tested with asan enabled. I suspect FSE dynamically downsizes reference sample statistics as they are generated from a small sample. Side effect : it doesn't test the limits of internal buffers.

If you witness some instability issues outside of fullbench, it may be related to something else.

from zstd.

drach avatar drach commented on April 20, 2024

Indeed, fullbench is fixed in "dev", thank you.

If you witness some instability issues outside of fullbench, it may be related to something else.

Agree, will check it closer.

from zstd.

drach avatar drach commented on April 20, 2024

In an existing quite a big iOS app the invocation of ZSTD_decompress (exactly this function, not its internals) may just crash after several dozens of successful calls, it looks like a problem in a corrupted stack.

I was wrong. We call ZSTD_decompress in different threads, some of them are set up with a 128Kb stack size that turns to be small for a ZSTD context. I was careless to notice it at first because the same code and data worked normally with LZ4 and Zlib's DEFLATE.

from zstd.

Cyan4973 avatar Cyan4973 commented on April 20, 2024

Thanks for feedback @drach.

Which reminds me : the decompression context is quite large currently, and it's part of my todo list to tame it.

In the meantime, should you need to keep your stack at 128K per thread, I suggest creating a context per thread with ZSTD_createDCtx(), and use ZSTD_decompressDCtx() (which is not yet exposed unfortunately, but present within zstd_decompress.c). It will avoid usage of the stack.

from zstd.

Related Issues (20)

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.