[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: fuzzing SDFs

From: Anuj Verma
Subject: Re: fuzzing SDFs
Date: Sun, 1 Aug 2021 19:43:24 +0530

> Looks good!  Please submit a Pull Request that Armin (and others) can
> review.


The fuzzer already found a bug in the renderer, I have attached a patch (diff.txt) to
fix this (
If you require a merge request instead, then I'll make it tomorrow.

Also, there is a timeout in one of the tests:
240 - truetype-render-i38:oss-fuzz/10398-5653547685773312-timeout (Timeout)
Is there a way to increase the timeout ? As I would like to test this locally
and confirm that it's just a timeout and not a bug.

> I think generally there are two options:
> - We do have a few screws at this point to steer how many glyphs we're
> looking at in a single fuzzer run.  We could either use one of the existing
> ones or, if they are to coarse, we could introduce new screws that
> specifically target the SDF mode.
> - Alternatively SDF could get it's very own fuzzer.  We've already done this
> to separate rendering from utility access;  we _could_ look into further
> separating fuzzing rendering from fuzzing SDF rendering.  Although, that
> would be quite a lot of work.

Seeing how slow the `sdf` renderer is I think the second option would
be better. Dominik recently added the COLRv1 fuzzer, if the process is
similar, then I can work on adding a new fuzzer for SDF.


Attachment: diff.txt
Description: Text document

reply via email to

[Prev in Thread] Current Thread [Next in Thread]