You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hi, I've been fuzzing torchaudio project with sydr-fuzz and found error at sox/src/xa.c:164.
I think that unsigned integer overflow occurs because we get xa struct as (priv_t *) ft->priv and then use xa->header.align and xa->header.sampleRate without any checks.
The thing is, I am about to land #3497, which switchs libsox integration to runtime linking, and these source code will not be used. Therefore unfortunately these reports become irrelevant to torchaudio soon. They need to be addressed at upstream. (and upstream project has made a lot of changes since 14.4.2, so these might not be relevant.)
Do you find anything needs fix in torchaudio codebase, instead of third party directory? I am aware couple of cases where signed integer and unsigned integer are compared, but did not have time to fix them. It would be nice if you could make PR to fix them.
#3497 has landed, and now torchaudio links to sox dynamically. We no longer compile the specified code by ourselves and therefore this issue is now irrelevant.
Feel free to fuzz the code under torchaudio/csrc, and report issues there. Thanks,
🐛 Describe the bug
Hi, I've been fuzzing torchaudio project with sydr-fuzz and found error at
sox/src/xa.c:164
.I think that unsigned integer overflow occurs because we get
xa
struct as(priv_t *) ft->priv
and then usexa->header.align
andxa->header.sampleRate
without any checks.How to reproduce
Versions
torchaudio version: 30afaa9
pytorch version: 0f1621df1a0a73956c7ce4e2f72f069e610e0137
sox version: 14.4.2
OS: Ubuntu 20.04
The text was updated successfully, but these errors were encountered: