Conversation

Q is using the

's Scudo. And I am new to it so out of curiosity, I have tested it on a sample heap overflow code and it is working fine. [1]

Replying to

and

It's still using jemalloc. It has support for Scudo but it hasn't replaced jemalloc as the default allocator. You should carefully qualify statements like this rather than presenting it as if they've fully moved over to using Scudo which they definitely aren't doing in Android Q.

Replying to

FYI it's enabled by default in Q in media{extractor,codex}. Working on getting more adoption in R.

Replying to

I was pointing out it's not the default allocator but rather something that has to be specified and loaded as a special case. It should be qualified where it's used rather than making it seem like it replaced jemalloc as the default allocator which people will misinterpret.

2:21 AM · Sep 9, 2019Twitter Web App

Replying to

I'm planning on trying it globally for legacy 32-bit code until that's fully stripped out and that's definitely a change. There a fair bit of 32-bit processes left and hardened_malloc is inherently based around having a larger address space with no attempt to support 32-bit.