Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Gmail smartphone app hacked by researchers

bbc.com

Post reply on HN
Print view View original

11–14 of 14 posts

Re: Gmail smartphone app hacked by researchers

#11

Post by void-star »

sjf wrote: post #9
fidotron wrote: post #8

Earlier quoted context omitted.

Actually it's not, and it would break enough that they wouldn't be able to do that without complaints. See the entire permission system for another case. Therein lies Google's problem, as once anything is usable in public it does get used/abused. There are certain VM behaviours in compatibility mode which are like this, where outright bugs have to be preserved in later versions for the benefit of apps which aren't go…

I would actually like to know what apps want in /proc that they can't get through the API.

Well... how do you suppose the API gets at all the things? (not everything is done across IPC/binder, much is in process)

Re: Gmail smartphone app hacked by researchers

#12

Post by void-star »

giovannibajo1 wrote: post #7
fidotron wrote: post #6

Earlier quoted context omitted.

Sounds to me like it detects the activity to activity transition (which on Android creates an Intent which is passed to the system and back via shared memory) and then does something else, i.e. display a fake password entry screen when the transition is detected. It sounds like the actual Intent contents remain secure, however, the creation of an Intent while a particular app is in the foreground and using a particul…

or just remove /proc access from apps, which is totally useless.

ptrace is actually a more reliable way to get at remote process memory on Android, I think. I'd suspect this before /proc.

Also, /proc nor ptrace is "useless". Both are actually fundamental to several aspects of the underlying system that zygote runs on.

Re: Gmail smartphone app hacked by researchers

#13

Post by stusmall »

fidotron wrote: post #8
giovannibajo1 wrote: post #7

Earlier quoted context omitted.

or just remove /proc access from apps, which is totally useless.

Actually it's not, and it would break enough that they wouldn't be able to do that without complaints. See the entire permission system for another case. Therein lies Google's problem, as once anything is usable in public it does get used/abused. There are certain VM behaviours in compatibility mode which are like this, where outright bugs have to be preserved in later versions for the benefit of apps which aren't go…

I've seen /proc permissions from version to version. For example the available of /proc/last_kmsg changed at some point. In kitkat its only available to UID system and apps with android.permission.READ_LOGS[1]. Not sure exactly when this changed but this line wasn't present in 4.1.2.

1. https://android.googlesource.com/platform/system/core/+/mast... line 218

Re: Gmail smartphone app hacked by researchers

#14

Post by giovannibajo1 »

void-star wrote: post #11
sjf wrote: post #9

Earlier quoted context omitted.

I would actually like to know what apps want in /proc that they can't get through the API.

Well... how do you suppose the API gets at all the things? (not everything is done across IPC/binder, much is in process)

The permission model in Android goes through the APIs. If you can bypass the API to go lower-level, then you're effectively giving up the security, and just waiting for some cool researcher (or attacker) to find ways to exploit them.

The above makes sense if you want to assume that you need /proc at least at some permission level. If you don't (which I think you don't), then it should simply be unmounted in the sandbox.

IPC is the standard way to get out of the sandbox, and should be used also for whatever features the APIs need /proc for.

Post reply on HN