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

Attacking the Qualcomm Adreno GPU

googleprojectzero.blogspot.com

Post reply on HN
Print view View original

11–20 of 23 posts

Re: Attacking the Qualcomm Adreno GPU

#11

Post by panpanna »

londons_explore wrote: post #8
octoberfranklin wrote: post #6

Earlier quoted context omitted.

Yeah I think it is pretty nuts that people are willing to use CPUs that have those modems on the same die.

The simple solution is memory isolation - let the modem be as insecure as you like, but anyone who breaks into the modem can only see your network traffic (hopefully all encrypted anyway) and nothing else. Sadly todays qualcomm hardware has no real memory isolation at all - any bit of on-chip hardware can see all memory. It isn't perfect, but it's far easier to do that than properly secure a multi-million lines of co…

There is definitely memory protection in the modem. If nothing else because the code is so horrible it crashes all the time.

Three is even a hypervisor for hexagon but I don't think it is used.

Re: Attacking the Qualcomm Adreno GPU

#12

Post by londons_explore »

saagarjha wrote: post #10
londons_explore wrote: post #9

Earlier quoted context omitted.

For comparison, Google Chrome tries to get security patches to most users within 24 hours . Yet most android devices are lucky to receive a patch within a few months... Don't worry though - that's only a window of a few months where an evil actor can drain your bank account and log your porn browsing sessions...

As someone using iOS, does Chrome really update multiple times a week for security patches? Am I overestimating how often security fixes go in?

Critical security issues 'in the wild' only come up once every few months, but yes, when they do, the Chrome team has someone on duty 24 hours per day whose responsibility is to patch the code and do a release to all users within a matter of hours.

If you submit a security issue to Chrome, they actually have a tickbox on the webform to say "this issue is important enough to get someone out of bed for", and if you tick that box, it will actually wake someone at 3am to deal with it...

Re: Attacking the Qualcomm Adreno GPU

#13

Post by fluffything »

panpanna wrote: post #5

At least the GPU stuff is getting some scrutiny. Their modem code is a security nightmare and outside Qualcomms modem teams nobody is allowed to see it.

I wonder how secure new nvidia "security" features like MIG actually are. With people running real-time audio/video transcoders on GPUs for multiple connections, I wonder whether it is possible particularly crafted video connections to leak the video of other channels being processed by the same GPU.

Even if one were to encrypt all connections, these will probably need to be decrypted on the GPU for processing.

Re: Attacking the Qualcomm Adreno GPU

#14

Post by panpanna »

nl wrote: post #7
panpanna wrote: post #4

Earlier quoted context omitted.

Maybe you have not used Android lately? I have two phones and a tablet, all mid-range devices from 3 different vendors and all are on Android 10 with at least August patches. Edit: both phones are also more than 2 years old.

I have a Pixel. My comment referred to the timeline outlined in the post, in particular this part: Qualcomm gives an update on the progress of a microcode based fix. The plan is that the fix will be available for OEMs by September 7, but Qualcomm will request an extension to patch integration and testing by OEMs.allow more time for patch integration and testing by OEMs. and for their multiple subsequent requests for…

I am fine with this schedule.

Unless someone is actively exploiting devices I would prefer a well tested patch to a rushed patch.

Note that this whole issue is due a previously rushed patch.

Re: Attacking the Qualcomm Adreno GPU

#15

Post by baybal2 »

londons_explore wrote: post #8
octoberfranklin wrote: post #6

Earlier quoted context omitted.

Yeah I think it is pretty nuts that people are willing to use CPUs that have those modems on the same die.

The simple solution is memory isolation - let the modem be as insecure as you like, but anyone who breaks into the modem can only see your network traffic (hopefully all encrypted anyway) and nothing else. Sadly todays qualcomm hardware has no real memory isolation at all - any bit of on-chip hardware can see all memory. It isn't perfect, but it's far easier to do that than properly secure a multi-million lines of co…

There is an IOMMU on snapdragons, as the article says, but it is the IOMMU mapping itself which they attack.

This itself is kind of mind boggling how they let the device overwrite its own IOMMU configuration, effectively nullifying IOMMU's purpose, and its provided safeties.

It's like fencing your house with 10 meter high walls, but leaving the key lying in front of the gate.

Re: Attacking the Qualcomm Adreno GPU

#16

Post by guerby »

"We can offer a few additional recommendations: Transparency and openness: (...) More generally, the competitive benefits of a closed platform approach to hardware internals should be reassessed in 2020. This balance may have been historically appropriate when the GPU was not in the critical path for security, but today billions of users are relying on the GPU to uphold the operating system security model. "

This.

Re: Attacking the Qualcomm Adreno GPU

#17

Post by londons_explore »

panpanna wrote: post #11
londons_explore wrote: post #8

Earlier quoted context omitted.

The simple solution is memory isolation - let the modem be as insecure as you like, but anyone who breaks into the modem can only see your network traffic (hopefully all encrypted anyway) and nothing else. Sadly todays qualcomm hardware has no real memory isolation at all - any bit of on-chip hardware can see all memory. It isn't perfect, but it's far easier to do that than properly secure a multi-million lines of co…

There is definitely memory protection in the modem. If nothing else because the code is so horrible it crashes all the time. Three is even a hypervisor for hexagon but I don't think it is used.

As someone else said, it's not a security barrier between the ARM core and the radio hardware bits... It's more a tool for remapping stuff to make system design easier, and as a way of protecting against evil hardware outside the SoC.

I believe it can prevent the ARM core tampering with private radio hardware memory, but not the other way round.

Re: Attacking the Qualcomm Adreno GPU

#18

Post by panpanna »

londons_explore wrote: post #17
panpanna wrote: post #11

Earlier quoted context omitted.

There is definitely memory protection in the modem. If nothing else because the code is so horrible it crashes all the time. Three is even a hypervisor for hexagon but I don't think it is used.

As someone else said, it's not a security barrier between the ARM core and the radio hardware bits... It's more a tool for remapping stuff to make system design easier, and as a way of protecting against evil hardware outside the SoC. I believe it can prevent the ARM core tampering with private radio hardware memory, but not the other way round.

I think Qualcomm XPUs can be used as barriers (pretty much the same way SMMUs are used).

XPUs + hypervisor should be enough, assuming Qualcomm enables them and configures them correctly.

Re: Attacking the Qualcomm Adreno GPU

#19

Post by GeekyBear »

nl wrote: post #2

Next time Project Zero finds an iOS bug and people suggest it is a commercial hitjob, point them at this. Qualcomm (and all Android vendors) look like they have been screwed by this. (To be clear - they are screwed because their processes are to slow to get security updates out).

You should probably wait for an example where Google didn't put off writing the issue up for a year after the initial report.
Post reply on HN