Support NeoGAF

[c0de]

The PS4 is hardware wise a PC with standard APIs except for parts of the MediaCON=Southbridge=Aeolia (Marvell name).

Sony does not allow anyone except Japan programmers to touch Southbridge because that is the ARM Trustzone managed Southbridge TEE (Trusted Execution Environment) for DRM. Sony just sandboxes some of the code running in the APU and hash checks the APU's OS just after it is booted by Southbridge.

For those interested; Sony uses Mono (C# JIT compiler) calling webkit native libraries as a framework to support commercial APPS and Webkit as well as Mono are always loaded. OpenGL ES2 is the Graphics framework and it supports 2D and 3D graphics. Why mono instead of Javascript; because Mono is smaller,faster and easier for them to insure it has no vulnerabilities. This was a speculation of mine a few years ago that I got right.

The above are from recent dumps of the PS4 booting.

So Hacking the APU is no big deal as that is the open world part of the PS4 and since it must support most of the HTML5 calls to support Vidipath including W3C extensions, it is considered vulnerable to hacks which is the reason for Trustzone managed boots, hash checks of the OS and all DRM is behind the ARM Trustzone managed TEE. Given this the hack is to an already running PS4 and at reboot the hack never existed.

So much text and yet nothing you say is confirmed by failoverflow. It's not a standard pc, it doesn't use standard apis at all and they even modified standard stuff like pci bus.
Also I don't know in any way how anyone can talk about html5 in this context.

 

[c0de]

Can't sony block this in new firmware update ? Also mostly new games require updated firmware to work. May be PS4 Slim revision expected this year can fix those security ssues and they might be stopping making original model, so will not harm them anymore even if there is real hacking for piracy which will take time before that Slim will be released.

Did they say anything about their Linux working on latest firmware and future firmware versions also ?

What do you want them to block? Linux? Why?

 

[Caayn]

For those who're interested. Here's the source code of the modified Linux kernel which they used to achieve this.

https://github.com/fail0verflow/ps4-linux

 

[jeff_rigby]

So much text and yet nothing you say is confirmed by failoverflow. It's not a standard pc, it doesn't use standard apis at all and they even modified standard stuff like pci bus.
Also I don't know in any way how anyone can talk about html5 in this context.

Your right, Features for HTML5 Vidipath will be the same as offered by a PC and that can be abstracted. Sony wants developers as comfortable as possible and exposes API calls nearly 100% like the PS4 was a AMD APU PC but the Sony compiler takes care of the differences. One example; Sony supports a VCE encoder call while the PS4 APU doesn't have a VCE, it's a Xtensa DPU in Southbridge.

This slide from the Fail Overflow video mentions 7443 differences between a PC and the PS4. But two slides before it says it's a PC, the next slide adds but it's not and the differences. I'm afraid my memory and/or I was not as through in reading all the slides are making be eat leather.

That being said there is no reason for Sony to have differences unless there is a legitimate reason beyond DRM as it has a Trustzone TEE. A Major difference is the ARM block moved out of the APU to southbridge and the PCI0 port on the APU allowing the Aeolia southbridge to "trusted boot" the APU and to hash check the OS in GDDR5 memeory...special IOMMU in the APU on the PCI0 port?

If you read the PS4 boot log only one hardware difference is mentioned but the calls mentioned, at a lower level, can be code unique to the PS4. Two days ago this youtube was posted of a PS4 exploit. Using keywords PS4 Aeolia and you can find this: http://www.psdevwiki.com/ps4/Bootprocess


Remember AMD saying the Semi-custom Sony APU could be sold to third parties....there shouldn't be any special "Sony" features in it just 2014 or later AMD GPU features in a 2013 GPU so they start with a GPU family that is close and have to find the differences.

 

[c0de]

Your right, Features for HTML5 Vidipath will be the same as offered by a PC and that can be abstracted. Sony wants developers as comfortable as possible and exposes API calls nearly 100% like the PS4 was a AMD APU PC but the Sony compiler takes care of the differences. One example; Sony supports a VCE encoder call while the PS4 APU doesn't have a VCE, it's a Xtensa DPU in Southbridge.

Even if, it has nothing to do with the exploit and/or running Linux on it.

This slide from the Fail Overflow video mentions 7443 differences between a PC and the PS4. But two slides before it says it's a PC, the next slide adds but it's not and the differences. I'm afraid my memory and/or I was not as through in reading all the slides are making be eat leather.

That being said there is no reason for Sony to have differences unless there is a legitimate reason beyond DRM as it has a Trustzone TEE.

There is no reason to make it a PC like at all. And I don't know whether DRM is a legit reason. It's only different, not really DRM or security in any way. So yes, it is different but the conclusion that is has to be because of what you think is just nothing but a wild guess.

A Major difference is the ARM block moved out of the APU to southbridge and the PCI0 port on the APU allowing the Aeolia southbridge to "trusted boot" the APU and to hash check the OS in GDDR5 memeory...special IOMMU in the APU on the PCI0 port?

Where is ARM mentioned in any way by fail0verflow. I mean you could look at their kernel commits if you are interested.

If you read the PS4 boot log only one hardware difference is mentioned but the calls mentioned, at a lower level, can be code unique to the PS4. Two days ago this youtube was posted of a PS4 exploit. Using keywords PS4 Aeolia and you can find this: http://www.psdevwiki.com/ps4/Bootprocess

If I read fail0verflow correctly, someone fucked up PCI but they made the changes to make it work with Linux.

Remember AMD saying the Semi-custom Sony APU could be sold to third parties....there shouldn't be any special "Sony" features in it just 2014 or later AMD GPU features in a 2013 GPU so they start with a GPU family that is close and have to find the differences.

Nobody says they can't sell it. Doesn't mean Sony can't alter the hardware afterwards while AMD ships "standard" chips.

 

[jeff_rigby]

Even if, it has nothing to do with the exploit and/or running Linux on it.

This slide from the Fail Overflow video mentions 7443 differences between a PC and the PS4. But two slides before it says it's a PC, the next slide adds but it's not and the differences. I'm afraid my memory and/or I was not as through in reading all the slides are making be eat leather.

There is no reason to make it a PC like at all. And I don't know whether DRM is a legit reason. It's only different, not really DRM or security in any way. So yes, it is different but the conclusion that is has to be because of what you think is just nothing but a wild guess.

Where is ARM mentioned in any way by fail0verflow. I mean you could look at their kernel commits if you are interested.


If I read fail0verflow correctly, someone fucked up PCI but they made the changes to make it work with Linux.

Nobody says they can't sell it. Doesn't mean Sony can't alter the hardware afterwards while AMD ships "standard" chips.

Two things I can't stress enough:
1) Sony wants the PS4 to be easy to develop on and for PC developers this means few changes to port to a PC. This is why the PS4 supports a True Audio block which I don't think is used yet because there are few PCs out there with a True Audio Block (Xtensa DPU on an AMD AXI buss and part of the audio codec to be used with media in southbridge including blu-ray DTS audio (Seen in the PS4 bootlog). That's coming with calls very similar to what a AMD PC would have. AMD want's games using their hardware as a standard. I believe the same is true for openVX support though Sony could use OpenCL rather than Xtensa DPU as I am guessing as their GPU is optimized for compute.
2) Sony wants secure DRM for next generation media and that requires a TEE where all commercial media enters DRM encrypted and exits HDCP2 encrypted. That impacts the hardware design and because Sony chose to use GDDR5 the ARM block was moved out of the APU to Southbridge with 256MB of DDR3 memory. ALL commercial media uses Southbridge as a TEE.

 

[biglittleps]

What do you want them to block? Linux? Why?

Linux may allow to study their internals which may help some people who want to pirate the games for profit.

It would be great if Sony allows Linux on PS4 as a feature officially.

 

[c0de]

Two things I can't stress enough:
1) Sony wants the PS4 to be easy to develop on and for PC developers this means few changes to port to a PC. This is why the PS4 supports a True Audio block which I don't think is used yet because there are few PCs out there with a True Audio Block (Xtensa DPU on an AMD AXI buss and part of the audio codec to be used with media in southbridge including blu-ray DTS audio (Seen in the PS4 bootlog). That's coming with calls very similar to what a AMD PC would have. AMD want's games using their hardware as a standard. I believe the same is true for openVX support though Sony could use OpenCL rather than Xtensa DPU as I am guessing as their GPU is optimized for compute.

Of course they want that and they do that. But that is on Sony's part, not developers and has nothing to do with the hack. Devs can already use Audio APIs. It's Sony's job to deliver APIs, not AMD's. Whatever they abstract in their APIs, we don't know because that's, well, the part of the API and nobody should care anyway.

2) Sony wants secure DRM for next generation media and that requires a TEE where all commercial media enters DRM encrypted and exits HDCP2 encrypted. That impacts the hardware design and because Sony chose to use GDDR5 the ARM block was moved out of the APU to Southbridge with 256MB of DDR3 memory. ALL commercial media uses Southbridge as a TEE.

And even if this was true (especially the conclusion), what does it have to do with this hack, like at all?
As it is now, we don't know whether Sony uses the "other" hardware, whether they are planning to do it or not.

 

[TheSeks]

It would be great if Sony allows Linux on PS4 as a feature officially.

And have it be gimped like the PS3's? No one would use it, then. The issue is either you allow everything, or lock it completely down and don't bother.

 

[luca_29_bg]

http://wololo.net/2016/01/04/linux-on-ps4-fail0verflow-get-3d-drivers-to-work

3d driver works 😁

 

[LordOfChaos]

> On top of that, it seems that despite the actual hardware using a SATA interface, the PS4 itself appears to communicate with the hard drive via USB - a curious state of affairs.

Bizarre. Might be part of why SSDs show little gain on the PS4. USB - more CPU overhead.

 

[c0de]

Bizarre. Might be part of why SSDs show little gain on the PS4. USB - more CPU overhead.

Well, it especially means more latency, not CPU overhead.

 

[Blanquito]

Bizarre. Might be part of why SSDs show little gain on the PS4. USB - more CPU overhead.

What? That is really weird. I wonder why.

 

[gundalf]

AFAIK the latest PS4 Model does not use USB internally anymore.

 

[LordOfChaos]

AFAIK the latest PS4 Model does not use USB internally anymore.

As of the CAH 1100, or 1200?

Well, it especially means more latency, not CPU overhead.

It's both, even with USB 3 there's more overhead than other interfaces. While it's generally negligible on modern desktops, if it's limited by the ARM core or a single Jaguar core that could explain why SSDs don't make it much faster.

I wonder if the new one would be faster than the old with an SSD then

 

[ViciousDS]

Bizarre. Might be part of why SSDs show little gain on the PS4. USB - more CPU overhead.

yes....thats a little messed up