This Week In Security: SWAPGS, Malicious Shaders, More IOS Woes, And WPA3

I’m sure you’ve heard of Spectre, which was the first of many speculative execution vulnerabilities found in modern processors. A new one just popped up this week. At Blackhat on Tuesday, CVE-2019-1125 was announced by Bitdefender as SWAPGS.

SWAPGS is an x86_64 instruction that is intended for use in context switching, that is when execution is transferred from a user-space program back into the kernel. Specifically, SWAPGS swaps the value of the GS register so that it refers to either a memory location in the running application, or a location in the kernel’s space. An unprivileged program can attempt to call this instruction and leak kernel memory contents as a result of the processor speculatively executing the instruction (this is similar to Spectre). Even though the instruction will ultimately not be executed, because a userspace program doesn’t have sufficient privilege to do so, the contents of the system cache have already been sufficiently altered, and an attack could feasibly leverage this to read arbitrary kernel memory.

While the initial reports have mentioned both AMD and Intel products, AMD has released a statement:

AMD is aware of new research claiming new speculative execution attacks that may allow access to privileged kernel data. Based on external and internal analysis, AMD believes it is not vulnerable to the SWAPGS variant attacks because AMD products are designed not to speculate on the new GS value following a speculative SWAPGS. For the attack that is not a SWAPGS variant, the mitigation is to implement our existing recommendations for Spectre variant 1.

Patches for Windows and Linux have been released, and Red Hat has an informative write-up on the vulnerability. I would have reviewed Bitdefender’s whitepaper on the vulnerability, but rather than make it freely available, they have opted to require a name and email address. While I would like to see their work, I refuse to sell my contact information in exchange for access.

A Malicious Shader?

This is the first time I can remember hearing of a malicious pixel shader. Cisco Talos announced a set of vulnerabilities targeting VMware and NVIDIA graphics drivers.

Shaders are specialized programs that run on a video card, and are generally used to apply effects like blur, lighting, bump mapping, and more. Most of the graphical improvements in the last few years of gaming is a result of shaders.

Talos researchers were specifically looking at how to compromise a VM Hyper-visor from inside a guest OS, and they discovered that when a host provides 3d acceleration to the guest, shaders are passed directly through to the system drivers without verification. Because the NVIDIA drivers are also vulnerable, this could allow a malicious program on the host to run arbitrary code on the hypervisor.

While this is troubling enough, the topper is that a malicious shader could potentially be run via WebGL. Taken together, this represents a real danger where simply loading a malicious WebGL enabled page could compromise not only a conventional machine, but could also compromise the bare-metal OS even when run on a guest instance.

Both NVIDIA and VMware have already released driver updates that fixes the flaw, so go update!

iOS Problems

Natalie Silvanovich of Google’s Project Zero released a set of 5 iOS vulnerabilities on Wednesday the 7th. These are not garden variety bugs, but so-called “zero click” problems where no user interaction is required for exploit.

The first exploit, for example, is a spoofed visual voicemail message. Visual voicemail notifications are sent as specially formatted text messages and contain information about the message and the address of an IMAP server to connect to and download the message. That information can be spoofed, leading a device to try to download a message from an IMAP server in the control of an attacker. From that point, finding a bug in the iOS IMAP handling code was relatively easy.

5 vulnerabilities have been fixed in iOS updates. There is a 6th vulnerability, CVE-2019-8641, that has yet to be fixed. While a few hints about this problem are given, the details have been withheld until an update has been released to fully fix the problem. One could be a bit cynical and point out that it’s the Google research team announcing these flaws. While there is certainly a self-serving angle to consider, it’s much better for iOS and consumers if flaws are fixed and publicized, rather than kept secret and sold to an offensive security vendor.

One more iOS story is Apple Bleee. Bluetooth Low Energy is an extremely useful communication protocol, allowing Apple devices to perform many of their seemingly magic functionality. The downside is that to make the magic happen, iOS devices are constantly sending BLE signals, probing for other devices. The researchers at Hexway realized that these signals leak lots of data about your device, potentially including your phone number.

iOS uses a SHA256 hash of the device’s phone number as an identifier when using AirDrop. A SHA256 is still a reasonably secure one-way hash, so there’s no problem, right? The clever realization is that while the hash is secure, and the output space is too large to attack, the input space is small enough to be manageable. An attacker could target the most common area codes in their area, limiting the target space further. From there, the SHA256 hashes for all valid numbers can be pre-calculated and stored in a lookup table.

More WPA3 Problems

We’ve discussed Dragonblood, a WPA3 analysis project. A new problem has been identified, a timing analysis attack that leaks information about the internal state of the encryption algorithm.

Spoiler, Use-After-Free, And Ghidra: This Week In Computer Security

The past few days have been busy if you’re trying to keep up with the pace of computer security news. Between a serious Chromium bug that’s actively being exploited on Windows 7 systems, the NSA releasing one of their tools as an open source project, and a new Spectre-like speculative execution flaw in Intel processors, there’s a lot to digest.
Continue reading “Spoiler, Use-After-Free, And Ghidra: This Week In Computer Security”

Peering Into A Running Brain: SDRAM Refresh Analyzed From Userspace

Over on the Cloudflare blog, [Marek] found himself wondering about computer memory, as we all sometimes do. Specifically, he pondered if he could detect the refresh of his SDRAM from within a running program. We’re probably not ruining the surprise by telling you that the answer is yes — with a little more than 100 lines of C and help from our old friend the Fast Fourier Transform (FFT), [Marek] was able to detect SDRAM refresh cycles every 7818.6 ns, lining right up with the expected result.

The “D” in SDRAM stands for dynamic, meaning that unless periodically refreshed by reading and writing, data in the memory will decay. In this kind of memory, each bit is stored as a charge on a tiny capacitor. Given enough time (which varies with ambient temperature), this charge can leak away to neighboring silicon, turning all the 1s to 0s, and destroying the data. To combat this process, the memory controller periodically issues a refresh command which reads the data before it decays, then writes the data back to fully charge the capacitors again. Done often enough, this will preserve the memory contents indefinitely. SDRAM is relatively inexpensive and available in large capacity compared to the alternatives, but the drawback is that the CPU can’t access the portion of memory being refreshed, so execution gets delayed a little whenever a memory access and refresh cycle collide.

Chasing the Correct Hiccup

[Marek] figured that he could detect this “hiccup,” as he calls it, by running some memory accesses and recording the current time in a tight loop. Of course, the cache on modern CPUs would mean that for a small amount of data, the SDRAM would never be accessed, so he flushes the cache each time. The source code, which is available on GitHub, outputs the time taken by each iteration of the inner loop. In his case, the loop typically takes around 140 ns.

Hurray! The first frequency spike is indeed what we were looking for, and indeed does correlate with the refresh times.

The other spikes at 256kHz, 384kHz, 512kHz and so on, are multiplies of our base frequency of 128kHz called harmonics. These are a side effect of performing FFT on something like a square wave and totally expected.

As [Marek] notes, the raw data doesn’t reveal too much. After all, there are a lot of things that can cause little delays in a modern multitasking operating system, resulting in very noisy data. Even thresholding and resampling the data doesn’t bring refresh hiccups to the fore. To detect the SDRAM refresh cycles, he turned to the FFT, an efficient algorithm for computing the discrete Fourier transform, which excels at revealing periodicity. A few lines of python produced the desired result: a plot of the frequency spectrum of the lengthened loop iterations. Zooming in, he found the first frequency spike at 127.9 kHz, corresponding to the SDRAMs refresh period of 7.81 us, along with a number of other spikes representing harmonics of this fundamental frequency. To facilitate others’ experiments, [Marek] has created a command line version of the tool you can run on your own machine.

If this technique seems familiar, it may be because it’s similar the the Rowhammer attack we covered back in 2015, which can actually change data in SDRAM on vulnerable machines by rapidly accessing adjacent rows. As [Marek] points out, the fact that you can make these kinds of measurements from a userspace program can have profound security implications, as we saw with the meltdown and spectre attacks. We have to wonder what other vulnerabilities are lying inside our machines waiting to be discovered.

Thanks to [anfractuosity] for the tip!

Hacking Your Way To A Custom TV Boot Screen

More and more companies are offering ways for customers to personalize their products, realizing that the increase in production cost will be more than made up for by the additional sales you’ll net by offering a bespoke product. It’s great for us as consumers, but unfortunately we’ve still got a ways to go before this attitude permeates all corners of the industry.

[Keegan Ryan] recently purchased a TV and wanted to replace its stock boot screen logo with something of his own concoction, but sadly the set offered no official way to make this happen. So naturally he decided to crack the thing open and do it the hard way The resulting write-up is a fascinating step by step account of the trials and tribulations that ultimately got him his coveted custom boot screen, and just might be enough to get you to take a screw driver to your own flat panel at home.

The TV [Keegan] brought was from a brand called SCEPTRE, but as a security researcher for NCC Group he thought it would be a fun spin to change the boot splash to say SPECTRE in honor of the infamous x86 microarchitecture attack. Practically speaking it meant just changing around two letters, but [Keegan] would still need to figure out where the image is stored, how it’s stored, and write a modified version to the TV without letting the magic smoke escape. Luckily the TV wasn’t a “smart” model, so he figured there wouldn’t be much in the way of security to keep him from poking around.

He starts by taking the TV apart and studying the main PCB. After identifying the principle components, he deduces where the device’s firmware must be stored: an 8 MB SPI flash chip from Macronix. He connects a logic analyzer up to the chip, and sure enough sees that the first few kilobytes are being read on startup. Confident in his assessment, he uses his hot air rework station to lift the chip off the board so that he can dive into its contents.

With the help of the trusty Bus Pirate, [Keegan] is able to pull the chip’s contents and verify its integrity by reading a few human-readable strings from it. Using the binwalk tool he’s able to identify a JPEG image within the firmware file, and by feeding its offset to dd, pull it out so he can view it. As hoped, it’s the full screen SCEPTRE logo. A few minutes in GIMP, and he’s ready to merge the modified image with the firmware and write it back to the chip.

He boots the TV back up and finds…nothing changed. A check of the datasheet for the SPI flash chip shows there are some protection bits used to prevent modifying particular regions of the chip. So after some modifications to the Bus Pirate script and another write, he boots the TV and hopes for the best. Finally he sees the object of his affection pop up on the big screen, a subtle change that reminds him every time the TV starts about the power of reverse engineering.

Foreshadow: The Sky Is Falling Again For Intel Chips

It’s been at least a month or two since the last vulnerability in Intel CPUs was released, but this time it’s serious. Foreshadow is the latest speculative execution attack that allows balaclava-wearing hackers to steal your sensitive information. You know it’s a real 0-day because it already has a domain, a logo, and this time, there’s a video explaining in simple terms anyone can understand why the sky is falling. The video uses ukuleles in the sound track, meaning it’s very well produced.

The Foreshadow attack relies on Intel’s Software Guard Extension (SGX) instructions that allow user code to allocate private regions of memory. These private regions of memory, or enclaves, were designed for VMs and DRM.

How Foreshadow Works

The Foreshadow attack utilizes speculative execution, a feature of modern CPUs most recently in the news thanks to the Meltdown and Spectre vulnerabilities. The Foreshadow attack reads the contents of memory protected by SGX, allowing an attacker to copy and read back private keys and other personal information. There is a second Foreshadow attack, called Foreshadow-NG, that is capable of reading anything inside a CPU’s L1 cache (effectively anything in memory with a little bit of work), and might also be used to read information stored in other virtual machines running on a third-party cloud. In the worst case scenario, running your own code on an AWS or Azure box could expose data that isn’t yours on the same AWS or Azure box. Additionally, countermeasures to Meltdown and Spectre attacks might be insufficient to protect from Foreshadown-NG

The researchers behind the Foreshadow attacks have talked with Intel, and the manufacturer has confirmed Foreshadow affects all SGX-enabled Skylake and Kaby Lake Core processors. Atom processors with SGX support remain unaffected. For the Foreshadow-NG attack, many more processors are affected, including second through eighth generation Core processors, and most Xeons. This is a significant percentage of all Intel CPUs currently deployed. Intel has released a security advisory detailing all the affected CPUs.

Hackaday Links: February 18, 2018

Hacker uses pineapple on unencrypted WiFi. The results are shocking! Film at 11.

Right on, we’ve got some 3D printing cons coming up. The first is MRRF, the Midwest RepRap Festival. It’s in Goshen, Indiana, March 23-25th. It’s a hoot. Just check out all the coverage we’ve done from MRRF over the years. Go to MRRF.

We got news this was going to happen last year, and now we finally have dates and a location. The East Coast RepRap Fest is happening June 22-24th in Bel Air, Maryland. What’s the East Coast RepRap Fest? Nobody knows; this is the first time it’s happening, and it’s not being produced by SeeMeCNC, the guys behind MRRF. There’s going to be a 3D printed Pinewood Derby, though, so that’s cool.

జ్ఞ‌ా. What the hell, Apple?

Defcon’s going to China. The CFP is open, and we have dates: May 11-13th in Beijing. Among the things that may be said: “Hello Chinese customs official. What is the purpose for my visit? Why, I’m here for a hacker convention. I’m a hacker.”

Intel hit with lawsuits over security flaws. Reuters reports Intel shareholders and customers had filed 32 class action lawsuits against the company because of Spectre and Meltdown bugs. Are we surprised by this? No, but here’s what’s interesting: the patches for Spectre and Meltdown cause a noticeable and quantifiable slowdown on systems. Electricity costs money, and companies (server farms, etc) can therefore put a precise dollar amount on what the Spectre and Meltdown patches cost them. Two of the lawsuits allege Intel and its officers violated securities laws by making statements or products that were false. There’s also the issue of Intel CEO Brian Krzanich selling shares after he knew about Meltdown, but before the details were made public. Luckily for Krzanich, the rule of law does not apply to the wealthy.

What does the Apollo Guidance Computer look like? If you think it has a bunch of glowey numbers and buttons, you’re wrong; that’s the DSKY — the user I/O device. The real AGC is basically just two 19″ racks. Still, the DSKY is very cool and a while back, we posted something about a DIY DSKY. Sure, it’s just 7-segment LEDs, but whatever. Now this project is a Kickstarter campaign. Seventy bucks gives you the STLs for the 3D printed parts, BOM, and a PCB. $250 is the base for the barebones kit.

Intel Forms New Security Group To Avoid Future Meltdowns

Intel just moved some high level people around to form a dedicated security group.

When news of Meltdown and Spectre broke, Intel’s public relations department applied maximum power to their damage control press release generators. The initial message was one of defiance, downplaying the impact and implying people are over reacting. This did not go over well. Since then, we’ve started seeing a trickle of information from engineering and even direct microcode updates for people who dare to live on the bleeding edge.

All the technical work to put out the immediate fire is great, but for the sake of Intel’s future they need to figure out how to avoid future fires. The leadership needs to change the company culture away from an attitude where speed is valued over all else. Will the new security group have the necessary impact? We won’t know for quite some time. For now, it is encouraging to see work underway. Fundamental problems in corporate culture require a methodical fix and not a hack.

Editor’s note: We’ve changed the title of this article to better reflect its content: that Intel is making changes to its corporate structure to allow a larger voice for security in the inevitable security versus velocity tradeoff.