Post B19fUDVIRe14P1PJrs by malwareminigun@infosec.exchange
(DIR) More posts by malwareminigun@infosec.exchange
(DIR) Post #B17YzRgyeemROAtYeW by karotte@chaos.social
2025-12-10T21:58:28Z
0 likes, 0 repeats
You may have heard of eUSB, the low-voltage cousin to the venerable USB 2.0 from 25 years ago. It's used on SoCs made on small process nodes that can't support standard USB 2.0 voltage levels.Turns out, it recently got an upgrade called eUSB2V2 (yes, really). It supports asymmetric operation (which is a first for USB) at up to 4.8Gbit/s and is aimed at connecting high-resolution webcams in laptops.
(DIR) Post #B17YzSgexK7WTTkonQ by malwareminigun@infosec.exchange
2025-12-10T22:56:24Z
0 likes, 0 repeats
@karotte @whitequark I can’t believe they still can’t get the names right even when it’s not about the consumer “make the names confusing so we can advertise the same garbage as new” scamI will keep protesting by saying USB <bitrate> until they choose something more sane.No: USB 3 Gen 1 Yes: USB 5GbpsNo: USB 3 Gen 2 Yes: USB 10GbpsNo: USB 4 2.0 Yes: USB 80GbpsAnd so on
(DIR) Post #B17YzTip6lRfgTm3o8 by ignaloidas@not.acu.lt
2025-12-10T23:40:06.329Z
0 likes, 0 repeats
@malwareminigun@infosec.exchange @karotte@chaos.social @whitequark@mastodon.social FWIW for users, that's exactly what the USB forum themselves recommendsSpec names aren't for users, but for developers
(DIR) Post #B17ZYJfIUW3id2jkC8 by fleaz@chaos.social
2025-12-10T23:07:53Z
0 likes, 0 repeats
@malwareminigun @karotte @whitequark You forgot the worst of them all "USB 3.2 Gen 2x2" 🤡
(DIR) Post #B17ZYLMw9kYpugA6r2 by malwareminigun@infosec.exchange
2025-12-10T23:10:35Z
0 likes, 0 repeats
@fleaz I think 'USB 4 2.0' is worse than that. At least the 'x2' part actually describes the operation of the interface and conveys that devices not supporting that protocol would still operate at 10Gbps (assuming USB 3 Gen 2 is understood as 10Gbps).In fact this protocol is the one big wrench in the gears of my 'just name then by bandwidth' strategy as my hypothetical 'USB 20Gbps' actually operate in 'USB 10Gbps' mode for most 'USB 40Gbps' implementations like Thunderbolt 3/4 controllers.
(DIR) Post #B17ZYMPoGYS99sVuyG by ignaloidas@not.acu.lt
2025-12-10T23:46:24.407Z
0 likes, 0 repeats
@malwareminigun@infosec.exchange @fleaz@chaos.social I don't get why your "USB 20Gbps" (I presume Gen 2 x 2) would operate in "USB 10Gbps"(Gen 1 x 2? Gen 2 x 1??) with "USB 40Gbps" (Gen 3 x 2)Anyways, USB4 2.0 is a spec version, not a technology
(DIR) Post #B19ZJIHxiNraKfZwVU by coelacanthus@infosec.exchange
2025-12-11T20:32:32Z
0 likes, 0 repeats
@malwareminigun @karotte @whitequark It's not enough, I think. Both USB 3.2 gen2x1 and 3.2 gen1x2 are 10Gbps, but they are not the same thing. Plug USB 3.2 gen2x1 into gen1x2, you will got the most common speed gen1x1 which is 5Gbps, vice versa. The same thing is also happened on USB4 v1 and USB 3.2 gen2x2, both are 20Gbps (ignore optional USB4 v1 40Gbps). But USB4 use a different PHY layer, so it need a tunnel to be compatible with USB 3. Unfortunately, only 10Gbps tunnel is required, 20Gbps tunnel is optional, so if you plug a USB 3.2 gen2x2 into USB 4, you may can only get 10Gbps instead of 20Gbps.
(DIR) Post #B19ZJK25EOLlk0AI2C by ignaloidas@not.acu.lt
2025-12-11T22:53:09.502Z
0 likes, 0 repeats
@coelacanthus@infosec.exchange @malwareminigun@infosec.exchange @karotte@chaos.social @whitequark@mastodon.social re Gen1x2 - afaict nobody implemented that as the highest speed for their device. It came in the same release as Gen2x2 (Gen2 was a thing before that) and basically everything that wanted x2 used Gen2x2. There's a reason there's no marketing guidance for 10Gbps USB Type-C calbles - the ones only capable of Gen1 speeds are marked as 5Gbps, even though they do have both pairs.Re USB4 - I don't get where you're getting that from? USB4 can use (and must support) the same Gen2 PHY layer as USB 3, and hosts and hubs must support USB 3. The devices technically don't need to support USB 3 but I haven't seen one that could but doesn't. The protocol is different, yes, and it is doing tunneling, but it will only be doing that if you use a hub anyways. The limitation is that the hub is only required to implement Gen2x1, and Gen2x2 is optional - but that won't be changing whether the prot is operating in USB4 or not, and I haven't heard of any host implementations that don't handle it as Gen2x2.If this fucking thread ends up spreading even more USB misinfo, I'm going to fucking snap
(DIR) Post #B19ZyThQCAAzaw7p8S by miek@mastodon.social
2025-12-10T23:05:23Z
0 likes, 0 repeats
@malwareminigun @karotte i bring good news https://www.usb.org/sites/default/files/USB-IF%20Integrators%20List%20Marketing%20Name%20Guidance_January%202023.pdf
(DIR) Post #B19ZyVPPq4xgtfiTLc by malwareminigun@infosec.exchange
2025-12-10T23:14:09Z
0 likes, 0 repeats
@miek I'll believe it when I see it. As far as I can tell the only reason their naming is such a complicated mess is that it's good for manufacturers when they can sell the slower port with the new name.
(DIR) Post #B19ZyWKqOYtnlmaKrQ by jernej__s@infosec.exchange
2025-12-10T23:55:17Z
0 likes, 0 repeats
@malwareminigun @miek Remember when USB 2 came out, and they renamed USB 1 to USB Full-Speed (USB 2 was USB High-Speed)?
(DIR) Post #B19ZyX8TQ2b8FhnyDY by ignaloidas@not.acu.lt
2025-12-11T23:00:34.186Z
0 likes, 0 repeats
@jernej__s@infosec.exchange @malwareminigun@infosec.exchange @miek@mastodon.social AFAICT USB 1 always had Lo-speed and Full-Speed? Here's a screenshot from a delightfully old presentation long before USB 2 was ever a thing
(DIR) Post #B19ao1ReUOZN1XKMGu by pmdj@mstdn.social
2025-12-11T23:07:00Z
0 likes, 0 repeats
@ignaloidas @whitequark @malwareminigun @karotte @coelacanthus As far as I‘m aware, the USB implementation in Apple‘s M series SoCs supports Gen2 and USB4 transports, but not Gen2x2 mode. So Gen2x2 devices fall back to 10Gbps. At least that‘s the case in M1-M3, not sure if the newest ones change that.
(DIR) Post #B19ao2IpIh6VgSCp9c by ignaloidas@not.acu.lt
2025-12-11T23:09:54.179Z
0 likes, 0 repeats
@pmdj@mstdn.social @whitequark@mastodon.social @malwareminigun@infosec.exchange @karotte@chaos.social @coelacanthus@infosec.exchange huh, I could've sworn that Intel required TB3/4 ports to support USB 3 Gen 2x2, considering that they require basically every other optional feature, that's surprising if they don't.
(DIR) Post #B19b8oVnlkn2zgBNCK by malwareminigun@infosec.exchange
2025-12-11T23:09:22Z
0 likes, 0 repeats
@ignaloidas That it's such a mess that such misinfo is possible is kinda at the feet of the USB-IF. Though I'm happy to see they are adopting the form I've been using in protest for years.
(DIR) Post #B19b8pmV39k0vl0Drc by ignaloidas@not.acu.lt
2025-12-11T23:13:41.561Z
0 likes, 0 repeats
@malwareminigun@infosec.exchange USB-IF have been saying "just do the speeds" since USB 3.1 / Gen 2 days (so around 2012 or so). But to use the marketing terms that USB-IF provides and recomends you gotta get your products certified, and literally no consumer hosts do for some fucking reasonIf Microsoft added a "certify your USB hosts" requirement to their ready for windows certification, this could've been different for a long time nowWith USB4 USB-IF is way more aggressive with their trademarks, but I'm still not really seeing host devices actually getting certifications so...
(DIR) Post #B19bQUrJZrDrFmg1tQ by malwareminigun@infosec.exchange
2025-12-11T23:10:43Z
0 likes, 0 repeats
@ignaloidas Thunderbolt 3 (2015) came out before USB 3 Gen 2 x2 (2017), and Thunderbolt 4 seems to be Thunderbolt 3 but where the host guarantees it can do IOMMU
(DIR) Post #B19bQVt7kcGQRgWzLs by ignaloidas@not.acu.lt
2025-12-11T23:16:53.169Z
0 likes, 0 repeats
@malwareminigun@infosec.exchange Gen 2x2 came out with Type-C, TB 3 so certainly did not come out before it. TB4 is USB4 with a bunch of optional features (PCIe, at least 2 displays with a certain resolution, must support 40Gbps speeds, etc.)
(DIR) Post #B19c6wrfqJOkH55Trk by malwareminigun@infosec.exchange
2025-12-11T23:18:51Z
0 likes, 0 repeats
@ignaloidas Example: x670e and x870e versions of more or less the same board; the one from 2022 uses the name soup, the one from 2024 uses just the bandwidth figure. Maybe USB-IF said to do that in the USB3 days but hosts didn't get the memo. Looks like they may now be getting the memo though...
(DIR) Post #B19c6xnSNTcRAI7cvo by ignaloidas@not.acu.lt
2025-12-11T23:24:33.604Z
0 likes, 0 repeats
@malwareminigun@infosec.exchange initially it was SuperSpeed 5Gbps and SuperSpeed 10GbpsIt took a while to drop the SuperSpeed brandingI have an old, very very bad, barely coherent rant of a video that goes over the history of USB-IF suggested marketing terms from a time when Linus of poor tech tips did a whole bunch of misinfo and blamed USB-IF for not using speeds (such shit has really annoyed me for a long while now). If you're really interested, I do go through a whole bunch of history in it, though it's also barely watchable, but if you wish it's here https://www.youtube.com/watch?v=Mgb6GCp_TZ0
(DIR) Post #B19d4wTARQ9cQKER8K by malwareminigun@infosec.exchange
2025-12-11T23:21:03Z
0 likes, 0 repeats
@ignaloidas No, the Type-C connector is USB 3.1 (2013); "x2" was not added until USB 3.2 (2017). Thunderbolt 3 is in the middle (2015).
(DIR) Post #B19d4xsNDBuWmn253o by ignaloidas@not.acu.lt
2025-12-11T23:35:23.583Z
0 likes, 0 repeats
@malwareminigun@infosec.exchange Ah shit, somehow mistook the revision dates in the documents, yeah, you're right.
(DIR) Post #B19fUDVIRe14P1PJrs by malwareminigun@infosec.exchange
2025-12-11T23:07:43Z
0 likes, 0 repeats
@pmdj @ignaloidas @whitequark @karotte @coelacanthus Also true for Thunderbolt 3 and 4. Not sure about 5.
(DIR) Post #B19fUFHtoQUJw39eGO by pmdj@mstdn.social
2025-12-11T23:35:36Z
0 likes, 0 repeats
@malwareminigun @ignaloidas @whitequark @karotte @coelacanthus Indeed, Intel‘s own TB3 controllers did not support Gen2x2 either. I think at least some of their TB4 ones don’t either.
(DIR) Post #B19fUGIe38g94eVl44 by ignaloidas@not.acu.lt
2025-12-12T00:02:20.438Z
0 likes, 0 repeats
@pmdj@mstdn.social @malwareminigun@infosec.exchange @whitequark@mastodon.social @karotte@chaos.social @coelacanthus@infosec.exchange Huh, I'm really surprised about that, considering that for TB4 they have to support most of the signaling stuff anyways, it's just the link layer that needs real updates and logical layer mostly just needs an upgrade in bandwidth. It's not like they'd need to completely rewrite the things that limit the bandwidth, just upgrade them somewhat.
(DIR) Post #B1AP9T93gvpsiv3BzM by malwareminigun@infosec.exchange
2025-12-12T08:29:34Z
0 likes, 0 repeats
@ignaloidas This is part of why I assumed TB3 mostly == TB4 but with a guarantee that the host can do IOMMU
(DIR) Post #B1AP9UIJQ0peHuO63E by ignaloidas@not.acu.lt
2025-12-12T08:34:03.363Z
0 likes, 0 repeats
@malwareminigun@infosec.exchange USB4 is more capable in some respects (e.g. because it does protocol tunneling instead of dedicating lanes to some protocol, if your display traffic doesn't use up what would be two full lanes of bandwidth, you're getting more PCIe/USB traffic bandwidth than you would with TB3), and TB4 is essentially just an Intel marketing certification of USB4 with a bunch of optional features and other requirements. IOMMU is one, two displays is another, there's a full list of that somewhere on Intel's site.