I just acquired a second Mac mini M2Pro specifically to run Fedora's Asahi Remix. Despite the setting up appearing to go according to plan, I was shocked to find that when I booted into the Fedora's OS (from RecoveryOS), all I can see is a completely blank display (i.e. identical to how it looks if switched off).
I cannot believe that there is no mention of this in the official docs or FAQs, although I've seen vague references to similar issues (or the same ones) in Community posts from early 2023, and which didn't seem to get resolved then either. Booting into macOS, everything works normally.
Can anyone help, please? This is really frustrating/upsetting. Thank you!
Andrew
By context I assume you acquired a "second hand" Mac mini M2 Pro.
There is no mention of this since that is in general not an issue.
This could be a bug triggered by your display (Even if the display works in macos). Can you try a different one?
How did you install it?
From your description it sounds unlikely but just to make sure: you follow the installation instructions? Especially the second step with powering the device on by pressing and holding the power button after the running the installer in macos?
Hi Jannau,
It is strictly second-hand but never used. I actually have two Thunderbolt-connected displays: one is the old Apple Cinema Display and the other is the LG 5K display (developed specifically for macOS, and the predecessor to the Studio Display).
In terms of installation, I wouldn't have had a clue how to do it were it not for the instructions given at curl https://alx.sh | sh (which I followed to the letter). Based on my knowledge of partitioning Windows SSDs (alongside Linux distros), the architecture now on my Mac is consistent with Linux installed correctly.
I hadn't thought of using an HDMI display (which I happen to have lying around), although I see at https://asahilinux.org/fedora/#device-support the USB-C and Thunderbolt/USB-4 displays are in red—which I guess means unstable or not ready. That's not mentioned anywhere else and if I'd seen this page I probably would have tried to get a MacBook Air. However, you didn't mention that either, so I'm not sure if you were getting at something else.
I appreciate the response regardless!
Andy
Thunderbolt displays are not supported currently.
Hej Andy,
red means on that page not supported. Only HDMI ports are currently supported for external displays. It should work with the HDMI display.
The second hand is not important, I just wrote it since you wrote acquired a second Mac mini M2Pro which made not much sense in the context.
acquired a second Mac mini M2Pro
Either way, not a bug (missing features are not considered bugs).
Metadata Update from @marcan: - Issue status updated to: Closed (was: Open)
Thank you all. Yes, I now see my folly of failing to realise that Thunderbolt is not supported (yet); my bad. The second I plugged in a monitor with an HDMI port, the screen came to life immediately!
Is it known whether a HDMI 2.0 Splitter 1-IN-2-Out 4K@60Hz (e,g, something like this would work to enable two displays from the Mac Mini's single HDMI port? https://www.inkstation.com.au/2601/hdmi-20-splitter-1in2out-4k60hz-hdr10-2-port-hdmi-duplicator-simplecom-cm412-p-21995.html
Those splitters only duplicate the incoming signal (id est, mirroring). They don't give you the ability to use two independent displays from one port. DisplayPort can do this via MST, however the display controller in Apple's SoCs do not support this functionality.
Thank you; I realised/remembered that was the case soon after I asked the question.
A USB-3.0 hub/dock is also out of the question since the only USB-A ports on Mac minis are USB-2.0 (which can only carry data at 5 Gbps—not nearly enough for a (second) 4K display. 4
Since all M-chipped Macs have only a single HDMI port and 2 or more USB-C/USB-4 ports, it's therefore not possible to use more than one 4K display for any Asahi Remix distro (on macOS)—until support for USB-C/USB-4/Thunderbolt is added.
I wish I'd realised that before "acquiring" a second Mac; I guess my excitement got the better of me. Of course, I don't blame anyone for that but myself. :-)
Your conclusion is (mostly) correct but you seem rather confused about the hardware. The USB-A ports are USB 3.0 Gen1 (5Gbps) on the original M1 Mac Mini, and USB 3.0 Gen2 (10Gbps) on all subsequent models (Studios and the M2/M2 Pro Minis). The Thunderbolt USB-C ports are supported on all machines for USB3.0 Gen2 as well as USB-PD charging (on laptops). Non-TB USB-C ports are also fully supported for USB3.0 Gen2 on Studios/iMacs that have them.
Display over USB transport (Type A or Type C, doesn't matter) is a thing, it's called DisplayLink, and it's proprietary and kind of crap, and we don't support it officially, but it does work (over both 5G and 10G ports) and people have been using it as a workaround for the missing DP/TB support. It is the only way to get multiple external displays on Macs right now. It works because it is entirely CPU-based and does not use any special hardware.
We also fully support using the HDMI port on laptops plus the built-in display (which is almost 4K on 16" models).
What isn't supported is Thunderbolt/USB4 (used by Thunderbolt/USB4 docks and displays) and DisplayPort Alternate Mode over Type C (used by non-DisplayLink USB3 hubs/docks and non-TB displays with Type C ports, most HDMI/DP dongles, etc.). Due to the way the protocols work, TB/USB4 docks won't work at all (except maybe USB2 480Mbps mode for USB4 docks specifically, but that's kind of useless), while USB3 hubs/docks with DisplayPort will work but only the USB/peripheral part, not the display output part, and obviously pure-DisplayPort devices won't work at all.
OK, let's break it down so no one's confused—including you, since you misidentified the names of the USB 3-series (USB 3.0 Gen 1/Gen 2" are referred to as USB 3.1 Gen 1/Gen 2), and you appear to have mischaracterised the USB-A port on the Mac mini.
The USB-A ports on the Mac mini M1/M2 models are indeed USB 3.0 ports (not USB 2.0, as I incorrectly noted). Whether the version and generation of the USB-A port changed from the M1 to the M2 model (i.e. from USB 3.1 Gen 1 to Gen 2) is not documented, and is especially vague on Apple's own literature (see https://support.apple.com/en-us/111837). Nevertheless, according to Apple, the transmission speeds of the USB-A port have remained at 5 Gbps; in other words, you're wrong about the 10 Gbps. But while you can quibble over the nomenclature of the USB-A ports, 5 Gbps (or even 10 Gbps) ain't going to light up a 4K display. But I'm rather confused about the hardware?
To be clear, I was NOT referring to what Apple (and you) call "support for", in reference to their Thunderbolt ports, and what they support—since it's immaterial to this discussion (as Neal commented above) since Asahi Remix does not support Thunderbolt of any kind.
I'm well aware of DisplayLink, having used it several times with different hubs, and while it serves the purpose of a workaround, it's absolute crap and a poor substitute for dedicated multiple-monitor support, where "multiple" is >1 or >2 (depending on the Mac Silicon model and M version). The Mac mini M2 Pro has dedicated support for 3 external displays; I'm running two 5K displays, each using one of the two Thunderbolt buses (2 ports per bus) and one 4K display (using its sole HDMI port).
Of course you support the HDMI port and a built-in display (if one exists, which it obviously does not on the Mac mini). What made you believe I thought otherwise?
I understand the point you tried to make (not very well) in your last paragraph, but that's also immaterial. The only material point is that you don't yet support USB-C/USB4/Thunderbolt to the extent that such ports are normally or typically used for powering high-resolution displays. The USB-A port on the Mac mini is a non-issue.
I trust that I've now disabused you of the erroneous belief that I was confused about "the hardware".
I did indeed say 3.0 instead of 3.1 (because they're essentially the same thing, 3.0 just got renamed to 3.1 Gen1, USB version naming is a hilarious mess), but other than that, I have no idea why you're arguing against the people literally reverse engineering and supporting the hardware.
[226500.377335] xhci_hcd 0000:04:00.0: Host supports USB 3.1 Enhanced SuperSpeed [...] [226536.315631] usb 2-1: new SuperSpeed Plus Gen 2x1 USB device number 2 using xhci_hcd
The Type A ports support Gen2x1 10Gbps on everything but the original M1 Mini (which used a different chipset). If Apple claims otherwise, their marketing is wrong, which wouldn't be the first time their marketing disagrees with their actual products. So yes, you're confused, and apparently so are Apple themselves.
1) The Type A ports and the Type C ports support exactly the same USB 3 protocol standard and both work on Asahi, so you bringing up the Type A ports specifically implied some sort of misunderstanding, perhaps a belief that the Type C ports don't work at all on Asahi. 2) Thinking the ports are merely USB 2.0 (which is 480Mbps, not 5 Gbps) further demonstrated more confusion. 3) Bringing up USB data rates at all in the context of display output again makes little sense unless talking about things like DisplayLink, which is why I mentioned that. It doesn't matter if a hypothetical USB3 standard supported 100Gbps because the USB3 protocol is not a display protocol and cannot directly drive displays because there is no hardware mechanism to tunnel display data over USB3, only software approaches like DisplayLink. The data rate is irrelevant. You need an alternate mode (which is only possible over Type C) and then the capabilities of that alternate mode determine what is supported, not whatever the USB3 mode does. The throughput of alternate modes is not the same as the throughput of the baseline USB3 mode and those capabilitles need to be documented separately.
each using one of the two Thunderbolt buses (2 ports per bus)
There are no aggregate Thunderbolt "buses" on Apple Silicon. Each port is a completely independent bus with completely separate PHY, USB3 controller, USB4/TB layer, PCIe backend, and everything, unlike the implementation on Intel machines. The only thing shared is the display controllers, of which there is a limited number (3 on the M2 Pro) but they can be routed independently to the Thunderbolt ports, including routing 2 display controllers to the same port for dual display over one Thunderbolt channel.
What made you believe I thought otherwise?
You said "it's therefore not possible to use more than one 4K display for any Asahi Remix distro" and I wanted to clarify that this limitation is specifically about external displays and the internal display on laptops does not count, on laptops that do indeed have one.
The USB-A port on the Mac mini is a non-issue.
Hence why you bringing it up in the context of display output demonstrated some kind of confusion/misunderstanding. USB-A docks/hubs with display output simply don't exist other than DisplayLink, so the entire premise behind your statement about display output, data rates, and USB A ports just made no sense whatsoever.
Put simply: If you're discounting DisplayLink, any discussion of USB A ports or USB3 data rates is entirely superfluous when talking about displays. The only way to get native display output over a USB standard port is using a Type C port with either DisplayPort Alternate Mode or Thunderbolt/USB4 Alternate Mode (which are the two things we don't support yet). Data rate does not matter, the link-level technology in use does and whether it supports DisplayPort video stream data (which USB3 does not). Native display over USB3 (any data rate, any resolution) does not exist, and alternate modes don't exist for Type A ports either.
What? I guess no, you wouldn=E2=80=99t have any idea=E2=80=94if you=E2=80= =99re inclined to be condescending to well-intentioned users=E2=80=94and be= wrong about the facts. Are =E2=80=9C=E2=80=98the =E2=80=98people=E2=80=99 = literally reverse engineering=E2=80=A6=E2=80=9D superior forms of life? In = any case, I wasn=E2=80=99t =E2=80=9Carguing=E2=80=9D; I was merely pointing= out that I wasn=E2=80=99t as dumb as you made me out to be.
On 15 Jul 2024, at 11:10=E2=80=AFAM, Hector Martin pagure@pagure.io wrote= :
marcan added a new comment to an issue you are following: `` I did indeed say 3.0 instead of 3.1 (because they're essentially the same t= hing, 3.0 just got renamed to 3.1 Gen1, USB version naming is a hilarious m= ess), but other than that, I have no idea why you're arguing against the pe= ople literally reverse engineering and supporting the hardware.
[226500.377335] xhci_hcd 0000:04:00.0: Host supports USB 3.1 Enhanced Super= Speed [...] [226536.315631] usb 2-1: new SuperSpeed Plus Gen 2x1 USB device number 2 us= ing xhci_hcd
The Type A ports support Gen2x1 10Gbps on everything but the original Mini = (which used a different chipset). If Apple claims otherwise, their marketin= g is wrong, which wouldn't be the first time their marketing disagrees with= their actual products. So yes, you're confused, and apparently so are Appl= e themselves.
1) The Type A ports and the Type C ports support exactly the same USB 3 pro= tocol standard and both work on Asahi, so you bringing up the Type A ports = specifically implied some sort of misunderstanding, perhaps a belief that t= he Type C ports don't work at all on Asahi. 2) Thinking the ports are merely USB 2.0 (which is 480Mbps, not 5 Gbps) fur= ther demonstrated more confusion. 3) Bringing up USB data rates at all in the context of display output again= makes little sense unless talking about things like DisplayLink, which is = why I mentioned that. It doesn't matter if a hypothetical USB3 standard sup= ported 100Gbps because the USB3 protocol is not a display protocol and c= annot directly drive displays because there is no hardware mechanism to tu= nnel display data over USB3, only software approaches like DisplayLink. The= data rate is irrelevant. You need an alternate mode (which is only possibl= e over Type C) and then the capabilities of that alternate mode determine w= hat is supported, not whatever the USB3 mode does. The throughput of altern= ate modes is not the same as the throughput of the baseline USB3 mode and= those capabilitles need to be documented separately.
There are no aggregate Thunderbolt "buses" on Apple Silicon. Each port is a= completely independent bus with completely separate PHY, USB3 controller, = USB4/TB layer, PCIe backend, and everything, unlike the implementation on I= ntel machines. The only thing shared is the display controllers, of which t= here is a limited number (3 on the M2 Pro) but they can be routed independe= ntly to the Thunderbolt ports, including routing 2 display controllers to t= he same port for dual display over one Thunderbolt channel.
Hence why you bringing it up in the context of display output demonstrated = some kind of confusion/misunderstanding.
``
To reply, visit the link below or just reply to this email https://pagure.io/fedora-asahi/remix-bugs/issue/10