Wow, talk about a good run. 😅
Still, I feel a little sad to see it go. I mean, it's a relatively easy framebuffer device to support. But other than hobbyist niche use, I totally get why there's no point in keeping it in the mainline Linux kernel source code.
Agreed. Sad day.
I didn't even know there was some support for HGC in the kernel itself...
I recall experimenting with Linux and Hercules about 30 years ago, using the XFree86 XF86_Mono server - it did work, but very inconvenient because of poor resolution and 1-bit depth.
In 2003, I voted in favour of joining the European Union. However, due to recent developments - especially the restrictions on cash usage - I'm hereby withdrawing my support. DOWN WITH THE EU!
Is this an fbdev hercules driver? I've kind of been confused what the driver can be used to do. I think some people in earlier days did use a driver in the kernel to get the dual monitor vga+mono setup that added a tty on the mono display. If for graphics, I don't know what can really make use of that.
When I last used a hercules card on linux, I pretty much just wanted a console for a server. All I did was stick the card into a computer for a primary display and let the BIOS initialize it. So this change should not really change that if you disable kms/fb, and you have BIOS support, it should still work.
If you had the old X Server, you could also do graphics without needing it in the kernel. Who knows if you can still compile it on a Linux Kernel 7.x system.
I kind of read the situation as the kernel maintainers are trying to delete all ISA support from the kernel. But this situation seems strange, and the article admitted that there are still industrial uses, including for the hercules driver...?
the3dfxdude wrote on Yesterday, 15:08:I kind of read the situation as the kernel maintainers are trying to delete all ISA support from the kernel. But this situation seems strange, and the article admitted that there are still industrial uses, including for the hercules driver...?
IIRC it's partly due to the flood of potential security hole reports due to AI. With no one to test, the kernel team has no idea whether their fixes break the drivers or not so they just want to delete drivers if there is no one to test. They did back off a bit on common ISA ethernet cards that VM software still presents as an emulated card, since users pointed out they're still using them.
😢
"Time, it seems, doesn't flow. For some it's fast, for some it's slow.
In what to one race is no time at all, another race can rise and fall..." - The Minstrel
//My video channel//
jakethompson1 wrote on Yesterday, 15:17:the3dfxdude wrote on Yesterday, 15:08:I kind of read the situation as the kernel maintainers are trying to delete all ISA support from the kernel. But this situation seems strange, and the article admitted that there are still industrial uses, including for the hercules driver...?
IIRC it's partly due to the flood of potential security hole reports due to AI. With no one to test, the kernel team has no idea whether their fixes break the drivers or not so they just want to delete drivers if there is no one to test. They did back off a bit on common ISA ethernet cards that VM software still presents as an emulated card, since users pointed out they're still using them.
Just another item on the ever expanding list of "things that AI makes worse"
jakethompson1 wrote on Yesterday, 15:17:IIRC it's partly due to the flood of potential security hole reports due to AI. With no one to test, the kernel team has no idea whether their fixes break the drivers or not so they just want to delete drivers if there is no one to test. They did back off a bit on common ISA ethernet cards that VM software still presents as an emulated card, since users pointed out they're still using them.
Hmmm... Considering the limited functionality of the HGC standard (a text mode with 2 intensities and a fixed resolution monochrome graphics mode with two frame buffers) and therefore the relatively limited attack surface, if we were to find a dedicated Vogoner willing to volunteer to maintain and regularly test the driver, perhaps Linus and his lieutenants could be convinced to keep the HGC fbdev driver in the mainline kernel? 🙂
It might be come part of a larger initiative to set up a vintage test farm that automatically tests old drivers on actual old hardware, remotely driven by CI/CD pipelines.
We could call it "Jurassic Park"! 😁
digger wrote on Today, 15:15:Hmmm... Considering the limited functionality of the HGC standard (a text mode with 2 intensities and a fixed resolution monochr […]
jakethompson1 wrote on Yesterday, 15:17:IIRC it's partly due to the flood of potential security hole reports due to AI. With no one to test, the kernel team has no idea whether their fixes break the drivers or not so they just want to delete drivers if there is no one to test. They did back off a bit on common ISA ethernet cards that VM software still presents as an emulated card, since users pointed out they're still using them.
Hmmm... Considering the limited functionality of the HGC standard (a text mode with 2 intensities and a fixed resolution monochrome graphics mode with two frame buffers) and therefore the relatively limited attack surface, if we were to find a dedicated Vogoner willing to volunteer to maintain and regularly test the driver, perhaps Linus and his lieutenants could be convinced to keep the HGC fbdev driver in the mainline kernel? 🙂
It might be come part of a larger initiative to set up a vintage test farm that automatically tests old drivers on actual old hardware, remotely driven by CI/CD pipelines.
We could call it "Jurassic Park"! 😁
That's sound all good and fine. But as jake said, they probably did drop this because they don't have an active maintainer right now to handle the flood of information coming from their AI overlords. So it was just easier to drop things, and this is kind of arbitrary even if someone could have maintained it, or that there are still users out there.
I just had a first hand experience with an AI security fix yesterday, as there are tell-tale signs, that got accepted into open source project streams, without anyone reviewing or testing it! Not only does it not work, but it completely broke the function and returns wrong! It changed code that was completely fine in the name of security because AI doesn't know what the function is for, and therefore how it is not actually vulnerable. What's worse, is it went through the largest distributions, very big corporate ones, already a YEAR ago, when we finally got it in the last few days in our little corner of the internet for us to see and figure this out. This means it is likely widespread but not many people are actually using it to have noticed.
So I don't think that even having a few users and keeping it in is going to remain unaffected, if they are just going to get people (or fully automated machines) pushing through this slop in the name of "security". It might be better to drop it from linux, and let more competent people maintain it out of tree. These are stable, simple enough drivers and programs, that don't need much maintenance -- well certainly not the careless kind. I'd rather them leave this stuff alone than pretend they have any clue what they are doing running AI.
digger wrote on Today, 15:15:Hmmm... Considering the limited functionality of the HGC standard (a text mode with 2 intensities and a fixed resolution monochr […]
jakethompson1 wrote on Yesterday, 15:17:IIRC it's partly due to the flood of potential security hole reports due to AI. With no one to test, the kernel team has no idea whether their fixes break the drivers or not so they just want to delete drivers if there is no one to test. They did back off a bit on common ISA ethernet cards that VM software still presents as an emulated card, since users pointed out they're still using them.
Hmmm... Considering the limited functionality of the HGC standard (a text mode with 2 intensities and a fixed resolution monochrome graphics mode with two frame buffers) and therefore the relatively limited attack surface, if we were to find a dedicated Vogoner willing to volunteer to maintain and regularly test the driver, perhaps Linus and his lieutenants could be convinced to keep the HGC fbdev driver in the mainline kernel? 🙂
It might be come part of a larger initiative to set up a vintage test farm that automatically tests old drivers on actual old hardware, remotely driven by CI/CD pipelines.
We could call it "Jurassic Park"! 😁
I don't know about the HGC specifically, but the text mode virtual console generally has been a source of security holes. If you notice Shift-PageUp scrollback on tty1-6 broke a while ago, that was a simplification of the virtual console as part of a security fix rather than fixing the root issue, because the devs believed that functionality belongs in user space rather than kernel anyway.
