Printable View
the thing about ditching other ancient archs that may still be in use (such as i486) is the modern kernel is likely too heavy for them, nearly none of the new features put into the kernel will actually aid old archs, and people who use them probably don't care to update to the latest anyway.
While it is a little sad that the original arch is being ditched, I'm glad to see the kernel getting cleaned up a bit and making future releases easier to work on.
Who still has the original 80386 machines around? As we know there are times when things come to an end and thus support for the old 386 is ending. It sure has been a great run though for Linux on the old machines though, considering that Linux kernel was born on that architecture.
It sure is a great time to do some spring cleaning on the kernel, so why not look at purging support for very old devices or at least mark them as Legacy within the tree so all it will take is a config option to turn on legacy device support for those rare cases there happens to be some old device or controller in a particular system
they already do this and this has been the case, for as long as as it has been applicable... When configuring a kernel (make menuconfig), if you have done that before...? (which i assume is probably not the case, based on your above comments), there are still legacy drivers (marked 'deprecated' and disabled). One such example;Quote:
Originally Posted by DeepDayze
besides that, all of the device drivers (or any other feature) of the linux kernel can be enabled/disabled (compiled or not, builtin or as a module, etc) via what you would call a 'config option'. it's all pretty standard stuff :)Quote:
Originally Posted by make menuconfig
Quote:
Originally Posted by DeepDayze
Dude, you don't say "its a great time to do _spring_ cleaning" in december. Its just not right.
Regardless, I find it also slightly wrong to drop support for arches. It's just the easy way out. arch support should be sufficiently pluggable that you wouldn't get limited by supporting more architectures, and the API to be reasonably stable enough, that you wouldn't need to maintain arches each release (that totally ISNT the case in Linux and probably the exact reason for the drop: nobody wants to support it)
are there real world examples of people really compiling the latest kernel releases and using them on 486's????
how? and why? what for?
Before switching to a full intel laptop I was trying pretty much all the latest distros in a pentium iv system with 1 gram.
It's an exercise in masochism, even in that system I had to settle for like kernel v2.6 at most 2.4 working the best.
with the addition of arm code and arm64 and all this new arches wouldn't it be better to drop anything prior to pentium4's??
Linux kernel + small env need 4MB of memory minimum. But this config is mostly under P1 minimum, then drop the 486 can be a good idea. But if not add amount of code and emulation... If the support is same as P4, ... why don't keep it.
I'm to keep the old compatilibity if newer than 10 years old (else you have older linux -> kernel + OS, or emulation) or if don't block the evolution.
Quote:
Wikipedia:
The instruction set of the i486 is very similar to its predecessor, the Intel 80386
The team could at least mark 80386 support Legacy for now though then totally remove the support for the 80386 cpu itself (not to mean the entire i386 architecture!)Quote:
Originally Posted by ninez
I have a hard time believing that anyone still using a 386 would be using anything newer than 2.4, maybe 2.6, let alone anything 3.0 or newer.
Dammit. :(
Now I have to go out and buy one of those four-thousand dollar Pentiums. :mad:
Might be a bit too close a shave. I am still running Linux quite well on an old AthlonXP.Quote:
Originally Posted by Pallidus