Skip to contentSkip to conversation

x86-64 2.4 tree in strict maintenance mode now

DISCUSSION Andi Kleen 2 participants 3 messages Apr 14, 2004 – Apr 16, 2004

lore
Share & tools
lore

Andi Kleen

author
#0 lore raw

2.4.26 has been finally released. It includes the patches for Intel support and a few other bug fixes for x86-64. This was about the latest feature I planned to merge (in fact one more than originally planned)

2.4/x86-64 is currently in a quite good shape. It is widely used and seems to generally work well. 2.6 is also good enough now to be used in production.

As far as I'm concerned the 2.4 port of x86-64 is in strict maintenance mode now. This means no features will not be accepted anymore; only very clear bug fixes that fix serious bugs (like oopses or security holes)

In particular this means that the 32bit emulation is frozen now; any missing ioctls etc. are not bugs and will not be fixed in mainline.

I will also not add workarounds for broken hardware to 2.4 anymore unless the change is very simple and obvious (and even then it may be not worth it)

For example the recent VIA IOMMU breakage workaround is already 2.6 only with no plans to backport. If you have buggy or very new hardware you may need to migrate to 2.6.

This also means that the 2.4 CVS kernel tree on x86-64.org will be retired. I expect that the small obvious bug fixes that should be still merged will be all sent directly to Marcelo and put into the main tree; so there is no need for an x86-64 private tree anymore.

If you want to do any features or cleanups for x86-64 work on 2.6.

-Andi

#1 lore raw

Hi!

Quoted 2 lines

I will also not add workarounds for broken hardware to 2.4 anymore unless the change is very simple and obvious (and even then it may be not worth it)

This looks even more strict than Marcelo's rules... Is it wise?

Signature · 1 line
64 bytes from 195.113.31.123: icmp_seq=28 ttl=51 time=448769.1 ms         
author
#2 lore raw
Quoted 6 lines · On Fri, 16 Apr 2004 17:00:27 +0200 Pavel Machek <[email protected]> wrote:

Hi!

Quoted 2 lines

I will also not add workarounds for broken hardware to 2.4 anymore unless the change is very simple and obvious (and even then it may be not worth it)

This looks even more strict than Marcelo's rules... Is it wise?

Marcelo's rules should be the same now for 2.4.27 and later (modulo SATA)

In fact it has been this way for some time already (except for the IA32e merge, which was a special case). This announcement just formalizes it.

-Andi