[gentoo-kernel] New genpatches numbering scheme

[gentoo-kernel] New genpatches numbering scheme

[Daniel Drake]

With linux-stable becoming more popular we are reaching the limit of the 
current numbering scheme, which is currently hacked to allow 100 patches 
from that tree.

The obvious solution is to stop requiring a unique 4 digit number for 
each patch, but I have grown to like that system, and it allows for some 
easy-to-use scripts.

I also feel restricted by the current scheme, pasted below for reference.

1XXX	critical fixes
  1XX	 security
  3XX	 arch-compat
  9XX	 other
2XXX	driver related patches
  1XX	 network
  3XX	 raid/storage
  5XX	 graphics/sound
  7XX	 motherboard
  9XX	 other
3XXX	performance patches
  1XX	 cpu-task schedular & related
  3XX	 disk/memory/swap I/O and management
  5XX	 graphics/sound
  9XX	 other
4XXX	additional features.
  1XX	 network
  3XX	 raid/storage
  5XX	 graphics/sound
  7XX	 filesystem
  9XX	 other
5XXX	experimental patches
  1XX	 network
  3XX	 raid/storage
  5XX	 graphics/sound
  7XX	 filesystem
  9XX	 other

I dislike having to file fixes for networking core and netfilter in the 
same 'category' as fixes for network drivers. Similar situation for 
storage. I also dislike the motherboard category which is very broad by 
nature, generally swallowing patches for USB, PCI, ACPI, ...

At the same time, I don't want to get too specific: for example, there's 
not much point differentiating between ALSA and OSS drivers, framebuffer 
video drivers and DRI video drivers, etc.

There's also no point in the 3XXX and 5XXX ranges with our current 
policy - we'd never put things under those categories.

Here's my new proposal. Any comments? I hope to put something similar to 
the following into action for the first 2.6.16 release.

FIXES
=====
1000-1400	linux-stable
1400-1500	linux-stable queue
1500-1600	architecture-related
1600-1700	security
1700-1800	mm/scheduling/misc
1800-1900	filesystems
1900-2000	networking core
2000-2100	storage core
2100-2000	power management (acpi, apm)
2200-2300	bus (usb, ieee1394, pci, pcmcia, ...)
2300-2400	network drivers
2400-2500	storage drivers
2500-2600	input
2600-2800	media (graphics, sound, tv)
2800-2900	other
2900-4000	reserved

FEATURES
========
4000-4100	network
4100-4200	storage
4200-4300	graphics
4300-4400	filesystem
4400-4500	other
-- 
gentoo-kernel@gentoo.org mailing list

Re: [gentoo-kernel] New genpatches numbering scheme

[Tim Yamin]

On Mon, Feb 13, 2006 at 10:41:14PM +0000, Daniel Drake wrote:
> FIXES
> =====
> 1000-1400	linux-stable
> 1400-1500	linux-stable queue
> 1500-1600	architecture-related
> 1600-1700	security
> 1700-1800	mm/scheduling/misc
> 1800-1900	filesystems
> 1900-2000	networking core
> 2000-2100	storage core
> 2100-2000	power management (acpi, apm)
> 2200-2300	bus (usb, ieee1394, pci, pcmcia, ...)
> 2300-2400	network drivers
> 2400-2500	storage drivers
> 2500-2600	input
> 2600-2800	media (graphics, sound, tv)
> 2800-2900	other
> 2900-4000	reserved
> 
> FEATURES
> ========
> 4000-4100	network
> 4100-4200	storage
> 4200-4300	graphics
> 4300-4400	filesystem
> 4400-4500	other

I like it :)
-- 
gentoo-kernel@gentoo.org mailing list

Re: [gentoo-kernel] New genpatches numbering scheme

[Greg KH]

On Mon, Feb 13, 2006 at 10:41:14PM +0000, Daniel Drake wrote:
> FIXES
> =====
> 1000-1400	linux-stable

Why does this need to be so big?  Why not just include the -stable
patches directly, all rolled up (2.6.15.4), instead of all broken out?

Other than that, I have no problem with the new scheme.

thanks,

greg k-h
-- 
gentoo-kernel@gentoo.org mailing list

Re: [gentoo-kernel] New genpatches numbering scheme

[Daniel Drake]

Greg KH wrote:
> On Mon, Feb 13, 2006 at 10:41:14PM +0000, Daniel Drake wrote:
>> FIXES
>> =====
>> 1000-1400	linux-stable
> 
> Why does this need to be so big?  Why not just include the -stable
> patches directly, all rolled up (2.6.15.4), instead of all broken out?

Sometimes (admittedly not very often) we get clashes. For example, the 
usermode-sources upstream patch sometimes includes UML fixes which are 
also present in -stable.

This allows us to work around that with minimal hassle in the ebuild, by 
setting UNIPATCH_EXCLUDE=1234 where 1234 is the genpatches ID of the 
duplicated patch.

Daniel
-- 
gentoo-kernel@gentoo.org mailing list

Re: [gentoo-kernel] New genpatches numbering scheme

[Luca Barbato]

Daniel Drake wrote:
> FIXES
> =====
> 1000-1400    linux-stable
> 1400-1500    linux-stable queue
> 1500-1600    architecture-related
> 1600-1700    security
> 1700-1800    mm/scheduling/misc
> 1800-1900    filesystems
> 1900-2000    networking core
> 2000-2100    storage core
> 2100-2000    power management (acpi, apm)
> 2200-2300    bus (usb, ieee1394, pci, pcmcia, ...)
> 2300-2400    network drivers
> 2400-2500    storage drivers
> 2500-2600    input
> 2600-2800    media (graphics, sound, tv)
> 2800-2900    other
> 2900-4000    reserved
> 
> FEATURES
> ========
> 4000-4100    network
> 4100-4200    storage
> 4200-4300    graphics
> 4300-4400    filesystem
> 4400-4500    other

Seems ok

-- 

Luca Barbato

Gentoo/linux Developer		Gentoo/PPC Operational Leader
http://dev.gentoo.org/~lu_zero

-- 
gentoo-kernel@gentoo.org mailing list

Re: [gentoo-kernel] New genpatches numbering scheme

[John Mylchreest]

[-- Attachment #1: Type: text/plain, Size: 1458 bytes --]

On Mon, Feb 13, 2006 at 10:41:14PM +0000, Daniel Drake <dsd@gentoo.org> wrote:
> FIXES
> =====
> 1000-1400	linux-stable
> 1400-1500	linux-stable queue
> 1500-1600	architecture-related
> 1600-1700	security
> 1700-1800	mm/scheduling/misc
> 1800-1900	filesystems
> 1900-2000	networking core
> 2000-2100	storage core
> 2100-2000	power management (acpi, apm)
> 2200-2300	bus (usb, ieee1394, pci, pcmcia, ...)
> 2300-2400	network drivers
> 2400-2500	storage drivers
> 2500-2600	input
> 2600-2800	media (graphics, sound, tv)
> 2800-2900	other
> 2900-4000	reserved
> 
> FEATURES
> ========
> 4000-4100	network
> 4100-4200	storage
> 4200-4300	graphics
> 4300-4400	filesystem
> 4400-4500	other

I assume the last number is exclusive ;)
I can't see anything wrong. Looks good.

For the best part the naming convention is trivial, afterall, its only
really the patch order that matters - which is likely why the original
lasted so long :)

I would maybe suggest making the security patch size that little bit bigger
(hell, we have the room) and shrink linux-stable a little bit. This is
directly relating to what we spoke about a while ago re: maintaining
aging genpatches. This isn't vital though.

Cheers for that.
- John

-- 
Role:            Gentoo Linux Kernel Lead
Gentoo Linux:    http://www.gentoo.org
Public Key:      gpg --recv-keys 9C745515
Key fingerprint: A0AF F3C8 D699 A05A EC5C  24F7 95AA 241D 9C74 5515


[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-kernel] New genpatches numbering scheme

[Greg KH]

On Tue, Feb 14, 2006 at 12:00:29AM +0000, Daniel Drake wrote:
> Greg KH wrote:
> >On Mon, Feb 13, 2006 at 10:41:14PM +0000, Daniel Drake wrote:
> >>FIXES
> >>=====
> >>1000-1400	linux-stable
> >
> >Why does this need to be so big?  Why not just include the -stable
> >patches directly, all rolled up (2.6.15.4), instead of all broken out?
> 
> Sometimes (admittedly not very often) we get clashes. For example, the 
> usermode-sources upstream patch sometimes includes UML fixes which are 
> also present in -stable.
> 
> This allows us to work around that with minimal hassle in the ebuild, by 
> setting UNIPATCH_EXCLUDE=1234 where 1234 is the genpatches ID of the 
> duplicated patch.

Ah, ok that makes sense.  Looks good to me.

thanks,

greg k-h
-- 
gentoo-kernel@gentoo.org mailing list