TuxOnIce in Ubuntu kernel?

[Nigel Cunningham]

Hi.

Chow Loong Jin wrote:
> On Monday 19,October,2009 08:32 PM, Amit Kucheria wrote:
>> This question has come up _several_ times in the past. We've refused to carry
>> the Tux-on-ice patch for the following reasons:
>>
>>  - Due to its size and complexity (distro kernel maintenance becomes harder)
>>  - Upstream compatibility (changing to TOI means a different set of bugs than
>>  any other distro, again leading to maintenance issues)
> I understand these two concerns, but unfortunately, I can't answer to those. A
> diffstat from my local git tree shows:
>  109 files changed, 18005 insertions(+), 158 deletions(-)
> which constitutes quite a lot of changes.

It should be mentioned, though, that the vast majority of those changes
are completely new files, and simple symbol exports that are used only
for building TuxOnIce as modules. There's very very little vanilla code
touched by TuxOnIce. And the vanilla 2.6.31 patch applies without any
rejects to Karmic.

>>  - It requires its own scripts (IIRC) (again divergence from upstream and
>>  hence new set of bugs and maintenance issues)
> There's a CONFIG_ parameter that makes TOI run in swsusp emulation mode, which
> makes TOI a drop-in replacement for swsusp which we currently use, if I'm not
> mistaken.

Yes. It doesn't require its own scripts. It's just that pm-utils has
inadequate support for TuxOnIce.

>> No doubt that TOI has some very nice features that our users would like.
>> Hopefully Nigel and Rafael can now extract those features from TOI and
>> integrated them into the upstream kernel.
> I'm fine either way, it's just that it'd be really awesome to see
> hibernate/resume work *fast* as it does with TOI, unlike our current methods.

It would be, but that's not going to happen overnight, unfortunately.

Regards,

Nigel