Re: Linux Kernel Update N900
 

I don't think the bootloader gives a damn, and any device firmware simply needs to have it provided to the kernel at build time. The firmware doesn't care about the kernel, after all.

Re: Linux Kernel Update N900
 

There seems to be a lot of naysayers in this thread. It is completely possible to port the changes up from the .28 kernel to a .33 kernel. If you read my signature, you can see I've been able to make most of the features of the n810 work with my .33 kernel, and the n810 is not mainlined, and it is not properly supported in linux-omap. If I owned an n900, I would be working on this, but I don't :P which is why I'm working on the n810 kernel.

Still, a .28 kernel isn't too bad - remember Android devices only run .29

Re: Linux Kernel Update N900
 

Ok, then go for it :)

Re: Linux Kernel Update N900
 

That's exactly my point :)

Sure the kernel my not be supported entierly by upstream linux-omap, but that doesn't mean we can't host our own kernel source. And sure newer kernels may not provide all of the drivers - but keep in mind that all of the kernel drivers/modules in maemo5 are open source, so there's nothing really stoping you from applying driver patches from the current kernel to newer kernels. ;)

Re: Linux Kernel Update N900
 

Unfortunately - NO. Change from PR1.0 to PR1.1 can't be reversed due to change in firmware - the old PR1.0 doesn't support the new firmware and Nokia doesn't distribute the old firmware. After you do upgrade from PR1.0 to PR1.1 once you have a new firmware in cell/3G and return to PR1.0 doesn't revert that firmware.

In short words - API between firmware and driver changes.

Re: Linux Kernel Update N900
 

"...so people who only care about throughput (ie, servers)..."

Basically, the "performance degradation" is that you get an interactive user interface, but copying some file might take 30 seconds, instead of let's say 25 seconds.

The thing is, no normal user cares whether some file copies a couple seconds faster or slower. What does matter is the perceived speed of the device. You don't want to be missing your phone calls, because the OS is currently busy with some disk operations and can't be bothered to show the phone UI. And even if that UI eventually is displayed, you can't use it, because all your input has 10 second lag.

For example, take a look at the two browsers of N900. While the Firefox may load and render the web page faster than Microb (when measured with a clock), it's the latter one that feels fast and responsive.

Start copying something to the phone using scp and try to use the phone at the same time. Or start downloading podcasts using gPodder.

Re: Linux Kernel Update N900
 

Well aware of this.

PR1.0 and PR1.1 are much, much more than just the kernel.

Unless the code for that is now a closed source module, there is no reason that a new kernel version is impossible. The incompatibility with the old firmware version lies at a level above the kernel.

Re: Linux Kernel Update N900
 

Sure.

Uhm...m-mm, technically it is not true but I think it is not a case for N900. For new device - may be (I point to DRM possibility).

However, at the end of day I think we can do and run a more newest kernel even with proprietary Nokia modules because the only possible problems here are:

1. module format - solvable, via conversion for exam
2. old kernel APIs which may be excluded from newest kernel (usually - some details) - we can keep it actually or support a conversion middle layer.

Re: Linux Kernel Update N900
 

The wifi kernel infrastructure changed from .29 to .30 introducing a much easier way to implement Infrastructure mode wifi, which was not dependent on certain wifi chipsets that work with hostap. Things like the PSP require Infrastructure rather than ad-hoc and won't work with the n900's wifi tether mode. I don't know how easy/possible it is to backport that in to .28 or try to make a kernel that works on the n900 work with .30 or higher.

Re: Linux Kernel Update N900
 

hmm.. aren't MeeGo using .33? I think i saw somewhere that we would go to .33 later. It where in some blog of a Nokia emploiee. So the work are probably already in the spin.