Looks like Suspend2 (http://www.suspend2.net/) is headed for the mm and then mainline kernel (finally!). It's been working rather good for a while now and it would be great if FC5 kernels had it once the release hits the mirrors. I'm guessing most notebook (even desktop) users would welcome it.
There are already excellent examples of those patches inside FC3/FC4/FC5 kernels as well as related stuff like mkinitrd, sysinit etc. here:
Would Fedora kernel maintainers be willing to roll Suspend2 into the development kernels?
Bojan Smojver wrote:
Looks like Suspend2 (http://www.suspend2.net/) is headed for the mm and then mainline kernel (finally!). It's been working rather good for a while now and it would be great if FC5 kernels had it once the release hits the mirrors. I'm guessing most notebook (even desktop) users would welcome it.
There are already excellent examples of those patches inside FC3/FC4/FC5 kernels as well as related stuff like mkinitrd, sysinit etc. here:
Would Fedora kernel maintainers be willing to roll Suspend2 into the development kernels?
I am pretty sure when it hits mainline kernel, it will be included in the development tree. If there are additional enhancements that can done in user space or in integration efforts, feel free to file them as enhancements when it reaches rawhide
regards Rahul
Quoting Rahul Sundaram sundaram@redhat.com:
I am pretty sure when it hits mainline kernel, it will be included in the development tree.
I'm sure that'll be the case. However, I wanted to do a bit of preemptive testing and bug fixing here, given the popularity of FC. I'm guessing this functionality is compiled out in current FC kernels due to bugs in mainline suspend code, so suspend2 could fill that hole. I hope I'm not the only person running FC on a notebook...
If you guys get this code into FC development kernels, more people will test it, more bugs will be reported and everyone will end up happier (I'm an eternal optimist, what can I do :-). Of course, if it turns out that this thing is still more trouble than it's worth before the release of FC5, it can simply be dropped.
Is my convincing working or should I go and cry in the corner for a while? ;-)
If there are additional enhancements that can done in user space or in integration efforts, feel free to file them as enhancements when it reaches rawhide
OK.
Bojan Smojver wrote:
Quoting Rahul Sundaram sundaram@redhat.com:
I am pretty sure when it hits mainline kernel, it will be included in the development tree.
I'm sure that'll be the case. However, I wanted to do a bit of preemptive testing and bug fixing here, given the popularity of FC. I'm guessing this functionality is compiled out in current FC kernels due to bugs in mainline suspend code, so suspend2 could fill that hole. I hope I'm not the only person running FC on a notebook...
Note that some work has gone into enabling the current mainline suspend code to work properly in the recent rawhide kernels. https://www.redhat.com/archives/fedora-devel-list/2005-August/msg00143.html So you can test it out on your notebook and see if you can be happy with that.
The -mm kernel as I am sure you are already aware is a soft development tree for the current 2.6.x kernels. Patches that ends in there has a reasonable oppurtunity of hitting mainline soon but also has a chance of being pulled out if it turns out to be sour . So I would wait till it reaches mainline and then rawhide before jumping up in excitement to test it
regards Rahul
Quoting Rahul Sundaram sundaram@redhat.com:
Note that some work has gone into enabling the current mainline suspend code to work properly in the recent rawhide kernels. https://www.redhat.com/archives/fedora-devel-list/2005-August/msg00143.html So you can test it out on your notebook and see if you can be happy with that.
OK. I tried the original suspend code before and it wasn't very good (slow, panics, disk corruption etc.), but I'll give this one a go. Who knows, maybe it'll be better this time.
I read (some of) the thread you pointed to and I see that Dave Jones actually had another (very reasonable) explanation for not using suspend2 in current Rawhide kernels, which has to do with deviation from the vanilla kernels. I understand that, makes perfect sense.
Oh well, I guess I'll have to roll my own for a while. Then again, stranger things were known happen, so maybe suspend2 code filters through mm and mainline in FC5 time (ah, that optimist side again :-).
Quoting Bojan Smojver bojan@rexursive.com:
OK. I tried the original suspend code before and it wasn't very good (slow, panics, disk corruption etc.), but I'll give this one a go. Who knows, maybe it'll be better this time.
Nah, no luck. Both 1552 and 1553 (which is 2.6.14-rc1, as I understand) can't bring the machine back properly (i.e. the whole thing hangs eventually). Suspend2 is also way faster, not to mention more pretty ;-)
Back to patching...
On Wed, 2005-08-24 at 09:59 +1000, Bojan Smojver wrote:
Oh well, I guess I'll have to roll my own for a while.
For instance, 2.6.13-1.1535_FC5 gets patched and suspends/resumes perfectly with suspend2-rc6 on my notebook. Interestingly enough, when the patching is attempted on 1553 (which is 2.6.14-rc1), there is a whole heap of messages about already applied patches. I guess a lot of that stuff from suspend2 already made it into mainline, so the actual suspend2 patch, when released against the 2.6.14, should be much smaller. Ah, the spoils... ;-)
-
Bojan Smojver -
Rahul Sundaram