Hi, Parag Warudkar <[email protected]> [20060319 09:06:25 -0500]: > > The codebase already seems identical to ondemand - Are your patches in > 2.6.16-rc6 or -mm? If they are - let me know which. If you posted them but > they haven't yet made it into either -mm or mainline can you please post > links to all your patches please? I can test them. > Well I submitted them back on 2006-02-24: http://marc.theaimsgroup.com/?l=linux-kernel&m=114079151404567&w=2 http://marc.theaimsgroup.com/?l=linux-kernel&m=114079151425558&w=2 http://marc.theaimsgroup.com/?l=linux-kernel&m=114079151417220&w=2 It has been pointed out I did forget to CC the cpufreq folk, so thats completely my fault; I thought I had. :-/ > Why do we even have conservative and ondemand as two separate modules given > they share huge amount of code - perhaps make conservative an optional > behaviour of ondemand or alteast make a common lib which both use? > Originally the 'conservative' feature was just a sysfs flag that could be set however it was rejected for a number of reasons; one of them quite rightly, we have a modular system that can take stacks of cpufreq governors so lets use that :) Also, more importantly, bugs in my bits do not affect the original author. It makes more sense to keep things this way as the internal code then does not need a bunch of if{}'s scattered around and also I have a number of extra sysfs files to tweak which would either do nothing in 'ondemand' mode or have to be magically created and destroyed. Either way, its probably neater this way and its *my* duty to make sure anything changing for ondemand is considered for conservative. If you look at a lot of the userland tools that have come out, it would be a pain to have them consider the exception class of handling the combined ondemand/conservative. Breaking out the thing into a library probably would be awkward as all the similar code is actually inline in functions, putting that in a seperate file would be pointless. Hopefully when you apply my patches and then do a diff between the ondemand and conservative governors you will see what I mean. Cheers Alex -- ________________________________________ / Fortune finishes the great quotations, \ | #3 | | | | Birds of a feather flock to a newly | \ washed car. / ---------------------------------------- \ ^__^ \ (oo)\_______ (__)\ )\/\ ||----w | || ||
Attachment:
signature.asc
Description: Digital signature
- References:
- OOPS: 2.6.16-rc6 cpufreq_conservative
- From: Parag Warudkar <[email protected]>
- Re: OOPS: 2.6.16-rc6 cpufreq_conservative
- From: Parag Warudkar <[email protected]>
- Re: OOPS: 2.6.16-rc6 cpufreq_conservative
- From: Alexander Clouter <[email protected]>
- Re: OOPS: 2.6.16-rc6 cpufreq_conservative
- From: Parag Warudkar <[email protected]>
- OOPS: 2.6.16-rc6 cpufreq_conservative
- Prev by Date: [PATCH] v9fs: update license boilerplate
- Next by Date: Re: [PATCH 001/001] PCI: PCI quirk for Asus A8V and A8V Deluxe motherboards
- Previous by thread: Re: OOPS: 2.6.16-rc6 cpufreq_conservative
- Next by thread: [PATCH] fix memory leak in mm/slab.c::alloc_kmemlist() (try #2)
- Index(es):