Skip to contentSkip to conversation

Enhanced PowerSaver driver

DISCUSSION Rafał Bilski 3 participants 6 messages Jan 28, 2007 – Jan 31, 2007

lore
Share & tools
lore
#0 lore raw

Hello,

Looks like VIA have add support for C7 processors to Longhaul patch available from www.viaarena.com. It is done using hardcoded tables with P-states. I wasn't aware that driver for C7 already exists when I posted my patch. Please ignore it.

Regards
Rafa³

----------------------------------------------------------------------
Wpadka w kosciele - zobacz >> http://link.interia.pl/f19ea

#1 lore raw
Quoted 6 lines · RafaÅ‚ Bilski

Hello,

Looks like VIA have add support for C7 processors to Longhaul patch available from www.viaarena.com. It is done using hardcoded tables with P-states. I wasn't aware that driver for C7 already exists when I posted my patch. Please ignore it.

I wish they'd send patches upstream instead of creating their own world to live in. It's not just cpufreq patches either, I've just had to sift through a 6.7MB AGP patch they released just to fish out some additional PCI IDs.

Annoying.

Are you going to sift through their C7 driver and send diffs, or shall I add to my todo ?

thanks,

Dave

Signature · 1 line
http://www.codemonkey.org.uk
#2 lore raw
Quoted 13 lines · Rafa³ Bilski
Quoted 6 lines

Hello,

Looks like VIA have add support for C7 processors to Longhaul patch available from www.viaarena.com. It is done using hardcoded tables with P-states. I wasn't aware that driver for C7 already exists when I posted my patch. Please ignore it.

I wish they'd send patches upstream instead of creating their own world to live in. It's not just cpufreq patches either, I've just had to sift through a 6.7MB AGP patch they released just to fish out some additional PCI IDs.

Quoted 8 lines

Annoying.

Are you going to sift through their C7 driver and send diffs, or shall I add to my todo ?

thanks,

Dave

Well, this driver is based on 2.6.16 Longhaul. I have a couple of small patches for C3 which I would like to send first. I would like to help, but I don't have access to any motherboard with C7 CPU. I can't test it. I just think that there is no need to add my version to kernel when there is official VIA driver available.

Regards
Rafa³

----------------------------------------------------------------------
Zalety czapki niewidki. Zobacz >> http://link.interia.pl/f19e3

#3 lore raw
Quoted 13 lines
Quoted 6 lines

Are you going to sift through their C7 driver and send diffs, or shall I add to my todo ?

thanks,

Dave

Well, this driver is based on 2.6.16 Longhaul. I have a couple of small patches for C3 which I would like to send first. I would like to help, but I don't have access to any motherboard with C7 CPU. I can't test it. I just think that there is no need to add my version to kernel when there is official VIA driver available.

Quoted 6 lines

Regards
Rafał

----------------------------------------------------------------------
Zalety czapki niewidki. Zobacz >> http://link.interia.pl/f19e3

I have a board (IB798) with a Via Eden CPU (a C7 deviate). On kernel 2.6.16.38 I tried to compile and load Via's longhaul driver. When I started cpufreq, the O/S crashed.

If e.g. Rafał create som patches for the original longhaul driver, from ideas based on Via's code, I will install them, test and give feedback.

Brgds Jorgen

#4 lore raw

[...]

Quoted 5 lines

I have a board (IB798) with a Via Eden CPU (a C7 deviate). On kernel 2.6.16.38 I tried to compile and load Via's longhaul driver. When I started cpufreq, the O/S crashed.

This is rather bad sign. Looks like I spoke to soon.

Quoted 3 lines

If e.g. Rafa³ create som patches for the original longhaul driver, from ideas based on Via's code, I will install them, test and give feedback.

Thanks.

After looking into code I have found two things:
- it is using "disable bus master bit on each PCI device"... for C7?!,
- it is using current CPU f and vcc to find table for it. I don't like this idea. If CPU has min/max info why not use it? Strange thing in C7 is that it is starting at lowest supported frequency. BIOS has to switch it to max. As You can read at viaarena, BIOS is only doing it when "optimized defaults" are loaded. Looks like it isn't default. What will happen if driver will search table for CPU running at lowest frequency? Probably processors with different max can have same min frequency.

Dave: Can I repost e_powersaver with "experimental" and some small changes? Sorry.

Quoted 1 line

Brgds Jorgen

Regards
Rafa³

----------------------------------------------------------------------
Fergie i jej biodra.... - zobacz >> http://link.interia.pl/f19ed

#5 lore raw
Quoted 22 lines · Rafał Bilski

[...]

Quoted 5 lines

I have a board (IB798) with a Via Eden CPU (a C7 deviate). On kernel 2.6.16.38 I tried to compile and load Via's longhaul driver. When I started cpufreq, the O/S crashed.

This is rather bad sign. Looks like I spoke to soon.

Quoted 3 lines

If e.g. Rafał create som patches for the original longhaul driver, from ideas based on Via's code, I will install them, test and give feedback.

Thanks.

After looking into code I have found two things:
- it is using "disable bus master bit on each PCI device"... for C7?!,
- it is using current CPU f and vcc to find table for it. I don't like this idea. If CPU has min/max info why not use it? Strange thing in C7 is that it is starting at lowest supported frequency. BIOS has to switch it to max. As You can read at viaarena, BIOS is only doing it when "optimized defaults" are loaded. Looks like it isn't default. What will happen if driver will search table for CPU running at lowest frequency? Probably processors with different max can have same min frequency.

Quoted 2 lines

Dave: Can I repost e_powersaver with "experimental" and some small changes? Sorry.

Sure, feel free.
I've got 'look at the via driver' on my TODO, but I've got a ton of stuff to dig through after having been away for the last two weeks, so it may take me a while to get around to it. If you beat me to it, all the better :-)

Dave

Signature · 1 line
http://www.codemonkey.org.uk