On 08/15/10 13:07, Isaac Dupree wrote:

On 08/14/10 17:46, Tobias Powalowski wrote:

...

Latest kernel is in testing, please signoff for both arches.

Signoff x86_64 , it seems to be working fine for me

Also, I think the ath9k regressions (randomly losing the wireless connection) that .33 and .34 kernels had for me are resolved in .35! Yay! -Isaac

On 08/15/2010 12:07 PM, Isaac Dupree wrote:

On 08/14/10 17:46, Tobias Powalowski wrote:

...

Latest kernel is in testing, please signoff for both arches.

Signoff x86_64 , it seems to be working fine for me

(I had the impression that it might be heating up my laptop more than normal, but then I was doing backups last night, and I'm feeling a part of my laptop that I'm not sure if it's normally warm anyway. Is there a way to measure this? The estimated battery life from GNOME looks pretty normal, if that's an indication of heating by way of power-usage.)

-Isaac

Isaac, I do think you have a valid point on the heat issue. The fans on my laptop are notably louder with 2.6.35. Not by a bunch, but from quiet to a low "ssshhhhsshhhsshhh" that you cannot mistake. But given it runs great, it is a big step forward from my 2.6.34 woes... What type of video card do you have? I have a radeon rs690m that linux has never quite figured out. I'm curious if the new way the module handling is done doesn't mess up what little advances in idle downclocking that have been made. I know for a fact that in the 2.6.33 timeframe, the video-ati driver and the way the kernel handled it was spot-on. No fan noise and near fglrx performance. Now I know glxgears is not a performance measure, but it is a valid comparison between software configs on the same box. With fglrx, I get ~1100 FPS with the default glxgears window and dead silent fans. With video-ati, I get ~685 FPS which is *great* for the video-ati driver but with 2.6.35, I also get the fan noise. ( a year ago, good with video-ati or video-radeonhd was 300 FPS and the laptop was so hot it was nearly unusable without the fglrx driver). So performance is good and usable with video ati, but the fan noise with 2.6.35 is annoying and I can feel the excess heat in the palm-rest. It isn't near as bad as the hand-frying temps from a year ago, but it is something that was much better in 2.6.33 and now the temps/fan noise has crept up again. The kicker is temp issues are hard to chase down. Something is putting more demand on the processor in the 2.6.35 setup, but what?? This is a desktop independent issue because I can use gnome, kde3, fluxbox, etc.. and the fan noise and temp is consistent across the desktp. It could be anywhere in the system from a regression in the video-ati driver to a runaway loop in the kernel itself. I have no idea where to even began isolating this one other than booting to maintenance mode and adding modules one-at-a-time (which I would probably have to get much smarter on) If you can identify anything on your box that seems to influence the idle temps or fan noise, please document what you see and I'll collect the information and add it to what I have to hopefully narrow down where the additional heat is coming from. -- David C. Rankin, J.D.,P.E. Rankin Law Firm, PLLC 510 Ochiltree Street Nacogdoches, Texas 75961 Telephone: (936) 715-9333 Facsimile: (936) 715-9339 www.rankinlawfirm.com

I also noticed that kernel 2.6.35 wake the fan in my notebook more often. watching sensors output, the fan wakes when temp reached 52°C, then cooled down to 45°C in 30 secs, then fan stops, and the temp soon go up to 50+ in another 30 secs. On Mon, Aug 16, 2010 at 7:46 AM, Isaac Dupree < ml@isaac.cedarswampstudios.org> wrote:

On 08/15/10 17:45, David C. Rankin wrote:

...

What type of video card do you have?

Intel 945 class -- it's one of the first to get KMS and the support is excellent by now. I never ever get slow graphics (although I don't play games much... IIRC extremetuxracer works fine...)

If you can identify anything on your box that seems to influence the

...

idle temps or fan noise

Actually, my fan mostly didn't speed up, instead the temp (if this is real) got hotter, so I manually adjusted it up a bit (via some macbook-fan-specific mechanism), but maybe it's the same thing (macbook linux are long known to run the fan a little slower than might be nice, by default, anyway).

Check your battery life. My battery life isn't noticeably lower than before, which makes me doubt whether more energy is actually being used (and correspondingly turned into heat).

-- Arch Linuxer, Pythoner, Geek --> Blog: http://apt-blog.net

On 08/17/2010 02:50 AM, David C. Rankin wrote:

Good point on power consumption. I'll see if I can find powertop in AUR. The instrumentation on my laptop sucks - few sensors, but my best sensors are my hands. The CPU and GPU are located in the left side of my laptop and you can feel the temp changes through the left palmrest. (best sensor of all :) FYI, Powertop is in community repository.

Cheers, Smith

On 09:59 Tue 17 Aug , Dennis Beekman wrote:

signoff here, works fine on both my desktop and my laptop (both i386)

On Sat, Aug 14, 2010 at 11:46 PM, Tobias Powalowski wrote:

...

Latest kernel is in testing, please signoff for both arches.

greetings tpowa -- Tobias Powalowski Archlinux Developer & Package Maintainer (tpowa) http://www.archlinux.org tpowa@archlinux.org

Some problems with ATI\KMS. I have HD4570, latest opensource drivers, and there are a couple of problems(first of all - radeon.dynpm stop working for me,and with any radeon.* options kms don't start). Second - if kms even start, X server freezes system. Modifications in xorg.conf(even moving it to xorg.conf.bak) do nothing. -- With best regards, Waylls

On 08/18/2010 01:24 PM, David C. Rankin wrote:

Call Trace: [<ffffffffa01c3ba5>] usb_kill_urb+0x85/0xc0 [usbcore] [<ffffffff810717c0>] ? autoremove_wake_function+0x0/0x40 [<ffffffffa01f6831>] usbhid_init_reports+0xb1/0x120 [usbhid]

Tobias, I'm sure you know, but 'usbhid' is the network-ups-tools driver. -- David C. Rankin, J.D.,P.E. Rankin Law Firm, PLLC 510 Ochiltree Street Nacogdoches, Texas 75961 Telephone: (936) 715-9333 Facsimile: (936) 715-9339 www.rankinlawfirm.com

Tobias Powalowski (2010-08-18 21:03):

Am Mittwoch 18 August 2010 schrieb David C. Rankin:

...

On 08/18/2010 01:24 PM, David C. Rankin wrote:

...

Call Trace: [<ffffffffa01c3ba5>] usb_kill_urb+0x85/0xc0 [usbcore] [<ffffffff810717c0>] ? autoremove_wake_function+0x0/0x40 [<ffffffffa01f6831>] usbhid_init_reports+0xb1/0x120 [usbhid]

Tobias,

I'm sure you know, but 'usbhid' is the network-ups-tools driver.

usbhid is for usb input devices, like keyboard mouse and such.

I know I've seen it mentioned that usbhid is used for UPS as well. Here, from drivers/hid/usbhid/Kconfig: config USB_HID tristate "USB Human Interface Device (full HID) support" default y depends on USB && INPUT select HID ---help--- Say Y here if you want full HID support to connect USB keyboards, mice, joysticks, graphic tablets, or any other HID based devices to your computer via USB, as well as Uninterruptible Power Supply (UPS) and monitor control devices. -- -- Rogutės Sparnuotos

On 08/21/2010 01:33 PM, Lukáš Jirkovský wrote:

...

noob question: what does this signoff means?

Basically it means that it works OK for you and you are willing to "sign off the declaration the package is OK and can be moved to [core]." Usually it's not used for packages from [extra] or [community].

Generally packages only need signoffs from developers on arch-dev mailing list. Sometimes (like with kernel26) public signoffs are wanted. -- Ape <Lauri Niskanen>

On Sat, 2010-08-21 at 18:51 +0530, Madhurya Kakati wrote:

On Sat, Aug 21, 2010 at 4:53 PM, Lauri Niskanen wrote:

...

On 08/21/2010 01:33 PM, Lukáš Jirkovský wrote:

...

...

noob question: what does this signoff means?

Basically it means that it works OK for you and you are willing to "sign off the declaration the package is OK and can be moved to [core]." Usually it's not used for packages from [extra] or [community].

Generally packages only need signoffs from developers on arch-dev mailing list. Sometimes (like with kernel26) public signoffs are wanted.

-- Ape <Lauri Niskanen>

Thanks! :D

I wasn't actually aware that public signoffs are wanted for kernel26. As far as I can remember some packages specifically request user signoffs due to lack of usage among devs (and those eventually get dropped from core in my observation). Any simple list as to which packages users should help signoff?