Month: August 2007

Categories
everything else

New wakeups from usb_hcd_poll_rh_status in 2.6.23-rc4

It seems other people are experiencing the same thing. Unfortunately, that guy didn’t get a response. Both of these readings were taken with no Xorg or Wifi running. The laptop is a Pentium M Centrino model.
kernel 2.6.22-hrt6

PowerTOP 1.8 (C) 2007 Intel Corporation
Collecting data for 15 seconds
Cn Avg residency
C0 (cpu running) ( 0.1%)
C1 0.0ms ( 0.0%)
C2 60.5ms ( 4.9%)
C3 0.0ms ( 0.0%)
C4 645.0ms (95.0%)
P-states (frequencies)
1.87 Ghz 0.0%
1.60 Ghz 0.0%
1333 Mhz 0.0%
800 Mhz 100.0%
Wakeups-from-idle per second : 2.3 interval: 15.0s
no ACPI power usage estimate available
Top causes for wakeups:
40.4% ( 1.5) laptop_mode : queue_delayed_work_on (delayed_work_timer_fn)
12.3% ( 0.5) <interrupt> : ide0
12.3% ( 0.5) <kernel> : queue_delayed_work_on (delayed_work_timer_fn)
7.0% ( 0.3) <kernel> : neigh_table_init_no_netlink (neigh_periodic_timer)
5.3% ( 0.2) ifplugd : schedule_timeout (process_timeout)
5.3% ( 0.2) <kernel> : page_writeback_init (wb_timer_fn)
5.3% ( 0.2) init : schedule_timeout (process_timeout)
3.5% ( 0.1) cupsd : schedule_timeout (process_timeout)
1.8% ( 0.1) <kernel> : end_that_request_last (laptop_timer_fn)
1.8% ( 0.1) <kernel> : init_nonfatal_mce_checker (delayed_work_timer_fn)
1.8% ( 0.1) cifsdnotifyd : schedule_timeout (process_timeout)
1.8% ( 0.1) <kernel> : seqgen_init (delayed_work_timer_fn)
1.8% ( 0.1) <kernel> : init_sunrpc (delayed_work_timer_fn)

Kernel 2.6.23-rc4-hrt1

PowerTOP 1.8 (C) 2007 Intel Corporation
Collecting data for 15 seconds
Cn Avg residency
C0 (cpu running) ( 0.1%)
C1 0.0ms ( 0.0%)
C2 35.7ms ( 2.4%)
C3 0.0ms ( 0.0%)
C4 169.9ms (97.5%)
P-states (frequencies)
1.87 Ghz 0.0%
1.60 Ghz 0.0%
1333 Mhz 0.0%
800 Mhz 100.0%
Wakeups-from-idle per second : 6.4 interval: 15.0s
no ACPI power usage estimate available
Top causes for wakeups:
89.4% ( 16.3) <kernel> : usb_hcd_poll_rh_status (rh_timer_func)
2.6% ( 0.5) <interrupt> : ide0
1.5% ( 0.3) <kernel> : neigh_table_init_no_netlink (neigh_periodic_timer)
1.1% ( 0.2) init : schedule_timeout (process_timeout)
1.1% ( 0.2) ifplugd : schedule_timeout (process_timeout)
1.1% ( 0.2) <kernel> : page_writeback_init (wb_timer_fn)
0.7% ( 0.1) <interrupt> : acpi
0.7% ( 0.1) cupsd : schedule_timeout (process_timeout)
0.4% ( 0.1) <kernel> : end_that_request_last (laptop_timer_fn)
0.4% ( 0.1) hald : schedule_timeout (process_timeout)
0.4% ( 0.1) <kernel> : ip_rt_init (rt_check_expire)
0.4% ( 0.1) cifsdnotifyd : schedule_timeout (process_timeout)
0.4% ( 0.1) <kernel> : init_nonfatal_mce_checker (delayed_work_timer_fn)

Categories
Linux

hrtimers is DA BOMB!

Last night, I patched a vanilla 2.6.22 kernel with the hrtimers code and I’m still finding all the magic things it can do. Two of the biggest things are the cpuidle governor and hpet support. cpuidle “menu” governor allows the cpu to drop into the lowest C-state possible without passing through any other states. The difference is dramatic, without hrtimers ipw2200 would not allow the cpu lower than C2, now the cpu spends 80% of it’s time in C4. HPET replaces the old timer code that woke up a minimum of 30 times a second. With HPET running, this floor is removed and the computer can idle at two wakeups per second if there are no heavy services running!

PowerTOP 1.8 has been released. Feaures like deferrable timers depend on Linux 2.6.23 but a new dump mode will be useful to capture data.

Categories
gentoo

gnome-games, Oh My!

Gentoo just stabilized Gnome 2.18, everybody should avoid the Expat landmine by running revdep-rebuild after the upgrade to avoid breakage from occuring.

However, the Gnome upgrade is only tangently related to my problem. Of course, after I –sync’ed there was ~70 packages waiting to be installed. Digging through the install list I found a lot of really large/odd dependencies for nautilus, totem, and more deps related to these two. Considering, I switched to Xfce these deps were unacceptable. Well, emerge -puDNt wasn’t very clear so I did it the hard way and put totem in package.mask to see what complaint it would have. Turns out, Totem is a dep of gnome-python-desktop, which is a dep of… gnome-games. This is like playing six degrees of Kevin Bacon. If I try to emerge gnome-games now it pulls in the whole Gnome desktop.

ebrake@laptop ~ $ emerge -p gnome-games
These are the packages that would be merged, in order:

Calculating dependencies… done!
[ebuild N ] media-plugins/gst-plugins-ogg-0.10.11
[ebuild N ] media-plugins/gst-plugins-gconf-0.10.3 USE=”-debug”
[ebuild N ] gnome-base/gnome-menus-2.18.2 USE=”python -debug”
[ebuild N ] gnome-base/gail-1.18.0 USE=”-debug -doc”
[ebuild N ] x11-libs/gtkglext-1.2.0 USE=”-debug -doc”
[ebuild N ] media-plugins/gst-plugins-pango-0.10.11
[ebuild N ] media-libs/gst-plugins-ugly-0.10.3 USE=”-debug”
[ebuild N ] media-plugins/gst-plugins-gnomevfs-0.10.11
[ebuild N ] media-plugins/gst-plugins-dvdread-0.10.4
[ebuild N ] media-plugins/gst-plugins-theora-0.10.11
[ebuild N ] media-libs/a52dec-0.7.4-r5 USE=”-djbfft -oss”
[ebuild N ] media-libs/libmpeg2-0.4.1 USE=”X sdl”
[ebuild N ] gnome-base/libgtop-2.14.9 USE=”X gdbm -debug”
[ebuild N ] dev-python/pyorbit-2.14.3 USE=”-debug”
[ebuild N ] app-admin/eselect-esd-20060719
[ebuild N ] x11-wm/metacity-2.18.5 USE=”-debug -xinerama”
[ebuild N ] x11-libs/gtksourceview-1.8.5 USE=”-debug -doc”
[ebuild N ] media-sound/esound-0.2.38 USE=”alsa tcpd -debug -ipv6″
[ebuild N ] gnome-base/eel-2.18.0.1 USE=”X -debug”
[ebuild N ] dev-python/pygtkglext-1.1.0
[ebuild N ] media-plugins/gst-plugins-a52dec-0.10.3
[ebuild N ] media-plugins/gst-plugins-mpeg2dec-0.10.3
[ebuild N ] dev-python/gnome-python-2.18.2 USE=”-debug -doc”
[ebuild N ] gnome-base/gnome-panel-2.18.3 USE=”-debug -doc -eds”
[ebuild N ] media-video/totem-2.18.2 USE=”dvd flac hal mad nsplugin ogg theora vorbis xv -a52 -debug -ffmpeg -gnome -lirc -mpeg -nvtv -seamonkey -xulrunner”
[ebuild N ] gnome-base/nautilus-2.18.1-r1 USE=”X -beagle -debug -gnome”
[ebuild N ] gnome-extra/nautilus-cd-burner-2.18.2 USE=”cdr dvdr -debug”
[ebuild N ] gnome-extra/gnome-media-2.18.0 USE=”mad ogg vorbis -debug -ipv6″
[ebuild N ] dev-python/gnome-python-desktop-2.18.0 USE=”X -debug -doc”
[ebuild N ] gnome-extra/gnome-games-2.18.2.1 USE=”opengl -artworkextra -debug -guile”

Categories
everything else

omg, my first paypal account!!!1!!1one

Up until a few days ago I did not have my own paypal account (shocking!). What changed my mind was the talk on the Security Now! podcast about PayPal’s new keychain device. Steve Gibson and Leo recently completed a long series on authentication (SN90, SN94, SN95, SN98, SN101, SN103). Which, looking back, the security key seems like the inevitable result of everything they discussed. Steve says as much in SN103 that he wants Verisign to support OpenID and tie this into the key functionality. The PayPal official, Michael Vergara, also said that a common network was being developed so that the same key could be used everywhere without the so called “necklace problem”. I feel like I’ve fallen for a very clever marketing ploy, but I did create the paypal account and order the security key all because of Security Now!. It’s all your fault Steve and Leo!