Revision history [back]

Starting with (probably) 3.2.0.11, connman does something periodically (scanning WLANs?) that causes lipstick to stutter every few seconds, network I/O to stall and constant background CPU usage when WLAN is enabled. This also drains the battery very quickly, to the point where an Xperia X with no running apps and the display off won’t last even one day.

This happens whether or not the device is connected to a network, but the effect is much worse when no known networks are nearby.

Disabling WLAN makes these problems go away.

Starting with (probably) 3.2.0.11, connman does something periodically (scanning WLANs?) that causes lipstick to stutter every few seconds, network I/O to stall and constant background CPU usage when WLAN is enabled. This also drains the battery very quickly, to the point where an Xperia X with no running apps and the display off won’t last even one day.

This happens whether or not the device is connected to a network, but the effect is much worse when no known networks are nearby.

Disabling WLAN makes these problems go away.

Starting with (probably) 3.2.0.11, connman does something periodically (scanning WLANs?) that causes lipstick to stutter every few seconds, network I/O to stall and constant background CPU usage when WLAN is enabled. This also drains the battery very quickly, to the point where an Xperia X with no running apps and the display off won’t last even one day.

This happens whether or not the device is connected to a network, but the effect is much worse when no known networks are nearby.

Disabling WLAN makes these problems go away.

Edit: I just checked dbus-monitor and indeed it looks like connman is scanning wifi networks every 3 seconds or so, and then dumps all the info, which is very slow. This explains the above observations.

Starting with (probably) 3.2.0.11, connman does something periodically (scanning WLANs?) that causes lipstick to stutter every few seconds, network I/O to stall and constant background CPU usage when WLAN is enabled. This also drains the battery very quickly, to the point where an Xperia X with no running apps and the display off won’t last even one day.

This happens whether or not the device is connected to a network, but the effect is much worse when no known networks are nearby.

Disabling WLAN makes these problems go away.

Edit: I just checked dbus-monitor and indeed it looks like connman is scanning wifi networks every 3 seconds or so, and then dumps all the info, which is very slow. This explains the above observations.

Edit 2: Setting breakpoints in connman with gdb points to connman/plugins/sailfish_wifi.c as the source of the problem. The autoscan timer appears to get registered too many times, which causes it to get triggered way too frequently. I have not tried to confirm this though.

Starting with (probably) 3.2.0.11, connman does something periodically (scanning WLANs?) that causes lipstick to stutter every few seconds, network I/O to stall and constant background CPU usage when WLAN is enabled. This also drains the battery very quickly, to the point where an Xperia X with no running apps and the display off won’t last even one day.

This happens whether or not the device is connected to a network, but the effect is much worse when no known networks are nearby.

Disabling WLAN makes these problems go away.

Edit: I just checked dbus-monitor and indeed it looks like connman is scanning wifi networks every 3 seconds or so, and then dumps all the info, which is very slow. This explains the above observations.

Edit 2: Setting breakpoints in connman with gdb points to connman/plugins/sailfish_wifi.c as the source of the problem. The autoscan timer appears to get registered too many times, which causes it to get triggered way too frequently. I have not tried to confirm this though.

Edit 3: Found the offender: offender, at least for the case where there’s an active connection: https://git.sailfishos.org/mer-core/connman/blob/master/connman/plugins/sailfish_wifi.c#L2645-2648

Commenting these lines fixes the problem but I have no idea why they’re there in the first place.

Starting with (probably) 3.2.0.11, connman does something periodically (scanning WLANs?) that causes lipstick to stutter every few seconds, network I/O to stall and constant background CPU usage when WLAN is enabled. This also drains the battery very quickly, to the point where an Xperia X with no running apps and the display off won’t last even one day.

This happens whether or not the device is connected to a network, but the effect is much worse when no known networks are nearby.

Disabling WLAN makes these problems go away.

Edit: I just checked dbus-monitor and indeed it looks like connman is scanning wifi networks every 3 seconds or so, and then dumps all the info, which is very slow. This explains the above observations.

Edit 2: Setting breakpoints in connman with gdb points to connman/plugins/sailfish_wifi.c as the source of the problem. The autoscan timer appears to get registered too many times, which causes it to get triggered way too frequently. I have not tried to confirm this though.

Edit 3: Found the offender, at least for the case where there’s an active connection: https://git.sailfishos.org/mer-core/connman/blob/master/connman/plugins/sailfish_wifi.c#L2645-2648

Commenting these lines fixes the problem but I have no idea why they’re there in the first place.

Starting with (probably) 3.2.0.11, connman Connman does something periodically (scanning WLANs?) that causes lipstick to stutter every few seconds, network I/O to stall and constant background CPU usage when WLAN is enabled. This also drains the battery very quickly, to the point where an Xperia X with no running apps and the display off won’t last even one day.

This happens whether or not the device is connected to a network, but the effect is much worse when no known networks are nearby.

Disabling WLAN makes these problems go away.

Edit: I just checked dbus-monitor and indeed it looks like connman is scanning wifi networks every 3 seconds or so, and then dumps all the info, which is very slow. This explains the above observations.

Edit 2: Setting breakpoints in connman with gdb points to connman/plugins/sailfish_wifi.c as the source of the problem. The autoscan timer appears to get registered too many times, which causes it to get triggered way too frequently. I have not tried to confirm this though.

Edit 3: Found the offender, at least for the case where there’s an active connection: https://git.sailfishos.org/mer-core/connman/blob/master/connman/plugins/sailfish_wifi.c#L2645-2648

Commenting these lines fixes the problem but I have no idea why they’re there in the first place.

Edit 4: Found what’s causing problems when not associated with a network: https://git.sailfishos.org/mer-core/connman/blob/master/connman/plugins/sailfish_wifi.c#L2859-2868

IMO this is completely unnecessary.

tl;dr connman has never been tested in densely populated areas with lots of wifi networks, and scans way too frequently, which drains the battery and makes the system less responsive