reassign 540613 network-manager retitle 540613 NM doesn't cope with new ssb bus type in hal 0.5.13 thanks Patrick Matthäi wrote: > Michael Biebl schrieb: >> Patrick Matthäi wrote: >>> Michael Biebl schrieb: >>>> Michael Biebl wrote: >>>>> Luís Guilherme Fernandes Pereira wrote: >>>>>> Package: hal >>>>>> Version: 0.5.13-2 >>>>>> Severity: normal >>>>>> >>>>>> >>>>>> I was submitting a bug to NetworkManager but I investigated further >>>>>> and found it to be a hal related. As said here, downgrading to 0.5.12 >>>>>> solves the issue. I am attaching the (incomplete) bug I was sending to >>>>>> NetworkManager. Maybe the logs I have put there can be of some >>>>>> assistance >>>>>> >>>>> Could you please send me a lshal dump. >>>> For both 0.5.13 and 0.5.12. >>> Sorry, a bit late, but here it is. >>> >>> With the new hal version, there are 3 devices missing. >>> >> Many thanks so far for your great help. >
So, it seems to be a NM problem after all. HAL added a new bus type (ssb) in [1], which NM doesn't cope with (yet) NM upstream has to say the following: <mbiebl> got more info on the b43 prob <dcbw> mbiebl: ok <dcbw> mbiebl: did I ever get an lshal for ssb ? <mbiebl> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=540613 <mbiebl> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=540613#48 <mbiebl> One difference is, that with old hal, rfkill-input is loaded <dcbw> mbiebl: if it doesn't show up in 'iwconfig' or the module doesn't get autoloaded, that's more of a kernel/driver problem <dcbw> mbiebl: that's odd <dcbw> mbiebl: I wonder what does that... <dcbw> mbiebl: fun <dcbw> I think the HAL ssb patch is, well, flawed <dcbw> first off, the 'ssb' bus is *always* owned by a specific hardware device <dcbw> it's on a PCI card, and it includes a PCI <-> SSB bridge as part of the chip <dcbw> thus, the ssb bus is a child of that PCI card (or USB) <dcbw> I read the lshal wrong, forget that <mbiebl> It's interesting to compare the diff between the old and new hal <mbiebl> dcbw: why does NM not find a device driver, btw <dcbw> mbiebl: because ssb is a new bus, and since the 'driver' key isn't on the device itself, but a parent the device, we have to walk around the device list and find it, and that's different for every bus type <mbiebl> /org/freedesktop/Hal/devices/pci_14e4_4312 has info.linux.driver = 'b43-pci-bridge' <dcbw> mbiebl: right, but info.bus == ssb for wlan0 <mbiebl> dcbw: so it is a bug/missing feature in NM then? <dagger> similar to http://bugs.gentoo.org/show_bug.cgi?id=279104 ? <dcbw> mbiebl: yes, it's a bug in NM, but it's caused by HAL changing behavior pretty dramatically <dcbw> mbiebl: HAL added the 'ssb' bus, and thus the device list no longer has the same relationships between devices that we had expected <dcbw> ewoerner: how much time would it take? <ewoerner> dcbw: i just hat 5 seconds (admittedly, i had some serious swapping problems at that time) <dagger> mbiebl: dcbw: I had a similar bug report with suggested patch, but I didn't have time to look at it yet. http://bugs.gentoo.org/attachment.cgi?id=199167&action=edit <mbiebl> dcbw: interesting that you haven't seen this issue show up in fedora <dcbw> mbiebl: we're in the process of getting rid of HAL, not udpating it :) <mbiebl> which hal/nm version do you ship there? <dagger> 0.7.1 (with some extra patches backported from 0.7 branch) + hal 0.5.13 <dcbw> mbiebl: F12 will have 0.5.13 <dcbw> mbiebl: F11 (current release) ships 0.5.12 <dcbw> mbiebl: but in F12, NetworkManager 0.8 doesn't use HAL anymore <dcbw> dagger: patch doesn't apply to current 0.7 branch of course, but I think I've got something that will work there Cheers, Michael [1]http://cgit.freedesktop.org/hal/commit/?id=36d076bf26350e022731cb7c2c9f4e840dd25e37 -- Why is it that all of the instruments seeking intelligent life in the universe are pointed away from Earth?
signature.asc
Description: OpenPGP digital signature
