Bug 139089
| Summary: | kudzu doesn't detect/configure madwifi/ath0 devices correctly | ||
|---|---|---|---|
| Product: | [Fedora] Fedora | Reporter: | Satish Balay <balay> |
| Component: | kudzu | Assignee: | Bill Nottingham <notting> |
| Status: | CLOSED NOTABUG | QA Contact: | |
| Severity: | medium | Docs Contact: | |
| Priority: | medium | ||
| Version: | 3 | CC: | rvokal |
| Target Milestone: | --- | Keywords: | FutureFeature |
| Target Release: | --- | ||
| Hardware: | i386 | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Fixed In Version: | Doc Type: | Enhancement | |
| Doc Text: | Story Points: | --- | |
| Clone Of: | Environment: | ||
| Last Closed: | 2004-11-12 21:08:22 UTC | Type: | --- |
| Regression: | --- | Mount Type: | --- |
| Documentation: | --- | CRM: | |
| Verified Versions: | Category: | --- | |
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
| Cloudforms Team: | --- | Target Upstream Version: | |
| Embargoed: | |||
|
Description
Satish Balay
2004-11-12 20:56:33 UTC
The madwifi driver needs to support ethtool GDRVINFO; that should clear most of this up. Looking at the madwifi source - it appears to support 'GDRVINFO'. What
is the exact filed that should be set to 'ath' for kudzu to work?
I'm guessing its probably a 1 line change to the driver.
Thanks,
--
ath_ioctl_ethtool(struct ath_softc *sc, int cmd, void *addr)
{
struct ethtool_drvinfo info;
if (cmd != ETHTOOL_GDRVINFO)
return -EOPNOTSUPP;
memset(&info, 0, sizeof(info));
info.cmd = cmd;
strncpy(info.driver, dev_info, sizeof(info.driver)-1);
strncpy(info.version, version, sizeof(info.version)-1);
return copy_to_user(addr, &info, sizeof(info)) ? -EFAULT : 0;
}
It needs to set the pci path:
strcpy(info.bus_info, pci_name(whatever its private pci struct
is));
The problem with kudzu seems to be not only the eth1 config files. Kudzu must do something else, which breaks the functionality of madwifi. Unfortunately I don'nt know enough details about how kudzu works. The Workaround, which works for me, is: - Install madwifi as described in it's readme ( make; make install) - check, if the line "alias ath0 ath_pci" is present in /etc/modprobe.conf - reboot - if kudzu comes up to tell you it found new atheros hardware, tell kudzu to *ignore* it completely, i.e. not to change any configuration and not to care about it in the future - open neat and create a new interface. With the current version (fc3) ath0 is listed in the hardware section, so you can easily do that and don't need to edit the files in /etc/sysconfig/network-scripts etc. manually If madwifi doesn't work, try the following: - undelete the files in /lib/modules/<version>/net/* - run depmod -ae - eventually reboot - reinstall madwifi using the procedure above. If you follow that procedure, an entry ath0 is generated in /etc/sysconfig/hwconfig. Eventually you have to delete a formerly generated entry for eth1. Bill, the following addition appears to work. strcpy(info.bus_info, pci_name((struct pci_dev *)sc->sc_bdev)); However kudzu is unable to configure 'ath0' (does it do any wireless devices?) - but 'neat' does configures it fine. thanks, Satish How exactly does it *not* configure it, on first detection - does it pop up that it's new, etc? 'not configure' refers to 'wireless settings' kudzu does pop-up a menu with configure option. Choosing configure lets me set dhcp/static ip adress settings (which makes sense for wired network cards). This creates /etc/sysconfig/network-scripts/ifcfg-ath0 with: [root@asterix etc]# cat /etc/sysconfig/network-scripts/ifcfg-ath0 # Atheros Communications, Inc. AR5211 802.11ab NIC DEVICE=ath0 ONBOOT=yes BOOTPROTO=dhcp HWADDR=xx:xx:xx:xx:xx:xx The wireless require at-least TYPE=Wireless/ESSID/KEY options - which kudzu currently doesn't set. And I'm not sure if kudzu is supporsed to do anything more (code/functionality overlap with system-config-network) So, for wireless devices - perhaps kudzu shouldn't even pop up this 'configure' option - since it can't configure it anyway? Correct; the tool kudzu calls doesn't do wireless options. |