Most recent kernel where this bug did not occur: do not know, can only test fairly recent kernels (drivers of IDE/SATA controller), but I think I had this problem before on a 2.6.12 kernel, but am not 100% sure. Distribution: Debian Unstable Hardware Environment: lspci -vvv (of network card): 0000:03:00.0 Ethernet controller: Broadcom Corporation BCM4401-B0 100Base-TX (rev 02) Subsystem: Dell Inspiron 6000 laptop Control: I/O- Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR+ FastB2B- Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=fast >TAbort- <TAbort- <MAbort- >SERR- <PERR- Latency: 64 Interrupt: pin A routed to IRQ 9 Region 0: Memory at dfdfe000 (32-bit, non-prefetchable) [size=8K] Capabilities: [40] Power Management version 2 Flags: PMEClk- DSI- D1+ D2+ AuxCurrent=0mA PME(D0+,D1+,D2+,D3hot+,D3cold+) Status: D0 PME-Enable- DSel=0 DScale=2 PME- Software Environment: mii-tool.c 1.9 2000/04/28 00:56:08 (David Hinds) ifconfig 1.42 (2001-04-13) gcc (GCC) 4.0.3 20051111 (prerelease) (Debian 4.0.2-4) Problem Description: Computer/kernel freezes/locks up solid every time when 'mii-tool ethX' is done, while no 'ifconfig ethX up' is done. It works properly when ifconfig is done before mii-tool. After some 'research' with printk's I found out that it happens when a SIOCGMIIPHY is sent to the ioctl. It finally freezes over 'u32 val = br32(bp, reg);' in b44_wait_bit. I presume that the hardware is not brought up properly, as I see Steps to reproduce: - start computer, but do not run 'ifconfig ethX up' (i.e. disable network scripts) - mii-tool ethX === computer freezes ===
Created attachment 6608 [details] complete dmesg output
Created attachment 6609 [details] complete lspci output
Created attachment 6610 [details] output of ver_linux
Created attachment 6641 [details] early return in dev->do_ioctl when the device is not up It's probably a bit late in the 2.6.15-rcX cycle to try to move the init sequence from the device open routine to the PCI probe one. -- Ueimor
The patch solves the freeze, although "SIOCGMIIPHY on 'eth0' failed: Invalid argument" is not a very clear explanation. I think it would be usefull to start the device, i.e. run the init sequence, when mii-tool is run, but the interface is still down.
Francois, did a patch get merged up for this?
> Francois, did a patch get merged up for this? Yes. It is gross but it works. I would not mind an extra state in bugzilla to turn the PR into a feature request (the patch is _really_ gross). -- Ueimor
But having such resolution type would be really helpful, I'll work on that. Was the problem resolved, should it be closed? Thanks.
Sorry, I closed it prematurely.