Re: bge problems when resuming

From: Gonzalo Nemmi <gnemmi_at_gmail.com>
Date: Tue, 21 Jul 2009 15:43:54 -0300
On Tue, Jul 21, 2009 at 5:46 AM, Paul B. Mahol <onemda_at_gmail.com> wrote:

> On 7/20/09, Gonzalo Nemmi <gnemmi_at_gmail.com> wrote:
> > On Sunday 19 July 2009 7:53:52 pm Paul B. Mahol wrote:
> >> On 7/20/09, Gonzalo Nemmi <gnemmi_at_gmail.com> wrote:
> >> > On Sat, Jul 18, 2009 at 12:09 AM, Paul B. Mahol <onemda_at_gmail.com>
> > wrote:
> >> >> On 7/17/09, Gonzalo Nemmi <gnemmi_at_gmail.com> wrote:
> >> >> > On Wednesday 15 July 2009 8:13:47 am Adam K Kirchhoff wrote:
> >> >> >> On Wednesday 15 July 2009 03:20:45 Paul B. Mahol wrote:
> >> >> >> > On 7/15/09, Adam K Kirchhoff <adamk_at_voicenet.com> wrote:
> >> >> >> > > Hello all,
> >> >> >> > >
> >> >> >> > > I have a Dell Latitude D610 laptop with 8.0-BETA1
> >> >> >> > > installed.  I hadn't tried suspend/resume for a while and
> >> >> >> > > decided to give it a shot.  I was pleasantly surprised to
> >> >> >> > > see that I could suspend to ram, resume, and have a
> >> >> >> > > (relatively) working system (previously the display would
> >> >> >> > > never come back up and the serial console I had hooked up
> >> >> >> > > remained dead). Great job to everyone who helped make that
> >> >> >> > > possible.
> >> >> >> > >
> >> >> >> > > The only real issue that I seem to have now is that bge is
> >> >> >> > > completely unusable after resume.  Another individual seems
> >> >> >> > > to have reported similar problems with bge and resume, but
> >> >> >> > > he also had other issues that apparently trumped his
> >> >> >> > > networking issues:
> >> >> >> > >
> >> >> >> > > http://lists.freebsd.org/pipermail/freebsd-current/2009-Jul
> >> >> >> > >y/0090 23.html
> >> >> >> > >
> >> >> >> > > Like him, resuming from suspend gives me:
> >> >> >> > >
> >> >> >> > > Jul 14 12:35:53 scroll kernel: bge0: PHY write timed out
> >> >> >> > > (phy 1, reg 0, val 32768)
> >> >> >> > > Jul 14 12:35:53 scroll kernel: bge0: PHY read timed out
> >> >> >> > > (phy 1, reg 0, val 0xffffffff)
> >> >> >> > > Jul 14 12:35:53 scroll kernel: bge0: PHY write timed out
> >> >> >> > > (phy 1, reg 24, val 3072)
> >> >> >> > > Jul 14 12:35:53 scroll kernel: bge0: PHY write timed out
> >> >> >> > > (phy 1, reg 23, val 10)
> >> >> >> > > Jul 14 12:35:53 scroll kernel: bge0: PHY write timed out
> >> >> >> > > (phy 1, reg 21, val 12555)
> >> >> >> > > Jul 14 12:35:53 scroll kernel: bge0: PHY write timed out
> >> >> >> > > (phy 1, reg 23, val 8223)
> >> >> >> > > Jul 14 12:35:53 scroll kernel: bge0: PHY write timed out
> >> >> >> > > (phy 1, reg 21, val 38150)
> >> >> >> > > Jul 14 12:35:53 scroll kernel: bge0: PHY write timed out
> >> >> >> > > (phy 1, reg 23, val 16415)
> >> >> >> > > Jul 14 12:35:53 scroll kernel: bge0: PHY write timed out
> >> >> >> > > (phy 1, reg 21, val 5346)
> >> >> >> > > Jul 14 12:35:53 scroll kernel: bge0: PHY write timed out
> >> >> >> > > (phy 1, reg 24, val 1024)
> >> >> >> > > Jul 14 12:35:53 scroll kernel: bge0: PHY write timed out
> >> >> >> > > (phy 1, reg 24, val 7)
> >> >> >> > >
> >> >> >> > > And so on and so forth.
> >> >> >> > >
> >> >> >> > > I thought that compiling if_bge as a module, unloading it
> >> >> >> > > before suspend, and reloading it after resume, might get
> >> >> >> > > this working. However, doing a "kldload if_bge" after the
> >> >> >> > > resume does nothing. Well, the module gets loaded, but the
> >> >> >> > > device doesn't show up.  No errors from kldload, and there
> >> >> >> > > is nothing new in dmesg.
> >> >> >> > >
> >> >> >> > > Before the suspend, the device shows up as:
> >> >> >> > >
> >> >> >> > > bge0_at_pci0:2:0:0:        class=0x020000 card=0x01821028
> >> >> >> > > chip=0x167714e4 rev=0x01 hdr=0x00
> >> >> >> > >     vendor     = 'Broadcom Corporation'
> >> >> >> > >     device     = 'NetXtreme Gigabit Ethernet PCI Express
> >> >> >> > > (BCM5750A1)' class      = network
> >> >> >> > >     subclass   = ethernet
> >> >> >> > >
> >> >> >> > > After resuming,  and reloading the module, it's:
> >> >> >> > >
> >> >> >> > > none1_at_pci0:2:0:0:       class=0x020000 card=0x01821028
> >> >> >> > > chip=0x167714e4 rev=0x01 hdr=0x00
> >> >> >> > >     vendor     = 'Broadcom Corporation'
> >> >> >> > >     device     = 'NetXtreme Gigabit Ethernet PCI Express
> >> >> >> > > (BCM5750A1)' class      = network
> >> >> >> > >     subclass   = ethernet
> >> >> >> > >
> >> >> >> > > If there are no ideas, I'll go ahead and open up a pr.  I
> >> >> >> > > assume this is just one bug, since both problems (the PHY
> >> >> >> > > issues and the inability to reload the driver) are both
> >> >> >> > > related to the network device.
> >> >> >> >
> >> >> >> > Put this lines into loader.conf and reboot.
> >> >> >> >
> >> >> >> > hw.pci.do_power_nodriver="3"
> >> >> >> > hw.pci.do_power_resume="1"
> >> >> >> >
> >> >> >> > Now, before suspend, unload if_bge and some another driver
> >> >> >> > (sound drivers are best candidate) and load sound driver
> >> >> >> > again, suspend and resume.
> >> >> >> > Now loading if_bge should make it succesfully attach.
> >> >> >>
> >> >> >> Unfortunately, after doing this, reloading the if_bge driver
> >> >> >> causes the laptop to completely lock up...  It gets as far as:
> >> >> >>
> >> >> >> bge0: <Broadcom NetXtreme Gigabit Ethernet Controller, unknown
> >> >> >> ASIC rev. 0xffff>
> >> >> >> mem 0xdfdf0000-0xdfdfffff irq 16 at device 0.0 on pci2
> >> >> >>
> >> >> >> And then the entire machine hangs.  I'm on ttyv0, so I'd see
> >> >> >> any kernel panic, but nothing like that happens.  The screen
> >> >> >> stays on, but nothing else happens till I force a reboot.
> >> >> >>
> >> >> >> Adam
> >> >> >
> >> >> > Hi Adam, Paul ...
> >> >> > I'm the "another individual" from you OP.
> >> >> > I have the same problems you have regarding bge, but they
> >> >> > weren't trumped .. I just had an order of priorities ;)
> >> >> >
> >> >> > Anyways, I tried the solution Paul posted and, just as in your
> >> >> > case, I got a hard lock too ...
> >> >> >
> >> >> > I tried loading if_bge through /boot/loader.conf
> >> >> > Then issued a:
> >> >> >
> >> >> > kldunload if_bge coretemp
> >> >>
> >> >> coretemp is wrong module, it must be one of modules that attach to
> >> >> pci.
> >> >
> >> > Sorry Paul!
> >> > I gave it a go with snd_hda and I got the same result except that
> >> > this time I also got the following message:
> >>
> >> After unloading snd_hda you loaded it again before suspending?
> >
> > Doing so yielded a Fatal trap 12 on BETA2. Yesterday I install BETA2 and
> > here are the results:
>

Sorry, I meant BETA1 yielded a Fatal trap12, not BETA2


> >
> > kldstat
> >
> > Id Refs Address    Size     Name
> >  1   28 0xc0400000 cf6c70   kernel
> >  2    1 0xc10f7000 11bc0    if_bge.ko
> >  3    1 0xc1109000 1ac4c    snd_hda.ko
> >  4    2 0xc1124000 61f78    sound.ko
> >  5    1 0xc1186000 2af4     coretemp.ko
> >  6    1 0xc1189000 a6d8     i915.ko
> >  7    2 0xc1194000 177d4    drm.ko
> >
> >
> > kldunload if_bge snd_hda
> >
> > Jul 20 17:50:49 gargoyle login: ROOT LOGIN (root) ON ttyv0
> > Jul 20 17:51:06 gargoyle kernel: brgphy0: detached
> > Jul 20 17:51:06 gargoyle kernel: lock order reversal:
> > Jul 20 17:51:06 gargoyle kernel: 1st 0xc0dba45c kernel linker (kernel
> > linker) _at_ /usr/src/sys/kern/kern_linker.c:1079
> > Jul 20 17:51:06 gargoyle kernel: 2nd 0xc0dbbc64 sysctl lock (sysctl
> > lock) _at_ /usr/src/sys/kern/kern_sysctl.c:257
> > Jul 20 17:51:06 gargoyle kernel: KDB: stack backtrace:
> > Jul 20 17:51:06 gargoyle kernel:
> > db_trace_self_wrapper(c0c6baf4,e6daba34,c08bc995,c08ad6db,c0c6e989,...)
> > at db_trace_self_wrapper+0x26
> > Jul 20 17:51:06 gargoyle kernel:
> > kdb_backtrace(c08ad6db,c0c6e989,c452bc88,c4529e10,e6daba90,...) at
> > kdb_backtrace+0x29
> > Jul 20 17:51:06 gargoyle kernel:
> > _witness_debugger(c0c6e989,c0dbbc64,c0c69667,c4529e10,c0c6956e,...) at
> > _witness_debugger+0x25
> > Jul 20 17:51:06 gargoyle kernel:
> > witness_checkorder(c0dbbc64,9,c0c6956e,101,0,...) at
> > witness_checkorder+0x839
> > Jul 20 17:51:06 gargoyle kernel:
> > _sx_xlock(c0dbbc64,0,c0c6956e,101,c4722c00,...) at _sx_xlock+0x85
> > Jul 20 17:51:06 gargoyle kernel:
> > sysctl_ctx_free(c4722c4c,c4722c00,e6dabb18,c08a3c85,c4722c00,...) at
> > sysctl_ctx_free+0x30
> > Jul 20 17:51:06 gargoyle kernel:
> > device_sysctl_fini(c4722c00,0,c0d4c848,c472a810,c4ab3400,...) at
> > device_sysctl_fini+0x1a
> > Jul 20 17:51:06 gargoyle kernel:
> > device_detach(c4722c00,c4722b80,e6dabb38,c06bc622,c4722b80,...) at
> > device_detach+0x1f5
> > Jul 20 17:51:06 gargoyle kernel:
> > bus_generic_detach(c4722b80,c4722b80,e6dabb64,c08a3b1c,c4722b80,...) at
> > bus_generic_detach+0x29
> > Jul 20 17:51:06 gargoyle kernel:
> > miibus_detach(c4722b80,c45d6060,c0d4ca68,a3c,c0c76f47,...) at
> > miibus_detach+0x12
> > Jul 20 17:51:06 gargoyle kernel:
> > device_detach(c4722b80,c472b008,e6dabb98,c10ff7ff,c4722300,...) at
> > device_detach+0x8c
> > Jul 20 17:51:06 gargoyle kernel:
> > bus_generic_detach(c4722300,1,c1104b66,aec,c4722300,...) at
> > bus_generic_detach+0x29
> > Jul 20 17:51:06 gargoyle kernel:
> > bge_detach(c4722300,c4677060,c0d4ca68,a3c,c4526300,...) at
> > bge_detach+0xbf
> > Jul 20 17:51:06 gargoyle kernel:
> > device_detach(c4722300,c086c843,c0dbb570,c1106c20,c456fb80,...) at
> > device_detach+0x8c
> > Jul 20 17:51:06 gargoyle kernel:
> > driver_module_handler(c4526300,1,c1106c20,109,0,...) at
> > driver_module_handler+0x29c
> > Jul 20 17:51:06 gargoyle kernel:
> > module_unload(c4526300,c0c652ef,273,270,c08604b6,...) at
> > module_unload+0x43
> > Jul 20 17:51:06 gargoyle kernel:
> > linker_file_unload(c4544200,0,c0c652ef,437,c10f7000,...) at
> > linker_file_unload+0x15e
> > Jul 20 17:51:06 gargoyle kernel:
> > kern_kldunload(c4b346c0,2,0,e6dabd2c,c0ba8dd3,...) at
> > kern_kldunload+0xd5
> > Jul 20 17:51:06 gargoyle kernel:
> > kldunloadf(c4b346c0,e6dabcf8,8,c0c6fa4b,c0d50450,...) at
> > kldunloadf+0x2b
> > Jul 20 17:51:06 gargoyle kernel: syscall(e6dabd38) at syscall+0x2a3
> > Jul 20 17:51:06 gargoyle kernel: Xint0x80_syscall() at
> > Xint0x80_syscall+0x20
> > Jul 20 17:51:06 gargoyle kernel: --- syscall (444, FreeBSD ELF32,
> > kldunloadf), eip = 0x280d516b, esp = 0xbfbfe47c, ebp = 0xbfbfecc8 ---
> > Jul 20 17:51:06 gargoyle kernel: miibus0: detached
> > Jul 20 17:51:06 gargoyle kernel: bge0: detached
> > Jul 20 17:51:06 gargoyle kernel: sysctl_unregister_oid: failed to
> > unregister sysctl
>
> if_bge driver looks very problematic to me. Probably it can not detach at
> all.
>

Could be .. will have to recompile the kernel and use if_bge as a module to
see what happens ...
I just read a mail from Øyvind Rakvåg saying he is experiencing a freeze
when setting bge0 to UP:
http://lists.freebsd.org/pipermail/freebsd-current/2009-July/009677.html
Maybe bge needs some care :s


> > Jul 20 17:51:06 gargoyle kernel: pcm0: detached
> > Jul 20 17:51:06 gargoyle kernel: hdac0: detached
> >
> >
> > kld snd_hda
>   ^^^
> You mean kldload.
>

Yes, sorry Paul, I guess I typed too fast :s


>
> >
> > Jul 20 17:52:16 gargoyle kernel: hdac0: <Intel 82801H High Definition
> > Audio Controller> mem 0xf6dfc000-0xf6dfffff irq 21 at device 27.0 on
> > pci0
> > Jul 20 17:52:16 gargoyle kernel: hdac0: HDA Driver Revision:
> > 20090624_0136
> > Jul 20 17:52:16 gargoyle kernel: hdac0: [ITHREAD]
> > Jul 20 17:52:16 gargoyle kernel: hdac0: HDA Codec #0: Sigmatel STAC9228X
> > Jul 20 17:52:16 gargoyle kernel: bge0: <Broadcom BCM5906 A2, ASIC rev.
> > 0xc002> mem 0xf69f0000-0xf69fffff irq 17 at device 0.0 on pci9
> > Jul 20 17:52:16 gargoyle kernel: miibus0: <MII bus> on bge0
> > Jul 20 17:52:16 gargoyle kernel: brgphy0: <BCM5906 10/100baseTX PHY> PHY
> > 1 on miibus0
> > Jul 20 17:52:16 gargoyle kernel: brgphy0:  10baseT, 10baseT-FDX,
> > 100baseTX, 100baseTX-FDX, auto
> > Jul 20 17:52:16 gargoyle kernel: bge0: Ethernet address:
> > 00:23:ae:04:ba:ca
> > Jul 20 17:52:16 gargoyle kernel: bge0: [ITHREAD]
> > Jul 20 17:52:16 gargoyle kernel: pcm0: <HDA Sigmatel STAC9228X PCM #0
> > Analog> at cad 0 nid 1 on hdac0
> > Jul 20 17:52:16 gargoyle kernel: bge0: link state changed to DOWN
> > Jul 20 17:52:18 gargoyle kernel: bge0: link state changed to UP
>
> Why bge0 appeared again?
>

Really don´t know ...
That´s what happened after issuing a kldload snd_hda.


>
> >
> > acpiconf -s 3
>
> After this command bge0 should not appear at all because it should not
> be attached to
> device.
>
> >
> > Jul 20 17:53:51 gargoyle acpi: suspend at 20090720 17:53:51
> > Jul 20 17:53:56 gargoyle kernel: fwohci0: fwohci_pci_suspend
> > Jul 20 17:54:25 gargoyle kernel: bge0: PHY write timed out (phy 1, reg
> > 0, val 32768)
> > Jul 20 17:54:25 gargoyle kernel: bge0: PHY read timed out (phy 1, reg 0,
> > val 0xffffffff)
> > Jul 20 17:54:25 gargoyle kernel: bge0: PHY read timed out (phy 1, reg
> > 24, val 0xffffffff)
> > Jul 20 17:54:25 gargoyle kernel: bge0: PHY read timed out (phy 1, reg
> > 16, val 0xffffffff)
> > Jul 20 17:54:25 gargoyle kernel: bge0: PHY write timed out (phy 1, reg
> > 16, val 0)
> > Jul 20 17:54:25 gargoyle kernel: bge0: PHY read timed out (phy 1, reg
> > 16, val 0xffffffff)
> > Jul 20 17:54:25 gargoyle kernel: bge0: PHY write timed out (phy 1, reg
> > 16, val 0)
> > Jul 20 17:54:25 gargoyle kernel: bge0: PHY write timed out (phy 1, reg
> > 23, val 18)
> > Jul 20 17:54:25 gargoyle kernel: bge0: flow-through queue init failed
> > Jul 20 17:54:25 gargoyle kernel: bge0: initialization failure
> > Jul 20 17:54:25 gargoyle kernel: fwohci0: Phy 1394a available S400, 1
> > ports.
> > Jul 20 17:54:25 gargoyle kernel: fwohci0: Link S400, max_rec 2048 bytes.
> > Jul 20 17:54:25 gargoyle kernel: fwohci0: Initiate bus reset
> > Jul 20 17:54:25 gargoyle kernel: fwohci0: fwohci_intr_core: BUS reset
> > Jul 20 17:54:25 gargoyle kernel: fwohci0: fwohci_intr_core:
> > node_id=0x00000000, SelfID Count=1, CYCLEMASTER mode
> > Jul 20 17:54:25 gargoyle kernel: firewire0: 1 nodes, maxhop <= 0 cable
> > IRM irm(0)  (me)
> > Jul 20 17:54:25 gargoyle kernel: firewire0: bus manager 0
> > Jul 20 17:54:25 gargoyle kernel: fwohci0: unrecoverable error
> > Jul 20 17:54:25 gargoyle kernel: wakeup from sleeping state (slept
> > 00:00:29)
> > Jul 20 17:54:25 gargoyle acpi: resumed at 20090720 17:54:25
> >
> > Should a PR on fwohci and firewire also be filed??
>
> Try with custom kernel with smaller number of drivers as possible.
> (use modules instead)


Will do. I´be been trying to avoid that in order to test with a custom
kernel.
Will recompile GENERIC commenting out if_bge only, load it via loader.conf
and test it to see what happens.


> From your mail I dont see where is problem with firewire.
>

OK ... I filed a PR on fwohhci (unrecoverable error) though, if I shouldn´t
have done that, sorry for the noise then, I don´t have firewire devices to
test it: http://www.freebsd.org/cgi/query-pr.cgi?pr=136946

Best Regards
Gonzalo
Received on Tue Jul 21 2009 - 16:43:57 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:39:52 UTC