Home » Mailing list archives » linux-netdev » 2008 » August » 6

Re: Realtek 8111C transmit timed out

Previous thread: [PATCH] correct pointer assignment in free_opts() by Denys Fedoryshchenko on Wednesday, August 6, 2008 - 10:47 am. (1 message)

Next thread: hash_list and rcu locking question by Tobias Koeck on Wednesday, August 6, 2008 - 11:56 am. (2 messages)
[view original message]
From: Indan Zupancic <indan@...>
To: <netdev@...>
Subject: Re: Realtek 8111C transmit timed out
Date: Wednesday, August 6, 2008 - 10:57 am

I'm running a git version just before 2.6.27-rc2 and while the card

worked fine in 2.6.26, after a resume from suspend to ram I now get:


[ 8367.073701] r8169 0000:01:00.0: restoring config space at offset 0x1

(was 0x100007, writing 0x100407)

[ 8367.073863] sd 2:0:0:0: [sda] Starting disk

[ 8367.077786] r8169: eth0: link up

[ 8367.265653] ata4.01: configured for UDMA/100

[ 8371.953804] ata3.00: configured for UDMA/133

[ 8371.964803] sd 2:0:0:0: [sda] 390721968 512-byte hardware sectors (200050 MB)

[ 8371.964833] sd 2:0:0:0: [sda] Write Protect is off

[ 8371.964836] sd 2:0:0:0: [sda] Mode Sense: 00 3a 00 00

[ 8371.964883] sd 2:0:0:0: [sda] Write cache: enabled, read cache: enabled,

doesn't support DPO or FUA

[ 8371.975538] pci 0000:00:02.0: PCI INT A -> GSI 16 (level, low) -> IRQ 16

[ 8371.975545] pci 0000:00:02.0: setting latency timer to 64

[ 8371.977770] PM: Finishing wakeup.

[ 8371.977772] Restarting tasks ... done.

[ 8397.700577] NETDEV WATCHDOG: eth0 (r8169): transmit timed out

[ 8397.700590] ------------[ cut here ]------------

[ 8397.700592] WARNING: at

/home/indan/src/git/linux-2.6/net/sched/sch_generic.c:219

dev_watchdog+0xfa/0x14a()

[ 8397.700594] Modules linked in: i915 drm

[ 8397.700600] Pid: 0, comm: swapper Not tainted 2.6.27-rc1 #8

[ 8397.700602]

[ 8397.700602] Call Trace:

[ 8397.700604]  <IRQ>  [<ffffffff80235512>] warn_on_slowpath+0x51/0x77

[ 8397.700612]  [<ffffffff80236043>] printk+0x4e/0x56

[ 8397.700615]  [<ffffffff804cda7d>] _spin_lock_irqsave+0x18/0x3a

[ 8397.700619]  [<ffffffff8023dd61>] lock_timer_base+0x26/0x4c

[ 8397.700622]  [<ffffffff804cdd80>] _spin_unlock_irqrestore+0x2d/0x39

[ 8397.700625]  [<ffffffff8023deee>] __mod_timer+0xbb/0xcd

[ 8397.700628]  [<ffffffff8022869a>] calc_delta_fair+0x1a/0x34

[ 8397.700632]  [<ffffffff8024448a>] queue_delayed_work_on+0xaa/0xba

[ 8397.700635]  [<ffffffff8047d14c>] dev_watchdog+0x0/0x14a

[ 8397.700637]  [<ffffffff8047d246>] dev_watchdog...
[ message continues ]
[view original message]
From: Francois Romieu <romieu@...>
To: Indan Zupancic <indan@...>
Cc: <netdev@...>
Subject: Re: Realtek 8111C transmit timed out
Date: Wednesday, August 6, 2008 - 2:12 pm

Indan Zupancic <indan@nul.nu> :


You cab try to revert the four (4) r8169.c related patches that went

in between 2.6.26 and 2.6.27-rc1. I doubt it will make a difference

though.


Is a git bisect an option ?


--

Ueimor

--
[view original message]
From: Indan Zupancic <indan@...>
To: Francois Romieu <romieu@...>
Cc: <netdev@...>
Subject: Re: Realtek 8111C transmit timed out
Date: Wednesday, August 6, 2008 - 3:37 pm

Hello,


"git log drivers/net/r8169.c" shows only two commits since May 11:


r8169: avoid thrashing PCI conf space above RTL_GIGA_MAC_VER_06


r8169: multicast register update


Then at May 11 comes:


r8169: remove non-napi code


But I think that was in 2.6.26 already.


Reverting the first two didn't help, so I guess it's messed up

somewhere else.


Are two patches missing or didn't they touch r8169.c?


I tried manually reverting patch


0004-r8169-new-phy-init-parameters-for-the-8168b.patch


Yes, but it takes quite a lot of time and I'm busy at the moment,

so I'll wait till rc2 to see if it's fixed then. If it isn't I'll

start digging.


Greetings,


Indan


--
[view original message]
From: Francois Romieu <romieu@...>
To: Indan Zupancic <indan@...>
Cc: <netdev@...>
Subject: Re: Realtek 8111C transmit timed out
Date: Wednesday, August 6, 2008 - 3:50 pm

Indan Zupancic <indan@nul.nu> :


$ git rev-list v2.6.26..v2.6.27-rc1 -- drivers/net/r8169.c

77332894c21165404496c56763d7df6c15c4bb09

f887cce8de019bb32917789379af89ae4c0294ee

865c652d6be9929927cabdc54b137b7541eb6612


No. :o)


Ok.


--

Ueimor

--
[view original message]
From: Indan Zupancic <indan@...>
To: Francois Romieu <romieu@...>
Cc: <netdev@...>
Subject: Re: Realtek 8111C transmit timed out
Date: Saturday, August 23, 2008 - 9:51 am

Running 2.6.27-rc3 now and that works fine. :-)

(Of course the next day rc4 came out...)


Let's use this opportunity to say a big thank you to all

you wonderful people who keep things running. The world

wouldn't be the same without you.


Greetings,


Indan


--
[view original message]
From: Indan Zupancic <indan@...>
To: Francois Romieu <romieu@...>
Cc: <netdev@...>
Subject: Re: Realtek 8111C transmit timed out
Date: Wednesday, August 6, 2008 - 6:50 pm

Oh...


I tried 2.6.27-rc2 with that patch reverted and NAPI disabled,

but that didn't work either, so it seems it's something a level

higher. (No way that that last patch could cause this...)


You said off-list that I may want to check if there is any aspm

related message in dmesg. Well, I get this:


  Pre-1.1 PCIe device detected, disable ASPM for 0000:00:1c.1.

  It can be enabled forcedly with 'pcie_aspm=force'


So I guess it's unlikely that it causes any problems, except

if the driver assumes it's always enabled. Booting with

pcie_aspm=force didn't change anything (though strangely

enough I still get the same message as above).


All in all I'll wait till rc3 and see how things are then.


Thank you for the help,


Indan


--
Previous thread: [PATCH] correct pointer assignment in free_opts() by Denys Fedoryshchenko on Wednesday, August 6, 2008 - 10:47 am. (1 message)

Next thread: hash_list and rcu locking question by Tobias Koeck on Wednesday, August 6, 2008 - 11:56 am. (2 messages)