Re: [RTnet-users] rtifconfig rteth0 up dumps irq event 52: bogus return value

Jan Kiszka Tue, 08 Sep 2009 14:29:05 -0700

Anze Zagar wrote:
> Hi all,
> 
> I am experiencing problems when trying to bring up Ethernet interface on
> RTnet enabled environment.
> 
> System is Scientific Linux 5.3 with custom built 2.6.30 x86_64 kernel
> and Xenomai 2.4.9. RTnet version was initially 0.9.11 and I tried the
> latest git snapshot as well.
> 
> System has three Ethernet controllers:
> 02:02.0 Ethernet controller: Intel Corporation 82541PI Gigabit Ethernet
> Controller (rev 05)
>       Subsystem: Intel Corporation PRO/1000 GT Desktop Adapter
>       Flags: bus master, 66MHz, medium devsel, latency 52, IRQ 24
>       Memory at d0220000 (32-bit, non-prefetchable) [size=128K]
>       Memory at d0200000 (32-bit, non-prefetchable) [size=128K]
>       I/O ports at 2000 [size=64]
>       [virtual] Expansion ROM at d1000000 [disabled] [size=128K]
>       Capabilities: [dc] Power Management version 2
>       Capabilities: [e4] PCI-X non-bridge device
> 
> 0d:00.0 Ethernet controller: Intel Corporation 82573E Gigabit Ethernet
> Controller (Copper) (rev 03)
>       Subsystem: Super Micro Computer Inc Unknown device 108c
>       Flags: bus master, fast devsel, latency 0, IRQ 52
>       Memory at d0300000 (32-bit, non-prefetchable) [size=128K]
>       I/O ports at 3000 [size=32]
>       Capabilities: [c8] Power Management version 2
>       Capabilities: [d0] Message Signalled Interrupts: 64bit+ Queue=0/0
> Enable+
>       Capabilities: [e0] Express Endpoint IRQ 0
> 
> 0f:00.0 Ethernet controller: Intel Corporation 82573L Gigabit Ethernet
> Controller
>       Subsystem: Super Micro Computer Inc Unknown device 109a
>       Flags: bus master, fast devsel, latency 0, IRQ 17
>       Memory at d0400000 (32-bit, non-prefetchable) [size=128K]
>       I/O ports at 4000 [size=32]
>       Capabilities: [c8] Power Management version 2
>       Capabilities: [d0] Message Signalled Interrupts: 64bit+ Queue=0/0
> Enable-
>       Capabilities: [e0] Express Endpoint IRQ 0
> 
> rteth0 that I am trying to bring up in this case is actually the PCIe
> 82573E. The PCI based 82541PI uses a non-rtnet e1000 driver for regular
> UDP/TCP traffic.
> 
> After issuing the following commands:
> /sbin/rmmod e1000e;
> /sbin/insmod /usr/rtnet/modules/rtnet.ko socket_rtskbs=64;
> /sbin/insmod /usr/rtnet/modules/rtipv4.ko;
> /sbin/insmod /usr/rtnet/modules/rtpacket.ko;
> /sbin/insmod /usr/rtnet/modules/rt_loopback.ko;
> /sbin/insmod /usr/rtnet/modules/rt_e1000_new.ko pciif=pcie;
> /sbin/insmod /usr/rtnet/modules/rtcap.ko;
> /usr/rtnet/sbin/rtifconfig rtlo up 127.0.0.1;
> /usr/rtnet/sbin/rtifconfig rteth0 up 192.168.5.1 promisc netmask
> 255.255.255.0;
> 
> The last line dumps the following output in the dmesg (see attached
> dump.txt for the entire dmesg dump):
> rt_e1000_new 0000:0d:00.0: irq 52 for MSI/MSI-X
> rt_e1000_new 0000:0d:00.0: irq 52 for MSI/MSI-X
> irq event 52: bogus return value 2fb07cb0
> Pid: 0, comm: swapper Not tainted 2.6.30-xenomai-2.4.9 #3
> Call Trace:
>  [<ffffffff8026c690>] ? __report_bad_irq+0x30/0x7d
>  [<ffffffff8026c743>] ? note_interrupt+0x66/0x153
>  [<ffffffff8026d398>] ? handle_edge_irq+0xe4/0x114
>  [<ffffffff8020e315>] ? handle_irq+0x81/0x8a
>  [<ffffffff8020da0c>] ? do_IRQ+0x5a/0xa3
>  [<ffffffff8020d9b2>] ? do_IRQ+0x0/0xa3
>  [<ffffffff802702f4>] ? __ipipe_sync_stage+0x166/0x16c
>  [<ffffffff802702fa>] ? __xirq_end+0x0/0x6b
>  [<ffffffff8021faef>] ? __ipipe_handle_irq+0x18a/0x275
>  [<ffffffff8020c493>] ? common_interrupt+0x13/0x2c
>  [<ffffffff802120c9>] ? default_idle+0x8f/0x105
>  [<ffffffff80505b22>] ? __atomic_notifier_call_chain+0x46/0x69
>  [<ffffffff8020aadb>] ? cpu_idle+0x59/0x8b
> handlers:
> [<ffffffffa0360f33>] (e1000_intr_msi_test+0x0/0x71 [rt_e1000_new])
> rt_e1000_new 0000:0d:00.0: irq 52 for MSI/MSI-X
> e1000: rteth0: e1000_watchdog_task: NIC Link is Up 1000 Mbps Full
> Duplex, Flow Control: RX/TX
>


Hmm, the msi-test IRQ seems to trigger this mess. Could you try if
disabling CONFIG_PCI_MSI changes the picture.

Mathias, did you successfully tested or even used your driver with MSI
interrupts?

> 
> 
> Note also that in order to build the experimental e1000 driver on this
> 64-bit system, I needed to patch the code a little:
> --- drivers/experimental/e1000/e1000_main.c   2009-08-29
> 17:41:04.000000000 +0200
> +++ drivers/experimental/e1000/e1000_main.c~  2009-02-28
> 14:34:03.000000000 +0100
> @@ -305,3 +305,3 @@
>  static int  e1000_intr(rtdm_irq_t *irq_handle);
> -static irqreturn_t e1000_intr_msi(rtdm_irq_t *irq_handle);
> +static int e1000_intr_msi(rtdm_irq_t *irq_handle);
>  static bool e1000_clean_tx_irq(struct e1000_adapter *adapter,

This doesn't apply to git head. What version are you using, what
patches?

> 
> because I was getting the following compile time errors:
>   CC
> [M]  /usr/src/rtnet-0.9.11.1/drivers/experimental/e1000/e1000_main.o
> /usr/src/rtnet-0.9.11.1/drivers/experimental/e1000/e1000_main.c: In
> function ‘e1000_test_msi_interrupt’:
> /usr/src/rtnet-0.9.11.1/drivers/experimental/e1000/e1000_main.c:1750:
> warning: passing argument 2 of ‘request_irq’ from incompatible pointer
> type
> /usr/src/rtnet-0.9.11.1/drivers/experimental/e1000/e1000_main.c: In
> function ‘e1000_tx_csum’:
> /usr/src/rtnet-0.9.11.1/drivers/experimental/e1000/e1000_main.c:3567:
> warning: unused variable ‘css’
> /usr/src/rtnet-0.9.11.1/drivers/experimental/e1000/e1000_main.c: At top
> level:
> /usr/src/rtnet-0.9.11.1/drivers/experimental/e1000/e1000_main.c:4496:
> error: conflicting types for ‘e1000_intr_msi’
> /usr/src/rtnet-0.9.11.1/drivers/experimental/e1000/e1000_main.c:306:
> error: previous declaration of ‘e1000_intr_msi’ was here
> /usr/src/rtnet-0.9.11.1/drivers/experimental/e1000/e1000_main.c: In
> function ‘e1000_intr_msi’:
> /usr/src/rtnet-0.9.11.1/drivers/experimental/e1000/e1000_main.c:4578:
> warning: ‘return’ with a value, in function returning void
> make[5]: ***
> [/usr/src/rtnet-0.9.11.1/drivers/experimental/e1000/e1000_main.o] Error
> 1

OK, please try commit 07fb8132 from git head:

diff --git a/drivers/experimental/e1000/e1000_main.c 
b/drivers/experimental/e1000/e1000_main.c
index db5d61b..918159a 100644
--- a/drivers/experimental/e1000/e1000_main.c
+++ b/drivers/experimental/e1000/e1000_main.c
@@ -4492,7 +4492,7 @@ static void e1000_phy_read_status(struct e1000_adapter 
*adapter)
  * @irq: interrupt number
  * @data: pointer to a network interface device structure
  **/
-static irqreturn_t e1000_intr_msi(rtdm_irq_t *irq_handle)
+static int e1000_intr_msi(rtdm_irq_t *irq_handle)
 {
     struct rtnet_device *netdev = rtdm_irq_get_arg(irq_handle, struct 
rtnet_device); 
        struct e1000_adapter *adapter = netdev_priv(netdev);


Jan

signature.asc
Description: OpenPGP digital signature

------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day 
trial. Simplify your report design, integration and deployment - and focus on 
what you do best, core application coding. Discover what's new with 
Crystal Reports now.  http://p.sf.net/sfu/bobj-july

_______________________________________________
RTnet-users mailing list
RTnet-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/rtnet-users

Re: [RTnet-users] rtifconfig rteth0 up dumps irq event 52: bogus return value

Reply via email to