Re: pcie-card issue with R-Car M3-W WS1.1 board (NETDEV WATCHDOG: eth0 (e1000e): transmit queue 0 timed out)

[Marek Vasut <marek.vasut@gmail.com>]

On 6/5/19 9:40 AM, Biju Das wrote:
> Hi All,
> 
>  
> 
> Looks like there is a pcie issue on R-Car M3-W  WS1.1 Salvator-XS board
> with upstream kernel.
> 
>  
> 
> As we know M3-W board has 2 memory banks
> 
>  
> 
> Bank #0: 0x048000000 - 0x0bfffffff, 1.9 GiB
> 
> Bank #1: 0x600000000 - 0x67fffffff, 2 GiB
> 
>  
> 
> I am using PCIE-ethernet  card(Ethernet controller: Intel Corporation
> 82574L Gigabit Network Connection) for pcie testing.
> 
>  
> 
> With default configuration, it detects the ethernet  card
> 
>  
> 
> root@salvator-x:~# lspci
> 
> 00:00.0 PCI bridge: Renesas Technology Corp. Device 0028
> 
> 01:00.0 Ethernet controller: Intel Corporation 82574L Gigabit Network
> Connection
> 
>  
> 
> However, if we enable kernel configuration for PCIE-ethernet
> card(CONFIG_E1000E=y)then i get below crash during boot
> 
>  
> 
> [   11.999119] ------------[ cut here ]------------
> 
> [   12.004979] NETDEV WATCHDOG: eth0 (e1000e): transmit queue 0 timed out
> 
> [   12.012734] WARNING: CPU: 0 PID: 0 at net/sched/sch_generic.c:443
> dev_watchdog+0x3e8/0x420
> 
> [   12.022206] CPU: 0 PID: 0 Comm: swapper/0 Not tainted
> 5.2.0-rc3-arm64-renesas-00004-g2e3a083-dirty #25
> 
> [   12.032728] Hardware name: Renesas Salvator-X 2nd version board based
> on r8a7796 (DT)
> 
> [   12.041772] pstate: 60000005 (nZCv daif -PAN -UAO)
> 
> [   12.047756] pc : dev_watchdog+0x3e8/0x420
> 
> [   12.052948] lr : dev_watchdog+0x3e8/0x420
> 
> [   12.058111] sp : ffff000010003d00
> 
> [   12.062557] x29: ffff000010003d00 x28: 0000000000000002
> 
> [   12.069000] x27: 00000000ffffffff x26: ffff000010f1c000
> 
> [   12.075426] x25: ffff800639324430 x24: ffff000010f4f740
> 
> [   12.081829] x23: ffff8006393243e0 x22: ffff00001104c000
> 
> [   12.088207] x21: ffff000010f4a000 x20: ffff800639324000
> 
> [   12.094564] x19: 0000000000000000 x18: ffffffffffffffff
> 
> [   12.100903] x17: 00000000000040f8 x16: 000000000000403c
> 
> [   12.107222] x15: 0000000000000001 x14: ffff000010baef68
> 
> [   12.113520] x13: 0000000000000000 x12: 0000000000000000
> 
> [   12.119809] x11: 0000000000000002 x10: 0000000000000002
> 
> [   12.126077] x9 : 0000000000000000 x8 : ffff000010f4f708
> 
> [   12.132331] x7 : ffff00001015d06c x6 : ffff000010f52000
> 
> [   12.138574] x5 : 0000000000000000 x4 : 0000000000000002
> 
> [   12.144798] x3 : ffffffffffffff50 x2 : ffff000010f6f7d8
> 
> [   12.151022] x1 : 6610066f85848800 x0 : 0000000000000000
> 
> [   12.157235] Call trace:
> 
> [   12.160547]  dev_watchdog+0x3e8/0x420
> 
> [   12.165081]  call_timer_fn+0xbc/0x400
> 
> [   12.169596]  expire_timers+0x110/0x230
> 
> [   12.174194]  run_timer_softirq+0xe4/0x1a0
> 
> [   12.179047]  __do_softirq+0x114/0x578
> 
> [   12.183542]  irq_exit+0x144/0x150
> 
> [   12.187665]  __handle_domain_irq+0x60/0xb8
> 
> [   12.192558]  gic_handle_irq+0x58/0xa8
> 
> [   12.197014]  el1_irq+0xbc/0x180
> 
> [   12.200913]  arch_cpu_idle+0x34/0x238
> 
> [   12.205308]  default_idle_call+0x1c/0x40
> 
> [   12.209961]  do_idle+0x1fc/0x2f0
> 
> [   12.213895]  cpu_startup_entry+0x24/0x28
> 
> [   12.218540]  rest_init+0x1a0/0x270
> 
> [   12.222671]  arch_call_rest_init+0xc/0x14
> 
> [   12.227417]  start_kernel+0x4a8/0x4d4
> 
> [   12.231818] irq event stamp: 40197
> 
> [   12.235962] hardirqs last  enabled at (40196): [<ffff00001015d12c>]
> console_unlock+0x3ec/0x5c0
> 
> [   12.245376] hardirqs last disabled at (40197): [<ffff0000100817a4>]
> do_debug_exception+0xbc/0x178
> 
> [   12.255042] softirqs last  enabled at (40174): [<ffff0000100f13a8>]
> _local_bh_enable+0x20/0x40
> 
> [   12.264434] softirqs last disabled at (40175): [<ffff0000100f17e4>]
> irq_exit+0x144/0x150
> 
> [   12.273278] ---[ end trace cfa1ae1709343a35 ]---
> 
> [   12.278880] e1000e 0000:01:00.0 eth0: Reset adapter unexpectedly
> 
>  
> 
> Q1) Has any one observed  this issue?  
> 
>  
> 
> It works without any issues for the following cases with upstream kernel.
> 
>  
> 
> Case1: Use the below commit[1] from renesas-bsp
> 
>  
> 
> [1].
> https://git.kernel.org/pub/scm/linux/kernel/git/horms/renesas-bsp.git/commit/?id=561e15db1ed8c5cc2fc340796e2d1c643111d27a
> 
>  
> 
> Case 2: Disable the bank#1 from device tree
> 
>  
> 
> +++ b/arch/arm64/boot/dts/renesas/r8a7796-salvator-xs.dts
> 
> +#if 0
> 
>         memory@600000000 {
> 
>                 device_type = "memory";
> 
>                 reg = <0x6 0x00000000 0x0 0x80000000>;
> 
>         };
> 
> +#endif
> 
> +
> 
> };
> 
>  
> 
> Q2) Could please share your thought on this issue?

I presume it's this [1]. Robin seems to be working on a fix, see [2],
although I suspect he won't deliver it, so I'll have to get back to this
topic myself.

[1]
https://git.kernel.org/pub/scm/linux/kernel/git/horms/renesas-bsp.git/commit/?id=561e15db1ed8c5cc2fc340796e2d1c643111d27a
[2] https://patchwork.kernel.org/patch/10842111/
-- 
Best regards,
Marek Vasut