Re: Oops on ehci_hcd when booting 3.0.0-rc2 on panda

[Rajendra Nayak <rnayak@ti.com>]

On 8/11/2011 1:37 PM, Ohad Ben-Cohen wrote:
> + Paul, Benoit, Rajendra
>
> On Tue, Aug 9, 2011 at 2:26 PM, Luciano Coelho<coelho@ti.com>  wrote:
>> I'm again getting a very similar oops with 3.1-rc1 on my pandaboard:
>>
>> [    2.054351] usbcore: registered new interface driver cdc_ncm
>> [    2.061431] ehci_hcd: USB 2.0 'Enhanced' Host Controller (EHCI) Driver
>> [    2.068664] Unhandled fault: imprecise external abort (0x1406) at 0x00000000
>> [    2.076110] Internal error: : 1406 [#1] SMP
>> [    2.080505] Modules linked in:
>> [    2.083709] CPU: 0    Not tainted  (3.1.0-rc1-wl+ #283)
>> [    2.089233] PC is at omap_usbhs_enable+0x148/0x590
>> [    2.094299] LR is at trace_hardirqs_off+0x14/0x18
> ...
>> [    2.310150] [<c02a8640>] (omap_usbhs_enable+0x148/0x590) from [<c0321d60>] (ehci_hcd_omap_probe+0x1b4/0x568)
>> [    2.320526] [<c0321d60>] (ehci_hcd_omap_probe+0x1b4/0x568) from [<c0298cfc>] (platform_drv_probe+0x24/0x28)
>> [    2.330780] [<c0298cfc>] (platform_drv_probe+0x24/0x28) from [<c0297650>] (driver_probe_device+0x158/0x27c)
>> [    2.341033] [<c0297650>] (driver_probe_device+0x158/0x27c) from [<c02977ec>] (__driver_attach+0x78/0x9c)
>> [    2.351043] [<c02977ec>] (__driver_attach+0x78/0x9c) from [<c0296c00>] (bus_for_each_dev+0x5c/0x8c)
>> [    2.360565] [<c0296c00>] (bus_for_each_dev+0x5c/0x8c) from [<c0297334>] (driver_attach+0x28/0x30)
>> [    2.369903] [<c0297334>] (driver_attach+0x28/0x30) from [<c02963ec>] (bus_add_driver+0xd8/0x260)
>> [    2.379180] [<c02963ec>] (bus_add_driver+0xd8/0x260) from [<c0297f00>] (driver_register+0xb8/0x144)
>> [    2.388702] [<c0297f00>] (driver_register+0xb8/0x144) from [<c02991e0>] (platform_driver_register+0x54/0x68)
>> [    2.399047] [<c02991e0>] (platform_driver_register+0x54/0x68) from [<c069904c>] (ehci_hcd_init+0xa8/0xfc)
>> [    2.409149] [<c069904c>] (ehci_hcd_init+0xa8/0xfc) from [<c0008854>] (do_one_initcall+0xa8/0x17c)
>> [    2.418487] [<c0008854>] (do_one_initcall+0xa8/0x17c) from [<c06792d4>] (kernel_init+0x88/0x134)
>> [    2.427764] [<c06792d4>] (kernel_init+0x88/0x134) from [<c0014ba0>] (kernel_thread_exit+0x0/0x8)
>
> I get this too.
>
>> Any clues?

Its quite expected as omap_usbhs_enable() still relies on clock
framework to enable the clocks.
Any driver still using clock framework on OMAP4 to enable "main" clocks
is expected to be broken. The only way to fix this is to adapt
the driver to runtime PM.

>
> Reverting 665d001338b494d6d62810aa99b4c0fa1a0884b9 "OMAP2+: hwmod:
> Follow the recommended PRCM module enable sequence" fixes this for me.
>
> More specifically, this hunk alone seems to do the trick:
>
> diff --git a/arch/arm/mach-omap2/clock44xx_data.c b/arch/arm/mach-omap2/clock44x
> index 2af0e3f..12d22a8 100644
> --- a/arch/arm/mach-omap2/clock44xx_data.c
> +++ b/arch/arm/mach-omap2/clock44xx_data.c
> @@ -3379,7 +3379,6 @@ int __init omap4xxx_clk_init(void)
>          }
>
>          clk_init(&omap2_clk_functions);
> -       omap2_clk_disable_clkdm_control();
>
>          for (c = omap44xx_clks; c<  omap44xx_clks + ARRAY_SIZE(omap44xx_clks);
>                                                                            c++)
>
> I'm not suggesting this is anyway near a real fix, but hopefully it
> will help pin-point the problem (clock44xx_data.c changes ?).