user-mode-linux-devel

Re: [uml-devel] [PATCH 02/21] usb: add HAS_IOMEM dependency to USB_NET2272

[Vegard N. <veg...@or...>]

On 02/10/2016 05:15 PM, Greg Kroah-Hartman wrote:
> On Wed, Feb 10, 2016 at 03:29:37PM +0100, Vegard Nossum wrote:
>> drivers/usb/gadget/udc/net2272.c: In function ‘net2272_remove’:
>> drivers/usb/gadget/udc/net2272.c:2232:2: error: implicit declaration of function ‘iounmap’ [-Werror=implicit-function-declaration]
>>    iounmap(dev->base_addr);
>>    ^
>> drivers/usb/gadget/udc/net2272.c: In function ‘net2272_plat_probe’:
>> drivers/usb/gadget/udc/net2272.c:2650:2: error: implicit declaration of function ‘ioremap_nocache’ [-Werror=implicit-function-declaration]
>>    dev->base_addr = ioremap_nocache(base, len);
>>    ^
>> drivers/usb/gadget/udc/net2272.c:2650:17: warning: assignment makes pointer from integer without a cast [enabled by default]
>>    dev->base_addr = ioremap_nocache(base, len);
>>                   ^
>>
>> Signed-off-by: Vegard Nossum <veg...@or...>
>> ---
>>   drivers/usb/gadget/udc/Kconfig | 1 +
>>   1 file changed, 1 insertion(+)
>>
>> diff --git a/drivers/usb/gadget/udc/Kconfig b/drivers/usb/gadget/udc/Kconfig
>> index 753c29b..ca19f6f 100644
>> --- a/drivers/usb/gadget/udc/Kconfig
>> +++ b/drivers/usb/gadget/udc/Kconfig
>> @@ -287,6 +287,7 @@ config USB_FSL_QE
>>   	   dynamically linked module called "fsl_qe_udc".
>>
>>   config USB_NET2272
>> +	depends on HAS_IOMEM
>
> Why not fix the root of the problem and provide the correct functions
> for this when HAS_IOMEM is not enabled?

I don't think there is a "correct function" for when HAS_IOMEM is not
enabled. There is no IO address space on UML, so it doesn't make sense
to compile these drivers in the first place.

Or do you mean to use the dummy implementation from asm-generic/io.h? (I
have to admit I don't know how that would work.)


Vegard

Re: [uml-devel] [PATCH 02/21] usb: add HAS_IOMEM dependency to USB_NET2272

[Richard W. <ri...@no...>]

Am 10.02.2016 um 17:28 schrieb Vegard Nossum:
> I don't think there is a "correct function" for when HAS_IOMEM is not
> enabled. There is no IO address space on UML, so it doesn't make sense
> to compile these drivers in the first place.
> 
> Or do you mean to use the dummy implementation from asm-generic/io.h? (I
> have to admit I don't know how that would work.)

We already had an discussion of having an ioremap() like this for UML and S390:

static inline void __iomem *ioremap(phys_addr_t offset, size_t size)
{
	BUG();
	return NULL;
}

But IMHO we should just fix the driver dependencies instead of providing dummy
interfaces just for the sake of making things somehow build...

Thanks,
//richard

Re: [uml-devel] [PATCH 20/21] usb: support building without CONFIG_HAS_DMA

[Greg Kroah-H. <gr...@li...>]

On Wed, Feb 10, 2016 at 03:29:55PM +0100, Vegard Nossum wrote:
> Some platforms don't have DMA, but we should still be able to build
> USB drivers for these platforms. They could still be used through
> vhci_hcd, usbip_host, or maybe something like USB passthrough in UML
> from a capable host.
> 
> This is admittedly ugly with all the #ifdefs, but it is necessary to
> get around linker errors like these:

Sorry, you are going to have to do something else here, I can't accept
this patch, it's just too ugly to read :(

Re: [uml-devel] [PATCH 20/21] usb: support building without CONFIG_HAS_DMA

[Vegard N. <veg...@or...>]

On 02/15/2016 02:09 AM, Greg Kroah-Hartman wrote:
> On Wed, Feb 10, 2016 at 03:29:55PM +0100, Vegard Nossum wrote:
>> Some platforms don't have DMA, but we should still be able to build
>> USB drivers for these platforms. They could still be used through
>> vhci_hcd, usbip_host, or maybe something like USB passthrough in UML
>> from a capable host.
>>
>> This is admittedly ugly with all the #ifdefs, but it is necessary to
>> get around linker errors like these:
>
> Sorry, you are going to have to do something else here, I can't accept
> this patch, it's just too ugly to read :(
>

Ok, I'll work on an alternative. Thanks for the feedback ;-)


Vegard