Re: [Minios-devel] [UNIKRAFT PATCH v4 2/9] plat/linuxu: Add scheduling support for x86_64

[Costin Lupu <costin.lup@xxxxxxxxx>]

Hi Florian,

On 08/30/2018 04:51 PM, Florian Schmidt wrote:
> Hi Costin, Sharan,
> 
> On 08/27/2018 02:43 PM, Costin Lupu wrote:
>> On 08/24/2018 04:03 PM, Sharan Santhanam wrote:
>>> Not related to this patch but ARCH_X86_32 replaced with
>>> CONFIG_ARCH_X86_32
>>> Not related to this patch but ARCH_ARM_64 replaced with
>>> CONFIG_ARCH_ARM_64
> 
> I'll change that.
> 
>>>> +ifeq ($(CONFIG_HAVE_SCHED),y)
>>>> +LIBLINUXUPLAT_SRCS-$(CONFIG_ARCH_X86_64) +=
>>>> $(UK_PLAT_COMMON_BASE)/x86/thread_start.S|common
>>>> +LIBLINUXUPLAT_SRCS-$(CONFIG_ARCH_X86_64) +=
>>>> $(UK_PLAT_COMMON_BASE)/thread.c|common
>>>> +LIBLINUXUPLAT_SRCS-$(CONFIG_ARCH_X86_64) +=
>>>> $(UK_PLAT_COMMON_BASE)/sw_ctx.c|common
>>>
>>> Why are thread.c and sw_ctx.c specific to the X86_64?
>>
>> You're right, they aren't.
>>
>>>
>>> If we only support x86_64 platform for now it might be wise to mention
>>> it as a comment or else it is better to add those files to the x86
>>> folder?
>>
>> For sure these files should stay in the common folder. That flag there
>> was set like that because at that time it was very fuzzy regarding the
>> ARM support on linuxu. Besides that, after adding ARM support that flag
>> would be changed to 'y' anyhow.
> 
> So... should these be set to LIBLINUXPLAT_SRCS-y then, because they're
> not x86-specific? Seems the right thing to do IMHO, and then deal with
> potential fallout of that later? Because as it stands (with
> LIBLINUXPLAT_SRCS--$(CONFIG_ARCH_X86_64)), it probably won't compile at
> all on ARM anyway if scheduling is enabled?

Right, let's replace it with 'y'.

Costin

_______________________________________________
Minios-devel mailing list
Minios-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/minios-devel