[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID:
<BL1PR12MB5849EFA99B7F0C55D201738AE7792@BL1PR12MB5849.namprd12.prod.outlook.com>
Date: Fri, 11 Oct 2024 09:20:44 +0000
From: "Chen, Jiqian" <Jiqian.Chen@....com>
To: Jan Beulich <jbeulich@...e.com>
CC: "xen-devel@...ts.xenproject.org" <xen-devel@...ts.xenproject.org>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
Marek Marczykowski-Górecki
<marmarek@...isiblethingslab.com>, Juergen Gross <jgross@...e.com>, Stefano
Stabellini <sstabellini@...nel.org>, Oleksandr Tyshchenko
<oleksandr_tyshchenko@...m.com>, "Chen, Jiqian" <Jiqian.Chen@....com>
Subject: Re: [PATCH v3] xen: Remove dependency between pciback and privcmd
On 2024/10/11 16:54, Jan Beulich wrote:
> On 11.10.2024 05:42, Jiqian Chen wrote:
>> @@ -1757,11 +1756,19 @@ static int __init xen_pcibk_init(void)
>> bus_register_notifier(&pci_bus_type, &pci_stub_nb);
>> #endif
>>
>> +#ifdef CONFIG_XEN_ACPI
>> + xen_acpi_register_get_gsi_func(pcistub_get_gsi_from_sbdf);
>> +#endif
>> +
>> return err;
>> }
>>
>> static void __exit xen_pcibk_cleanup(void)
>> {
>> +#ifdef CONFIG_XEN_ACPI
>> + xen_acpi_register_get_gsi_func(NULL);
>> +#endif
>
> Just wondering - instead of these two #ifdef-s, ...
>
>> --- a/include/xen/acpi.h
>> +++ b/include/xen/acpi.h
>> @@ -91,13 +91,9 @@ static inline int xen_acpi_get_gsi_info(struct pci_dev *dev,
>> }
>> #endif
>>
>> -#ifdef CONFIG_XEN_PCI_STUB
>> -int pcistub_get_gsi_from_sbdf(unsigned int sbdf);
>> -#else
>> -static inline int pcistub_get_gsi_from_sbdf(unsigned int sbdf)
>> -{
>> - return -1;
>> -}
>> -#endif
>> +typedef int (*get_gsi_from_sbdf_t)(u32 sbdf);
>> +
>> +void xen_acpi_register_get_gsi_func(get_gsi_from_sbdf_t func);
>> +int xen_acpi_get_gsi_from_sbdf(u32 sbdf);
>
> ... wouldn't a static inline stub (for the !XEN_ACPI case) aid overall readability?
I'm not sure if other files do this. But for me, it feels a little strange to use "#ifdef CONFIG_XEN_ACPI #else" in apci.h, like self-containment.
And "#include apci.h" in pic_stub.c is also wraped with CONFIG_XEN_ACPI.
>
> Jan
--
Best regards,
Jiqian Chen.
Powered by blists - more mailing lists