[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <DB6PR0501MB270922D61D8F228A31A664CDB3420@DB6PR0501MB2709.eurprd05.prod.outlook.com>
Date: Thu, 21 Mar 2019 14:30:55 +0000
From: Junhan Zhou <Junhan@...lanox.com>
To: Randy Dunlap <rdunlap@...radead.org>,
Borislav Petkov <bp@...en8.de>,
Mauro Carvalho Chehab <mchehab@...nel.org>
CC: Liming Sun <lsun@...lanox.com>,
"linux-edac@...r.kernel.org" <linux-edac@...r.kernel.org>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: RE: [PATCH v2] EDAC, mellanox: Add ECC support for BlueField DDR4
> > +config EDAC_BLUEFIELD
> > + tristate "Mellanox BlueField Memory ECC"
> > + depends on (MELLANOX_PLATFORM && ARM64 && ACPI) ||
> COMPILE_TEST
>
> Hi,
>
> While I am in favor of using COMPILE_TEST whenever possible, I don't see
> how that is possible here if:
>
> # MELLANOX_PLATFORM is not set
> # ARM64 is not set
> # ACPI is not set
> COMPILE_TEST=y
>
> Can you please explain?
>
> Thanks.
>
> ~Randy
Good point. I tried compiling it against the x86 architecture and it fails because it can't find the definition for arm_smccc_smc(). But it would compile for the ARM64 architecture with both ACPI and MELLANOX _PLATFORM not set, so I'll change it to ARM64 && ((MELLANOX_PLATFORM && ACPI) || COMPILE_TEST) instead.
The dependency on ACPI is because our BlueField SoC firmware only uses ACPI to pass the device mappings, and MELLANOX_PLATFORM because this device is only seen on the Mellanox BlueField SoC so it would be pointless to be used elsewhere, but these doesn't prevent it from being compiled.
-Junhan
Powered by blists - more mailing lists