[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <87cy87zx87.ffs@tglx>
Date: Wed, 03 Sep 2025 11:44:40 +0200
From: Thomas Gleixner <tglx@...utronix.de>
To: Guillaume La Roque <glaroque@...libre.com>, vigneshr@...com, Nishanth
Menon <nm@...com>, Santosh Shilimkar <ssantosh@...nel.org>, Tero Kristo
<kristo@...nel.org>
Cc: linux-kernel@...r.kernel.org, linux-arm-kernel@...ts.infradead.org,
Guillaume La Roque <glaroque@...libre.com>
Subject: Re: [PATCH RESEND v2 2/2] soc: ti: ti_sci_inta_msi: Enable module
compilation support
On Tue, Sep 02 2025 at 16:43, Guillaume La Roque wrote:
> Add module support to the TI SCI INTA MSI driver:
> - Change Kconfig from bool to tristate to allow module compilation
> - Add linux/module.h include for module functionality
> - Add MODULE_LICENSE, MODULE_DESCRIPTION, and MODULE_AUTHOR macros
>
> This allows the driver to be compiled as a loadable kernel module
> named ti_sci_inta_msi.
No objection per se, but looking at this "driver" just makes me
cry. There is zero justification for this to be separate from the actual
ti-sci-inta irq chip driver, which selects this monstrosity
unconditionally.
All of this code can be moved into the irq chip driver and thereby
consolidated to do the chip initialization and the domain fiddling in
the related data structures instead of doing it programatically.
Consolidating it reduces the number of exports and makes it one module.
No?
Thanks,
tglx
Powered by blists - more mailing lists