Received-SPF: pass (google.com: domain of linux-kernel+bounces-62008-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:4601:e00::3 as permitted sender) client-ip=2604:1380:4601:e00::3;
Date: Mon, 12 Feb 2024 18:49:49 +0200
From: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
To: Robin van der Gracht <robin@protonic.nl>
Cc: devicetree@vger.kernel.org, linux-kernel@vger.kernel.org,
	Miguel Ojeda <ojeda@kernel.org>, Rob Herring <robh@kernel.org>,
	Krzysztof Kozlowski <krzysztof.kozlowski+dt@linaro.org>,
	Conor Dooley <conor+dt@kernel.org>,
	Paul Burton <paulburton@kernel.org>,
	Geert Uytterhoeven <geert+renesas@glider.be>
Subject: Re: [PATCH v1 10/15] auxdisplay: linedisp: Provide a small buffer in
 the struct linedisp
Message-ID: <ZcpMLXTfc95fGTae@smile.fi.intel.com>
References: <20240208184919.2224986-1-andriy.shevchenko@linux.intel.com>
 <20240208184919.2224986-11-andriy.shevchenko@linux.intel.com>
 <20240212092500.62f006cc@ERD993>
 <ZcoF9ZxPBkVS_6Da@smile.fi.intel.com>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <ZcoF9ZxPBkVS_6Da@smile.fi.intel.com>
Organization: Intel Finland Oy - BIC 0357606-4 - Westendinkatu 7, 02160 Espoo

On Mon, Feb 12, 2024 at 01:50:13PM +0200, Andy Shevchenko wrote:
> On Mon, Feb 12, 2024 at 09:25:00AM +0100, Robin van der Gracht wrote:
> > On Thu,  8 Feb 2024 20:48:08 +0200
> > Andy Shevchenko <andriy.shevchenko@linux.intel.com> wrote:
> 
> > > +	linedisp->num_chars = buf ? num_chars : min(num_chars, LINEDISP_DEFAULT_BUF_SZ);
> > 
> > It's not a big buffer, but now it's always there even if it's not used.
> > And even if it's used, it might be only partially used.
> > Why not used a malloc instead?
> 
> malloc() infra takes more than this IIRC (something like up to 32 bytes on
> 64-bit platforms) or comparable sizes. Yes, the malloc() along with the
> linedisp structure might make sense, but will require more invasive change.
> 
> Do you want me to drop this one from the set?
> (I have no hard feelings about it, as I see better way and just having no
>  time for taking care about, as it's not the main point of the series.)

Looking again into it, the allocation separately with linedisp structure
is indeed too much invasive. I prefer (as we save lines of code and deduplicate
the buffer at least for two drivers, including a new one) to leave this patch
for now. We may rework it later on. Do you agree?

-- 
With Best Regards,
Andy Shevchenko