From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jean Delvare Subject: Re: [PATCH] i2c-i801: Don't depend on other kernel driver config options Date: Sun, 15 May 2011 13:12:28 +0200 Message-ID: <20110515131228.59e14178@endymion.delvare> References: <20110513120100.4d47410a@endymion.delvare> <1305282187.30435.343.camel@i7.infradead.org> Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <1305282187.30435.343.camel-W2I5cNIroUsVm/YvaOjsyQ@public.gmane.org> Sender: linux-i2c-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org To: "Woodhouse, David" Cc: Linux I2C , Hans de Goede List-Id: linux-i2c@vger.kernel.org Hi David, On Fri, 13 May 2011 11:23:06 +0100, Woodhouse, David wrote: > On Fri, 2011-05-13 at 11:01 +0100, Jean Delvare wrote: > > Don't let other driver config options influence us, as it makes the > > code more complex and fragile for a small benefit. There's nothing > > wrong with instantiating I2C devices even if they don't have a driver. > > And we're talking about 835 extra bytes in the binary on x86-64, > > that's hardly worth arguing about. > > Looks sane to me; thanks. Should it be CONFIG_DMI instead of (or in > addition to) CONFIG_X86? Instead of, no. IA64 has DMI but doesn't want this code. In addition to, yes, that would make sense. I expected the extra code to transparently vanish when CONFIG_DMI isn't set, but gcc is apparently not smart enough for this, so it could use some help. BTW, it may be desirable to make the apanel part depend on DMI, even if DMI is technically not needed to find the slave address. Scanning the BIOS memory for an arbitrary string is costly (a minimum of 4096 calls to check_signature(), which in turn calls readb() at least once, in the negative case.) Checking for the DMI vendor first should be way less costly for the negative case. I'll submit a new patch set later today, thanks for the suggestions. -- Jean Delvare