From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S933203AbeCOQEA convert rfc822-to-8bit (ORCPT ); Thu, 15 Mar 2018 12:04:00 -0400 Received: from coyote.holtmann.net ([212.227.132.17]:43224 "EHLO mail.holtmann.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932094AbeCOQDz (ORCPT ); Thu, 15 Mar 2018 12:03:55 -0400 Content-Type: text/plain; charset=utf-8 Mime-Version: 1.0 (Mac OS X Mail 11.2 \(3445.5.20\)) Subject: Re: [PATCH] Bluetooth: btrsi: rework dependencies From: Marcel Holtmann In-Reply-To: Date: Thu, 15 Mar 2018 17:03:52 +0100 Cc: Johan Hedberg , Kalle Valo , Linux Bluetooth mailing list , LKML , linux-wireless , Networking Content-Transfer-Encoding: 8BIT Message-Id: References: <20180315154312.3941446-1-arnd@arndb.de> To: Arnd Bergmann X-Mailer: Apple Mail (2.3445.5.20) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Hi Arnd, >>> diff --git a/drivers/bluetooth/Makefile b/drivers/bluetooth/Makefile >>> index 03cfc1b20c4a..9e8d22712ff3 100644 >>> --- a/drivers/bluetooth/Makefile >>> +++ b/drivers/bluetooth/Makefile >>> @@ -28,7 +28,7 @@ obj-$(CONFIG_BT_QCA) += btqca.o >>> >>> obj-$(CONFIG_BT_HCIUART_NOKIA) += hci_nokia.o >>> >>> -obj-$(CONFIG_BT_HCIRSI) += btrsi.o >>> +obj-$(CONFIG_BT_HCIRSI_MODULE) += btrsi.o >> >> do we need this new option? I have avoided these kind of complex things multi config entries. Can we not just select the RSI_91X? >> > > I couldn't come up with a simpler way to do this. > Selecting RSI_91X is not possible unless we make the BT > driver 'depend on WLAN_VENDOR_RSI && MAC80211', > which is even more backwards. > > The problem here is that it's actually a reverse dependency: > the wlan driver calls into the bt driver. > > What we could do is to make BT_HCIRSI a silent symbol > and have that selected by RSI_COEX, which can then > be user-visible. With that, the Kconfig structure would follow > what the code does. that sounds a bit better to me. If the RSI driver maintainers don’t like that, then they should re-architect the code to make this more dynamic and flexible. Regards Marcel