From mboxrd@z Thu Jan 1 00:00:00 1970 From: Baruch Siach Subject: Re: [PATCH 2/2] PCI: armada8k: don't toggle reset twice Date: Sun, 13 Jan 2019 20:42:39 +0200 Message-ID: <87o98kv1pc.fsf@tarshish> References: <024cc24efa7b99186750f90c91880b29357d379d.1547123182.git.baruch@tkos.co.il> <94cd23a60c647020dd87a923684b59255b89f02c.1547123182.git.baruch@tkos.co.il> <20190110125548.GA29612@lunn.ch> <8736q0irwy.fsf@tarshish> <20190110131912.GB29612@lunn.ch> <87zhs8h5ej.fsf@tarshish> <87pnt0vil3.fsf@tarshish> <20190113154028.GB4635@lunn.ch> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: In-reply-to: <20190113154028.GB4635@lunn.ch> List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: "linux-arm-kernel" Errors-To: linux-arm-kernel-bounces+linux-arm-kernel=m.gmane.org@lists.infradead.org To: Andrew Lunn Cc: linux-pwm@vger.kernel.org, Lorenzo Pieralisi , Sven Auhagen , Jason Cooper , Bartosz Golaszewski , linux-pci@vger.kernel.org, Linus Walleij , linux-gpio@vger.kernel.org, Thierry Reding , Thomas Petazzoni , Bjorn Helgaas , Gregory Clement , linux-arm-kernel@lists.infradead.org, Sebastian Hesselbarth List-Id: linux-gpio@vger.kernel.org Hi Andrew, On Sun, Jan 13 2019, Andrew Lunn wrote: >> Another look at this comment made me realize that we need comphy >> initialization support in the kernel for PCIe reset to work correctly. >> >> The workaround that this patch proposes will not solve the problem for >> v5.0, since the GPIO get_direction patch will only appear in v5.1. > > You could ask for it to be applied to v5.0 since a real fix depends on > it. > > We have a bit of a balancing act to sort out. The revert will fix > Macchiatobin. But i assume it breaks some other platforms which just > started to work. Ideally we want the best of both worlds. So maybe > getting the GPIO change into stable is the correct thing to do? Support for that other board (Clearfog GT-8K) has only been introduced in v5.0. The kernel never supported PCIe on that platform. So there is no regression for CF GT-8K like there is for the Macchiatobin. The issue is quite easy to deal with at the U-Boot level even for v5.0. The workaround that this patch suggests is pretty ugly, as you noted. I think we can live with no PCIe reset support in the kernel for another release or two. baruch -- http://baruch.siach.name/blog/ ~. .~ Tk Open Systems =}------------------------------------------------ooO--U--Ooo------------{= - baruch@tkos.co.il - tel: +972.52.368.4656, http://www.tkos.co.il -