From mboxrd@z Thu Jan 1 00:00:00 1970 From: Frank Rowand Subject: Re: [PATCH] tty: serial: msm_serial: Use DT aliases Date: Mon, 10 Nov 2014 17:56:03 -0800 Message-ID: <54616CB3.5000405@gmail.com> References: <1414024381-12204-1-git-send-email-sboyd@codeaurora.org> <5461150B.2010907@codeaurora.org> Reply-To: frowand.list@gmail.com Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <5461150B.2010907@codeaurora.org> Sender: linux-arm-msm-owner@vger.kernel.org To: Stephen Boyd Cc: Kevin Hilman , Greg Kroah-Hartman , lkml , linux-arm-msm , "linux-arm-kernel@lists.infradead.org" , linux-serial@vger.kernel.org, Olof Johansson , Arnd Bergmann , Tyler Baker List-Id: linux-serial@vger.kernel.org On 11/10/2014 11:42 AM, Stephen Boyd wrote: > On 11/10/2014 10:54 AM, Kevin Hilman wrote: >> On Wed, Oct 22, 2014 at 5:33 PM, Stephen Boyd wrote: >>> We rely on probe order of this driver to determine the line number for >>> the uart port. This makes it impossible to know the line number >>> when these devices are populated via DT. Use the DT alias >>> mechanism to assign the line based on the aliases node. >>> >>> Signed-off-by: Stephen Boyd >> FYI... this patch hit linux-next and caused multiple boot failures on >> qcom platforms[1] as of next-20141110. I'm assuming this is because >> the corresponding DTS changes have not hit linux-next yet. >> >> Kevin >> >> [1] http://status.armcloud.us/boot/?qcom > > Hmm the intention was to make it optional so that dts changes aren't > necessary unless you want deterministic numbering. I screwed that up > badly :/ Thanks for finding this. > > Greg, can you also apply this patch or squash it into the bad one? > > ----8<----- > > From: Stephen Boyd > Subject: [PATCH] tty: serial: msm_serial: Don't required DT aliases > > If there isn't a DT alias then of_alias_get_id() will return > -ENODEV. This will cause the msm_serial driver to fail probe, > when we want to keep the previous behavior where we generated a > dynamic line number at probe time. Restore this behavior by > generating a dynamic id if the line number is still negative > after checking for an alias or (in the non-DT case) looking at the > .id field of the platform device. > > Reported-by: Kevin Hilman > Signed-off-by: Stephen Boyd > --- > drivers/tty/serial/msm_serial.c | 6 +++--- > 1 file changed, 3 insertions(+), 3 deletions(-) > > diff --git a/drivers/tty/serial/msm_serial.c b/drivers/tty/serial/msm_serial.c > index 09364dd8cf3a..d1bc6b6cbc70 100644 > --- a/drivers/tty/serial/msm_serial.c > +++ b/drivers/tty/serial/msm_serial.c > @@ -1046,14 +1046,14 @@ static int msm_serial_probe(struct platform_device *pdev) > const struct of_device_id *id; > int irq, line; > > - if (pdev->id == -1) > - pdev->id = atomic_inc_return(&msm_uart_next_id) - 1; > - > if (pdev->dev.of_node) > line = of_alias_get_id(pdev->dev.of_node, "serial"); > else > line = pdev->id; > > + if (line < 0) > + line = atomic_inc_return(&msm_uart_next_id) - 1; > + > if (unlikely(line < 0 || line >= UART_NR)) Then this original check for "line < 0" can also be removed. > return -ENXIO; > >