From mboxrd@z Thu Jan 1 00:00:00 1970 Date: Sun, 15 Jan 2017 15:33:45 -0800 (PST) From: jmrodriguezhw88@gmail.com Message-Id: In-Reply-To: <1484094731-15965-1-git-send-email-logang@deltatee.com> References: <1484094731-15965-1-git-send-email-logang@deltatee.com> Subject: Re: [PATCH 0/2] Style fixes: open code obfuscating macros MIME-Version: 1.0 Content-Type: multipart/mixed; boundary="----=_Part_4321_99687317.1484523225270" To: linux-ntb Cc: Xiangliang.Yu@amd.com, jdmason@kudzu.us, dave.jiang@intel.com, Allen.Hubbe@emc.com, joe@perches.com, gregkh@linuxfoundation.org, linux-kernel@vger.kernel.org, logang@deltatee.com List-ID: ------=_Part_4321_99687317.1484523225270 Content-Type: multipart/alternative; boundary="----=_Part_4322_1055698061.1484523225271" ------=_Part_4322_1055698061.1484523225271 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 7bit On Tuesday, January 10, 2017 at 4:33:17 PM UTC-8, Logan Gunthorpe wrote: > > Hi, > > I had copied some poor code style from the NTB drivers into an unrelated > driver. Upon review of my new code, I learned it was not a good idea > to sweep dirty things under the rug^W macro. See [1], where Gregg k-h > suggested I fix the problem at the source. > > Thus, please merge the following two patches to improve the style of > the NTB drivers. > > Thanks, > > Logan > > [1] http://www.spinics.net/lists/linux-pci/msg56904.html > > > Logan Gunthorpe (2): > ntb_hw_amd: Style fixes: open code macros that just obfuscate code > ntb_hw_intel: Style fixes: open code macros that just obfuscate code > > drivers/ntb/hw/amd/ntb_hw_amd.c | 59 ++++++++------- > drivers/ntb/hw/amd/ntb_hw_amd.h | 3 - > drivers/ntb/hw/intel/ntb_hw_intel.c | 145 > ++++++++++++++++++------------------ > drivers/ntb/hw/intel/ntb_hw_intel.h | 3 - > 4 files changed, 104 insertions(+), 106 deletions(-) > > -- > 2.1.4 > On Tuesday, January 10, 2017 at 4:33:17 PM UTC-8, Logan Gunthorpe wrote: > > Hi, > > I had copied some poor code style from the NTB drivers into an unrelated > driver. Upon review of my new code, I learned it was not a good idea > to sweep dirty things under the rug^W macro. See [1], where Gregg k-h > suggested I fix the problem at the source. > > Thus, please merge the following two patches to improve the style of > the NTB drivers. > > Thanks, > > Logan > > [1] http://www.spinics.net/lists/linux-pci/msg56904.html > > > Logan Gunthorpe (2): > ntb_hw_amd: Style fixes: open code macros that just obfuscate code > ntb_hw_intel: Style fixes: open code macros that just obfuscate code > > drivers/ntb/hw/amd/ntb_hw_amd.c | 59 ++++++++------- > drivers/ntb/hw/amd/ntb_hw_amd.h | 3 - > drivers/ntb/hw/intel/ntb_hw_intel.c | 145 > ++++++++++++++++++------------------ > drivers/ntb/hw/intel/ntb_hw_intel.h | 3 - > 4 files changed, 104 insertions(+), 106 deletions(-) > > -- > 2.1.4 > ------=_Part_4322_1055698061.1484523225271 Content-Type: text/html; charset=utf-8 Content-Transfer-Encoding: quoted-printable


On Tuesday, January 10, 2017 at 4:33:17 PM UTC-8, = Logan Gunthorpe wrote:
Hi,

I had copied some poor code style from the NTB drivers into an unrelate= d
driver. Upon review of my new code, I learned it was not a good idea
to sweep dirty things under the rug^W macro. See [1], where Gregg k-h
suggested I fix the problem at the source.

Thus, please merge the following two patches to improve the style of
the NTB drivers.

Thanks,

Logan

[1] http://www.spinics.net/lists/linux-pci/msg56904.html


Logan Gunthorpe (2):
=C2=A0 ntb_hw_amd: Style fixes: open code macros that just obfuscate co= de
=C2=A0 ntb_hw_intel: Style fixes: open code macros that just obfuscate = code

=C2=A0drivers/ntb/hw/amd/ntb_hw_amd.c =C2=A0 =C2=A0 | =C2=A059 +++= +++++-------
=C2=A0drivers/ntb/hw/amd/ntb_hw_amd.h =C2=A0 =C2=A0 | =C2=A0 3 -
=C2=A0drivers/ntb/hw/intel/ntb_hw_intel.c | 145 ++++++++++++++++++= ------------------
=C2=A0drivers/ntb/hw/intel/ntb_hw_intel.h | =C2=A0 3 -
=C2=A04 files changed, 104 insertions(+), 106 deletions(-)

--
2.1.4

On Tuesday, January 10, 2017 at 4:33:17 PM UTC-8, Loga= n Gunthorpe wrote:
Hi,

I had copied some poor code style from the NTB drivers into an unrelate= d
driver. Upon review of my new code, I learned it was not a good idea
to sweep dirty things under the rug^W macro. See [1], where Gregg k-h
suggested I fix the problem at the source.

Thus, please merge the following two patches to improve the style of
the NTB drivers.

Thanks,

Logan

[1] http://www.spinics.net/lists/linux-pci/msg56904.html


Logan Gunthorpe (2):
=C2=A0 ntb_hw_amd: Style fixes: open code macros that just obfuscate co= de
=C2=A0 ntb_hw_intel: Style fixes: open code macros that just obfuscate = code

=C2=A0drivers/ntb/hw/amd/ntb_hw_amd.c =C2=A0 =C2=A0 | =C2=A059 +++= +++++-------
=C2=A0drivers/ntb/hw/amd/ntb_hw_amd.h =C2=A0 =C2=A0 | =C2=A0 3 -
=C2=A0drivers/ntb/hw/intel/ntb_hw_intel.c | 145 ++++++++++++++++++= ------------------
=C2=A0drivers/ntb/hw/intel/ntb_hw_intel.h | =C2=A0 3 -
=C2=A04 files changed, 104 insertions(+), 106 deletions(-)

--
2.1.4
------=_Part_4322_1055698061.1484523225271-- ------=_Part_4321_99687317.1484523225270-- From mboxrd@z Thu Jan 1 00:00:00 1970 Date: Sun, 15 Jan 2017 14:38:15 -0800 (PST) From: jmrodriguezhw88@gmail.com Message-Id: <9a71e132-ec50-4e5d-bd29-1ad735acec07@googlegroups.com> In-Reply-To: <1483407036-20697-1-git-send-email-logang@deltatee.com> References: <1483407036-20697-1-git-send-email-logang@deltatee.com> Subject: Re: [PATCH 0/2] Style fixes: open code obfuscating macros MIME-Version: 1.0 Content-Type: multipart/mixed; boundary="----=_Part_4493_1823960954.1484519895838" To: linux-ntb Cc: Xiangliang.Yu@amd.com, jdmason@kudzu.us, dave.jiang@intel.com, Allen.Hubbe@emc.com, gregkh@linuxfoundation.org, linux-kernel@vger.kernel.org, logang@deltatee.com List-ID: ------=_Part_4493_1823960954.1484519895838 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 7bit On Monday, January 2, 2017 at 5:31:15 PM UTC-8, Logan Gunthorpe wrote: > Hi, > > I had copied some poor code style from the NTB drivers into an unrelated > driver. Upon review of my new code, I learned it was not a good idea > to sweep dirty things under the rug^W macro. See [1], where Gregg k-h > suggested I fix the problem at the source. > > Thus, please merge the following two patches to improve the style of > the NTB drivers. > > Thanks, > > Logan > > [1] http://www.spinics.net/lists/linux-pci/msg56904.html > > > Logan Gunthorpe (2): > ntb_hw_amd: Style fixes: open code macros that just obfuscate code > ntb_hw_intel: Style fixes: open code macros that just obfuscate code > > drivers/ntb/hw/amd/ntb_hw_amd.c | 59 ++++++++------- > drivers/ntb/hw/amd/ntb_hw_amd.h | 3 - > drivers/ntb/hw/intel/ntb_hw_intel.c | 145 ++++++++++++++++++------------------ > drivers/ntb/hw/intel/ntb_hw_intel.h | 3 - > 4 files changed, 104 insertions(+), 106 deletions(-) > > -- > 2.1.4 ------=_Part_4493_1823960954.1484519895838-- From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from ale.deltatee.com (ale.deltatee.com. [207.54.116.67]) by gmr-mx.google.com with ESMTPS id r1si1412032pfg.4.2017.01.11.09.48.55 for (version=TLS1_2 cipher=ECDHE-RSA-AES128-GCM-SHA256 bits=128/128); Wed, 11 Jan 2017 09:48:56 -0800 (PST) References: <1483407036-20697-1-git-send-email-logang@deltatee.com> <20170110215527.GD13381@kudzu.us> <1484089612.12006.11.camel@perches.com> <20170111065036.GA4111@kroah.com> From: Logan Gunthorpe Message-ID: Date: Wed, 11 Jan 2017 10:48:51 -0700 MIME-Version: 1.0 In-Reply-To: <20170111065036.GA4111@kroah.com> Content-Type: text/plain; charset=windows-1252 Content-Transfer-Encoding: 7bit Subject: Re: [PATCH 0/2] Style fixes: open code obfuscating macros To: Greg Kroah-Hartman Cc: Joe Perches , Jon Mason , Xiangliang Yu , Dave Jiang , Allen Hubbe , linux-ntb@googlegroups.com, linux-kernel List-ID: On 10/01/17 11:50 PM, Greg Kroah-Hartman wrote: > You should be printing against the ntb struct device, as that's the > "correct" structure here, and not the PCI device at all, as that's too > far up the device chain from what the driver is supposed to be doing. Yup, agreed and thanks. My v2 patchset already makes this change. Logan From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail.linuxfoundation.org (mail.linuxfoundation.org. [140.211.169.12]) by gmr-mx.google.com with ESMTPS id r8si1040281pfa.5.2017.01.10.22.50.15 for (version=TLS1_2 cipher=ECDHE-RSA-AES128-GCM-SHA256 bits=128/128); Tue, 10 Jan 2017 22:50:15 -0800 (PST) Date: Wed, 11 Jan 2017 07:50:36 +0100 From: Greg Kroah-Hartman Subject: Re: [PATCH 0/2] Style fixes: open code obfuscating macros Message-ID: <20170111065036.GA4111@kroah.com> References: <1483407036-20697-1-git-send-email-logang@deltatee.com> <20170110215527.GD13381@kudzu.us> <1484089612.12006.11.camel@perches.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: To: Logan Gunthorpe Cc: Joe Perches , Jon Mason , Xiangliang Yu , Dave Jiang , Allen Hubbe , linux-ntb@googlegroups.com, linux-kernel List-ID: On Tue, Jan 10, 2017 at 04:34:41PM -0700, Logan Gunthorpe wrote: > > > On 10/01/17 04:06 PM, Joe Perches wrote: > > It might also be better to add and use YA > > logging macro like ndev_dbg > > > > #define ndev_dvg(ndev, fmt, ...) \ > > dev_dbg(&(ndev)->ntb.pdev->dev, fmt, ##__VA_ARGS__) > > > > so that the new temporaries that are used just > > for debugging are not required. > > > > I feel like that's going in the exact opposite direction from what Greg > was looking for... I think the point is to hide less of where the device > is coming from, not to create a new and different macro that hides > things in a different way. > > Or, more likely, I _think_ Greg would suggest that this is showing that > the driver model isn't really being used properly. Perhaps, we should be > printing against the ntb struct device instead of the pci device so > there is no dereferencing. You should be printing against the ntb struct device, as that's the "correct" structure here, and not the PCI device at all, as that's too far up the device chain from what the driver is supposed to be doing. thanks, greg k-h From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from ale.deltatee.com (ale.deltatee.com. [207.54.116.67]) by gmr-mx.google.com with ESMTPS id r1si835706pfg.4.2017.01.10.16.34.14 for (version=TLS1_2 cipher=ECDHE-RSA-AES128-GCM-SHA256 bits=128/128); Tue, 10 Jan 2017 16:34:14 -0800 (PST) References: <1484094731-15965-1-git-send-email-logang@deltatee.com> From: Logan Gunthorpe Message-ID: <2f7c2161-5e5b-13b7-4970-4c9c25f0e0c9@deltatee.com> Date: Tue, 10 Jan 2017 17:34:13 -0700 MIME-Version: 1.0 In-Reply-To: <1484094731-15965-1-git-send-email-logang@deltatee.com> Content-Type: text/plain; charset=windows-1252 Content-Transfer-Encoding: 7bit Subject: Re: [PATCH 0/2] Style fixes: open code obfuscating macros To: Xiangliang Yu , Jon Mason , Dave Jiang , Allen Hubbe , Joe Perches Cc: Greg Kroah-Hartman , linux-ntb@googlegroups.com, linux-kernel@vger.kernel.org List-ID: Sorry, sorry. I resent the wrong set :( Logan On 10/01/17 05:32 PM, Logan Gunthorpe wrote: > Hi, > > I had copied some poor code style from the NTB drivers into an unrelated > driver. Upon review of my new code, I learned it was not a good idea > to sweep dirty things under the rug^W macro. See [1], where Gregg k-h > suggested I fix the problem at the source. > > Thus, please merge the following two patches to improve the style of > the NTB drivers. > > Thanks, > > Logan > > [1] http://www.spinics.net/lists/linux-pci/msg56904.html > > > Logan Gunthorpe (2): > ntb_hw_amd: Style fixes: open code macros that just obfuscate code > ntb_hw_intel: Style fixes: open code macros that just obfuscate code > > drivers/ntb/hw/amd/ntb_hw_amd.c | 59 ++++++++------- > drivers/ntb/hw/amd/ntb_hw_amd.h | 3 - > drivers/ntb/hw/intel/ntb_hw_intel.c | 145 ++++++++++++++++++------------------ > drivers/ntb/hw/intel/ntb_hw_intel.h | 3 - > 4 files changed, 104 insertions(+), 106 deletions(-) > > -- > 2.1.4 > From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from ale.deltatee.com (ale.deltatee.com. [207.54.116.67]) by gmr-mx.google.com with ESMTPS id r8si872083pfa.5.2017.01.10.16.33.17 for (version=TLS1_2 cipher=ECDHE-RSA-AES128-GCM-SHA256 bits=128/128); Tue, 10 Jan 2017 16:33:17 -0800 (PST) From: Logan Gunthorpe Date: Tue, 10 Jan 2017 17:32:09 -0700 Message-Id: <1484094731-15965-1-git-send-email-logang@deltatee.com> Subject: [PATCH 0/2] Style fixes: open code obfuscating macros To: Xiangliang Yu , Jon Mason , Dave Jiang , Allen Hubbe , Joe Perches Cc: Greg Kroah-Hartman , linux-ntb@googlegroups.com, linux-kernel@vger.kernel.org, Logan Gunthorpe List-ID: Hi, I had copied some poor code style from the NTB drivers into an unrelated driver. Upon review of my new code, I learned it was not a good idea to sweep dirty things under the rug^W macro. See [1], where Gregg k-h suggested I fix the problem at the source. Thus, please merge the following two patches to improve the style of the NTB drivers. Thanks, Logan [1] http://www.spinics.net/lists/linux-pci/msg56904.html Logan Gunthorpe (2): ntb_hw_amd: Style fixes: open code macros that just obfuscate code ntb_hw_intel: Style fixes: open code macros that just obfuscate code drivers/ntb/hw/amd/ntb_hw_amd.c | 59 ++++++++------- drivers/ntb/hw/amd/ntb_hw_amd.h | 3 - drivers/ntb/hw/intel/ntb_hw_intel.c | 145 ++++++++++++++++++------------------ drivers/ntb/hw/intel/ntb_hw_intel.h | 3 - 4 files changed, 104 insertions(+), 106 deletions(-) -- 2.1.4 From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from ale.deltatee.com (ale.deltatee.com. [207.54.116.67]) by gmr-mx.google.com with ESMTPS id w2si867625pfg.0.2017.01.10.15.34.44 for (version=TLS1_2 cipher=ECDHE-RSA-AES128-GCM-SHA256 bits=128/128); Tue, 10 Jan 2017 15:34:44 -0800 (PST) References: <1483407036-20697-1-git-send-email-logang@deltatee.com> <20170110215527.GD13381@kudzu.us> <1484089612.12006.11.camel@perches.com> From: Logan Gunthorpe Message-ID: Date: Tue, 10 Jan 2017 16:34:41 -0700 MIME-Version: 1.0 In-Reply-To: <1484089612.12006.11.camel@perches.com> Content-Type: text/plain; charset=windows-1252 Content-Transfer-Encoding: 7bit Subject: Re: [PATCH 0/2] Style fixes: open code obfuscating macros To: Joe Perches , Jon Mason Cc: Xiangliang Yu , Dave Jiang , Allen Hubbe , Greg Kroah-Hartman , linux-ntb@googlegroups.com, linux-kernel List-ID: On 10/01/17 04:06 PM, Joe Perches wrote: > It might also be better to add and use YA > logging macro like ndev_dbg > > #define ndev_dvg(ndev, fmt, ...) \ > dev_dbg(&(ndev)->ntb.pdev->dev, fmt, ##__VA_ARGS__) > > so that the new temporaries that are used just > for debugging are not required. > I feel like that's going in the exact opposite direction from what Greg was looking for... I think the point is to hide less of where the device is coming from, not to create a new and different macro that hides things in a different way. Or, more likely, I _think_ Greg would suggest that this is showing that the driver model isn't really being used properly. Perhaps, we should be printing against the ntb struct device instead of the pci device so there is no dereferencing. Logan From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from smtprelay.hostedemail.com (smtprelay0121.hostedemail.com. [216.40.44.121]) by gmr-mx.google.com with ESMTPS id c195si317412itb.0.2017.01.10.15.06.55 for (version=TLS1_2 cipher=ECDHE-RSA-AES128-GCM-SHA256 bits=128/128); Tue, 10 Jan 2017 15:06:55 -0800 (PST) Message-ID: <1484089612.12006.11.camel@perches.com> Subject: Re: [PATCH 0/2] Style fixes: open code obfuscating macros From: Joe Perches Date: Tue, 10 Jan 2017 15:06:52 -0800 In-Reply-To: References: <1483407036-20697-1-git-send-email-logang@deltatee.com> <20170110215527.GD13381@kudzu.us> Content-Type: text/plain; charset="ISO-8859-1" Mime-Version: 1.0 Content-Transfer-Encoding: 7bit To: Jon Mason , Logan Gunthorpe Cc: Xiangliang Yu , Dave Jiang , Allen Hubbe , Greg Kroah-Hartman , linux-ntb@googlegroups.com, linux-kernel List-ID: On Tue, 2017-01-10 at 16:59 -0500, Jon Mason wrote: > On Tue, Jan 10, 2017 at 4:55 PM, Jon Mason wrote: > > On Mon, Jan 02, 2017 at 06:30:34PM -0700, Logan Gunthorpe wrote: > > > Hi, > > > > > > I had copied some poor code style from the NTB drivers into an unrelated > > > driver. Upon review of my new code, I learned it was not a good idea > > > to sweep dirty things under the rug^W macro. See [1], where Gregg k-h > > > suggested I fix the problem at the source. > > > > > > Thus, please merge the following two patches to improve the style of > > > the NTB drivers. > > > > > > Thanks, > > > > > > Logan > > > > Applying the series, thanks > > Gaaa, these don't apply cleanly. Can you rebase on my > https://github.com/jonmason/ntb (branch ntb) and resubmit? It might also be better to add and use YA logging macro like ndev_dbg #define ndev_dvg(ndev, fmt, ...) \ dev_dbg(&(ndev)->ntb.pdev->dev, fmt, ##__VA_ARGS__) so that the new temporaries that are used just for debugging are not required. From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-oi0-x242.google.com (mail-oi0-x242.google.com. [2607:f8b0:4003:c06::242]) by gmr-mx.google.com with ESMTPS id c195si305606itb.0.2017.01.10.14.14.14 for (version=TLS1_2 cipher=ECDHE-RSA-AES128-GCM-SHA256 bits=128/128); Tue, 10 Jan 2017 14:14:14 -0800 (PST) Received: by mail-oi0-x242.google.com with SMTP id u143so23419017oif.3 for ; Tue, 10 Jan 2017 14:14:14 -0800 (PST) MIME-Version: 1.0 In-Reply-To: <20170110215527.GD13381@kudzu.us> References: <1483407036-20697-1-git-send-email-logang@deltatee.com> <20170110215527.GD13381@kudzu.us> From: Jon Mason Date: Tue, 10 Jan 2017 16:59:09 -0500 Message-ID: Subject: Re: [PATCH 0/2] Style fixes: open code obfuscating macros Content-Type: text/plain; charset=UTF-8 To: Logan Gunthorpe Cc: Xiangliang Yu , Dave Jiang , Allen Hubbe , Greg Kroah-Hartman , linux-ntb@googlegroups.com, linux-kernel List-ID: On Tue, Jan 10, 2017 at 4:55 PM, Jon Mason wrote: > On Mon, Jan 02, 2017 at 06:30:34PM -0700, Logan Gunthorpe wrote: >> Hi, >> >> I had copied some poor code style from the NTB drivers into an unrelated >> driver. Upon review of my new code, I learned it was not a good idea >> to sweep dirty things under the rug^W macro. See [1], where Gregg k-h >> suggested I fix the problem at the source. >> >> Thus, please merge the following two patches to improve the style of >> the NTB drivers. >> >> Thanks, >> >> Logan > > Applying the series, thanks Gaaa, these don't apply cleanly. Can you rebase on my https://github.com/jonmason/ntb (branch ntb) and resubmit? Thanks, Jon > >> >> [1] http://www.spinics.net/lists/linux-pci/msg56904.html >> >> >> Logan Gunthorpe (2): >> ntb_hw_amd: Style fixes: open code macros that just obfuscate code >> ntb_hw_intel: Style fixes: open code macros that just obfuscate code >> >> drivers/ntb/hw/amd/ntb_hw_amd.c | 59 ++++++++------- >> drivers/ntb/hw/amd/ntb_hw_amd.h | 3 - >> drivers/ntb/hw/intel/ntb_hw_intel.c | 145 ++++++++++++++++++------------------ >> drivers/ntb/hw/intel/ntb_hw_intel.h | 3 - >> 4 files changed, 104 insertions(+), 106 deletions(-) >> >> -- >> 2.1.4 From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-qt0-x241.google.com (mail-qt0-x241.google.com. [2607:f8b0:400d:c0d::241]) by gmr-mx.google.com with ESMTPS id v77si259268ywa.7.2017.01.10.13.55.30 for (version=TLS1_2 cipher=ECDHE-RSA-AES128-GCM-SHA256 bits=128/128); Tue, 10 Jan 2017 13:55:30 -0800 (PST) Received: by mail-qt0-x241.google.com with SMTP id f4so16839656qte.2 for ; Tue, 10 Jan 2017 13:55:30 -0800 (PST) Return-Path: Date: Tue, 10 Jan 2017 16:55:28 -0500 From: Jon Mason Subject: Re: [PATCH 0/2] Style fixes: open code obfuscating macros Message-ID: <20170110215527.GD13381@kudzu.us> References: <1483407036-20697-1-git-send-email-logang@deltatee.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <1483407036-20697-1-git-send-email-logang@deltatee.com> To: Logan Gunthorpe Cc: Xiangliang Yu , Dave Jiang , Allen Hubbe , Greg Kroah-Hartman , linux-ntb@googlegroups.com, linux-kernel@vger.kernel.org List-ID: On Mon, Jan 02, 2017 at 06:30:34PM -0700, Logan Gunthorpe wrote: > Hi, > > I had copied some poor code style from the NTB drivers into an unrelated > driver. Upon review of my new code, I learned it was not a good idea > to sweep dirty things under the rug^W macro. See [1], where Gregg k-h > suggested I fix the problem at the source. > > Thus, please merge the following two patches to improve the style of > the NTB drivers. > > Thanks, > > Logan Applying the series, thanks > > [1] http://www.spinics.net/lists/linux-pci/msg56904.html > > > Logan Gunthorpe (2): > ntb_hw_amd: Style fixes: open code macros that just obfuscate code > ntb_hw_intel: Style fixes: open code macros that just obfuscate code > > drivers/ntb/hw/amd/ntb_hw_amd.c | 59 ++++++++------- > drivers/ntb/hw/amd/ntb_hw_amd.h | 3 - > drivers/ntb/hw/intel/ntb_hw_intel.c | 145 ++++++++++++++++++------------------ > drivers/ntb/hw/intel/ntb_hw_intel.h | 3 - > 4 files changed, 104 insertions(+), 106 deletions(-) > > -- > 2.1.4 From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from ale.deltatee.com (ale.deltatee.com. [207.54.116.67]) by gmr-mx.google.com with ESMTPS id o85si9450414pfj.2.2017.01.02.17.31.17 for (version=TLS1_2 cipher=ECDHE-RSA-AES128-GCM-SHA256 bits=128/128); Mon, 02 Jan 2017 17:31:17 -0800 (PST) From: Logan Gunthorpe Date: Mon, 2 Jan 2017 18:30:36 -0700 Message-Id: <1483407036-20697-3-git-send-email-logang@deltatee.com> In-Reply-To: <1483407036-20697-1-git-send-email-logang@deltatee.com> References: <1483407036-20697-1-git-send-email-logang@deltatee.com> Subject: [PATCH 2/2] ntb_hw_intel: Style fixes: open code macros that just obfuscate code To: Xiangliang Yu , Jon Mason , Dave Jiang , Allen Hubbe , Greg Kroah-Hartman Cc: linux-ntb@googlegroups.com, linux-kernel@vger.kernel.org, Logan Gunthorpe List-ID: As per a comments in [1] by Greg Kroah-Hartman, the ndev_* macros should be cleaned up. This makes it more clear what's actually going on when reading the code. [1] http://www.spinics.net/lists/linux-pci/msg56904.html Signed-off-by: Logan Gunthorpe --- drivers/ntb/hw/intel/ntb_hw_intel.c | 145 ++++++++++++++++++------------------ drivers/ntb/hw/intel/ntb_hw_intel.h | 3 - 2 files changed, 73 insertions(+), 75 deletions(-) diff --git a/drivers/ntb/hw/intel/ntb_hw_intel.c b/drivers/ntb/hw/intel/ntb_hw_intel.c index 7310a26..3c8ef1d 100644 --- a/drivers/ntb/hw/intel/ntb_hw_intel.c +++ b/drivers/ntb/hw/intel/ntb_hw_intel.c @@ -254,12 +254,12 @@ static inline int ndev_db_addr(struct intel_ntb_dev *ndev, if (db_addr) { *db_addr = reg_addr + reg; - dev_dbg(ndev_dev(ndev), "Peer db addr %llx\n", *db_addr); + dev_dbg(&ndev->ntb.pdev->dev, "Peer db addr %llx\n", *db_addr); } if (db_size) { *db_size = ndev->reg->db_size; - dev_dbg(ndev_dev(ndev), "Peer db size %llx\n", *db_size); + dev_dbg(&ndev->ntb.pdev->dev, "Peer db size %llx\n", *db_size); } return 0; @@ -352,7 +352,8 @@ static inline int ndev_spad_addr(struct intel_ntb_dev *ndev, int idx, if (spad_addr) { *spad_addr = reg_addr + reg + (idx << 2); - dev_dbg(ndev_dev(ndev), "Peer spad addr %llx\n", *spad_addr); + dev_dbg(&ndev->ntb.pdev->dev, "Peer spad addr %llx\n", + *spad_addr); } return 0; @@ -390,7 +391,7 @@ static irqreturn_t ndev_interrupt(struct intel_ntb_dev *ndev, int vec) vec_mask = ndev_vec_mask(ndev, vec); - dev_dbg(ndev_dev(ndev), "vec %d vec_mask %llx\n", vec, vec_mask); + dev_dbg(&ndev->ntb.pdev->dev, "vec %d vec_mask %llx\n", vec, vec_mask); ndev->last_ts = jiffies; @@ -416,7 +417,7 @@ static irqreturn_t ndev_irq_isr(int irq, void *dev) { struct intel_ntb_dev *ndev = dev; - return ndev_interrupt(ndev, irq - ndev_pdev(ndev)->irq); + return ndev_interrupt(ndev, irq - ndev->ntb.pdev->irq); } static int ndev_init_isr(struct intel_ntb_dev *ndev, @@ -426,7 +427,7 @@ static int ndev_init_isr(struct intel_ntb_dev *ndev, struct pci_dev *pdev; int rc, i, msix_count, node; - pdev = ndev_pdev(ndev); + pdev = ndev->ntb.pdev; node = dev_to_node(&pdev->dev); @@ -465,7 +466,7 @@ static int ndev_init_isr(struct intel_ntb_dev *ndev, goto err_msix_request; } - dev_dbg(ndev_dev(ndev), "Using msix interrupts\n"); + dev_dbg(&pdev->dev, "Using msix interrupts\n"); ndev->db_vec_count = msix_count; ndev->db_vec_shift = msix_shift; return 0; @@ -493,7 +494,7 @@ static int ndev_init_isr(struct intel_ntb_dev *ndev, if (rc) goto err_msi_request; - dev_dbg(ndev_dev(ndev), "Using msi interrupts\n"); + dev_dbg(&pdev->dev, "Using msi interrupts\n"); ndev->db_vec_count = 1; ndev->db_vec_shift = total_shift; return 0; @@ -511,7 +512,7 @@ static int ndev_init_isr(struct intel_ntb_dev *ndev, if (rc) goto err_intx_request; - dev_dbg(ndev_dev(ndev), "Using intx interrupts\n"); + dev_dbg(&pdev->dev, "Using intx interrupts\n"); ndev->db_vec_count = 1; ndev->db_vec_shift = total_shift; return 0; @@ -525,7 +526,7 @@ static void ndev_deinit_isr(struct intel_ntb_dev *ndev) struct pci_dev *pdev; int i; - pdev = ndev_pdev(ndev); + pdev = ndev->ntb.pdev; /* Mask all doorbell interrupts */ ndev->db_mask = ndev->db_valid_mask; @@ -559,7 +560,7 @@ static ssize_t ndev_debugfs_read(struct file *filp, char __user *ubuf, union { u64 v64; u32 v32; u16 v16; u8 v8; } u; ndev = filp->private_data; - pdev = ndev_pdev(ndev); + pdev = ndev->ntb.pdev; mmio = ndev->self_mmio; buf_size = min(count, 0x800ul); @@ -820,7 +821,8 @@ static void ndev_init_debugfs(struct intel_ntb_dev *ndev) ndev->debugfs_info = NULL; } else { ndev->debugfs_dir = - debugfs_create_dir(ndev_name(ndev), debugfs_dir); + debugfs_create_dir(pci_name(ndev->ntb.pdev), + debugfs_dir); if (!ndev->debugfs_dir) ndev->debugfs_info = NULL; else @@ -1007,13 +1009,13 @@ static int intel_ntb_link_enable(struct ntb_dev *ntb, if (ndev->ntb.topo == NTB_TOPO_SEC) return -EINVAL; - dev_dbg(ndev_dev(ndev), + dev_dbg(&ntb->pdev->dev, "Enabling link with max_speed %d max_width %d\n", max_speed, max_width); if (max_speed != NTB_SPEED_AUTO) - dev_dbg(ndev_dev(ndev), "ignoring max_speed %d\n", max_speed); + dev_dbg(&ntb->pdev->dev, "ignoring max_speed %d\n", max_speed); if (max_width != NTB_WIDTH_AUTO) - dev_dbg(ndev_dev(ndev), "ignoring max_width %d\n", max_width); + dev_dbg(&ntb->pdev->dev, "ignoring max_width %d\n", max_width); ntb_ctl = ioread32(ndev->self_mmio + ndev->reg->ntb_ctl); ntb_ctl &= ~(NTB_CTL_DISABLE | NTB_CTL_CFG_LOCK); @@ -1036,7 +1038,7 @@ static int intel_ntb_link_disable(struct ntb_dev *ntb) if (ndev->ntb.topo == NTB_TOPO_SEC) return -EINVAL; - dev_dbg(ndev_dev(ndev), "Disabling link\n"); + dev_dbg(&ntb->pdev->dev, "Disabling link\n"); /* Bring NTB link down */ ntb_cntl = ioread32(ndev->self_mmio + ndev->reg->ntb_ctl); @@ -1243,30 +1245,33 @@ static int atom_link_is_err(struct intel_ntb_dev *ndev) static inline enum ntb_topo atom_ppd_topo(struct intel_ntb_dev *ndev, u32 ppd) { + struct device *dev = &ndev->ntb.pdev->dev; + switch (ppd & ATOM_PPD_TOPO_MASK) { case ATOM_PPD_TOPO_B2B_USD: - dev_dbg(ndev_dev(ndev), "PPD %d B2B USD\n", ppd); + dev_dbg(dev, "PPD %d B2B USD\n", ppd); return NTB_TOPO_B2B_USD; case ATOM_PPD_TOPO_B2B_DSD: - dev_dbg(ndev_dev(ndev), "PPD %d B2B DSD\n", ppd); + dev_dbg(dev, "PPD %d B2B DSD\n", ppd); return NTB_TOPO_B2B_DSD; case ATOM_PPD_TOPO_PRI_USD: case ATOM_PPD_TOPO_PRI_DSD: /* accept bogus PRI_DSD */ case ATOM_PPD_TOPO_SEC_USD: case ATOM_PPD_TOPO_SEC_DSD: /* accept bogus SEC_DSD */ - dev_dbg(ndev_dev(ndev), "PPD %d non B2B disabled\n", ppd); + dev_dbg(dev, "PPD %d non B2B disabled\n", ppd); return NTB_TOPO_NONE; } - dev_dbg(ndev_dev(ndev), "PPD %d invalid\n", ppd); + dev_dbg(dev, "PPD %d invalid\n", ppd); return NTB_TOPO_NONE; } static void atom_link_hb(struct work_struct *work) { struct intel_ntb_dev *ndev = hb_ndev(work); + struct device *dev = &ndev->ntb.pdev->dev; unsigned long poll_ts; void __iomem *mmio; u32 status32; @@ -1304,30 +1309,30 @@ static void atom_link_hb(struct work_struct *work) /* Clear AER Errors, write to clear */ status32 = ioread32(mmio + ATOM_ERRCORSTS_OFFSET); - dev_dbg(ndev_dev(ndev), "ERRCORSTS = %x\n", status32); + dev_dbg(dev, "ERRCORSTS = %x\n", status32); status32 &= PCI_ERR_COR_REP_ROLL; iowrite32(status32, mmio + ATOM_ERRCORSTS_OFFSET); /* Clear unexpected electrical idle event in LTSSM, write to clear */ status32 = ioread32(mmio + ATOM_LTSSMERRSTS0_OFFSET); - dev_dbg(ndev_dev(ndev), "LTSSMERRSTS0 = %x\n", status32); + dev_dbg(dev, "LTSSMERRSTS0 = %x\n", status32); status32 |= ATOM_LTSSMERRSTS0_UNEXPECTEDEI; iowrite32(status32, mmio + ATOM_LTSSMERRSTS0_OFFSET); /* Clear DeSkew Buffer error, write to clear */ status32 = ioread32(mmio + ATOM_DESKEWSTS_OFFSET); - dev_dbg(ndev_dev(ndev), "DESKEWSTS = %x\n", status32); + dev_dbg(dev, "DESKEWSTS = %x\n", status32); status32 |= ATOM_DESKEWSTS_DBERR; iowrite32(status32, mmio + ATOM_DESKEWSTS_OFFSET); status32 = ioread32(mmio + ATOM_IBSTERRRCRVSTS0_OFFSET); - dev_dbg(ndev_dev(ndev), "IBSTERRRCRVSTS0 = %x\n", status32); + dev_dbg(dev, "IBSTERRRCRVSTS0 = %x\n", status32); status32 &= ATOM_IBIST_ERR_OFLOW; iowrite32(status32, mmio + ATOM_IBSTERRRCRVSTS0_OFFSET); /* Releases the NTB state machine to allow the link to retrain */ status32 = ioread32(mmio + ATOM_LTSSMSTATEJMP_OFFSET); - dev_dbg(ndev_dev(ndev), "LTSSMSTATEJMP = %x\n", status32); + dev_dbg(dev, "LTSSMSTATEJMP = %x\n", status32); status32 &= ~ATOM_LTSSMSTATEJMP_FORCEDETECT; iowrite32(status32, mmio + ATOM_LTSSMSTATEJMP_OFFSET); @@ -1494,7 +1499,7 @@ static inline enum ntb_topo xeon_ppd_topo(struct intel_ntb_dev *ndev, u8 ppd) static inline int xeon_ppd_bar4_split(struct intel_ntb_dev *ndev, u8 ppd) { if (ppd & XEON_PPD_SPLIT_BAR_MASK) { - dev_dbg(ndev_dev(ndev), "PPD %d split bar\n", ppd); + dev_dbg(&ndev->ntb.pdev->dev, "PPD %d split bar\n", ppd); return 1; } return 0; @@ -1524,11 +1529,11 @@ static int xeon_setup_b2b_mw(struct intel_ntb_dev *ndev, int b2b_bar; u8 bar_sz; - pdev = ndev_pdev(ndev); + pdev = ndev->ntb.pdev; mmio = ndev->self_mmio; if (ndev->b2b_idx == UINT_MAX) { - dev_dbg(ndev_dev(ndev), "not using b2b mw\n"); + dev_dbg(&pdev->dev, "not using b2b mw\n"); b2b_bar = 0; ndev->b2b_off = 0; } else { @@ -1536,24 +1541,21 @@ static int xeon_setup_b2b_mw(struct intel_ntb_dev *ndev, if (b2b_bar < 0) return -EIO; - dev_dbg(ndev_dev(ndev), "using b2b mw bar %d\n", b2b_bar); + dev_dbg(&pdev->dev, "using b2b mw bar %d\n", b2b_bar); bar_size = pci_resource_len(ndev->ntb.pdev, b2b_bar); - dev_dbg(ndev_dev(ndev), "b2b bar size %#llx\n", bar_size); + dev_dbg(&pdev->dev, "b2b bar size %#llx\n", bar_size); if (b2b_mw_share && XEON_B2B_MIN_SIZE <= bar_size >> 1) { - dev_dbg(ndev_dev(ndev), - "b2b using first half of bar\n"); + dev_dbg(&pdev->dev, "b2b using first half of bar\n"); ndev->b2b_off = bar_size >> 1; } else if (XEON_B2B_MIN_SIZE <= bar_size) { - dev_dbg(ndev_dev(ndev), - "b2b using whole bar\n"); + dev_dbg(&pdev->dev, "b2b using whole bar\n"); ndev->b2b_off = 0; --ndev->mw_count; } else { - dev_dbg(ndev_dev(ndev), - "b2b bar size is too small\n"); + dev_dbg(&pdev->dev, "b2b bar size is too small\n"); return -EIO; } } @@ -1565,7 +1567,7 @@ static int xeon_setup_b2b_mw(struct intel_ntb_dev *ndev, * offsets are not in a consistent order (bar5sz comes after ppd, odd). */ pci_read_config_byte(pdev, XEON_PBAR23SZ_OFFSET, &bar_sz); - dev_dbg(ndev_dev(ndev), "PBAR23SZ %#x\n", bar_sz); + dev_dbg(&pdev->dev, "PBAR23SZ %#x\n", bar_sz); if (b2b_bar == 2) { if (ndev->b2b_off) bar_sz -= 1; @@ -1574,11 +1576,11 @@ static int xeon_setup_b2b_mw(struct intel_ntb_dev *ndev, } pci_write_config_byte(pdev, XEON_SBAR23SZ_OFFSET, bar_sz); pci_read_config_byte(pdev, XEON_SBAR23SZ_OFFSET, &bar_sz); - dev_dbg(ndev_dev(ndev), "SBAR23SZ %#x\n", bar_sz); + dev_dbg(&pdev->dev, "SBAR23SZ %#x\n", bar_sz); if (!ndev->bar4_split) { pci_read_config_byte(pdev, XEON_PBAR45SZ_OFFSET, &bar_sz); - dev_dbg(ndev_dev(ndev), "PBAR45SZ %#x\n", bar_sz); + dev_dbg(&pdev->dev, "PBAR45SZ %#x\n", bar_sz); if (b2b_bar == 4) { if (ndev->b2b_off) bar_sz -= 1; @@ -1587,10 +1589,10 @@ static int xeon_setup_b2b_mw(struct intel_ntb_dev *ndev, } pci_write_config_byte(pdev, XEON_SBAR45SZ_OFFSET, bar_sz); pci_read_config_byte(pdev, XEON_SBAR45SZ_OFFSET, &bar_sz); - dev_dbg(ndev_dev(ndev), "SBAR45SZ %#x\n", bar_sz); + dev_dbg(&pdev->dev, "SBAR45SZ %#x\n", bar_sz); } else { pci_read_config_byte(pdev, XEON_PBAR4SZ_OFFSET, &bar_sz); - dev_dbg(ndev_dev(ndev), "PBAR4SZ %#x\n", bar_sz); + dev_dbg(&pdev->dev, "PBAR4SZ %#x\n", bar_sz); if (b2b_bar == 4) { if (ndev->b2b_off) bar_sz -= 1; @@ -1599,10 +1601,10 @@ static int xeon_setup_b2b_mw(struct intel_ntb_dev *ndev, } pci_write_config_byte(pdev, XEON_SBAR4SZ_OFFSET, bar_sz); pci_read_config_byte(pdev, XEON_SBAR4SZ_OFFSET, &bar_sz); - dev_dbg(ndev_dev(ndev), "SBAR4SZ %#x\n", bar_sz); + dev_dbg(&pdev->dev, "SBAR4SZ %#x\n", bar_sz); pci_read_config_byte(pdev, XEON_PBAR5SZ_OFFSET, &bar_sz); - dev_dbg(ndev_dev(ndev), "PBAR5SZ %#x\n", bar_sz); + dev_dbg(&pdev->dev, "PBAR5SZ %#x\n", bar_sz); if (b2b_bar == 5) { if (ndev->b2b_off) bar_sz -= 1; @@ -1611,7 +1613,7 @@ static int xeon_setup_b2b_mw(struct intel_ntb_dev *ndev, } pci_write_config_byte(pdev, XEON_SBAR5SZ_OFFSET, bar_sz); pci_read_config_byte(pdev, XEON_SBAR5SZ_OFFSET, &bar_sz); - dev_dbg(ndev_dev(ndev), "SBAR5SZ %#x\n", bar_sz); + dev_dbg(&pdev->dev, "SBAR5SZ %#x\n", bar_sz); } /* SBAR01 hit by first part of the b2b bar */ @@ -1628,7 +1630,7 @@ static int xeon_setup_b2b_mw(struct intel_ntb_dev *ndev, else return -EIO; - dev_dbg(ndev_dev(ndev), "SBAR01 %#018llx\n", bar_addr); + dev_dbg(&pdev->dev, "SBAR01 %#018llx\n", bar_addr); iowrite64(bar_addr, mmio + XEON_SBAR0BASE_OFFSET); /* Other SBAR are normally hit by the PBAR xlat, except for b2b bar. @@ -1639,26 +1641,26 @@ static int xeon_setup_b2b_mw(struct intel_ntb_dev *ndev, bar_addr = addr->bar2_addr64 + (b2b_bar == 2 ? ndev->b2b_off : 0); iowrite64(bar_addr, mmio + XEON_SBAR23BASE_OFFSET); bar_addr = ioread64(mmio + XEON_SBAR23BASE_OFFSET); - dev_dbg(ndev_dev(ndev), "SBAR23 %#018llx\n", bar_addr); + dev_dbg(&pdev->dev, "SBAR23 %#018llx\n", bar_addr); if (!ndev->bar4_split) { bar_addr = addr->bar4_addr64 + (b2b_bar == 4 ? ndev->b2b_off : 0); iowrite64(bar_addr, mmio + XEON_SBAR45BASE_OFFSET); bar_addr = ioread64(mmio + XEON_SBAR45BASE_OFFSET); - dev_dbg(ndev_dev(ndev), "SBAR45 %#018llx\n", bar_addr); + dev_dbg(&pdev->dev, "SBAR45 %#018llx\n", bar_addr); } else { bar_addr = addr->bar4_addr32 + (b2b_bar == 4 ? ndev->b2b_off : 0); iowrite32(bar_addr, mmio + XEON_SBAR4BASE_OFFSET); bar_addr = ioread32(mmio + XEON_SBAR4BASE_OFFSET); - dev_dbg(ndev_dev(ndev), "SBAR4 %#010llx\n", bar_addr); + dev_dbg(&pdev->dev, "SBAR4 %#010llx\n", bar_addr); bar_addr = addr->bar5_addr32 + (b2b_bar == 5 ? ndev->b2b_off : 0); iowrite32(bar_addr, mmio + XEON_SBAR5BASE_OFFSET); bar_addr = ioread32(mmio + XEON_SBAR5BASE_OFFSET); - dev_dbg(ndev_dev(ndev), "SBAR5 %#010llx\n", bar_addr); + dev_dbg(&pdev->dev, "SBAR5 %#010llx\n", bar_addr); } /* setup incoming bar limits == base addrs (zero length windows) */ @@ -1666,26 +1668,26 @@ static int xeon_setup_b2b_mw(struct intel_ntb_dev *ndev, bar_addr = addr->bar2_addr64 + (b2b_bar == 2 ? ndev->b2b_off : 0); iowrite64(bar_addr, mmio + XEON_SBAR23LMT_OFFSET); bar_addr = ioread64(mmio + XEON_SBAR23LMT_OFFSET); - dev_dbg(ndev_dev(ndev), "SBAR23LMT %#018llx\n", bar_addr); + dev_dbg(&pdev->dev, "SBAR23LMT %#018llx\n", bar_addr); if (!ndev->bar4_split) { bar_addr = addr->bar4_addr64 + (b2b_bar == 4 ? ndev->b2b_off : 0); iowrite64(bar_addr, mmio + XEON_SBAR45LMT_OFFSET); bar_addr = ioread64(mmio + XEON_SBAR45LMT_OFFSET); - dev_dbg(ndev_dev(ndev), "SBAR45LMT %#018llx\n", bar_addr); + dev_dbg(&pdev->dev, "SBAR45LMT %#018llx\n", bar_addr); } else { bar_addr = addr->bar4_addr32 + (b2b_bar == 4 ? ndev->b2b_off : 0); iowrite32(bar_addr, mmio + XEON_SBAR4LMT_OFFSET); bar_addr = ioread32(mmio + XEON_SBAR4LMT_OFFSET); - dev_dbg(ndev_dev(ndev), "SBAR4LMT %#010llx\n", bar_addr); + dev_dbg(&pdev->dev, "SBAR4LMT %#010llx\n", bar_addr); bar_addr = addr->bar5_addr32 + (b2b_bar == 5 ? ndev->b2b_off : 0); iowrite32(bar_addr, mmio + XEON_SBAR5LMT_OFFSET); bar_addr = ioread32(mmio + XEON_SBAR5LMT_OFFSET); - dev_dbg(ndev_dev(ndev), "SBAR5LMT %#05llx\n", bar_addr); + dev_dbg(&pdev->dev, "SBAR5LMT %#05llx\n", bar_addr); } /* zero incoming translation addrs */ @@ -1711,23 +1713,23 @@ static int xeon_setup_b2b_mw(struct intel_ntb_dev *ndev, bar_addr = peer_addr->bar2_addr64; iowrite64(bar_addr, mmio + XEON_PBAR23XLAT_OFFSET); bar_addr = ioread64(mmio + XEON_PBAR23XLAT_OFFSET); - dev_dbg(ndev_dev(ndev), "PBAR23XLAT %#018llx\n", bar_addr); + dev_dbg(&pdev->dev, "PBAR23XLAT %#018llx\n", bar_addr); if (!ndev->bar4_split) { bar_addr = peer_addr->bar4_addr64; iowrite64(bar_addr, mmio + XEON_PBAR45XLAT_OFFSET); bar_addr = ioread64(mmio + XEON_PBAR45XLAT_OFFSET); - dev_dbg(ndev_dev(ndev), "PBAR45XLAT %#018llx\n", bar_addr); + dev_dbg(&pdev->dev, "PBAR45XLAT %#018llx\n", bar_addr); } else { bar_addr = peer_addr->bar4_addr32; iowrite32(bar_addr, mmio + XEON_PBAR4XLAT_OFFSET); bar_addr = ioread32(mmio + XEON_PBAR4XLAT_OFFSET); - dev_dbg(ndev_dev(ndev), "PBAR4XLAT %#010llx\n", bar_addr); + dev_dbg(&pdev->dev, "PBAR4XLAT %#010llx\n", bar_addr); bar_addr = peer_addr->bar5_addr32; iowrite32(bar_addr, mmio + XEON_PBAR5XLAT_OFFSET); bar_addr = ioread32(mmio + XEON_PBAR5XLAT_OFFSET); - dev_dbg(ndev_dev(ndev), "PBAR5XLAT %#010llx\n", bar_addr); + dev_dbg(&pdev->dev, "PBAR5XLAT %#010llx\n", bar_addr); } /* set the translation offset for b2b registers */ @@ -1745,7 +1747,7 @@ static int xeon_setup_b2b_mw(struct intel_ntb_dev *ndev, return -EIO; /* B2B_XLAT_OFFSET is 64bit, but can only take 32bit writes */ - dev_dbg(ndev_dev(ndev), "B2BXLAT %#018llx\n", bar_addr); + dev_dbg(&pdev->dev, "B2BXLAT %#018llx\n", bar_addr); iowrite32(bar_addr, mmio + XEON_B2B_XLAT_OFFSETL); iowrite32(bar_addr >> 32, mmio + XEON_B2B_XLAT_OFFSETU); @@ -1764,6 +1766,7 @@ static int xeon_setup_b2b_mw(struct intel_ntb_dev *ndev, static int xeon_init_ntb(struct intel_ntb_dev *ndev) { + struct device *dev = &ndev->ntb.pdev->dev; int rc; u32 ntb_ctl; @@ -1779,7 +1782,7 @@ static int xeon_init_ntb(struct intel_ntb_dev *ndev) switch (ndev->ntb.topo) { case NTB_TOPO_PRI: if (ndev->hwerr_flags & NTB_HWERR_SDOORBELL_LOCKUP) { - dev_err(ndev_dev(ndev), "NTB Primary config disabled\n"); + dev_err(dev, "NTB Primary config disabled\n"); return -EINVAL; } @@ -1797,7 +1800,7 @@ static int xeon_init_ntb(struct intel_ntb_dev *ndev) case NTB_TOPO_SEC: if (ndev->hwerr_flags & NTB_HWERR_SDOORBELL_LOCKUP) { - dev_err(ndev_dev(ndev), "NTB Secondary config disabled\n"); + dev_err(dev, "NTB Secondary config disabled\n"); return -EINVAL; } /* use half the spads for the peer */ @@ -1822,18 +1825,17 @@ static int xeon_init_ntb(struct intel_ntb_dev *ndev) ndev->b2b_idx = b2b_mw_idx; if (ndev->b2b_idx >= ndev->mw_count) { - dev_dbg(ndev_dev(ndev), + dev_dbg(dev, "b2b_mw_idx %d invalid for mw_count %u\n", b2b_mw_idx, ndev->mw_count); return -EINVAL; } - dev_dbg(ndev_dev(ndev), - "setting up b2b mw idx %d means %d\n", + dev_dbg(dev, "setting up b2b mw idx %d means %d\n", b2b_mw_idx, ndev->b2b_idx); } else if (ndev->hwerr_flags & NTB_HWERR_B2BDOORBELL_BIT14) { - dev_warn(ndev_dev(ndev), "Reduce doorbell count by 1\n"); + dev_warn(dev, "Reduce doorbell count by 1\n"); ndev->db_count -= 1; } @@ -1874,7 +1876,7 @@ static int xeon_init_dev(struct intel_ntb_dev *ndev) u8 ppd; int rc, mem; - pdev = ndev_pdev(ndev); + pdev = ndev->ntb.pdev; switch (pdev->device) { /* There is a Xeon hardware errata related to writes to SDOORBELL or @@ -1950,14 +1952,14 @@ static int xeon_init_dev(struct intel_ntb_dev *ndev) return -EIO; ndev->ntb.topo = xeon_ppd_topo(ndev, ppd); - dev_dbg(ndev_dev(ndev), "ppd %#x topo %s\n", ppd, + dev_dbg(&pdev->dev, "ppd %#x topo %s\n", ppd, ntb_topo_string(ndev->ntb.topo)); if (ndev->ntb.topo == NTB_TOPO_NONE) return -EINVAL; if (ndev->ntb.topo != NTB_TOPO_SEC) { ndev->bar4_split = xeon_ppd_bar4_split(ndev, ppd); - dev_dbg(ndev_dev(ndev), "ppd %#x bar4_split %d\n", + dev_dbg(&pdev->dev, "ppd %#x bar4_split %d\n", ppd, ndev->bar4_split); } else { /* This is a way for transparent BAR to figure out if we are @@ -1967,7 +1969,7 @@ static int xeon_init_dev(struct intel_ntb_dev *ndev) mem = pci_select_bars(pdev, IORESOURCE_MEM); ndev->bar4_split = hweight32(mem) == HSX_SPLIT_BAR_MW_COUNT + 1; - dev_dbg(ndev_dev(ndev), "mem %#x bar4_split %d\n", + dev_dbg(&pdev->dev, "mem %#x bar4_split %d\n", mem, ndev->bar4_split); } @@ -2004,7 +2006,7 @@ static int intel_ntb_init_pci(struct intel_ntb_dev *ndev, struct pci_dev *pdev) rc = pci_set_dma_mask(pdev, DMA_BIT_MASK(32)); if (rc) goto err_dma_mask; - dev_warn(ndev_dev(ndev), "Cannot DMA highmem\n"); + dev_warn(&pdev->dev, "Cannot DMA highmem\n"); } rc = pci_set_consistent_dma_mask(pdev, DMA_BIT_MASK(64)); @@ -2012,7 +2014,7 @@ static int intel_ntb_init_pci(struct intel_ntb_dev *ndev, struct pci_dev *pdev) rc = pci_set_consistent_dma_mask(pdev, DMA_BIT_MASK(32)); if (rc) goto err_dma_mask; - dev_warn(ndev_dev(ndev), "Cannot DMA consistent highmem\n"); + dev_warn(&pdev->dev, "Cannot DMA consistent highmem\n"); } ndev->self_mmio = pci_iomap(pdev, 0, 0); @@ -2038,7 +2040,7 @@ static int intel_ntb_init_pci(struct intel_ntb_dev *ndev, struct pci_dev *pdev) static void intel_ntb_deinit_pci(struct intel_ntb_dev *ndev) { - struct pci_dev *pdev = ndev_pdev(ndev); + struct pci_dev *pdev = ndev->ntb.pdev; if (ndev->peer_mmio && ndev->peer_mmio != ndev->self_mmio) pci_iounmap(pdev, ndev->peer_mmio); @@ -2336,4 +2338,3 @@ static void __exit intel_ntb_pci_driver_exit(void) debugfs_remove_recursive(debugfs_dir); } module_exit(intel_ntb_pci_driver_exit); - diff --git a/drivers/ntb/hw/intel/ntb_hw_intel.h b/drivers/ntb/hw/intel/ntb_hw_intel.h index 3ec149c..f0227a4 100644 --- a/drivers/ntb/hw/intel/ntb_hw_intel.h +++ b/drivers/ntb/hw/intel/ntb_hw_intel.h @@ -334,9 +334,6 @@ struct intel_ntb_dev { struct dentry *debugfs_info; }; -#define ndev_pdev(ndev) ((ndev)->ntb.pdev) -#define ndev_name(ndev) pci_name(ndev_pdev(ndev)) -#define ndev_dev(ndev) (&ndev_pdev(ndev)->dev) #define ntb_ndev(__ntb) container_of(__ntb, struct intel_ntb_dev, ntb) #define hb_ndev(__work) container_of(__work, struct intel_ntb_dev, \ hb_timer.work) -- 2.1.4 From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from ale.deltatee.com (ale.deltatee.com. [207.54.116.67]) by gmr-mx.google.com with ESMTPS id z62si16558947pfz.1.2017.01.02.17.31.16 for (version=TLS1_2 cipher=ECDHE-RSA-AES128-GCM-SHA256 bits=128/128); Mon, 02 Jan 2017 17:31:16 -0800 (PST) From: Logan Gunthorpe Date: Mon, 2 Jan 2017 18:30:35 -0700 Message-Id: <1483407036-20697-2-git-send-email-logang@deltatee.com> In-Reply-To: <1483407036-20697-1-git-send-email-logang@deltatee.com> References: <1483407036-20697-1-git-send-email-logang@deltatee.com> Subject: [PATCH 1/2] ntb_hw_amd: Style fixes: open code macros that just obfuscate code To: Xiangliang Yu , Jon Mason , Dave Jiang , Allen Hubbe , Greg Kroah-Hartman Cc: linux-ntb@googlegroups.com, linux-kernel@vger.kernel.org, Logan Gunthorpe List-ID: As per a comments in [1] by Greg Kroah-Hartman, the ndev_* macros should be cleaned up. This makes it more clear what's actually going on when reading the code. [1] http://www.spinics.net/lists/linux-pci/msg56904.html Signed-off-by: Logan Gunthorpe --- drivers/ntb/hw/amd/ntb_hw_amd.c | 59 ++++++++++++++++++++++------------------- drivers/ntb/hw/amd/ntb_hw_amd.h | 3 --- 2 files changed, 31 insertions(+), 31 deletions(-) diff --git a/drivers/ntb/hw/amd/ntb_hw_amd.c b/drivers/ntb/hw/amd/ntb_hw_amd.c index 6ccba0d..85a9a4f 100644 --- a/drivers/ntb/hw/amd/ntb_hw_amd.c +++ b/drivers/ntb/hw/amd/ntb_hw_amd.c @@ -98,10 +98,10 @@ static int amd_ntb_mw_get_range(struct ntb_dev *ntb, int idx, return bar; if (base) - *base = pci_resource_start(ndev->ntb.pdev, bar); + *base = pci_resource_start(ntb->pdev, bar); if (size) - *size = pci_resource_len(ndev->ntb.pdev, bar); + *size = pci_resource_len(ntb->pdev, bar); if (align) *align = SZ_4K; @@ -126,7 +126,7 @@ static int amd_ntb_mw_set_trans(struct ntb_dev *ntb, int idx, if (bar < 0) return bar; - mw_size = pci_resource_len(ndev->ntb.pdev, bar); + mw_size = pci_resource_len(ntb->pdev, bar); /* make sure the range fits in the usable mw size */ if (size > mw_size) @@ -135,7 +135,7 @@ static int amd_ntb_mw_set_trans(struct ntb_dev *ntb, int idx, mmio = ndev->self_mmio; peer_mmio = ndev->peer_mmio; - base_addr = pci_resource_start(ndev->ntb.pdev, bar); + base_addr = pci_resource_start(ntb->pdev, bar); if (bar != 1) { xlat_reg = AMD_BAR23XLAT_OFFSET + ((bar - 2) << 3); @@ -226,7 +226,7 @@ static int amd_ntb_link_is_up(struct ntb_dev *ntb, if (width) *width = NTB_LNK_STA_WIDTH(ndev->lnk_sta); - dev_dbg(ndev_dev(ndev), "link is up.\n"); + dev_dbg(&ntb->pdev->dev, "link is up.\n"); ret = 1; } else { @@ -235,7 +235,7 @@ static int amd_ntb_link_is_up(struct ntb_dev *ntb, if (width) *width = NTB_WIDTH_NONE; - dev_dbg(ndev_dev(ndev), "link is down.\n"); + dev_dbg(&ntb->pdev->dev, "link is down.\n"); } return ret; @@ -255,7 +255,7 @@ static int amd_ntb_link_enable(struct ntb_dev *ntb, if (ndev->ntb.topo == NTB_TOPO_SEC) return -EINVAL; - dev_dbg(ndev_dev(ndev), "Enabling Link.\n"); + dev_dbg(&ntb->pdev->dev, "Enabling Link.\n"); ntb_ctl = readl(mmio + AMD_CNTL_OFFSET); ntb_ctl |= (PMM_REG_CTL | SMM_REG_CTL); @@ -276,7 +276,7 @@ static int amd_ntb_link_disable(struct ntb_dev *ntb) if (ndev->ntb.topo == NTB_TOPO_SEC) return -EINVAL; - dev_dbg(ndev_dev(ndev), "Enabling Link.\n"); + dev_dbg(&ntb->pdev->dev, "Enabling Link.\n"); ntb_ctl = readl(mmio + AMD_CNTL_OFFSET); ntb_ctl &= ~(PMM_REG_CTL | SMM_REG_CTL); @@ -467,18 +467,19 @@ static void amd_ack_smu(struct amd_ntb_dev *ndev, u32 bit) static void amd_handle_event(struct amd_ntb_dev *ndev, int vec) { void __iomem *mmio = ndev->self_mmio; + struct device *dev = &ndev->ntb.pdev->dev; u32 status; status = readl(mmio + AMD_INTSTAT_OFFSET); if (!(status & AMD_EVENT_INTMASK)) return; - dev_dbg(ndev_dev(ndev), "status = 0x%x and vec = %d\n", status, vec); + dev_dbg(dev, "status = 0x%x and vec = %d\n", status, vec); status &= AMD_EVENT_INTMASK; switch (status) { case AMD_PEER_FLUSH_EVENT: - dev_info(ndev_dev(ndev), "Flush is done.\n"); + dev_info(dev, "Flush is done.\n"); break; case AMD_PEER_RESET_EVENT: amd_ack_smu(ndev, AMD_PEER_RESET_EVENT); @@ -502,7 +503,7 @@ static void amd_handle_event(struct amd_ntb_dev *ndev, int vec) status = readl(mmio + AMD_PMESTAT_OFFSET); /* check if this is WAKEUP event */ if (status & 0x1) - dev_info(ndev_dev(ndev), "Wakeup is done.\n"); + dev_info(dev, "Wakeup is done.\n"); amd_ack_smu(ndev, AMD_PEER_D0_EVENT); @@ -511,14 +512,14 @@ static void amd_handle_event(struct amd_ntb_dev *ndev, int vec) AMD_LINK_HB_TIMEOUT); break; default: - dev_info(ndev_dev(ndev), "event status = 0x%x.\n", status); + dev_info(dev, "event status = 0x%x.\n", status); break; } } static irqreturn_t ndev_interrupt(struct amd_ntb_dev *ndev, int vec) { - dev_dbg(ndev_dev(ndev), "vec %d\n", vec); + dev_dbg(&ndev->ntb.pdev->dev, "vec %d\n", vec); if (vec > (AMD_DB_CNT - 1) || (ndev->msix_vec_count == 1)) amd_handle_event(ndev, vec); @@ -540,7 +541,7 @@ static irqreturn_t ndev_irq_isr(int irq, void *dev) { struct amd_ntb_dev *ndev = dev; - return ndev_interrupt(ndev, irq - ndev_pdev(ndev)->irq); + return ndev_interrupt(ndev, irq - ndev->ntb.pdev->irq); } static int ndev_init_isr(struct amd_ntb_dev *ndev, @@ -549,7 +550,7 @@ static int ndev_init_isr(struct amd_ntb_dev *ndev, struct pci_dev *pdev; int rc, i, msix_count, node; - pdev = ndev_pdev(ndev); + pdev = ndev->ntb.pdev; node = dev_to_node(&pdev->dev); @@ -591,7 +592,7 @@ static int ndev_init_isr(struct amd_ntb_dev *ndev, goto err_msix_request; } - dev_dbg(ndev_dev(ndev), "Using msix interrupts\n"); + dev_dbg(&pdev->dev, "Using msix interrupts\n"); ndev->db_count = msix_min; ndev->msix_vec_count = msix_max; return 0; @@ -618,7 +619,7 @@ static int ndev_init_isr(struct amd_ntb_dev *ndev, if (rc) goto err_msi_request; - dev_dbg(ndev_dev(ndev), "Using msi interrupts\n"); + dev_dbg(&pdev->dev, "Using msi interrupts\n"); ndev->db_count = 1; ndev->msix_vec_count = 1; return 0; @@ -635,7 +636,7 @@ static int ndev_init_isr(struct amd_ntb_dev *ndev, if (rc) goto err_intx_request; - dev_dbg(ndev_dev(ndev), "Using intx interrupts\n"); + dev_dbg(&pdev->dev, "Using intx interrupts\n"); ndev->db_count = 1; ndev->msix_vec_count = 1; return 0; @@ -650,7 +651,7 @@ static void ndev_deinit_isr(struct amd_ntb_dev *ndev) void __iomem *mmio = ndev->self_mmio; int i; - pdev = ndev_pdev(ndev); + pdev = ndev->ntb.pdev; /* Mask all doorbell interrupts */ ndev->db_mask = ndev->db_valid_mask; @@ -776,7 +777,8 @@ static void ndev_init_debugfs(struct amd_ntb_dev *ndev) ndev->debugfs_info = NULL; } else { ndev->debugfs_dir = - debugfs_create_dir(ndev_name(ndev), debugfs_dir); + debugfs_create_dir(pci_name(ndev->ntb.pdev), + debugfs_dir); if (!ndev->debugfs_dir) ndev->debugfs_info = NULL; else @@ -811,7 +813,7 @@ static int amd_poll_link(struct amd_ntb_dev *ndev) reg = readl(mmio + AMD_SIDEINFO_OFFSET); reg &= NTB_LIN_STA_ACTIVE_BIT; - dev_dbg(ndev_dev(ndev), "%s: reg_val = 0x%x.\n", __func__, reg); + dev_dbg(&ndev->ntb.pdev->dev, "%s: reg_val = 0x%x.\n", __func__, reg); if (reg == ndev->cntl_sta) return 0; @@ -893,7 +895,8 @@ static int amd_init_ntb(struct amd_ntb_dev *ndev) break; default: - dev_err(ndev_dev(ndev), "AMD NTB does not support B2B mode.\n"); + dev_err(&ndev->ntb.pdev->dev, + "AMD NTB does not support B2B mode.\n"); return -EINVAL; } @@ -922,10 +925,10 @@ static int amd_init_dev(struct amd_ntb_dev *ndev) struct pci_dev *pdev; int rc = 0; - pdev = ndev_pdev(ndev); + pdev = ndev->ntb.pdev; ndev->ntb.topo = amd_get_topo(ndev); - dev_dbg(ndev_dev(ndev), "AMD NTB topo is %s\n", + dev_dbg(&pdev->dev, "AMD NTB topo is %s\n", ntb_topo_string(ndev->ntb.topo)); rc = amd_init_ntb(ndev); @@ -934,7 +937,7 @@ static int amd_init_dev(struct amd_ntb_dev *ndev) rc = amd_init_isr(ndev); if (rc) { - dev_err(ndev_dev(ndev), "fail to init isr.\n"); + dev_err(&pdev->dev, "fail to init isr.\n"); return rc; } @@ -972,7 +975,7 @@ static int amd_ntb_init_pci(struct amd_ntb_dev *ndev, rc = pci_set_dma_mask(pdev, DMA_BIT_MASK(32)); if (rc) goto err_dma_mask; - dev_warn(ndev_dev(ndev), "Cannot DMA highmem\n"); + dev_warn(&pdev->dev, "Cannot DMA highmem\n"); } rc = pci_set_consistent_dma_mask(pdev, DMA_BIT_MASK(64)); @@ -980,7 +983,7 @@ static int amd_ntb_init_pci(struct amd_ntb_dev *ndev, rc = pci_set_consistent_dma_mask(pdev, DMA_BIT_MASK(32)); if (rc) goto err_dma_mask; - dev_warn(ndev_dev(ndev), "Cannot DMA consistent highmem\n"); + dev_warn(&pdev->dev, "Cannot DMA consistent highmem\n"); } ndev->self_mmio = pci_iomap(pdev, 0, 0); @@ -1003,7 +1006,7 @@ static int amd_ntb_init_pci(struct amd_ntb_dev *ndev, static void amd_ntb_deinit_pci(struct amd_ntb_dev *ndev) { - struct pci_dev *pdev = ndev_pdev(ndev); + struct pci_dev *pdev = ndev->ntb.pdev; pci_iounmap(pdev, ndev->self_mmio); diff --git a/drivers/ntb/hw/amd/ntb_hw_amd.h b/drivers/ntb/hw/amd/ntb_hw_amd.h index 2eac3cd..b6169d9 100644 --- a/drivers/ntb/hw/amd/ntb_hw_amd.h +++ b/drivers/ntb/hw/amd/ntb_hw_amd.h @@ -208,9 +208,6 @@ struct amd_ntb_dev { struct dentry *debugfs_info; }; -#define ndev_pdev(ndev) ((ndev)->ntb.pdev) -#define ndev_name(ndev) pci_name(ndev_pdev(ndev)) -#define ndev_dev(ndev) (&ndev_pdev(ndev)->dev) #define ntb_ndev(__ntb) container_of(__ntb, struct amd_ntb_dev, ntb) #define hb_ndev(__work) container_of(__work, struct amd_ntb_dev, hb_timer.work) -- 2.1.4 From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from ale.deltatee.com (ale.deltatee.com. [207.54.116.67]) by gmr-mx.google.com with ESMTPS id r1si16263679pfg.4.2017.01.02.17.31.15 for (version=TLS1_2 cipher=ECDHE-RSA-AES128-GCM-SHA256 bits=128/128); Mon, 02 Jan 2017 17:31:15 -0800 (PST) From: Logan Gunthorpe Date: Mon, 2 Jan 2017 18:30:34 -0700 Message-Id: <1483407036-20697-1-git-send-email-logang@deltatee.com> Subject: [PATCH 0/2] Style fixes: open code obfuscating macros To: Xiangliang Yu , Jon Mason , Dave Jiang , Allen Hubbe , Greg Kroah-Hartman Cc: linux-ntb@googlegroups.com, linux-kernel@vger.kernel.org, Logan Gunthorpe List-ID: Hi, I had copied some poor code style from the NTB drivers into an unrelated driver. Upon review of my new code, I learned it was not a good idea to sweep dirty things under the rug^W macro. See [1], where Gregg k-h suggested I fix the problem at the source. Thus, please merge the following two patches to improve the style of the NTB drivers. Thanks, Logan [1] http://www.spinics.net/lists/linux-pci/msg56904.html Logan Gunthorpe (2): ntb_hw_amd: Style fixes: open code macros that just obfuscate code ntb_hw_intel: Style fixes: open code macros that just obfuscate code drivers/ntb/hw/amd/ntb_hw_amd.c | 59 ++++++++------- drivers/ntb/hw/amd/ntb_hw_amd.h | 3 - drivers/ntb/hw/intel/ntb_hw_intel.c | 145 ++++++++++++++++++------------------ drivers/ntb/hw/intel/ntb_hw_intel.h | 3 - 4 files changed, 104 insertions(+), 106 deletions(-) -- 2.1.4