[PATCH rdma-core 00/21] Fixes for packaging

[PATCH rdma-core 00/21] Fixes for packaging

[Jason Gunthorpe]

This is a grab bag of things I noticed when working on the systemd bootup
process:

- Various cases where multiple RPMs owned the same file
- Incorrect /run/ directory for Debian builds
- Consolidate more things into rdma-core under the theme of 'kernel boot
  support' notably rxe_cfg moves in
- Use the 00-xxx.rules format for all udev rules
- Get srp_daemon running on Debian and fixup some logging issues I noticed
- Move iwpmd to /usr/sbin since it is not user runnable

See https://github.com/linux-rdma/rdma-core/pull/172

Jason Gunthorpe (21):
  srp: Fix redhat packaging
  cbuild: Switch to Fedora Core 26
  cbuild: Add debian stretch
  cbuild: Use opensuse leap 42.2
  Make 'cbuild travis pkg' run sparse like real travis
  Tidy travis sparse check
  redhat: Fix redhat packaging of /etc/rdma/ibacm_opts.cfg
  redhat: Fix duplicated udev rules in redhat packaging
  redhat: Fix duplicated libmlx*.so in redhat packaging
  debian: Set the run directory path to /run/
  debian: Remove infiniband-diags Depends from iwpmd
  debian: Move truescale setup into rdma-core
  Prefix udev rules with a number
  Revise the umad SYSTEMD_ALIAS fake path
  umad: Match only umad kernel devices in the udev rule
  srp_daemon: Do not use LOG_PERROR
  srp_deamon: Using v formatters directly for pr_err
  srp_daemon: Call systemctl properly from udev
  iwpmd: Fix install path for iwpmd
  Move rxe_cfg to the rdma-core package
  Make /etc/modprobe.d/mlx4.conf common

 debian/control                                     |  5 ++---
 debian/ibverbs-providers.install                   |  6 -----
 debian/iwpmd.install                               |  2 +-
 debian/libibumad3.install                          |  2 +-
 debian/rdma-core.install                           |  9 +++++++-
 debian/rules                                       |  1 +
 debian/srptools.install                            |  2 +-
 iwpmd/CMakeLists.txt                               |  8 ++++---
 iwpmd/{iwpmd.service => iwpmd.service.in}          |  2 +-
 iwpmd/{iwpmd_init => iwpmd_init.in}                |  2 +-
 libibumad/CMakeLists.txt                           |  4 ++--
 libibumad/libibumad.udev-rules                     |  2 +-
 providers/mlx4/CMakeLists.txt                      |  2 ++
 .../mlx4/mlx4.conf                                 |  0
 rdma-ndd/CMakeLists.txt                            |  4 +++-
 redhat/rdma-core.spec                              | 26 ++++++++++++----------
 srp_daemon/CMakeLists.txt                          |  4 +++-
 srp_daemon/srp_daemon.c                            | 13 ++++-------
 srp_daemon/srp_daemon.rules                        |  2 +-
 srp_daemon/srp_daemon_port@.service.in             |  4 ++--
 20 files changed, 53 insertions(+), 47 deletions(-)
 rename iwpmd/{iwpmd.service => iwpmd.service.in} (80%)
 rename iwpmd/{iwpmd_init => iwpmd_init.in} (97%)
 mode change 100755 => 100644
 rename redhat/rdma.mlx4.user.modprobe => providers/mlx4/mlx4.conf (100%)

-- 
2.7.4

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH rdma-core 01/21] srp: Fix redhat packaging

[Jason Gunthorpe]

Build fails with:

error: Installed (but unpackaged) file(s) found:
   /usr/sbin/srp_daemon.sh
    Installed (but unpackaged) file(s) found:
   /usr/sbin/srp_daemon.sh

Since we do not install the init.d script for srp on RH, also delete
the srp_daemon.sh, which is now only used by init.d.

Fixes: 1c7fe513e3e9 ("srp_daemon: One systemd service per port")
Signed-off-by: Jason Gunthorpe <jgunthorpe-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
---
 redhat/rdma-core.spec | 1 +
 1 file changed, 1 insertion(+)

diff --git a/redhat/rdma-core.spec b/redhat/rdma-core.spec
index 817013bc4905f8..80a78265d02bfb 100644
--- a/redhat/rdma-core.spec
+++ b/redhat/rdma-core.spec
@@ -274,6 +274,7 @@ install -D -m0644 ibacm_opts.cfg %{buildroot}%{_sysconfdir}/rdma/
 
 # Delete the package's init.d scripts
 rm -rf %{buildroot}/%{_initrddir}/
+rm -rf %{buildroot}/%{_sbindir}/srp_daemon.sh
 
 # libibverbs
 %post -n libibverbs -p /sbin/ldconfig
-- 
2.7.4

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH rdma-core 02/21] cbuild: Switch to Fedora Core 26

[Jason Gunthorpe]

Latest release.

Signed-off-by: Jason Gunthorpe <jgunthorpe-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
---
 buildlib/cbuild | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/buildlib/cbuild b/buildlib/cbuild
index 24136b4d8cc49c..78c9d7a8176ae4 100755
--- a/buildlib/cbuild
+++ b/buildlib/cbuild
@@ -126,10 +126,10 @@ class centos7_epel(centos7):
         res.lines.append("RUN ln -s /usr/bin/cmake3 /usr/local/bin/cmake");
         return res;
 
-class fc25(Environment):
-    docker_parent = "fedora:25";
+class fc26(Environment):
+    docker_parent = "fedora:26";
     pkgs = (centos7.pkgs - {"make"}) | {"ninja-build"};
-    name = "fc25";
+    name = "fc26";
     specfile = "redhat/rdma-core.spec";
     ninja_cmd = "ninja-build";
     is_rpm = True;
@@ -318,7 +318,7 @@ environments = [centos6(),
                 trusty(),
                 xenial(),
                 jessie(),
-                fc25(),
+                fc26(),
                 harlequin(),
                 malachite(),
                 tumbleweed(),
-- 
2.7.4

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH rdma-core 03/21] cbuild: Add debian stretch

[Jason Gunthorpe]

Latest stable debian release

Signed-off-by: Jason Gunthorpe <jgunthorpe-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
---
 buildlib/cbuild | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/buildlib/cbuild b/buildlib/cbuild
index 78c9d7a8176ae4..48a8c1ea48ea5e 100755
--- a/buildlib/cbuild
+++ b/buildlib/cbuild
@@ -182,6 +182,12 @@ class jessie(APTEnvironment):
     name = "debian-8";
     aliases = {"jessie"};
 
+class stretch(APTEnvironment):
+    docker_parent = "debian:9"
+    pkgs = trusty.pkgs;
+    name = "debian-9";
+    aliases = {"stretch"};
+
 class debian_experimental(APTEnvironment):
     docker_parent = "debian:experimental"
     pkgs = (trusty.pkgs ^ {"gcc"}) | {"gcc-7"};
@@ -318,6 +324,7 @@ environments = [centos6(),
                 trusty(),
                 xenial(),
                 jessie(),
+                stretch(),
                 fc26(),
                 harlequin(),
                 malachite(),
-- 
2.7.4

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH rdma-core 04/21] cbuild: Use opensuse leap 42.2

[Jason Gunthorpe]

Newest release available on dockerhub

Signed-off-by: Jason Gunthorpe <jgunthorpe-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
---
 buildlib/cbuild | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/buildlib/cbuild b/buildlib/cbuild
index 48a8c1ea48ea5e..50e478154b0295 100755
--- a/buildlib/cbuild
+++ b/buildlib/cbuild
@@ -304,11 +304,11 @@ class harlequin(ZypperEnvironment):
     name = "opensuse-13.2";
     aliases = {"harelequin"};
 
-class malachite(ZypperEnvironment):
-    docker_parent = "opensuse:42.1";
+class leap(ZypperEnvironment):
+    docker_parent = "opensuse:42.2";
     pkgs = harlequin.pkgs;
-    name = "opensuse-42.1";
-    aliases = {"malachite"};
+    name = "opensuse-42.2";
+    aliases = {"leap"};
 
 class tumbleweed(ZypperEnvironment):
     docker_parent = "opensuse:tumbleweed";
@@ -327,7 +327,7 @@ environments = [centos6(),
                 stretch(),
                 fc26(),
                 harlequin(),
-                malachite(),
+                leap(),
                 tumbleweed(),
                 debian_experimental(),
 ];
-- 
2.7.4

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH rdma-core 05/21] Make 'cbuild travis pkg' run sparse like real travis

[Jason Gunthorpe]

Provide a git dir in the container and compute the package set
relative to the current top of tree on github.

Signed-off-by: Jason Gunthorpe <jgunthorpe-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
---
 .travis.yml     |  1 +
 buildlib/cbuild | 12 ++++++++++++
 2 files changed, 13 insertions(+)

diff --git a/.travis.yml b/.travis.yml
index 4eba65be3353ca..c190a2418b651a 100644
--- a/.travis.yml
+++ b/.travis.yml
@@ -26,6 +26,7 @@ addons:
       - fakeroot
       - gcc
       - gcc-7
+      - git
       - libnl-3-dev
       - libnl-route-3-dev
       - libudev-dev
diff --git a/buildlib/cbuild b/buildlib/cbuild
index 50e478154b0295..888fae6ccca692 100755
--- a/buildlib/cbuild
+++ b/buildlib/cbuild
@@ -550,6 +550,16 @@ def run_travis_build(args,env):
             subprocess.check_call(["git",
                                    "--git-dir",os.path.join(opwd,".git"),
                                    "reset","--hard","HEAD"]);
+            subprocess.check_call(["git",
+                                   "--git-dir",os.path.join(opwd,".git"),
+                                   "fetch",
+                                   "--no-tags",
+                                   "https://github.com/linux-rdma/rdma-core.git","HEAD",
+                                   "master"]);
+            base = subprocess.check_output(["git",
+                                            "--git-dir",os.path.join(opwd,".git"),
+                                            "merge-base",
+                                            "HEAD","FETCH_HEAD"]).strip();
 
         home = os.path.join(os.path.sep,"home","travis");
         home_build = os.path.join(os.path.sep,home,"build");
@@ -559,8 +569,10 @@ def run_travis_build(args,env):
             "--read-only",
             "--rm=true",
             "-v","%s:%s"%(tmpdir, home_build),
+            "-v","%s:%s:ro"%(os.path.join(opwd,".git"),os.path.join(home_build,"src",".git")),
             "-w",os.path.join(home_build,"src"),
             "-u",str(os.getuid()),
+            "-e","TRAVIS_COMMIT_RANGE=%s..HEAD"%(base),
             "-e","HOME=%s"%(home),
             "-e","TMPDIR=%s"%(os.path.join(home_build,"tmp")),
         ];
-- 
2.7.4

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH rdma-core 06/21] Tidy travis sparse check

[Jason Gunthorpe]

- Use "x$FOO" notation to avoid failures if TRAVIS_COMMIT_RANGE is not
  defined
- Use wget -q to be quieter
- Disable CONST_STRUCT since it does not work

Signed-off-by: Jason Gunthorpe <jgunthorpe-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
---
 buildlib/travis-checkpatch | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/buildlib/travis-checkpatch b/buildlib/travis-checkpatch
index 8f80d025a96828..ef7df4e62df6d6 100755
--- a/buildlib/travis-checkpatch
+++ b/buildlib/travis-checkpatch
@@ -6,13 +6,13 @@
 # And the rdma-core code is not mature enough to be warning safe
 # set -e
 
-if [ $TRAVIS_COMMIT_RANGE != "" ]; then
+if [ "x$TRAVIS_COMMIT_RANGE" != "x" ]; then
 	cd buildlib/
-	wget https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/plain/scripts/checkpatch.pl
-	wget https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/plain/scripts/spelling.txt
+	wget -q https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/plain/scripts/checkpatch.pl \
+	        https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/plain/scripts/spelling.txt
 	DIR_FOR_PATCHES_TO_CHECK=$(mktemp -d)
 	git format-patch --no-cover-letter $TRAVIS_COMMIT_RANGE -o $DIR_FOR_PATCHES_TO_CHECK/
-	CHECKPATCH_OPT="--no-tree --ignore PREFER_KERNEL_TYPES,FILE_PATH_CHANGES,EXECUTE_PERMISSIONS,USE_NEGATIVE_ERRNO $DIR_FOR_PATCHES_TO_CHECK/*"
+	CHECKPATCH_OPT="--no-tree --ignore PREFER_KERNEL_TYPES,FILE_PATH_CHANGES,EXECUTE_PERMISSIONS,USE_NEGATIVE_ERRNO,CONST_STRUCT $DIR_FOR_PATCHES_TO_CHECK/*"
 	perl checkpatch.pl $CHECKPATCH_OPT
 	if [ $? -ne 0 ]; then
 		# We rerun checkpatch to simplify parsing and to understand if we failed for errors
-- 
2.7.4

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH rdma-core 07/21] redhat: Fix redhat packaging of /etc/rdma/ibacm_opts.cfg

[Jason Gunthorpe]

It was ending up in rdma-core.rpm and ibacm.rpm, it belongs in
ibacm only.

Fixes: 0576db26f998 ("redhat spec: Avoid conflict due to implicit directory add")
Signed-off-by: Jason Gunthorpe <jgunthorpe-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
---
 redhat/rdma-core.spec | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/redhat/rdma-core.spec b/redhat/rdma-core.spec
index 817013bc4905f8..6dbba733fe3236 100644
--- a/redhat/rdma-core.spec
+++ b/redhat/rdma-core.spec
@@ -319,7 +319,9 @@ rm -rf %{buildroot}/%{_initrddir}/
 %dir %{_sysconfdir}/rdma
 %dir %{_docdir}/%{name}-%{version}
 %doc %{_docdir}/%{name}-%{version}/README.md
-%config(noreplace) %{_sysconfdir}/rdma/*
+%config(noreplace) %{_sysconfdir}/rdma/mlx4.conf
+%config(noreplace) %{_sysconfdir}/rdma/rdma.conf
+%config(noreplace) %{_sysconfdir}/rdma/sriov-vfs
 %config(noreplace) %{_sysconfdir}/udev/rules.d/*
 %config(noreplace) %{_sysconfdir}/modprobe.d/mlx4.conf
 %config(noreplace) %{_sysconfdir}/modprobe.d/truescale.conf
-- 
2.7.4

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH rdma-core 08/21] redhat: Fix duplicated udev rules in redhat packaging

[Jason Gunthorpe]

/usr/lib/udev/rules.d/libibumad.rules was claimed by rdma-core and
libibumad.rpm

/usr/lib/udev/rules.d/srp_daemon.rules was claimed by rdma-core and
srp_daemon.rpm

Fixes: 5a6f287a3cd ("libibumad: Introduce the /dev/infiniband/umad/${ibdev}:${port}.device alias")
Fixes: d35e5c4f716 ("srp_daemon.service: Add support for hot-plugging")
Signed-off-by: Jason Gunthorpe <jgunthorpe-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
---
 redhat/rdma-core.spec | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/redhat/rdma-core.spec b/redhat/rdma-core.spec
index 6dbba733fe3236..f6cf1027e404e4 100644
--- a/redhat/rdma-core.spec
+++ b/redhat/rdma-core.spec
@@ -329,7 +329,8 @@ rm -rf %{buildroot}/%{_initrddir}/
 %{_unitdir}/rdma.service
 %dir %{dracutlibdir}/modules.d/05rdma
 %{dracutlibdir}/modules.d/05rdma/module-setup.sh
-%{_udevrulesdir}/*
+%{_udevrulesdir}/98-rdma.rules
+%{_udevrulesdir}/rdma-ndd.rules
 %{sysmodprobedir}/libmlx4.conf
 %{sysmodprobedir}/cxgb3.conf
 %{sysmodprobedir}/cxgb4.conf
-- 
2.7.4

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH rdma-core 09/21] redhat: Fix duplicated libmlx*.so in redhat packaging

[Jason Gunthorpe]

/usr/lib64/libmlx4.so and 5 were being claimed by both rdma-core-devel.rpm
and libibverbs.rpm

The *.so symlinks belong only in the -devel package.

Fixes: a61f2a6850b54 ("mlx4: Export mlx4 direct verbs interface")
Fixes: 07764bc805ba8 ("mlx5: Export mlx5 direct verbs interface")
Signed-off-by: Jason Gunthorpe <jgunthorpe-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
---
 redhat/rdma-core.spec | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/redhat/rdma-core.spec b/redhat/rdma-core.spec
index f6cf1027e404e4..673ae0da1e5525 100644
--- a/redhat/rdma-core.spec
+++ b/redhat/rdma-core.spec
@@ -366,8 +366,8 @@ rm -rf %{buildroot}/%{_initrddir}/
 %dir %{_libdir}/libibverbs
 %{_libdir}/libibverbs*.so.*
 %{_libdir}/libibverbs/*.so
-%{_libdir}/libmlx5.so*
-%{_libdir}/libmlx4.so*
+%{_libdir}/libmlx5.so.*
+%{_libdir}/libmlx4.so.*
 %config(noreplace) %{_sysconfdir}/libibverbs.d/*.driver
 %doc %{_docdir}/%{name}-%{version}/libibverbs.md
 %doc %{_docdir}/%{name}-%{version}/rxe.md
-- 
2.7.4

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH rdma-core 10/21] debian: Set the run directory path to /run/

[Jason Gunthorpe]

cmake defaults this to /usr/var/run if left alone. This causes srp_daemon to
not work properly on Debian derived.

Signed-off-by: Jason Gunthorpe <jgunthorpe-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
---
 debian/rules | 1 +
 1 file changed, 1 insertion(+)

diff --git a/debian/rules b/debian/rules
index 6dc8a59bd6a9f1..847194d750845f 100755
--- a/debian/rules
+++ b/debian/rules
@@ -23,6 +23,7 @@ override_dh_auto_configure:
 			-DCMAKE_INSTALL_SYSTEMD_SERVICEDIR:PATH=/lib/systemd/system \
 			-DCMAKE_INSTALL_INITDDIR:PATH=/etc/init.d \
 			-DCMAKE_INSTALL_LIBEXECDIR:PATH=/usr/lib \
+			-DCMAKE_INSTALL_RUNDIR:PATH=/run \
 			-DCMAKE_INSTALL_UDEV_RULESDIR:PATH=/lib/udev/rules.d
 
 override_dh_auto_build:
-- 
2.7.4

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH rdma-core 11/21] debian: Remove infiniband-diags Depends from iwpmd

[Jason Gunthorpe]

This was a copy and paste error.

Fixes: bba93a9ce37 ("Debian Packaging")
Signed-off-by: Jason Gunthorpe <jgunthorpe-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
---
 debian/control | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/debian/control b/debian/control
index 34b275dfeb8c62..8434bbb23693f8 100644
--- a/debian/control
+++ b/debian/control
@@ -47,8 +47,7 @@ Description: InfiniBand Communication Manager Assistant (ACM)
 
 Package: iwpmd
 Architecture: linux-any
-Depends: infiniband-diags,
-         lsb-base (>= 3.2-14~),
+Depends: lsb-base (>= 3.2-14~),
          ${misc:Depends},
          ${shlibs:Depends}
 Description: Userspace component for iWarp RDMA services
-- 
2.7.4

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH rdma-core 12/21] debian: Move truescale setup into rdma-core

[Jason Gunthorpe]

This matches what RedHat is doing and makes sense, the rdma-core package
contains the kernel support elements.

Also add a Debian dependency for dmidecode as the truescale.cmd script
uses it.

Signed-off-by: Jason Gunthorpe <jgunthorpe-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
---
 debian/control                   | 2 +-
 debian/ibverbs-providers.install | 2 --
 debian/rdma-core.install         | 2 ++
 3 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/debian/control b/debian/control
index 8434bbb23693f8..9d80e38bb07864 100644
--- a/debian/control
+++ b/debian/control
@@ -22,7 +22,7 @@ Homepage: https://github.com/linux-rdma/rdma-core
 
 Package: rdma-core
 Architecture: linux-any
-Depends: ${misc:Depends}, ${shlibs:Depends}
+Depends: dmidecode, ${misc:Depends}, ${shlibs:Depends}
 Breaks: infiniband-diags (<< 2.0.0)
 Replaces: infiniband-diags (<< 2.0.0)
 Description: RDMA core userspace infrastructure and documentation
diff --git a/debian/ibverbs-providers.install b/debian/ibverbs-providers.install
index e477cd7873b8bd..93144527be04ef 100644
--- a/debian/ibverbs-providers.install
+++ b/debian/ibverbs-providers.install
@@ -1,10 +1,8 @@
 etc/libibverbs.d/
-etc/modprobe.d/truescale.conf
 usr/bin/rxe_cfg
 usr/lib/*/libibverbs/lib*-rdmav2.so
 usr/lib/*/libmlx4.so.*
 usr/lib/*/libmlx5.so.*
-usr/lib/truescale-serdes.cmds
 usr/share/doc/rdma-core/rxe.md usr/share/doc/ibverbs-providers/
 usr/share/man/man7/rxe.7
 usr/share/man/man8/rxe_cfg.8
diff --git a/debian/rdma-core.install b/debian/rdma-core.install
index 371aaadc7483d3..871f5ff543f2bf 100644
--- a/debian/rdma-core.install
+++ b/debian/rdma-core.install
@@ -1,5 +1,7 @@
+etc/modprobe.d/truescale.conf
 lib/systemd/system/rdma-ndd.service
 lib/udev/rules.d/rdma-ndd.rules
+usr/lib/truescale-serdes.cmds
 usr/sbin/rdma-ndd
 usr/share/doc/rdma-core/MAINTAINERS
 usr/share/doc/rdma-core/README.md
-- 
2.7.4

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH rdma-core 13/21] Prefix udev rules with a number

[Jason Gunthorpe]

This follows the standard convention for udev.

I could not find an outline of where different activities should fall
within the numbering scheme, so these are a guess:

- 90 for libibumad rules because it only sets SYSTEMD_ALIAS and the only
  other rules to do that are labeled 99
- 60 for srp_daemon/ndd because other rules dealing with matching
  hardware seem to be around there as well

Fixes: 5a6f287a3cdf ("libibumad: Introduce the /dev/infiniband/umad/${ibdev}:${port}.device alias")
Fixes: d35e5c4f7169 ("srp_daemon.service: Add support for hot-plugging")
Fixes: 4910f1eb752e ("rdma-ndd: add rdma-ndd from infiniband-diags")
Signed-off-by: Jason Gunthorpe <jgunthorpe-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
---
 debian/libibumad3.install | 2 +-
 debian/rdma-core.install  | 2 +-
 debian/srptools.install   | 2 +-
 libibumad/CMakeLists.txt  | 4 ++--
 rdma-ndd/CMakeLists.txt   | 4 +++-
 redhat/rdma-core.spec     | 6 +++---
 srp_daemon/CMakeLists.txt | 4 +++-
 7 files changed, 14 insertions(+), 10 deletions(-)

diff --git a/debian/libibumad3.install b/debian/libibumad3.install
index 80a40982833e63..41339ebfa2f229 100644
--- a/debian/libibumad3.install
+++ b/debian/libibumad3.install
@@ -1,2 +1,2 @@
-lib/udev/rules.d/libibumad.rules
+lib/udev/rules.d/90-libibumad.rules
 usr/lib/*/libibumad*.so.*
diff --git a/debian/rdma-core.install b/debian/rdma-core.install
index 871f5ff543f2bf..4bab264208b691 100644
--- a/debian/rdma-core.install
+++ b/debian/rdma-core.install
@@ -1,6 +1,6 @@
 etc/modprobe.d/truescale.conf
 lib/systemd/system/rdma-ndd.service
-lib/udev/rules.d/rdma-ndd.rules
+lib/udev/rules.d/60-rdma-ndd.rules
 usr/lib/truescale-serdes.cmds
 usr/sbin/rdma-ndd
 usr/share/doc/rdma-core/MAINTAINERS
diff --git a/debian/srptools.install b/debian/srptools.install
index d1c9a42c890fcd..f780d663c9113b 100644
--- a/debian/srptools.install
+++ b/debian/srptools.install
@@ -1,7 +1,7 @@
 etc/srp_daemon.conf
 lib/systemd/system/srp_daemon.service
 lib/systemd/system/srp_daemon_port@.service
-lib/udev/rules.d/srp_daemon.rules
+lib/udev/rules.d/60-srp_daemon.rules
 usr/lib/srp_daemon/start_on_all_ports
 usr/sbin/ibsrpdm
 usr/sbin/srp_daemon
diff --git a/libibumad/CMakeLists.txt b/libibumad/CMakeLists.txt
index 150ef59b920227..84673ee01271cb 100644
--- a/libibumad/CMakeLists.txt
+++ b/libibumad/CMakeLists.txt
@@ -15,7 +15,7 @@ rdma_library(ibumad libibumad.map
   umad_str.c
   )
 
-rdma_subst_install(FILES libibumad.udev-rules
-  RENAME libibumad.rules
+install(FILES libibumad.udev-rules
+  RENAME 90-libibumad.rules
   DESTINATION "${CMAKE_INSTALL_UDEV_RULESDIR}")
 
diff --git a/rdma-ndd/CMakeLists.txt b/rdma-ndd/CMakeLists.txt
index 942d90b8943fe8..99e183e34ef758 100644
--- a/rdma-ndd/CMakeLists.txt
+++ b/rdma-ndd/CMakeLists.txt
@@ -14,7 +14,9 @@ rdma_man_pages(
   rdma-ndd.8.in
   )
 
-install(FILES "rdma-ndd.rules" DESTINATION "${CMAKE_INSTALL_UDEV_RULESDIR}")
+install(FILES "rdma-ndd.rules"
+  RENAME "60-rdma-ndd.rules"
+  DESTINATION "${CMAKE_INSTALL_UDEV_RULESDIR}")
 
 rdma_subst_install(FILES "rdma-ndd.service.in"
   DESTINATION "${CMAKE_INSTALL_SYSTEMD_SERVICEDIR}"
diff --git a/redhat/rdma-core.spec b/redhat/rdma-core.spec
index 673ae0da1e5525..8961d33e618b1d 100644
--- a/redhat/rdma-core.spec
+++ b/redhat/rdma-core.spec
@@ -330,7 +330,7 @@ rm -rf %{buildroot}/%{_initrddir}/
 %dir %{dracutlibdir}/modules.d/05rdma
 %{dracutlibdir}/modules.d/05rdma/module-setup.sh
 %{_udevrulesdir}/98-rdma.rules
-%{_udevrulesdir}/rdma-ndd.rules
+%{_udevrulesdir}/60-rdma-ndd.rules
 %{sysmodprobedir}/libmlx4.conf
 %{sysmodprobedir}/cxgb3.conf
 %{sysmodprobedir}/cxgb4.conf
@@ -405,7 +405,7 @@ rm -rf %{buildroot}/%{_initrddir}/
 
 %files -n libibumad
 %{_libdir}/libibumad*.so.*
-%{_udevrulesdir}/libibumad.rules
+%{_udevrulesdir}/90-libibumad.rules
 
 %files -n librdmacm
 %{_libdir}/librdmacm*.so.*
@@ -450,7 +450,7 @@ rm -rf %{buildroot}/%{_initrddir}/
 %{_sbindir}/ibsrpdm
 %{_sbindir}/srp_daemon
 %{_sbindir}/run_srp_daemon
-%{_udevrulesdir}/srp_daemon.rules
+%{_udevrulesdir}/60-srp_daemon.rules
 %{_mandir}/man1/ibsrpdm.1*
 %{_mandir}/man1/srp_daemon.1*
 %{_mandir}/man5/srp_daemon.service.5*
diff --git a/srp_daemon/CMakeLists.txt b/srp_daemon/CMakeLists.txt
index 62e91bb0f321d8..92a1d709364b0b 100644
--- a/srp_daemon/CMakeLists.txt
+++ b/srp_daemon/CMakeLists.txt
@@ -43,7 +43,9 @@ rdma_subst_install(FILES srp_daemon_port@.service.in
 
 install(FILES srp_daemon.conf DESTINATION "${CMAKE_INSTALL_SYSCONFDIR}")
 
-install(FILES srp_daemon.rules DESTINATION "${CMAKE_INSTALL_UDEV_RULESDIR}")
+install(FILES "srp_daemon.rules"
+  RENAME "60-srp_daemon.rules"
+  DESTINATION "${CMAKE_INSTALL_UDEV_RULESDIR}")
 
 # FIXME: The ib init.d file should really be included in rdma-core as well.
 set(RDMA_SERVICE "openibd" CACHE STRING "init.d file service name to order srpd after")
-- 
2.7.4

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH rdma-core 14/21] Revise the umad SYSTEMD_ALIAS fake path

[Jason Gunthorpe]

The only other users of SYSTEMD_ALIAS I could find uses the
/sys/subsystem/ prefix, so we should do the same instead of using
a fake /dev/ prefix.

Fixes: 5a6f287a3cdf ("libibumad: Introduce the /dev/infiniband/umad/${ibdev}:${port}.device alias")
Fixes: d35e5c4f7169 ("srp_daemon.service: Add support for hot-plugging")
Signed-off-by: Jason Gunthorpe <jgunthorpe-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
---
 libibumad/libibumad.udev-rules         | 2 +-
 srp_daemon/srp_daemon_port@.service.in | 4 ++--
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/libibumad/libibumad.udev-rules b/libibumad/libibumad.udev-rules
index 8b1ab6b5142009..b9f565eceaf3a3 100644
--- a/libibumad/libibumad.udev-rules
+++ b/libibumad/libibumad.udev-rules
@@ -1 +1 @@
-ACTION=="add", SUBSYSTEM=="infiniband_mad", TAG+="systemd", ENV{SYSTEMD_ALIAS}="/dev/infiniband/umad/$attr{ibdev}:$attr{port}"
+ACTION=="add", SUBSYSTEM=="infiniband_mad", TAG+="systemd", ENV{SYSTEMD_ALIAS}="/sys/subsystem/rdma/devices/$attr{ibdev}:$attr{port}/umad"
diff --git a/srp_daemon/srp_daemon_port@.service.in b/srp_daemon/srp_daemon_port@.service.in
index 0ec966f912aec8..03b9cb9a70119a 100644
--- a/srp_daemon/srp_daemon_port@.service.in
+++ b/srp_daemon/srp_daemon_port@.service.in
@@ -3,8 +3,8 @@ Description=SRP daemon that monitors port %i
 Documentation=man:srp_daemon file:/etc/rdma/rdma.conf file:/etc/srp_daemon.conf
 DefaultDependencies=false
 Conflicts=emergency.target emergency.service
-After=srp_daemon.service dev-infiniband-umad-%i.device network.target
-BindsTo=srp_daemon.service dev-infiniband-umad-%i.device
+After=srp_daemon.service sys-subsystem-rdma-devices-%i-umad.device network.target
+BindsTo=srp_daemon.service sys-subsystem-rdma-devices-%i-umad.device
 Before=remote-fs-pre.target
 
 [Service]
-- 
2.7.4

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH rdma-core 15/21] umad: Match only umad kernel devices in the udev rule

[Jason Gunthorpe]

Otherwise udev matches issm and umad for each resulting in systemd
warnings:

 systemd[1]: sys-subsystem-rdma-devices-mlx4_0:1-umad.device: Dev sys-subsystem-rdma-devices-mlx4_0:1-umad.device appeared twice with different sysfs paths /sys/devices/pci0000:00/0000:00:04.0/infiniband_mad/issm0 and /sys/devices/pci0000:00/0000:00:04.0/infiniband_mad/umad0

Fixes: d35e5c4f7169 ("srp_daemon.service: Add support for hot-plugging")
Fixes: 5a6f287a3cdf ("libibumad: Introduce the /dev/infiniband/umad/${ibdev}:${port}.device alias")
Signed-off-by: Jason Gunthorpe <jgunthorpe-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
---
 libibumad/libibumad.udev-rules | 2 +-
 srp_daemon/srp_daemon.rules    | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/libibumad/libibumad.udev-rules b/libibumad/libibumad.udev-rules
index b9f565eceaf3a3..cfb790e051ced7 100644
--- a/libibumad/libibumad.udev-rules
+++ b/libibumad/libibumad.udev-rules
@@ -1 +1 @@
-ACTION=="add", SUBSYSTEM=="infiniband_mad", TAG+="systemd", ENV{SYSTEMD_ALIAS}="/sys/subsystem/rdma/devices/$attr{ibdev}:$attr{port}/umad"
+ACTION=="add", SUBSYSTEM=="infiniband_mad", KERNEL=="*umad*", TAG+="systemd", ENV{SYSTEMD_ALIAS}="/sys/subsystem/rdma/devices/$attr{ibdev}:$attr{port}/umad"
diff --git a/srp_daemon/srp_daemon.rules b/srp_daemon/srp_daemon.rules
index 1210adef5b592b..6780d5421de7b4 100644
--- a/srp_daemon/srp_daemon.rules
+++ b/srp_daemon/srp_daemon.rules
@@ -1 +1 @@
-ACTION=="add", SUBSYSTEM=="infiniband_mad", PROGRAM:="/usr/bin/systemctl show srp_daemon -p ActiveState", RESULT=="ActiveState=active", ENV{SYSTEMD_WANTS}+="srp_daemon_port@$attr{ibdev}:$attr{port}.service"
+ACTION=="add", SUBSYSTEM=="infiniband_mad", KERNEL=="*umad*", PROGRAM:="/usr/bin/systemctl show srp_daemon -p ActiveState", RESULT=="ActiveState=active", ENV{SYSTEMD_WANTS}+="srp_daemon_port@$attr{ibdev}:$attr{port}.service"
-- 
2.7.4

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH rdma-core 16/21] srp_daemon: Do not use LOG_PERROR

[Jason Gunthorpe]

When running from systemd syslog will send messages to the log, but
so will stderr. Thus using LOG_PERROR doubles every log message.

Signed-off-by: Jason Gunthorpe <jgunthorpe-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
---
 srp_daemon/srp_daemon.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/srp_daemon/srp_daemon.c b/srp_daemon/srp_daemon.c
index c0e8d23d58faf5..f0dfe260945574 100644
--- a/srp_daemon/srp_daemon.c
+++ b/srp_daemon/srp_daemon.c
@@ -2090,7 +2090,7 @@ int main(int argc, char *argv[])
 		goto restore_sig;
 	}
 
-	openlog("srp_daemon", LOG_PID | LOG_PERROR, LOG_DAEMON);
+	openlog("srp_daemon", LOG_PID, LOG_DAEMON);
 
 	config = calloc(1, sizeof(*config));
 	if (!config) {
-- 
2.7.4

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH rdma-core 17/21] srp_deamon: Using v formatters directly for pr_err

[Jason Gunthorpe]

No reason to format to a string buffer and then just print that string,
use vsyslog and vfprintf directly instead.

Signed-off-by: Jason Gunthorpe <jgunthorpe-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
---
 srp_daemon/srp_daemon.c | 11 +++--------
 1 file changed, 3 insertions(+), 8 deletions(-)

diff --git a/srp_daemon/srp_daemon.c b/srp_daemon/srp_daemon.c
index f0dfe260945574..a75f3af085cbde 100644
--- a/srp_daemon/srp_daemon.c
+++ b/srp_daemon/srp_daemon.c
@@ -295,22 +295,17 @@ void pr_debug(const char *fmt, ...)
 void pr_err(const char *fmt, ...)
 {
 	va_list args;
-	int pos;
-	char str[1000];
 
 	va_start(args, fmt);
-	pos = vsnprintf(str, sizeof(str), fmt, args);
-	va_end(args);
-	if (pos >= sizeof(str))
-		str[sizeof(str) - 1] = '\0';
 	switch (s_log_dest) {
 	case log_to_syslog:
-		syslog(LOG_DAEMON | LOG_ERR, "%s", str);
+		vsyslog(LOG_DAEMON | LOG_ERR, fmt, args);
 		break;
 	case log_to_stderr:
-		fprintf(stderr, "%s", str);
+		vfprintf(stderr, fmt, args);
 		break;
 	}
+	va_end(args);
 }
 
 static int check_not_equal_str(const char *dir_name, const char *attr,
-- 
2.7.4

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH rdma-core 18/21] srp_daemon: Call systemctl properly from udev

[Jason Gunthorpe]

Use /bin/ as the call path for systemctl. For some reason Debian
derived distros do not have /usr/bin/systemctl, only /bin/systemctl.
Other distros seem to have both paths.

Also invoke PROGRAM using == which matches what the udev test suite does.

Fixes: d35e5c4f7169 ("srp_daemon.service: Add support for hot-plugging")
Signed-off-by: Jason Gunthorpe <jgunthorpe-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
---
 srp_daemon/srp_daemon.rules | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/srp_daemon/srp_daemon.rules b/srp_daemon/srp_daemon.rules
index 6780d5421de7b4..4cfeabc762da2a 100644
--- a/srp_daemon/srp_daemon.rules
+++ b/srp_daemon/srp_daemon.rules
@@ -1 +1 @@
-ACTION=="add", SUBSYSTEM=="infiniband_mad", KERNEL=="*umad*", PROGRAM:="/usr/bin/systemctl show srp_daemon -p ActiveState", RESULT=="ActiveState=active", ENV{SYSTEMD_WANTS}+="srp_daemon_port@$attr{ibdev}:$attr{port}.service"
+ACTION=="add", SUBSYSTEM=="infiniband_mad", KERNEL=="*umad*", PROGRAM=="/bin/systemctl show srp_daemon -p ActiveState", RESULT=="ActiveState=active", ENV{SYSTEMD_WANTS}+="srp_daemon_port@$attr{ibdev}:$attr{port}.service"
-- 
2.7.4

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH rdma-core 19/21] iwpmd: Fix install path for iwpmd

[Jason Gunthorpe]

iwpmd is not user runnable, so it should be installed in sbin.

Signed-off-by: Jason Gunthorpe <jgunthorpe-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
---
 debian/iwpmd.install                      | 2 +-
 iwpmd/CMakeLists.txt                      | 8 +++++---
 iwpmd/{iwpmd.service => iwpmd.service.in} | 2 +-
 iwpmd/{iwpmd_init => iwpmd_init.in}       | 2 +-
 redhat/rdma-core.spec                     | 2 +-
 5 files changed, 9 insertions(+), 7 deletions(-)
 rename iwpmd/{iwpmd.service => iwpmd.service.in} (80%)
 rename iwpmd/{iwpmd_init => iwpmd_init.in} (97%)
 mode change 100755 => 100644

diff --git a/debian/iwpmd.install b/debian/iwpmd.install
index bff17f15abf7d9..cd9f7adbdd287d 100644
--- a/debian/iwpmd.install
+++ b/debian/iwpmd.install
@@ -1,6 +1,6 @@
 etc/init.d/iwpmd
 etc/iwpmd.conf
 lib/systemd/system/iwpmd.service
-usr/bin/iwpmd
+usr/sbin/iwpmd
 usr/share/man/man1/iwpmd.1
 usr/share/man/man5/iwpmd.conf.5
diff --git a/iwpmd/CMakeLists.txt b/iwpmd/CMakeLists.txt
index ee9e5a3dfeb24e..200b14c1786422 100644
--- a/iwpmd/CMakeLists.txt
+++ b/iwpmd/CMakeLists.txt
@@ -1,4 +1,4 @@
-rdma_executable(iwpmd
+rdma_sbin_executable(iwpmd
   iwarp_pm_common.c
   iwarp_pm_helper.c
   iwarp_pm_server.c
@@ -13,8 +13,10 @@ rdma_man_pages(
   iwpmd.conf.5.in
   )
 
-install(FILES "iwpmd.service" DESTINATION "${CMAKE_INSTALL_SYSTEMD_SERVICEDIR}")
-install(FILES "iwpmd_init"
+rdma_subst_install(FILES "iwpmd.service.in"
+  RENAME "iwpmd.service"
+  DESTINATION "${CMAKE_INSTALL_SYSTEMD_SERVICEDIR}")
+rdma_subst_install(FILES "iwpmd_init.in"
   DESTINATION "${CMAKE_INSTALL_INITDDIR}"
   RENAME "iwpmd"
   PERMISSIONS OWNER_WRITE OWNER_READ GROUP_READ WORLD_READ OWNER_EXECUTE GROUP_EXECUTE WORLD_EXECUTE)
diff --git a/iwpmd/iwpmd.service b/iwpmd/iwpmd.service.in
similarity index 80%
rename from iwpmd/iwpmd.service
rename to iwpmd/iwpmd.service.in
index 6d093fb2fda5a7..47ca6518ee127c 100644
--- a/iwpmd/iwpmd.service
+++ b/iwpmd/iwpmd.service.in
@@ -4,7 +4,7 @@ Documentation=man:iwpmd file:/etc/iwpmd.conf
 After=network.target
 
 [Service]
-ExecStart=/usr/bin/iwpmd
+ExecStart=@CMAKE_INSTALL_FULL_SBINDIR@/iwpmd
 LimitNOFILE=102400
 KillMode=process
 
diff --git a/iwpmd/iwpmd_init b/iwpmd/iwpmd_init.in
old mode 100755
new mode 100644
similarity index 97%
rename from iwpmd/iwpmd_init
rename to iwpmd/iwpmd_init.in
index 0d2559e9de4a98..f3ae269c795c89
--- a/iwpmd/iwpmd_init
+++ b/iwpmd/iwpmd_init.in
@@ -16,7 +16,7 @@
 # Description: iWarp Port Mapper Daemon for opening sockets to claim TCP ports from userspace
 ### END INIT INFO
 
-IWPMD_BIN="/usr/bin/iwpmd"
+IWPMD_BIN="@CMAKE_INSTALL_FULL_SBINDIR@/iwpmd"
 LOCK="/var/lock/subsys/iwpmd"
 IWPMD_PID=0
 RETVAL=0
diff --git a/redhat/rdma-core.spec b/redhat/rdma-core.spec
index 8961d33e618b1d..fb6d7ffc4f29fa 100644
--- a/redhat/rdma-core.spec
+++ b/redhat/rdma-core.spec
@@ -393,7 +393,7 @@ rm -rf %{buildroot}/%{_initrddir}/
 %doc %{_docdir}/%{name}-%{version}/ibacm.md
 
 %files -n iwpmd
-%{_bindir}/iwpmd
+%{_sbindir}/iwpmd
 %{_unitdir}/iwpmd.service
 %config(noreplace) %{_sysconfdir}/iwpmd.conf
 %{_mandir}/man1/iwpmd.*
-- 
2.7.4

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH rdma-core 20/21] Move rxe_cfg to the rdma-core package

[Jason Gunthorpe]

rx_cfg is used to setup the kernel rxe device, it belongs with the
rest of the kernel support items.

Signed-off-by: Jason Gunthorpe <jgunthorpe-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
---
 debian/ibverbs-providers.install | 4 ----
 debian/rdma-core.install         | 4 ++++
 redhat/rdma-core.spec            | 8 ++++----
 3 files changed, 8 insertions(+), 8 deletions(-)

diff --git a/debian/ibverbs-providers.install b/debian/ibverbs-providers.install
index 93144527be04ef..73088d82f6e0e8 100644
--- a/debian/ibverbs-providers.install
+++ b/debian/ibverbs-providers.install
@@ -1,8 +1,4 @@
 etc/libibverbs.d/
-usr/bin/rxe_cfg
 usr/lib/*/libibverbs/lib*-rdmav2.so
 usr/lib/*/libmlx4.so.*
 usr/lib/*/libmlx5.so.*
-usr/share/doc/rdma-core/rxe.md usr/share/doc/ibverbs-providers/
-usr/share/man/man7/rxe.7
-usr/share/man/man8/rxe_cfg.8
diff --git a/debian/rdma-core.install b/debian/rdma-core.install
index 4bab264208b691..3d8cddab9297a5 100644
--- a/debian/rdma-core.install
+++ b/debian/rdma-core.install
@@ -1,8 +1,12 @@
 etc/modprobe.d/truescale.conf
 lib/systemd/system/rdma-ndd.service
 lib/udev/rules.d/60-rdma-ndd.rules
+usr/bin/rxe_cfg
 usr/lib/truescale-serdes.cmds
 usr/sbin/rdma-ndd
 usr/share/doc/rdma-core/MAINTAINERS
 usr/share/doc/rdma-core/README.md
+usr/share/doc/rdma-core/rxe.md usr/share/doc/ibverbs-providers/
+usr/share/man/man7/rxe.7
 usr/share/man/man8/rdma-ndd.8
+usr/share/man/man8/rxe_cfg.8
diff --git a/redhat/rdma-core.spec b/redhat/rdma-core.spec
index fb6d7ffc4f29fa..40797f268e3d18 100644
--- a/redhat/rdma-core.spec
+++ b/redhat/rdma-core.spec
@@ -319,6 +319,7 @@ rm -rf %{buildroot}/%{_initrddir}/
 %dir %{_sysconfdir}/rdma
 %dir %{_docdir}/%{name}-%{version}
 %doc %{_docdir}/%{name}-%{version}/README.md
+%doc %{_docdir}/%{name}-%{version}/rxe.md
 %config(noreplace) %{_sysconfdir}/rdma/mlx4.conf
 %config(noreplace) %{_sysconfdir}/rdma/rdma.conf
 %config(noreplace) %{_sysconfdir}/rdma/sriov-vfs
@@ -339,9 +340,12 @@ rm -rf %{buildroot}/%{_initrddir}/
 %{_libexecdir}/rdma-fixup-mtrr.awk
 %{_libexecdir}/mlx4-setup.sh
 %{_libexecdir}/truescale-serdes.cmds
+%{_bindir}/rxe_cfg
 %{_sbindir}/rdma-ndd
 %{_unitdir}/rdma-ndd.service
+%{_mandir}/man7/rxe*
 %{_mandir}/man8/rdma-ndd.*
+%{_mandir}/man8/rxe*
 %license COPYING.*
 
 %files devel
@@ -370,10 +374,6 @@ rm -rf %{buildroot}/%{_initrddir}/
 %{_libdir}/libmlx4.so.*
 %config(noreplace) %{_sysconfdir}/libibverbs.d/*.driver
 %doc %{_docdir}/%{name}-%{version}/libibverbs.md
-%doc %{_docdir}/%{name}-%{version}/rxe.md
-%{_bindir}/rxe_cfg
-%{_mandir}/man7/rxe*
-%{_mandir}/man8/rxe*
 
 %files -n libibverbs-utils
 %{_bindir}/ibv_*
-- 
2.7.4

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH rdma-core 21/21] Make /etc/modprobe.d/mlx4.conf common

[Jason Gunthorpe]

Move it out of the redhat directory and into providers/mlx4, this follows
the existing pattern of truescale.conf.

Signed-off-by: Jason Gunthorpe <jgunthorpe-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
---
 debian/rdma-core.install                                   | 1 +
 providers/mlx4/CMakeLists.txt                              | 2 ++
 redhat/rdma.mlx4.user.modprobe => providers/mlx4/mlx4.conf | 0
 redhat/rdma-core.spec                                      | 1 -
 4 files changed, 3 insertions(+), 1 deletion(-)
 rename redhat/rdma.mlx4.user.modprobe => providers/mlx4/mlx4.conf (100%)

diff --git a/debian/rdma-core.install b/debian/rdma-core.install
index 3d8cddab9297a5..c19d766fc0f48e 100644
--- a/debian/rdma-core.install
+++ b/debian/rdma-core.install
@@ -1,3 +1,4 @@
+etc/modprobe.d/mlx4.conf
 etc/modprobe.d/truescale.conf
 lib/systemd/system/rdma-ndd.service
 lib/udev/rules.d/60-rdma-ndd.rules
diff --git a/providers/mlx4/CMakeLists.txt b/providers/mlx4/CMakeLists.txt
index f6739d2dd0a7a2..2cd668cbbe213e 100644
--- a/providers/mlx4/CMakeLists.txt
+++ b/providers/mlx4/CMakeLists.txt
@@ -12,3 +12,5 @@ rdma_shared_provider(mlx4 libmlx4.map
 publish_headers(infiniband
   mlx4dv.h
 )
+
+install(FILES "mlx4.conf" DESTINATION "${CMAKE_INSTALL_SYSCONFDIR}/modprobe.d/")
diff --git a/redhat/rdma.mlx4.user.modprobe b/providers/mlx4/mlx4.conf
similarity index 100%
rename from redhat/rdma.mlx4.user.modprobe
rename to providers/mlx4/mlx4.conf
diff --git a/redhat/rdma-core.spec b/redhat/rdma-core.spec
index 40797f268e3d18..2565290e72c04e 100644
--- a/redhat/rdma-core.spec
+++ b/redhat/rdma-core.spec
@@ -257,7 +257,6 @@ install -D -m0755 redhat/rdma.ifup-ib %{buildroot}/%{_sysconfdir}/sysconfig/netw
 install -D -m0755 redhat/rdma.ifdown-ib %{buildroot}/%{_sysconfdir}/sysconfig/network-scripts/ifdown-ib
 install -D -m0644 redhat/rdma.service %{buildroot}%{_unitdir}/rdma.service
 install -D -m0644 redhat/rdma.udev-ipoib-naming.rules %{buildroot}%{_sysconfdir}/udev/rules.d/70-persistent-ipoib.rules
-install -D -m0644 redhat/rdma.mlx4.user.modprobe %{buildroot}%{_sysconfdir}/modprobe.d/mlx4.conf
 install -D -m0755 redhat/rdma.modules-setup.sh %{buildroot}%{dracutlibdir}/modules.d/05rdma/module-setup.sh
 install -D -m0644 redhat/rdma.udev-rules %{buildroot}%{_udevrulesdir}/98-rdma.rules
 install -D -m0644 redhat/rdma.mlx4.sys.modprobe %{buildroot}%{sysmodprobedir}/libmlx4.conf
-- 
2.7.4

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH rdma-core 16/21] srp_daemon: Do not use LOG_PERROR

[Bart Van Assche]

On Mon, 2017-07-24 at 14:00 -0600, Jason Gunthorpe wrote:
> When running from systemd syslog will send messages to the log, but
> so will stderr. Thus using LOG_PERROR doubles every log message.
> 
> Signed-off-by: Jason Gunthorpe <jgunthorpe-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
> ---
>  srp_daemon/srp_daemon.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/srp_daemon/srp_daemon.c b/srp_daemon/srp_daemon.c
> index c0e8d23d58faf5..f0dfe260945574 100644
> --- a/srp_daemon/srp_daemon.c
> +++ b/srp_daemon/srp_daemon.c
> @@ -2090,7 +2090,7 @@ int main(int argc, char *argv[])
>  		goto restore_sig;
>  	}
>  
> -	openlog("srp_daemon", LOG_PID | LOG_PERROR, LOG_DAEMON);
> +	openlog("srp_daemon", LOG_PID, LOG_DAEMON);
>  
>  	config = calloc(1, sizeof(*config));
>  	if (!config) {

Hello Jason,

LOG_PERROR is very convenient when debugging srp_daemon. How about using
the result of isatty(STDERR_FILENO) to decide whether or not to enable
LOG_PERROR?

All the other srp_daemon-related patches in this series look fine to me.
Thank you for having done this work!

Bart.--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH rdma-core 16/21] srp_daemon: Do not use LOG_PERROR

[Jason Gunthorpe]

On Tue, Jul 25, 2017 at 04:47:59PM +0000, Bart Van Assche wrote:

> LOG_PERROR is very convenient when debugging srp_daemon. How about using
> the result of isatty(STDERR_FILENO) to decide whether or not to enable
> LOG_PERROR?

How about a --debug option that sets

  s_log_dest = log_to_stderr;

?

> All the other srp_daemon-related patches in this series look fine to me.
> Thank you for having done this work!

Great, thanks, I'll add your Reviewed-by to SRP patches then?

Jason
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH rdma-core 16/21] srp_daemon: Do not use LOG_PERROR

[Bart Van Assche]

On Tue, 2017-07-25 at 11:17 -0600, Jason Gunthorpe wrote:
> On Tue, Jul 25, 2017 at 04:47:59PM +0000, Bart Van Assche wrote:
> > LOG_PERROR is very convenient when debugging srp_daemon. How about using
> > the result of isatty(STDERR_FILENO) to decide whether or not to enable
> > LOG_PERROR?
> 
> How about a --debug option that sets
> 
>   s_log_dest = log_to_stderr;

Sorry but I prefer that LOG_PERROR is retained when srp_daemon is started in
another way than by systemd. How about adding StandardError=null to the
srp_daemon unit file?

> > All the other srp_daemon-related patches in this series look fine to me.
> > Thank you for having done this work!
> 
> Great, thanks, I'll add your Reviewed-by to SRP patches then?

Please go ahead.

Bart.--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH rdma-core 16/21] srp_daemon: Do not use LOG_PERROR

[Jason Gunthorpe]

On Tue, Jul 25, 2017 at 05:28:10PM +0000, Bart Van Assche wrote:
> On Tue, 2017-07-25 at 11:17 -0600, Jason Gunthorpe wrote:
> > On Tue, Jul 25, 2017 at 04:47:59PM +0000, Bart Van Assche wrote:
> > > LOG_PERROR is very convenient when debugging srp_daemon. How about using
> > > the result of isatty(STDERR_FILENO) to decide whether or not to enable
> > > LOG_PERROR?
> > 
> > How about a --debug option that sets
> > 
> >   s_log_dest = log_to_stderr;
> 
> Sorry but I prefer that LOG_PERROR is retained when srp_daemon is started in
> another way than by systemd. How about adding StandardError=null to the
> srp_daemon unit file?

No, then emergency prints from libraries are lost forever eg glibc
malloc corruption prints.

How about a --systemd option? I think we are going to want that anyhow
down the road.

Jason
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH rdma-core 16/21] srp_daemon: Do not use LOG_PERROR

[Bart Van Assche]

On Tue, 2017-07-25 at 11:44 -0600, Jason Gunthorpe wrote:
> On Tue, Jul 25, 2017 at 05:28:10PM +0000, Bart Van Assche wrote:
> > On Tue, 2017-07-25 at 11:17 -0600, Jason Gunthorpe wrote:
> > > On Tue, Jul 25, 2017 at 04:47:59PM +0000, Bart Van Assche wrote:
> > > > LOG_PERROR is very convenient when debugging srp_daemon. How about using
> > > > the result of isatty(STDERR_FILENO) to decide whether or not to enable
> > > > LOG_PERROR?
> > > 
> > > How about a --debug option that sets
> > > 
> > >   s_log_dest = log_to_stderr;
> > 
> > Sorry but I prefer that LOG_PERROR is retained when srp_daemon is started in
> > another way than by systemd. How about adding StandardError=null to the
> > srp_daemon unit file?
> 
> No, then emergency prints from libraries are lost forever eg glibc
> malloc corruption prints.
> 
> How about a --systemd option? I think we are going to want that anyhow
> down the road.

Hello Jason,

That sounds fine to me.

Thanks,

Bart.--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH rdma-core 16/21] srp_daemon: Do not use LOG_PERROR

[Jason Gunthorpe]

On Tue, Jul 25, 2017 at 06:09:46PM +0000, Bart Van Assche wrote:
> That sounds fine to me.

Okay, I dropped the perror patch and replaced it with these two:

https://github.com/linux-rdma/rdma-core/pull/172/commits/f1f2e1b1bc3e90abdacdd06b3afe7ca4b107f9e3
https://github.com/linux-rdma/rdma-core/pull/172/commits/588c60a33ac5b846fee039a1cb8e622681fb980e

To implement a --systemd option to control the log behavior.

Jason
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH rdma-core 16/21] srp_daemon: Do not use LOG_PERROR

[Bart Van Assche]

On Tue, 2017-07-25 at 17:08 -0600, Jason Gunthorpe wrote:
> On Tue, Jul 25, 2017 at 06:09:46PM +0000, Bart Van Assche wrote:
> > That sounds fine to me.
> 
> Okay, I dropped the perror patch and replaced it with these two:
> 
> https://github.com/linux-rdma/rdma-core/pull/172/commits/f1f2e1b1bc3e90abdacdd06b3afe7ca4b107f9e3
> https://github.com/linux-rdma/rdma-core/pull/172/commits/588c60a33ac5b846fee039a1cb8e622681fb980e
> 
> To implement a --systemd option to control the log behavior.

Hello Jason,

Since both is_systemd() and get_config() ignore the local variable opt_idx,
how about passing NULL as the fifth argument to getopt_long()?

>From the getopt_long() man page: "The variable optind is the index of the
next element to be processed in argv. The system initializes this value to
1. The caller can reset it to 1 to restart scanning of the same argv, or
when scanning a new argument vector."

Since getopt_long() is called twice by srp_daemon, shouldn't get_config()
reset the optind global variable before calling getopt_long()?

Thanks,

Bart.--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH rdma-core 16/21] srp_daemon: Do not use LOG_PERROR

[Jason Gunthorpe]

On Tue, Jul 25, 2017 at 11:50:51PM +0000, Bart Van Assche wrote:
> On Tue, 2017-07-25 at 17:08 -0600, Jason Gunthorpe wrote:
> > On Tue, Jul 25, 2017 at 06:09:46PM +0000, Bart Van Assche wrote:
> > > That sounds fine to me.
> > 
> > Okay, I dropped the perror patch and replaced it with these two:
> > 
> > https://github.com/linux-rdma/rdma-core/pull/172/commits/f1f2e1b1bc3e90abdacdd06b3afe7ca4b107f9e3
> > https://github.com/linux-rdma/rdma-core/pull/172/commits/588c60a33ac5b846fee039a1cb8e622681fb980e
> > 
> > To implement a --systemd option to control the log behavior.
> 
> Hello Jason,
> 
> Since both is_systemd() and get_config() ignore the local variable opt_idx,
> how about passing NULL as the fifth argument to getopt_long()?

sure, done

> Since getopt_long() is called twice by srp_daemon, shouldn't get_config()
> reset the optind global variable before calling getopt_long()?

Oops! Certainly! done

Jason
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH rdma-core 11/21] debian: Remove infiniband-diags Depends from iwpmd

[Benjamin Drung]

Am Montag, den 24.07.2017, 14:00 -0600 schrieb Jason Gunthorpe:
> This was a copy and paste error.
> 
> Fixes: bba93a9ce37 ("Debian Packaging")
> Signed-off-by: Jason Gunthorpe <jgunthorpe-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
> ---
>  debian/control | 3 +--
>  1 file changed, 1 insertion(+), 2 deletions(-)
> 
> diff --git a/debian/control b/debian/control
> index 34b275dfeb8c62..8434bbb23693f8 100644
> --- a/debian/control
> +++ b/debian/control
> @@ -47,8 +47,7 @@ Description: InfiniBand Communication Manager
> Assistant (ACM)
>  
>  Package: iwpmd
>  Architecture: linux-any
> -Depends: infiniband-diags,
> -         lsb-base (>= 3.2-14~),
> +Depends: lsb-base (>= 3.2-14~),
>           ${misc:Depends},
>           ${shlibs:Depends}
>  Description: Userspace component for iWarp RDMA services

These reduced dependencies can go into one line now (you could run
'wrap-and-sort' to do it for you).

-- 
Benjamin Drung
System Developer
Debian & Ubuntu Developer

ProfitBricks GmbH
Greifswalder Str. 207
D - 10405 Berlin

Email: benjamin.drung-EIkl63zCoXaH+58JC4qpiA@public.gmane.org
Web: https://www.profitbricks.com

Sitz der Gesellschaft: Berlin.
Registergericht: Amtsgericht Charlottenburg, HRB 125506B.
Geschäftsführer: Achim Weiss.

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH rdma-core 20/21] Move rxe_cfg to the rdma-core package

[Benjamin Drung]

Am Montag, den 24.07.2017, 14:00 -0600 schrieb Jason Gunthorpe:
> rx_cfg is used to setup the kernel rxe device, it belongs with the
> rest of the kernel support items.
> 
> Signed-off-by: Jason Gunthorpe <jgunthorpe-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
> ---
>  debian/ibverbs-providers.install | 4 ----
>  debian/rdma-core.install         | 4 ++++
>  redhat/rdma-core.spec            | 8 ++++----
>  3 files changed, 8 insertions(+), 8 deletions(-)
> 
> diff --git a/debian/ibverbs-providers.install b/debian/ibverbs-
> providers.install
> index 93144527be04ef..73088d82f6e0e8 100644
> --- a/debian/ibverbs-providers.install
> +++ b/debian/ibverbs-providers.install
> @@ -1,8 +1,4 @@
>  etc/libibverbs.d/
> -usr/bin/rxe_cfg
>  usr/lib/*/libibverbs/lib*-rdmav2.so
>  usr/lib/*/libmlx4.so.*
>  usr/lib/*/libmlx5.so.*
> -usr/share/doc/rdma-core/rxe.md usr/share/doc/ibverbs-providers/
> -usr/share/man/man7/rxe.7
> -usr/share/man/man8/rxe_cfg.8
> diff --git a/debian/rdma-core.install b/debian/rdma-core.install
> index 4bab264208b691..3d8cddab9297a5 100644
> --- a/debian/rdma-core.install
> +++ b/debian/rdma-core.install
> @@ -1,8 +1,12 @@
>  etc/modprobe.d/truescale.conf
>  lib/systemd/system/rdma-ndd.service
>  lib/udev/rules.d/60-rdma-ndd.rules
> +usr/bin/rxe_cfg
>  usr/lib/truescale-serdes.cmds
>  usr/sbin/rdma-ndd
>  usr/share/doc/rdma-core/MAINTAINERS
>  usr/share/doc/rdma-core/README.md
> +usr/share/doc/rdma-core/rxe.md usr/share/doc/ibverbs-providers/
> +usr/share/man/man7/rxe.7
>  usr/share/man/man8/rdma-ndd.8
> +usr/share/man/man8/rxe_cfg.8

If you place rxe.md into rdma-core, then please keep it in
/usr/share/doc/rdma-core instead of moving it to
/usr/share/doc/ibverbs-providers/

-- 
Benjamin Drung
System Developer
Debian & Ubuntu Developer

ProfitBricks GmbH
Greifswalder Str. 207
D - 10405 Berlin

Email: benjamin.drung-EIkl63zCoXaH+58JC4qpiA@public.gmane.org
Web: https://www.profitbricks.com

Sitz der Gesellschaft: Berlin.
Registergericht: Amtsgericht Charlottenburg, HRB 125506B.
Geschäftsführer: Achim Weiss.

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html