summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorWilliam A. Kennington III <wak@google.com>2019-10-23 21:06:30 +0300
committerBrad Bishop <bradleyb@fuzziesquirrel.com>2019-11-22 05:30:25 +0300
commit9be47ffb27ff726e768c8f8155415c8d39544124 (patch)
tree2b833a4dabaccdd0b54b4206de448817c774ea3f
parent989ac308dedd12c913da60ba168f3c7a4ef0cbbb (diff)
downloadopenbmc-9be47ffb27ff726e768c8f8155415c8d39544124.tar.xz
phosphor-hwmon: Udev rules should never block
If we create a dependency in the hwmon service that makes it take extra time to start, this will block up all of udev and prevent other devices from being discovered as we wait on the specific hwmon instance to start. This can result in deadlocks. Typical convention in udev is to block as little as possible to avoid these type of scenarios. udev can't handle failures so it's pointless to wait anyway. The service start will still be queued up as expected. Tested: Ran on a machine configured to use hwmon. Configured sensors can still be read and the daemons start as expected. (From meta-phosphor rev: 1a21d0092ab2f07f5530f8b1167cbe5af12b9516) Change-Id: I358566ce124a2bcd44306df3d579beca65133e83 Signed-off-by: William A. Kennington III <wak@google.com> Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
-rwxr-xr-xmeta-phosphor/recipes-phosphor/sensors/phosphor-hwmon/start_hwmon.sh2
1 files changed, 1 insertions, 1 deletions
diff --git a/meta-phosphor/recipes-phosphor/sensors/phosphor-hwmon/start_hwmon.sh b/meta-phosphor/recipes-phosphor/sensors/phosphor-hwmon/start_hwmon.sh
index c1ec42e94..2f35865b3 100755
--- a/meta-phosphor/recipes-phosphor/sensors/phosphor-hwmon/start_hwmon.sh
+++ b/meta-phosphor/recipes-phosphor/sensors/phosphor-hwmon/start_hwmon.sh
@@ -20,4 +20,4 @@ fi
path="${path//:/--}"
path="${path//-/\\x2d}"
-systemctl $action 'xyz.openbmc_project.Hwmon@'$path'.service'
+systemctl --no-block $action 'xyz.openbmc_project.Hwmon@'$path'.service'