phosphor-user-manager[470]: Locking Shadow failed

[geissonator]

https://gerrit.openbmc-project.xyz/c/openbmc/phosphor-user-manager/+/43174 was merged recently to fix an issue found with an updated c++ compiler. It seems that before this commit, the shadowlock class objects were not actually being created or utilized. The fact they are now has shown there is a bug in this logic:

May 14 14:08:12 witherspoon systemd[1]: Starting Phosphor User Manager...
May 14 14:08:13 witherspoon phosphor-user-manager[470]: Locking Shadow failed
May 14 14:08:13 witherspoon phosphor-user-manager[470]: The operation failed internally.
May 14 14:08:13 witherspoon phosphor-user-manager[470]: terminate called after throwing an instance of 'sdbusplus::xyz::openbmc_project::Common::Error::InternalFailure'
May 14 14:08:13 witherspoon phosphor-user-manager[470]:   what():  xyz.openbmc_project.Common.Error.InternalFailure: The operation failed internally.
May 14 14:08:13 witherspoon systemd[1]: Started Process Core Dump (PID 471/UID 0).

@williamspatrick noted there appears to be a logic bug in this line:
if (!lckpwdf()) { elog<InternalFailure>();

lckpwdf() returns 0 on success so the "!" needs to be removed.

@ratagupt @rthomaiy

[geissonator]

This is blocking us from being able to pick up the latest yocto from upstream.

[geissonator]

https://gerrit.openbmc-project.xyz/c/openbmc/phosphor-user-manager/+/43232 to in essence put us back to where we were before the commit that "fixed" things but also allow us to move on with bringing in new yocto.

[rthomaiy]

Can you please confirm whether /etc/.pwd.lock is writeable (I meant able to create a file there and able to have write access to it)?