On Wed, Oct 02, 2019 at 01:39:03PM +0300, Nikolay Borisov wrote: > > > On 2.10.19 г. 13:30 ч., Anand Jain wrote: > > Its very helpful if we had logged the device scanner process name > > to debug the race condition between the systemd-udevd scan and the > > user initiated device forget command. > > > > This patch adds scanned-by process name to the scan message. > > > > Signed-off-by: Anand Jain <anand.j...@oracle.com> > > Same effect can be achieved (for debugging purposes) if you have used > ftrace on device_list_add without needing to patch the kernel.
For reproducible issues adding a debugging hooks is fine but races can be tricky and device scanning depends on the state of the system and other processes so I understand the need to document what happens for post-mortem analysis. > I'm somewhat indifferent whether this will be merged or not but I > personally don't see much value in it. We have messages for many administrative tasks, either requested by users or by developers so I don't object in principle against adding more.
