Hi!
On Mon 28 Feb 2022 23:28, Xiao Ni wrote:
> Hi Peter
>
> In rhel, we have a rhel only udev rule that checks
> DM_UDEV_DISABLE_OTHER_RULES_FLAG. Maybe it's the reason why you don't
> notice this. Besides DM_UDEV_DISABLE_OTHER_RULES_FLAG, it still checks
> other flags.
>
Ah yes, that's it! I've been still recalling this to be patched once.
So looks like it just didn't get propagated upstream :-/
> # Next make sure that this isn't a dm device we should skip for some reason
> ENV{DM_UDEV_RULES_VSN}!="?*", GOTO="dm_change_end"
> ENV{DM_UDEV_DISABLE_OTHER_RULES_FLAG}=="1", GOTO="dm_change_end"
> ENV{DM_SUSPENDED}=="1", GOTO="dm_change_end"
> KERNEL=="dm-*", SUBSYSTEM=="block", ENV{ID_FS_TYPE}=="linux_raid_member", \
> ACTION=="change", RUN+="/sbin/mdadm -I $env{DEVNAME}"
> LABEL="dm_change_end"
>
> In 10-dm.rules, if DM_SUSPENDED is 1, it sets
> DM_UDEV_DISABLE_OTHER_RULES_FLAG to 1. So we don't need the check of
> DM_SUSPENDED. But how DM_UDEV_RULES_VSN? Do we need to check it?
>
Yes, right, the check for DM_SUSPENDED is superfluous here so we don't
actually need that one. The single check for DM_UDEV_DISABLE_OTHER_RULES_FLAG
covers it.
The DM_UDEV_RULES_VSN - this was meant for future changes in case a new
set of DM udev variables is used or existing set changed so the other rules
know what exact variable set is available for checking. But I think the rules
are settled down for a few years now and I don't expect any more radical
changes here, so we can remove that check for DM_UDEV_RULES_VSN as well.
--
Peter
