[nrf fromtree] OpenThread: Clear SRP host and services during RevertConfiguration #573
Conversation
adigie
force-pushed
the
clear-srp-after-reset
branch
from
b9ab15f to
29917b6
Compare
There was a problem hiding this comment.
I think now that you delay RevertConfiguration, it needs to be considered what happens if BackupConfiguration is called before kServerReady is processed (which may happen if SRP response gets delayed or is lost and we receive a Matter command to modify Thread network). Then we should probably do the revert without waiting for the SRP response.
| // must be restored on the boot. If there's no backup, the below function is a no-op. | ||
| RevertConfiguration(); | ||
| // must be restored. | ||
| if (BackupExists()) |
There was a problem hiding this comment.
I don't think this check is necessary. RevertConfiguration takes care on its own whether the backup exists or not and kServerReady is generated only once per boot, so mRevertOnStartup seems redundant, right?
There was a problem hiding this comment.
Discussed offline - we will keep mRevertOnStartup to mark that during initialization backup was present. Additionally, we have to clear it when creating a new backup.
adigie
force-pushed
the
clear-srp-after-reset
branch
2 times, most recently
from
66ba0cc to
0003f04
Compare
adigie
force-pushed
the
clear-srp-after-reset
branch
from
0003f04 to
1a76106
Compare
…onfiguration (#38053) * Postpone `RevertConfiguration` until Server is initialized. * Clear SRP host and services before reverting dataset. * Abort revert if creating new backup. Signed-off-by: Adrian Gielniewski <adrian.gielniewski@nordicsemi.no> (cherry picked from commit 132a7a8aed99b6f8dda15c749bc2b182bf82ffed) Signed-off-by: Adrian Gielniewski <adrian.gielniewski@nordicsemi.no>
adigie
force-pushed
the
clear-srp-after-reset
branch
from
1a76106 to
e043125
Compare
adigie
changed the title
RevertConfigurationuntil OpenThread is initialized.