The Firmware Security panel exposes the host security levels
and details. The information is generated by fwupd. The panel
also exposes hardware configuration changes to pinpoint the
configuration changing time.
Currently this panel shows:
- HSI and secure boot status
- Details of HSI and secure boot
- Configuration changelog
- Digested security level
- Extended protection
These are not real modes, but just as place holders when generating
'mirror' configurations. The clone modes will be just to match the
flag/dimension, while the actual mode applied will be individual for
each monitor.
This allows monitors to have their own refresh rates, which is possible
since a few mutter versions back. This also matches how mutter itself
generates mirror modes when doing so via the key binding.
It is perfectly possible for mm_sim_dup_identifier() to return NULL if
the SIM ID wasn't provided by the modem for any reason, leading to an
assertion failure:
(gnome-control-center:910641): cc-wwan-data-CRITICAL **: 12:43:51.573:
cc_wwan_data_set_default_apn: assertion 'self->sim_id != NULL' failed
Handle the NULL SIM ID gracefully.
We need to cancel internal operations before starting to create a new
panel, otherwise panels might be creating the same object for the cache
in parallel and this is not supported by the object cache.
The alternative to this would be to handle this inside the object
store to allow parallel creation of the same key.
Fixes: #1685
The previous code was prone to race conditions if the child already
started writing to stdout before the dup2 call happened. This has
been detected in postmarketOS[1] and I also reproduced it in Alpine
Linux. Since passwd writes to stderr and linux-pam to stdout, the
redirection was needed. However, linux-pam was failing with
"Conversation error" since an fprintf(stdout, ...) call wasn't
able to write to an already-closed stdout.
This problem is fixed by setting the redirection in the child setup
function and ignoring the stderr pipe. It also fixes a leak, where
the stderr fd was simply ignored and never closed.
[1] https://gitlab.com/postmarketOS/pmaports/-/issues/1449
Commit 5e0fc5c1 ("network: prevent crash by disconnecting device")
introduced two new line breaks after the final call of
remove_wifi_device. Those line breaks are unnecessary, as pointed out by
Robert (@robert.ancell). They should be removed for the MR to be merged.
This patch removes those line breaks, which should allow for !1359 to be
properly merged.
When a device is registered with add_wifi_device, a corresponding signal
is registered for events. The associated remove_wifi_device clears the
associated pages, but does not disconnect the signal. This causes the
assertion NET_IS_DEVICE_WIFI to fail.
This patch corrects this error, resolving #1889 as per the suggested fix
from Benjamin (@bberg).
Implement handling for OWE-TM networks.
* Correctly display encryption type when connected to a OWE-TM network
with an open profile (Previously showed as WPA2
* Mask out hidden transition SSID when connected to the OWE-TM network
* Don't display OWE-TM network with lock icon
Signed-off-by: David Bauer <mail@david-bauer.net>
