FWIW, I've just submitted patch for this to the list: https://sourceforge.net/p/gnu-efi/mailman/message/37684742/
Ah, interesting. Another data point: This used to work fine with Enigmail, but I missed the exact point when it became broken. It changed in the past months, maybe 6 months ago most, but there was a distro update from me in that window as well (SUSE 42.2 -> 15.0). Let me know if you need further details about my settings. I originally had OpenPGP enabled for that account as well, just defaulting to S/MIME. And "reply encryptedly" (or how it's called was on then as well. That had no effect any more...
Enigmail affects default encryption policy of s/mime-only accounts
* fixed test loop counter bug
*added wait_sof and delta_t support for the master
* fixed cleanup_module
* RTmac event example added
* added module license
* clean-up
* increased cache locality while looking up UDP...
* moved RTmac examples from rtmac/examples to e...
* fixed warnings of gcc 3.3.3
* ported socket API to RTDM
* ported socket API to RTDM
* added volatiles to rtsocket struct where needed
* made pool_size atomic_t
* fixed race-conditions in socket API
* fixed warnings on old kernels
* removed debug messages (may run under irq lock)
* fixed loopback-rt
* loopback-rt beautified
* updated and enabled lookback-rt
*added a comment to README
* fixed parameter check in several API functions
* applied new header file structure
* applied new header file structure
* applied route-delete and arp-crash patches by...
* applied PPC driver patch by Wolfgang Grandegger
* fixed module usage counter
* applied new header file structure
*** empty log message ***
* fixed skb freeing upon transmission error
* fixed rteth0 in docs and scripts
* fixed cleanup code
* fixed some compiler warnings by gcc3 (mostly ...
* applied fragmentation patch by Mathias Koehrer
* renaming: tdma_get_by_device
* added ./autogen.sh output
* cosmetical changes
* added VNIC support to RTmac
* added non-realtime tx queue to RTmac/TDMA
* implemented new pool mechanisms
* restructured rtmac to prepare separated disci...
* applied new pool mechanisms
* fixed RTmac-awareness of rt_ip_build_xmit_slow
* disable only device irq instead of all irqs i...
* fixed error handling in IOC_RT_IFUP
* added warning about potential long locks
* prepared 0.2.10 release
* revised xmit and interupt locking
* added a does-not-work warning
* provide optional RES_SEM for hard_start_xmit
* removed semaphore protection from rtdev_xmit
* added rt_enable_irq to end of interrupt handl...
* applied priorities to outgoing skbs
* adapted to new pool mechanisms
* applied new pool mechanisms
* rt_ip_sockopt is now called from generic sets...
* implemented separate reply pool
* applied new pool mechanisms
* added generic setsockopt function
* added rt_socket_setsockopt
*** empty log message ***
* added new socket option RT_SO_TIMEOUT
* implemented packet transmission without bindi...
* applied new pool mechanisms
* updated setsockopt call
* removed wrong MOD_INC_USE_COUNT
* updated includes
* added per-socket pools
* added global pool initialization
* a better fix of the dependency problem ;)
* implemented socket priorities
* dependency detection workaround for rtnet.o
* added per-socket pools
* priority queues now always have 32 levels
* extend global rtskb-pool for additional devices
* updated to conform with the new blocking rece...
* reordered includes (avoids conflict with mess...
* updated to reflect latest API modifications
* fixed multiple IFUP problem
* removed useless memset in kfree_rtskb
* we set CFLAGS default to '-s'
* removed unneeded kernel includes
* applied new pool mechanisms
* dropped the need to specify local IP
* forgot to save my last modifcation...
* committed the internally used script and conf...
* added rtskb-kmalloc switch with check against...
* added missing rt_sem_signal in xmit function
* decided to call the release 0.5.0
* added MSG_PEEK support for UDP sockets
* ported to new skb pool architecture
* fixed some errors
* fixes kfree bug in dispose_rtskb
* corrected RealTek module name
* fixed compilation issue on older kernels than...
* added lock protection for delta_t
* removed forgotten debug code
* extended skb size for pcnet32 requirements