He's working around the crash by forcing PipeA to be enabled, without an external screen attached. This is a step in the right direction to debugging the issue. We may be able to get a better solution still by identifying where this specific crash occurs, and either enabling PIpeA for that portion of code, or avoiding that code-path when PipeA is disabled.
Raúl Sánchez Siles has a good lead on this in the fd.o bug 11432, /bugs.freedeskt op.org/ show_bug. cgi?id= 11432
https:/
He's working around the crash by forcing PipeA to be enabled, without an external screen attached. This is a step in the right direction to debugging the issue. We may be able to get a better solution still by identifying where this specific crash occurs, and either enabling PIpeA for that portion of code, or avoiding that code-path when PipeA is disabled.