> > It's probably still possible to have a CPU creating an EFI variable via the sysfs files while another CPU is in the pstore write path, but I'm > not convinced making the query_variable_info() and set_variable() atomic is something that we need to worry about. When I talked about query_variable_info() with Tony, Matthew commented as follows. http://marc.info/?l=linux-kernel&m=134305325801789&w=2 <snip> Running out of space in EFI isn't a well-tested scenario, and I wouldn't expect all firmware to handle it gracefully. This is made worse by EFI 1 not providing any information about available storage. <snip> I'm just concerned that set_variable() may run in the out of space situation if query_variable_info() and set_variable() are not atomic. Seiji > > -- > Matt Fleming, Intel Open Source Technology Center ��.n��������+%������w��{.n�����{����* jg��� �����ݢj����G���� ���j:+v���w�m������w�������h�����٥
