Ingo Schwarze <schwa...@usta.de> wrote: > Hi Theo, > > Theo de Raadt wrote on Thu, Jul 09, 2020 at 09:47:26AM -0600: > > > This time, they become invisible. > [...] > > There are many circumstances where ltrace is not suitable. > [...] > > We fiddle with programs all the time, to inspect them. > > Fair enough, then. > > The following variables already exist according to man -ak Ev~'\<LD_': > > * LD_PRELOAD > * LD_LIBRARY_PATH > * LD_BIND_NOW > * LD_TRACE_LOADED_OBJECTS > * LD_DEBUG > > I susspect the name should: > > * start with "LD_" > * contain "TIME" > * contain "SYSCALL" or "KTRACE" > > If i understand correctly, what the user really wants is that > time-related API calls appear in ktrace; that they cause system > calls is merely how that is implemented. Consequently, from the > user perspective, "KTRACE" is more relevant than "SYSCALL". > It is also clearer what it does; i assume using the word "SYSCALL" > would require a lengthy wording like "FORCE_SYSCALL". > > So, what about > > LD_KTRACE_GETTIME > > or a similar environment variable name?
That naming seems logical. If it is mostly hidden behind a ktrace flag (-T ?) then I think we're in good shape.
