Old doc bug 6207856 and new bug (6958434)

[Michael Ernest]

I filed the following bug yesterday:

6958434 Table 25-2 missing two probes

which is related to:

6207856 Line is repeated in table 25-2

Filing a third bug against table 25-1 seemed silly (probes not listed in
order). I'd like to set both tables straight: order the probes, add the
missing ones, and remove the duplicate noted in 6207856. Bugster says a fix
is in place for 6207856 but I gather it's fallen through the cracks.

Michael

[Michael Ernest]

After a deeper review of the tables in this section of wiki, I noticed the following more persnickety elements:

The descriptions for the probes exec-failure and exec-success state that they fire "only after the exec probe has fired in the same thread." I am not aware of a case where one probe firing is contingent on another firing. It's certainly not common, in any event.

In general it seems difficult to state, without ambiguity, the order in which two or more probes would fire if all were invoked, without either a) repeating the "if invoked" business [i]ad nauseam[/i]; or b) implying a lock-step relationship.

It seems to me an ordering condition like "only after" or "immediately before" should only apply when describing a probe's immediate relationship to the event it traces. When applied to the order of two or more probes, it allows for doubt in a close reading.

I don't have a completely satisfying suggestion but the term "follows" seems to help. The sentence:

The exec-failure probe fires only after the exec probe has fired in the same thread.

would be clearer as

The exec-failure probe fires some time after the exec probe, assuming both are invoked.

The sense of indefinite time used to describe the exec probe makes sense. Readers who need to determine how long is "some time" can go to the source; all others can live blithely in the mist that separates the description from the implementation itself.

Other typos:

-> In Table 25-1, the entry for the signal-clear probe points to sigtimedwait(3RT) and sigtimedwait(3RT).

The correct library is 3C.

-> In Table 25-1, the entry for the signal-send probe states "[t]he lwpsinfo_t and psinfo_t of the receiving process and thread are in args[0] and args[1], respectively."

"process and thread" should be "thread and process"

I propose to make these corrections in addition to the ones noted above if there are no objections.