1130782 - RunWatchdog should not fire when a debugger is present

Status: NEW

(Core :: XPCOM, defect, P5)

Comment 1

[Kyle Huey (Exited; not receiving bugmail, old account, do not use)]

Uh, why not?

Comment 2

[Honza Bambas (:mayhemer)]

(In reply to Kyle Huey [:khuey] ([e-mail]) from comment #1)
> Uh, why not?

Maybe I should rephrase the title: the watch dog should ignore delays caused by debugger sessions.  I'm currently not sure if this is an issue.  I recall I've reported this bug after debugging some shutdown code and just got hit by this artificial crash.  Very annoying.

Comment 3

[Kyle Huey (Exited; not receiving bugmail, old account, do not use)]

Ah, ok, that is a reasonable ask.  Resummarizing.

Comment 4

[Honza Bambas (:mayhemer)]

On windows, it can be easily discovered with a call to BOOL IsDebuggerPresent() function.

Comment 5

[Jens Stutte [:jstutte]]

So IIUC IsDebuggerPresent is a WinAPI function we have no general equivalent for. It checks if there is an external debugger attached (probably Visual Studio?).

There is an easy workaround for this: just set the pref toolkit.asyncshutdown.crash_timeout to something much higher than the default. Checking IsDebuggerPresent would be easy to do but is it worth to have a special case here just for Windows? If someone is debugging shutdown I hope they know how it works in general...

Comment 6

[Nika Layzell [:nika] (ni? for response)]

I don't really see the harm in adding a special case for Windows to the code, it seems straightforward enough to add by setting the timeout to INT32_MAX if ::IsDebuggerPresent(), and like it'd be unlikely to cause problems. Might be worth asking if it would be handy for folks in #windev on matrix?

Comment 7

[Jens Stutte [:jstutte]]

Asking (and keeping the ni?)

Comment 8

[Jens Stutte [:jstutte]]

By now no response on #windev. If someone stumbles again over this, we are happy to look at a patch.