[01:14:38] --- haba has become available
[01:16:25] <Simon Wilkinson> So, in today's tour of the Linux CM's bugs, let's consider the fact that we can return AOP_WRITEPAGE_ACTIVATE from write(). Which just looks to the kernel like we're claiming to have written 0x8000 bytes. When in fact we wrote nothing at all. 
[01:42:30] --- dev-zero@jabber.org has left
[02:01:09] --- abo has left
[02:06:15] --- haba has left
[02:07:46] <Simon Wilkinson> Gerrit upgrade about to happen. Expect some downtime ...
[02:31:38] --- dev-zero@jabber.org has become available
[02:43:44] --- dev-zero@jabber.org has left
[02:43:56] --- dev-zero@jabber.org has become available
[02:44:30] --- haba has become available
[03:22:08] --- manfred has become available
[03:25:04] --- abo has become available
[03:26:55] --- manfred has left
[03:47:39] <Simon Wilkinson> gerrit upgrade (finally) done. Sorry about the delay
[03:53:26] --- abo has left
[03:58:44] --- manfred has become available
[04:15:49] <shadow@gmail.com/owl8E1009BB> should i be able to sign in now?
[04:16:01] <shadow@gmail.com/owl8E1009BB> cuz, like, not so much
[04:16:02] <Simon Wilkinson> You should be, yes.
[04:16:26] <Simon Wilkinson> Wkl
[04:16:29] <Simon Wilkinson> Works for me (TM)
[04:16:40] <Simon Wilkinson> What error are you getting?
[04:16:42] <shadow@gmail.com/owl8E1009BB> signing in bring up that i am no longer signed in; then, i sign in
again, try to comment, and *that* brings up a no longer signed in. back
to the start
[04:16:55] <shadow@gmail.com/owl8E1009BB> session expired
[04:17:11] <shadow@gmail.com/owl8E1009BB> lemme guess, i need to toss my cookies
[04:17:27] <Simon Wilkinson> Yeh. Delete your cookies and see what happens...
[04:18:14] <Simon Wilkinson> The release notes say "Users may need to sign out and back in again following the upgrade"
[04:20:02] <shadow@gmail.com/owl8E1009BB> there's no "sign out"
[04:20:16] <Simon Wilkinson> Because you're not signed in?
[04:20:29] <shadow@gmail.com/owl8E1009BB> ding
[04:20:45] <Simon Wilkinson> Does deleting cookies help at all?
[04:21:11] <shadow@gmail.com/owl8E1009BB> i dunno. i don't want to delete *all* my cookies, so i'm fussing.
[04:21:23] <Simon Wilkinson> Also, are you in private browsing mode?
[04:24:04] <shadow@gmail.com/owl8E1009BB> no. i never use private browse mode
[04:24:18] <Simon Wilkinson> Nothing else that eats cookies running?
[04:25:20] <shadow@gmail.com/owl8E1009BB> i have a / and a /gerrit/rpc cookie, both for GerritAccount
[04:25:37] <Simon Wilkinson> Okay. Delete both of those, and see what happens.
[04:25:59] <shadow@gmail.com/owl8E1009BB> i did. seems to work
[04:26:27] <Simon Wilkinson> Excellent.
[04:26:55] <shadow@gmail.com/owl8E1009BB> and the resulting cookie is /
[04:27:15] <shadow@gmail.com/owl8E1009BB> (add n edit cookies addon for 3.5)
[04:27:19] <Simon Wilkinson> Yeh. I suspect its the /gerrit/rpc cookie that was causing you pain.
[04:44:22] --- meffie has become available
[06:04:49] <haba> Good afternoon (or apropritate time)
[06:05:10] <Simon Wilkinson> Afternoon is good for me!
[06:05:21] <haba> I have seen the following behaviour of vos dump | something-eating-dumps :
[06:05:52] <haba> When something-eating-dumps dies, vos either dumps core itself of just leaves the transaction hanging.
[06:06:28] <haba> Or I'd say it allways leaves the transaction hanging, sometimes in addition to that it dumps core.
[06:06:53] <haba> Is this consistent with what you are seeing or is it "just me"? 
[06:08:02] <haba> (something eating dumps is typically something involving TSM which of course can fail for no apparent reason *doh* )
[06:08:36] --- abo has become available
[06:13:09] <shadow@gmail.com/owl8E1009BB> afternoon may be good for me when it rolls around
[06:13:53] <shadow@gmail.com/owl8E1009BB> vos not dying: well, the sad thing is i can't say "get a backtrace"
because it being lwp means the odds of you *getting* a backtrace are
very low
[06:42:36] <haba> Does vos even try to do something when it gets a signal or goes it just boom?
[06:58:09] --- Jeffrey Altman has become available
[07:09:50] --- deason has become available
[07:13:58] <deason> depends on the signal
[07:14:08] <deason> it does for INT, looks like it does for PIPE as well for dumps
[07:14:24] <deason> it involves a longjmp which I've seen behave weirdly
[07:16:21] <Simon Wilkinson> Is there a pthreaded vos?
[07:16:24] --- meffie has left
[07:32:04] <haba> longjump mixed with threads on Linux ... mmmmm.
[07:33:12] <Simon Wilkinson> Does the longjmp not come with LWP. If we're doing longjmp with pthreads then, eeek.
[07:34:12] <deason> actually, I have a backtrace; may or may not be the same thing, but 125535
[07:36:02] --- stevenjenkins has left
[07:36:27] --- stevenjenkins has become available
[07:44:28] <haba> If you are in a syscall, nothing happens anyway until you end it, so on sigpipe setting a flag that has the same effect as a write error on the output file handle (in ReceiveFile()) maybe is better than a longjump. But, it's friday, so don't be hard if I'm not making sense.
[08:01:25] --- abo has left
[09:30:39] --- haba has left
[10:00:14] --- meffie has become available
[10:53:53] --- manfred has left
[12:46:35] --- manfred has become available
[12:49:50] --- manfred has left
[12:51:02] --- manfred has become available
[12:55:05] --- manfred has left
[12:59:10] --- manfred has become available
[13:08:05] --- jaltman has left: Disconnected
[13:23:55] --- dev-zero@jabber.org has left
[13:49:38] --- manfred has left
[14:39:26] --- jaltman has become available
[14:49:34] --- sxw has become available
[14:55:04] --- dev-zero@jabber.org has become available
[15:26:34] --- jaltman has left: Disconnected
[15:36:45] --- dev-zero@jabber.org has left
[15:38:53] --- deason has left
[15:56:01] --- deason has become available
[17:13:40] --- meffie has left
[19:24:23] --- jaltman has become available
[19:42:53] --- dev-zero@jabber.org has become available
[19:59:41] --- jaltman has left: Disconnected
[22:03:02] --- deason has left
[23:46:26] --- jaltman has become available