[00:22:48] --- Simon Wilkinson has become available [00:53:33] --- Simon Wilkinson has left [01:25:09] --- dev-zero@jabber.org has become available [02:10:43] --- dev-zero@jabber.org has left [02:24:53] --- dev-zero@jabber.org has become available [03:30:50] --- dev-zero@jabber.org has left [06:04:18] --- SecureEndpoints has become available [06:14:49] --- mmeffie has become available [06:22:32] --- dev-zero@jabber.org has become available [06:40:58] --- dmontuori has become available [06:43:11] --- summatusmentis has left [06:49:57] --- dev-zero@jabber.org has left: Lost connection [06:55:30] --- dev-zero@jabber.org has become available [07:43:01] --- dmontuori has left [07:43:01] --- dmontuori has become available [07:43:02] --- mmeffie has left [08:14:11] --- haba has become available [08:15:16] After the 3rd time I managed to type "conference" (in conference.openafs.org) correctly. [08:21:03] --- dmontuori has left [08:21:23] # find . -type f | xargs egrep -- 'struct vrequest [a-z]*;' | wc 93 372 4445 [08:22:46] yes, what about it? i find that i have grep on my machine also [08:23:56] are all these places where we might get to an uninitalized struct vrequest *areq in the end? [08:24:22] *areq are not places where we allocate one [08:24:46] however, if you want to know, adding asserts seems like it'd work [08:25:06] nope, but many of the the 93 struct vrequest treq lead to areq=&treq. [08:25:28] sure [08:25:56] i understand what you're getting at, but i think you're a bit... off? [08:27:47] shouldn't the first thing happening after any struct vrequest treq; be some function that gives the struct some safe start values? [08:28:58] not necessarily [08:29:07] "it depends" [08:29:42] As it is now, it is really hard to find if a struct member is first assigned a value or used. [08:32:07] Or (areq->inted && ......) everywhere [08:32:20] inited [08:32:36] (can't even spell it :) [08:33:07] char initd; /* if non-zero, non-uid fields meaningful */ (but I can cutnd paste) [08:33:16] :) [08:33:20] um, that would be dumb [08:33:32] you seem to not understand how it's working [08:33:42] you want to, after allocation, InitReq [08:41:30] sorry, in a hospital, disstracted [08:41:52] every allocation of a vrequest save one does the InitReq it should. that caller isn't called on linux. [08:43:38] actually, i think marc dionne's patch is wrong, despite it working. you able to try an alternate patch, harald? [08:43:58] yes, could try that [08:44:34] I got another patch from you, but that did a lot of other stuff and I have not tried that yet. [08:46:22] remove his patch. instead, patch afs_InitReq as such: /afs/andrew.cmu.edu/usr/shadow/initreq.diff [08:46:30] ignore the other patch [08:51:17] Init skipserver[i] at the same time as uid is "plan b" compared to "plan a" doing it in afs_FinalizeReq() when setting initd=1. [08:51:37] in english? [08:52:07] You want to do the skipserver[i] = 0 loop in InitReq, FinalizeReq or both? [08:52:18] i want you to apply the patch i gave you [08:52:32] That would be both :) [08:53:27] --- reuteras has left [08:53:34] basically, it's not that his patch is wrong per se. it's that checks for initd in more places spreads things logic which is probably best concentrated where we work with the vrequests [08:53:43] like, it will work fine [08:54:44] but we can avoid spraying checks for initd around that's better. (note that there are none outside the *Req and afs_Analyze functions otherwise) [08:55:07] I just was curious if the loop in FinalizeReq is redundant if you place it in InitReq, too. [08:55:30] But will shut up now and start patching :) [08:56:03] at this point in time i'd rather have it in both in case anyone else is dumb and misses an InitReq. i should fix the missing one, too, but that requires looking harder. it's in afs_osi_vget.c and doesn't apply in most cases [09:00:50] --- dev-zero@jabber.org has left [09:11:58] + int i = 0; [09:12:14] thanks [09:19:34] What about setting all the char accessError to char idleError to zero, too? Just to give them SOME value that is better than a random stack value? [09:21:38] nothing uses them before a caller to FinalizeReq [09:23:06] Ok, let's see if we "sob". [09:28:02] Wrote 30000 files in 113.662 s: Best result so far. [09:29:26] Read 30000 files in 10.536 s [09:32:22] (I'm not quite convinced by the "nothing uses them before..." argument.) Now doing continous sob runs..... [09:32:39] Looks better and better. [09:34:26] grep. [09:34:35] you don't need to be convinced. the code doesn't lie [09:36:23] for completeness, the "documentation": # diff -u afs.h.orig afs.h --- afs.h.orig 2008-10-20 18:34:10.000000000 +0200 +++ afs.h 2008-10-20 18:35:16.000000000 +0200 @@ -183,7 +183,7 @@ afs_int32 uid; /* user id making the request */ afs_int32 busyCount; /* how many busies we've seen so far */ afs_int32 flags; /* things like O_SYNC, O_NONBLOCK go here */ - char initd; /* if non-zero, non-uid fields meaningful */ + char initd; /* if non-zero, the Error fields are meaningful */ char accessError; /* flags for overriding error return code */ char volumeError; /* encountered a missing or busy volume */ char networkError; /* encountered network problems */ [09:39:18] Ok, should we try to get a test-round out of the europeans? As in "today" MET? I can write a "Derrick says test this" email or do you want? [09:42:12] well, you can tell them to just grab the head of openafs-stable-1_4_x [09:42:26] if i get time to look at the mac issue i will issue 1.4.8pre3 [10:03:36] --- dev-zero@jabber.org has become available [10:22:16] --- dmontuori has become available [10:43:05] --- haba has left [10:59:25] --- shadow@gmail.com/owl08984534 has left [11:22:36] --- dragos.tatulea has become available [11:53:39] --- shadow@gmail.com/owl09EC79F7 has become available [12:43:12] --- dev-zero@jabber.org has left [13:00:36] --- summatusmentis has become available [13:16:05] --- shadow@gmail.com/owlA0C12A38 has become available [13:20:08] --- shadow@gmail.com/owl09EC79F7 has left [13:46:38] --- summatusmentis has left [13:55:32] --- summatusmentis has become available [14:10:13] --- summatusmentis has left [14:10:21] --- summatusmentis has become available [14:10:48] --- Simon Wilkinson has become available [14:33:01] --- summatusmentis has left [14:59:43] --- dragos.tatulea has left [15:52:52] --- summatusmentis has become available [16:22:54] hi all [16:23:02] hi [16:23:12] you're always around [16:23:24] not always [16:25:39] spent hours at the hosptal today for instance [16:26:04] :-/ hospitals aren't fun? you personally? or visiting someone? [16:27:44] took someone, waited for them to have an organ removed, took them home [16:28:00] oh wow... that's a big deal. [16:52:17] I want a G1 :) [16:53:01] a decent android phone might be neat. i'm not sure the g1 is it [16:53:21] I'm a big fan of touch screen phones [16:53:36] me too. i had treo. the iphone is much better [16:54:02] I'd be a fan of the iPhone, if it weren't so locked down. Not really ok with the "walled garden" idea of it [16:55:30] it's less locked down if you start unlocking things [16:56:27] but then you get in trouble with Apple :) [16:56:52] i am not in trouble [16:57:06] Apple likes shadow [16:57:37] I don't know if Apple likes me. Apple hard drives don't [16:58:47] I do agree, however, that the iPhone has a very nice UI [17:02:28] --- summatusmentis has left [17:02:42] --- Simon Wilkinson has left [17:40:21] --- summatusmentis has become available [17:44:00] --- summatusmentis has left [19:05:05] --- summatusmentis has become available [19:07:44] --- summatusmentis24476 has become available [19:07:44] --- summatusmentis24476 has left [19:07:44] --- summatusmentis72356 has become available [19:07:44] --- summatusmentis72356 is now known as summatusmentis47248 [19:07:44] --- summatusmentis47248 has left [19:07:44] --- summatusmentis79447 has become available [19:07:44] --- summatusmentis79447 has left [19:07:45] --- summatusmentis72356 has become available [19:07:57] --- summatusmentis has left [19:08:04] --- summatusmentis72356 has left [19:08:13] --- summatusmentis has become available [19:10:54] --- summatusmentis has left [19:22:18] --- summatusmentis has become available [21:10:09] --- shadow@gmail.com/owlA0C12A38 has left [21:18:42] --- dev-zero@jabber.org has become available [22:32:48] --- dev-zero@jabber.org has left [22:59:49] --- reuteras has become available [23:46:08] --- dev-zero@jabber.org has become available