[00:13:21] --- lars.malinowsky has become available [00:13:42] --- lars.malinowsky has left [00:13:47] --- lama has become available [00:19:44] --- jaltman/FrogsLeap has left: Disconnected [00:22:12] --- Russ has left: Disconnected [00:27:57] --- lama has left [00:28:06] --- lars.malinowsky has become available [04:18:17] --- jaltman/FrogsLeap has become available [04:49:33] --- mdionne has become available [05:41:27] --- jaltman/FrogsLeap has left: Disconnected [05:42:46] --- mdionne has left [06:23:25] --- mdionne has become available [06:29:22] --- jaltman/FrogsLeap has become available [06:59:22] well, i guess now we look harder at deleted clients [07:01:29] --- lars.malinowsky has left [07:01:37] --- lars.malinowsky has become available [07:03:35] phalenor: willing to run a debug patch for that bypass problem [07:10:42] mdionne: sure. *might* get to it today. definitely will by the end of the week. [07:11:12] if you could provide a patch that'll apply against 1.6.0pre4 that I can just include in the spec file, even better. [07:14:41] the issue with the client->deleted is simple: [07:15:08] old h_Release_r: h_Decrement_r(host); if ((host->hostFlags & HOSTDELETED) || (host->hostFlags & CLIENTDELETED)) { h_TossStuff_r(host); [07:15:19] new h_Release_r: h_Decrement_r(x); \ if (((x)->refCount < 1) && \ (((x)->hostFlags & HOSTDELETED) || \ ((x)->hostFlags & CLIENTDELETED))) h_TossStuff_r((x)); \ [07:15:32] easy answer: don't *also* check for (x)->refCount < 1 [07:15:42] --- matt has become available [07:16:26] --- tkeiser has become available [07:17:33] (since TossStuff does its own checks to see if it's safe to toss other things) [07:20:37] actually that's not it. [07:21:28] unless somehow we have an extra ref in the same thread. [07:31:11] --- deason has become available [07:33:41] old h_Release_r had a !h_OtherHolds_r check, which is the equivalent of the refCount < 1 check... [07:34:12] the comment in old h_Release_r is wrong. it's "we have a hold, and no one else does". [07:34:23] but yeah [07:42:16] --- Simon Wilkinson has become available [07:42:59] --- mfelliott has become available [07:43:39] deason: If you have a spare moment, could you take a look over 4573 in gerrit. It's a patch for review for refresh (it modifies the old function names for clarity of review, we'd obviously be creating new names for all of these when we actually push them to the tree) [07:47:32] okay [08:06:41] --- lars.malinowsky has left [09:09:14] --- tkeiser has left [09:16:43] --- Russ has become available [09:27:32] --- matt has left [09:33:58] at the end of MultiBreakVolumeCallBack_r we get a hold if we have more work to do. i wonder if in any cases we leak that [09:35:11] that hold is new, btw, I added that; it used to be just a "return 1" to tell h_Enumerate_r to not release the hold [09:37:55] we call MultiBreakCallBack_r immediately afterwards, and it does seem to always release all of the hosts... [09:38:40] > we call MultiBreakCallBack_r immediately afterwards, and it does seem to yeah, it should. i'm just wondering if it always does. still reading. [09:57:07] --- tkeiser has become available [10:48:43] --- Russ has left: Disconnected [10:50:23] --- steven.jenkins has left [10:55:50] --- matt has become available [11:41:05] --- Simon Wilkinson has left [11:43:38] --- jaltman/FrogsLeap has left: Disconnected [11:50:32] --- mdionne has left [12:07:26] --- matt has left [12:11:42] --- ksumner has left [12:14:54] --- mdionne has become available [12:17:24] --- steven.jenkins has become available [12:21:44] --- mdionne has left [12:41:07] --- tkeiser has left [12:55:13] --- mdionne has become available [13:21:02] --- Simon Wilkinson has become available [13:40:29] --- Simon Wilkinson has left [15:19:57] --- jaltman/FrogsLeap has become available [15:23:51] --- deason has left [16:56:13] --- Russ has become available [17:54:06] --- jaltman/FrogsLeap has left: Disconnected [18:29:50] --- jaltman/FrogsLeap has become available [18:43:39] --- steven.jenkins has left [19:21:25] --- mdionne has left [20:18:05] --- jaltman/FrogsLeap has left: Replaced by new connection [20:18:11] --- jaltman/FrogsLeap has become available [22:08:39] --- jaltman/FrogsLeap has left: Replaced by new connection [22:08:40] --- jaltman/FrogsLeap has become available [22:45:13] --- reuteras has become available [23:43:42] --- Russ has left: Disconnected