[05:42:57] I'm out probably all of today but bwalker or drv let me know when the 100% external is ready to go and I'll push a few small changes for vagrant readme/setup [05:56:01] *** Quits: guerby (~guerby@april/board/guerby) (Ping timeout: 255 seconds) [06:16:56] *** Joins: guerby (~guerby@april/board/guerby) [06:22:08] *** Quits: guerby (~guerby@april/board/guerby) (Ping timeout: 240 seconds) [06:47:03] *** Joins: guerby (~guerby@april/board/guerby) [08:09:42] peluse: yeah - there's a bug there, the build.sh calls setup.sh in the vm meaning it will work when you first provision the VM but subsequent boots won't get it run [08:10:51] so this needs to move to the Vagrantfile [08:52:30] jimharris, cool, I'll make that part of the patch [09:09:59] *** Joins: johnmeneghini (d8f01e17@gateway/web/freenode/ip.216.240.30.23) [09:10:10] Hi. [09:14:17] I just found the following change to the check_format.sh script. [09:14:22] ssan-rx2560-03:spdk(remote_histogram_changes3) > ./scripts/check_format.sh Checking coding style... OK Checking comment style... OK Checking blank lines at end of file... OK Checking for POSIX includes...POSIX includes detected. Please include spdk/stdinc.h instead. include/spdk/cmn_utils.h:#include include/spdk/histogram.h:#include include/spdk/histogram.h:#include include/spdk/histogram.h:#includ [09:14:38] This is awesome. [09:15:28] Thanks for the good work in implementing spdk/stdinc.h [09:19:34] *** Joins: johnmeneghini1 (~johnmeneg@216.240.30.5) [09:39:20] *** Joins: jstern_ (~jstern@192.55.54.42) [09:48:15] drv, sethhowe: i noticed there is no Signed-off column on the gerrithub review page? [09:53:59] Is spdk using [-Werror=unused-const-variable=] [10:04:07] johnmeneghini: we do not explicitly enable that error [10:05:17] jimharris: Gerrit should be enforcing Signed-off-by when reviews are being uploaded [10:05:22] if not, let me know and I can look into it [10:05:40] ok - i just noticed the column was missing [10:05:56] yeah, that was a custom rule we added to our internal Gerrit instance [10:06:26] but now Gerrit should be able to enforce it automatically [10:21:42] Apparentely compiling with CONFIG_WERROR=y turns this on. [10:22:20] jimharris: just to verify what Daniel said, I tried to upload a change with my intel e-mail before it was associated with my github account this morning and gerrithub threw an error at me. [10:23:37] well, what we are talking about is the Signed-off-by: line in the commit message, but that is good too [10:23:54] make CONFIG_RDMA=y CONFIG_DEBUG=y CONFIG_ASAN=n CONFIG_UBSAN=y CONFIG_COVERAGE=y CONFIG_WERROR=y CONFIG_BCM_FC=n CONFIG_VHOST=y CONFIG_DPDK_DIR=../dpdk/build [10:24:12] CC lib/util/cmn_utils.o cmn_utils.c:5:23: error: âP_0â defined but not used [-Werror=unused-const-variable=] static const uint64_t P_0 = 1; [10:25:17] johnmeneghini: which compiler are you using? maybe that warning is turned on by default in some toolchains? [10:26:07] also, I'm not sure why you would need such a variable - we don't have a cmn_utils.c in our code, so this is presumably something you've added [10:28:37] good question. [10:29:00] I'm using the gcc compiler that comes with the Fedora 25 ditro. [10:29:14] ssan-rx2560-03:spdk(remote_histogram_changes3) > gcc --version gcc (GCC) 6.2.1 20160916 (Red Hat 6.2.1-2) Copyright (C) 2016 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. [10:29:42] ok, good, that's what we're using too - however, we don't have that static const unused variable [10:32:34] I just tried building the same code with gcc 4.9.2 and I don't see this error. [10:54:57] *** Quits: jstern_ (~jstern@192.55.54.42) (Ping timeout: 240 seconds) [12:39:54] *** Joins: jstern_ (~jstern@134.134.139.83) [13:38:01] *** Quits: jstern_ (~jstern@134.134.139.83) (Remote host closed the connection) [13:38:18] *** Joins: jstern_ (~jstern@134.134.139.83) [14:03:08] *** Quits: jstern_ (~jstern@134.134.139.83) (Ping timeout: 260 seconds) [14:07:29] *** Joins: jstern_ (~jstern@134.134.139.70) [15:33:28] *** Quits: johnmeneghini1 (~johnmeneg@216.240.30.5) (Remote host closed the connection) [15:33:34] *** Quits: johnmeneghini (d8f01e17@gateway/web/freenode/ip.216.240.30.23) (Ping timeout: 260 seconds) [15:55:57] *** Quits: jstern_ (~jstern@134.134.139.70) (Ping timeout: 240 seconds) [16:27:39] bwalker: i'm thinking these interrmittent blobfs failures in the test pool are related to a slew of unmap requests, not read/write [16:31:12] thinking we add a new bdev io status - BDEV_IO_STATUS_RETRY? [16:31:50] then we keep a queue of I/O in the common bdev.c code to handle these I/O [16:32:14] since you already added a channel in the generic bdev layer, we have a pretty convenient place to keep them