Conversation with #inferno at Wed Mar 27 05:18:23 2013 on powerman@irc.freenode.net (irc)
(05:18:23) #inferno: Topic for #inferno set by mennis at 17:30:16 on 12/28/09
(05:18:23) pratchett.freenode.net: (notice) [freenode-info] if you're at a conference and other people are having trouble connecting, please mention it to staff: http://freenode.net/faq.shtml#gettinghelp
(06:25:43) dreadlorde left the room (quit: Ping timeout: 264 seconds).
(06:26:04) dreadlorde [~dreadlord@c-68-42-82-10.hsd1.mi.comcast.net] entered the room.
(07:43:06) Gegemon [~ynv@mx1.airis.ru] entered the room.
(09:42:45) yshurik [~Adium@193.69.63.210] entered the room.
(15:58:35) Gegemon left the room (quit: Quit: Leaving.).
(17:10:09) yshurik left the room (quit: Quit: Leaving.).
(17:31:36) yshurik [~Adium@30.162.202.84.customer.cdi.no] entered the room.
(00:01:24) Fish left the room (quit: Ping timeout: 260 seconds).
(00:12:55) Fish [~Fish@9fans.fr] entered the room.
(01:21:46) yshurik left the room (quit: Quit: Leaving.).
(03:50:35) crank left the room (quit: Ping timeout: 255 seconds).
(03:55:38) crank [~crank@shell.tuxshells.com] entered the room.
(03:55:56) qrstuv left the room (quit: Ping timeout: 252 seconds).
(05:49:07) anth_r left the room (quit: Ping timeout: 264 seconds).
(05:54:00) anth_x1 [~a@99.40.132.132] entered the room.
(05:55:33) anth_x left the room (quit: Ping timeout: 245 seconds).
(06:27:11) dreadlorde left the room (quit: Ping timeout: 252 seconds).
(06:43:01) anth_x1 is now known as anth_x
(07:39:53) Gegemon [~ynv@mx1.airis.ru] entered the room.
(09:44:07) yshurik [~Adium@193.69.63.210] entered the room.
(10:11:53) qrstuv [root@50.122.19.219] entered the room.
(10:38:27) powerman: is it possible to control emu's exit status?
(10:38:27) powerman: i.e. I run emu-g to execute just one .dis and wanna know in host os is that .dis exits normally or raises.
(10:38:27) powerman: that file output a lot of text, so printing it exit status to stdout isn't really nice idea
(13:23:02) leetspete left the room (quit: Ping timeout: 260 seconds).
(13:34:47) leetspete [~pete@76.91.33.110] entered the room.
(13:55:40) rogpeppe: powerman: that's an excellent question. i seem to remember doing something related to that before.
(13:56:30) rogpeppe: powerman: i have a feeling you're probably out of luck
(13:56:45) powerman: yep :(
(13:57:02) powerman: btw, wrt 9p linux kernel driver
(13:57:42) powerman: I've tried to use it, and got same issue as many years ago - it's impossible to re-mount after network issues
(13:58:07) powerman: many years ago when I see it first time I just hope it will be fixed soon
(13:58:17) powerman: this time I've added report to linux kernel's bugzilla
(13:58:49) powerman: lack of reliable way to detect and handle this situation make mounting 9p by network useless in production :(
(14:00:26) powerman: probably this can be worked around by running inferno on each host, mount remote /mnt/registry using inferno, and then export it from inferno to 127.0.0.1, to let kernel mount it from localhost. and handle all network/remount issues within inferno
(14:00:36) powerman: but that's sounds somewhat overcomplicated
(14:02:38) powerman: https://bugzilla.kernel.org/show_bug.cgi?id=55791
(14:21:58) qrstuv: what's wrong with using 9pfuse?
(14:24:13) powerman: qrstuv: when I tried it last time (about few years ago) it was SLOOOOOOW
(14:25:41) powerman: if I remember correctly it send about 2 times more 9P packets than actually needed
(14:26:07) powerman: keeping in mind 9P latency issue it result in really bad performance
(14:27:57) qrstuv: how is the kernel module clever enough to avoid that?
(14:30:12) powerman: I don't think it's "clever enough". I just hope it won't stupid enough to send more packets than really needed.
(14:30:49) powerman: actually may be this issue already fixed in 9pfuse, I've tried it long time ago
(14:32:06) powerman: it's ease to check - use inferno monitoring tool to see 9P packets inferno recv/send on exported fs. then mount that fs using 9pfuse, try to ls/stat/cat files from linux and check which 9P packets inferno process
(14:33:06) powerman: then mount same exported fs in another inferno, do same ls/cat operations and compare amount of used 9P packets for these operations 
(15:11:29) powerman: qrstuv: actually, 9pfuse hangs in exactly same way
(15:12:13) qrstuv: yeah, but it's easier to modify
(15:13:11) powerman: only difference was ability to kill 9pfuse process, which result in returning ENOTCONN from syscall in hang `stat` process
(15:24:40) powerman: qrstuv: actually, ability to kill 9pfuse probably enough to implement reliable mount. so thank you for reminding me about 9pfuse!
(15:25:47) qrstuv: i think someone might have recently added aan to p9p
(15:26:01) powerman: aan?
(15:26:50) qrstuv: http://plan9.bell-labs.com/magic/man2html/8/aan
(15:30:11) Gegemon left the room (quit: Quit: Leaving.).
(15:47:38) powerman: qrstuv: there is no aan in plan9port-20130320 :(
(15:48:02) qrstuv: oh well
(15:48:15) qrstuv: couldn't be much work to do yourself
(16:21:38) yshurik left the room (quit: Quit: Leaving.).
(16:32:11) crank left the room (quit: Ping timeout: 252 seconds).
(16:37:27) crank [~crank@shell.tuxshells.com] entered the room.
(16:44:36) yshurik [~Adium@30.162.202.84.customer.cdi.no] entered the room.
(17:08:29) rogpeppe: powerman: the problem is probably that the calls are coming from the kernel which assumes that the fs caches everything so all these ops are very cheap
(17:08:41) rogpeppe: powerman: except with 9p they're not so cheap
(17:09:29) rogpeppe: powerman: for the record, i won't see your messages unless you mention my irc nick.
(17:09:43) rogpeppe: powerman: unless i explicitly go looking, like i did this time
(20:03:00) anth_r [none@99.40.132.129] entered the room.
(21:10:20) anth_x left the room (quit: Quit: Leaving.).
(21:33:53) powerman: looks like it's impossible to use epoll for non-blocking read from /mnt/registry/event mounted using 9pfuse
(21:34:44) powerman: epoll report this file can be read immediately, but on read attempt my app hangs in syscall because there is no data yet
(21:49:17) anth_x [~a@rrcs-24-142-134-226.central.biz.rr.com] entered the room.
(22:39:01) anth_x left the room (quit: Quit: Leaving.).
(23:57:36) dreadlorde [~dreadlord@35.16.33.227] entered the room.
(01:18:49) dreadlorde left the room (quit: Ping timeout: 256 seconds).
(02:19:38) dreadlorde [dreadlorde@c-68-42-82-10.hsd1.mi.comcast.net] entered the room.
(02:38:03) The account has disconnected and you are no longer in this chat. You will automatically rejoin the chat when the account reconnects.