Back
[07:36:58] <jepler> this looks interesting
https://lkml.org/lkml/2015/4/30/196 "[RFD 0/5] Add latency histogram"
[12:57:02] <andypugh> Ref:
http://git.linuxcnc.org/gitweb?p=linuxcnc.git;a=commit;h=3133713553da54cf3a7e89072b107882cae36403
[12:57:43] <andypugh> It looks like the #<_current_pocket> predefined parameter needs to be fixed too
[21:02:53] <jepler> hmm halrun is quite confused after being ctrl-z'd, you can't type into it anymore after fg
[21:04:13] <jepler> [pid 26324] ioctl(0, SNDCTL_TMR_STOP or TCSETSW, {B38400 opost isig icanon echo ...}) = ? ERESTARTSYS (To be restarted if SA_RESTART is set)
[21:04:16] <jepler> [pid 26324] --- SIGTTOU {si_signo=SIGTTOU, si_code=SI_KERNEL} ---
[21:04:18] <jepler> [pid 26324] rt_sigreturn({mask=[]}) = -1 EINTR (Interrupted system call)
[21:04:22] <jepler> [pid 26324] ioctl(0, TCGETS, {B38400 opost isig -icanon -echo ...}) = 0
[21:04:24] <jepler> [pid 26324] ioctl(0, SNDCTL_TMR_STOP or TCSETSW, {B38400 opost isig icanon echo ...}) = ? ERESTARTSYS (To be restarted if SA_RESTART is set)
[21:04:36] <jepler> in fact, halcmd loops like this the whole time it's ctrl-z'd, at 100% CPU
[21:05:52] <jepler> confusingly, when *not* launched from halrun, halcmd doesn't behave like this
[21:17:16] <jepler> https://lists.gnu.org/archive/html/bug-readline/2015-03/msg00000.html
[21:26:03] <jepler> $ bash -c 'bash ; :'
[21:26:08] <jepler> $ suspend
[21:26:10] <jepler> [bash is hung]
[21:26:17] <jepler> $ strace -e signal bash
[21:26:50] <jepler> $ suspend
[21:26:53] <jepler> [bash is hung
[21:26:54] <jepler> ]
[21:27:08] <jepler> wow you don't have to look very far for completely effed up behavior out of readline, at least on debian jessie
[21:28:39] <cradek> I wonder if recent kernels broke it or something. it's so old it's hard to believe it's just all busted.
[21:29:24] <jepler> I don't know
[21:29:48] <jepler> same thing happens on my jessie kfreebsd machine, which says to me it's not the kernel
[21:30:11] <jepler> in the land of consistency..
[21:30:12] <jepler> $ ssh zaphod
[21:30:12] <jepler> Enter passphrase for key '/home/jepler/.ssh/id_rsa':
[21:30:17] <jepler> $ ssh-add
[21:30:18] <jepler> Enter passphrase for /home/jepler/.ssh/id_rsa:
[21:32:45] <jepler> FreeBSD 6.2-RELEASE with bash 3.1 (not even dynamic-linked with readline, just ncurses) is affected
[21:32:56] <jepler> so no it's not recent breakage
[21:33:04] <jepler> -rwxr-xr-x 1 root wheel 594692 Oct 15 2006 /usr/local/bin/bash
[21:35:22] <jepler> -rwxr-xr-x 1 root root 626028 Feb 11 2003 /bin/bash*
[21:35:25] <jepler> also affected (redhat 9)
[21:35:37] <cradek> wow
[21:36:01] <jepler> is there some way it could be my terminal emulator, even though this bash is running on the wrong side of 2 sshes?
[21:37:13] <jepler> xterm, linux console affected
[21:40:49] <PCW> That's pretty ugly
[21:41:26] <cradek> you'd think we'd all suspend readline things all the time
[21:42:35] <PCW> Look's like Michael did not do the DPLL encoder latch support but will have time to work on it this weekend (If i understood him correctly)
[21:43:43] <jepler> cradek: I think it has something to do with being the grandchild, rather than the direct grandchild, of the shell that it suspends back to
[21:44:05] <PCW> I optimistically mis-remembered that it had been done :-(
[21:44:08] <cradek> oh, ok
[21:59:11] <seb_kuzminsky> PCW: yep, that's what he told me, that encoder-dpll was coming in the next week or two, and we decided to hold off the 2.7.0 release until then