summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorhttps://www.google.com/accounts/o8/id?id=AItOawkDt2vwi7MECPF4_fVO75ZNRtTj0Z0bY2g <harish@web>2011-08-24 08:07:15 +0200
committerGNU Hurd web pages engine <web-hurd@gnu.org>2011-08-24 08:07:15 +0200
commita83964bbd21dde0ca52a92cd070f60d4fc2a7e7e (patch)
treee79e5e7186e9272809e1add92ce3449a400c6e39
parent8d29e149a8cbbf4503bb5ee259566a1da864bc6c (diff)
more info added
-rw-r--r--open_issues/runit.mdwn42
1 files changed, 42 insertions, 0 deletions
diff --git a/open_issues/runit.mdwn b/open_issues/runit.mdwn
index c7a0962c..6b336ef7 100644
--- a/open_issues/runit.mdwn
+++ b/open_issues/runit.mdwn
@@ -17,3 +17,45 @@ report is just from his memory, and his memory is dim... The problem *might*
either be a time stamping issue (which might be fixed by now) or it *might* be
the `select` call failing issue we're seeing from time to time. Or something
else.
+
+[[Harish Badrinath|harishbadrinath]]
+Originally answered by Samuel Thibault:
+> 120->proc_dostop_request ( 138) = 0
+>
+> </snip>
+
+Usual issue with rpctrace: it does not support fork().
+
+I've checked a backtrace in gdb, got this:
+
+ 0x0105af6c in mach_msg_trap ()
+ at /build/eglibc-jWVnRE/eglibc-2.13/build-tree/hurd-i386-libc/mach/mach_msg_trap.S:2
+
+1 0x0105b769 in __mach_msg (msg=0x1024af8, option=258, send_size=0, rcv_size=40, rcv_name=140,
+ timeout=1000020, notify=0) at msg.c:110
+
+2 0x01062251 in _hurd_select (nfds=2, pollfds=0x1024dc0, readfds=0x0, writefds=0x0, exceptfds=0x0,
+ timeout=0x1024bbc, sigmask=0x0) at hurdselect.c:324
+
+3 0x0114427b in __poll (fds=0x1024dc0, nfds=2, timeout=1000020) at ../sysdeps/mach/hurd/poll.c:48
+
+4 0x0804b770 in iopause (x=0x1024dc0, len=2, deadline=0x1024dd8, stamp=0x1024de8) at iopause.c:29
+
+5 0x08048efc in main (argc=2, argv=0x1024e94) at runsv.c:543
+
+
+and main() shows up as:
+
+ sig_unblock(sig_term);
+
+ sig_unblock(sig_child);
+
+ -> iopause(x, 2 +haslog, &deadline, &now);
+
+ sig_block(sig_term);
+
+ sig_block(sig_child);
+
+
+So it simply looks like the known "signals don't interrupt select" bug.
+