IRC.

1 files changed, 54 insertions, 0 deletions

diff --git a/open_issues/subhurd_vs_proc_server.mdwn b/open_issues/subhurd_vs_proc_server.mdwn
new file mode 100644
index 00000000..36d150f8
--- /dev/null
+++ b/open_issues/subhurd_vs_proc_server.mdwn
@@ -0,0 +1,54 @@
+[[!meta copyright="Copyright © 2013 Free Software Foundation, Inc."]]
+
+[[!meta license="""[[!toggle id="license" text="GFDL 1.2+"]][[!toggleable
+id="license" text="Permission is granted to copy, distribute and/or modify this
+document under the terms of the GNU Free Documentation License, Version 1.2 or
+any later version published by the Free Software Foundation; with no Invariant
+Sections, no Front-Cover Texts, and no Back-Cover Texts.  A copy of the license
+is included in the section entitled [[GNU Free Documentation
+License|/fdl]]."]]"""]]
+
+[[!tag open_issue_hurd]]
+
+
+# IRC, freenode, #hurd, 2013-02-09
+
+    <youpi> also, can you actually gdb a process of another subhurd?
+    <braunr> yes
+    <youpi> but you need to talk to its proc server, don't you?
+    <braunr> i don't know
+    <braunr> but i did it several times
+    <youpi> how?
+    <braunr> the usual way
+    <braunr> gdb /path/to/bin pid
+    <youpi> but which pid?
+    <braunr> the hard part was finding the right pid
+    <youpi> well, gdb still needs to talk with the right proc too
+    <braunr> i don't think it does
+    <youpi> btw about the "unable to adjust libports thread priority" errors
+      I'm seeing on the buildd consoles
+    <braunr> from what i've seen, proc "creates" tasks when it first sees them
+      too
+    <youpi> it's about the destination port
+    <braunr> yes
+    <braunr> i have those when starting a subhurd too
+    <youpi> so it would mean that proc somehow got bogus
+    <youpi> ah
+    <youpi> so  you can actually use your own proc
+    <braunr> yes
+    <braunr> and it feels bogus to me
+    <youpi> and I guess mach lets that proc access the task because your proc
+      is privileged
+    <braunr> probably
+    <braunr> it feels bogus because, you can't rely on pids being allocated per
+      task
+    <braunr> what i mean is that, if some tasks spawn and die quickly
+    <braunr> and you start another application running long enough to see it in
+      ps
+    <braunr> it's pid will be +1, not +the number of created tasks
+    <braunr> which means the proc server will never have seen those previous
+      tasks
+    <braunr> it's minor but a bit confusing
+    <braunr> i personally don't like seeing the tasks of other systems in ps :/
+    <braunr> and despite the ability to use gdb from another hurd, i think we
+      should improve the intra system debugging tools