path: root/hurd/virtual_file_system
diff options
authorThomas Schwinge <>2011-11-30 21:21:45 +0100
committerThomas Schwinge <>2011-11-30 21:21:45 +0100
commitbe4193108513f02439a211a92fd80e0651f6721b (patch)
treea8fa187c9a6d4ba806a1b7799fa82f712f667c4e /hurd/virtual_file_system
parentbe49aa7ddec52e121d562e14d4d93fd301b05fbb (diff)
Diffstat (limited to 'hurd/virtual_file_system')
1 files changed, 39 insertions, 0 deletions
diff --git a/hurd/virtual_file_system/discussion.mdwn b/hurd/virtual_file_system/discussion.mdwn
new file mode 100644
index 00000000..9e12d01e
--- /dev/null
+++ b/hurd/virtual_file_system/discussion.mdwn
@@ -0,0 +1,39 @@
+[[!meta copyright="Copyright © 2011 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
+[[!tag open_issue_documentation]]
+IRC, freenode, #hurd, 2011-11-12:
+ <sea4ever> So hurd implements a 'transparent translator' somewhere which
+ just passes all IO calls to the posix IO I'm used to? (i.e. read, write,
+ open, close, etc.?)
+ <youpi> it's the normal way of operation
+ <youpi> glibc's read() doesn't do a system call, it always does an RPC to
+ the underlying translator
+ <youpi> be it ext2fs for /, or your foobarfs for your node
+ <sea4ever> Ok that makes sense. How does one program know which translator
+ it should refer to though?
+ <sea4ever> the read() call magically knows which process to invoke?
+ <youpi> the / translator is always known
+ <youpi> and then you ask /'s translator about /home, then /home/you, then
+ /home/you/foobar
+ <youpi> it tells you which other translator tyou have to contact
+ <youpi> that's on open
+ <sea4ever> It's a tree! Ok.
+ <youpi> the notion of fd is then simply knowing the translator
+ <sea4ever> Right. 'file descriptor' is now 'translator address descriptor'
+ maybe.
+ <youpi> it's glibc which knows about FDs, nothing else knows
+ <youpi> yes
+ <youpi> actually an RPC port, simply
+ <sea4ever> I want to try out the new RPC mechanism that mach implements
+ <youpi> err, which "new" RPC ?
+ <youpi> mach's RPCs are very old actually :)