GNU Hurd/ open issues/ virtualization/ remap root translator

Welcome to... ... the GNU Hurd!

/!\ Does this completely resolve server overriding?

IRC, freenode, #hurd, 2013-01-05

<youpi> so we have a "remap" root translator?
<youpi> I  mean this:
<youpi> I'd run my shell in a subhurd whose only difference is that the
  root is not the system's root, but my own
<youpi> which catches accesses to /servers/socket/2 for instance
<youpi> but leaves the rest flow through the system's root
<braunr> there is just boot, i don't think it can do that
<youpi> it'd be useful to have that
<youpi> it'd be a very useful feature
<youpi> to use another tcp/ip stack etc.
<braunr> what happens when translators need to locate other translators
  used by the client ?
<youpi> can't it tell the client to ask the real system's root?
<youpi> (with the same path)
<youpi> I don't remember the exact reply name
<braunr> hum, it's getting too fuzzy for my head :p
<youpi> well, I mean it's just like translator entries in an ext2fs
<youpi> ext2fs replies "not me, this one"
<braunr> but what if e.g. a user has its own pflocal, and when calling
  another translator, that one wants to contact the pflocal used by the
  client ?
<youpi> ah, that won't work of course
<braunr> do we actually have such cases btw ?
<braunr> procfs perhaps
<youpi> I don't think we'd want it actually
<braunr> but isn't that required sometimes ?
<youpi> inside a shell script, yes
<braunr> for example, a storeio translator could ask about the priority
  properties of the client to proc
<youpi> but I don't remember a case where an external translator would need
  the access
<youpi> well, that's actually what we want
<youpi> we don't want to fool the storeio with user-provided data :)
<braunr> yes
<youpi> unless the user starts the storeio himself, in which case he will
  have to re-root it
<braunr> so it has to locate the right translator, despite not using the
  remap root translator
<youpi> err, it will already
<youpi> by just using the system's path
<braunr> ?
<youpi> maybe you need to say exactly what "it" and "right" are :)
<braunr> ok, let's imagine your previous example with a subhurd and pfinet
<braunr> the remap translator would imply that users from the subhurd
  *directly* access all services from the main hurd, except when routed
  otherwise by the remap translator to pfinet
<youpi> by "directly", I mean asking the remap translator, which gives as
  answer "not me, the root"
<braunr> now, what if a translator in the main hurd wants e.g. network
  stats from pfinet, it will ask the main one, not the one obtained through
  remap
<braunr> yes
<youpi> that's completely fine
<braunr> ah
<braunr> that's fine if the results don't matter
<youpi> to get network stats from the user pfinet you'd have to be inside
  the shell using the remap translator
<braunr> otherwise they're inconsistent
<braunr> yes
<youpi> I don't see why you'd want to get the pfinet stats from outside
<youpi> you mean ethernet board usage?
<braunr> service interactions
<braunr> i can't think of anything relevant with pfinet
<braunr> but imagine pflocal and credentials
<youpi> I believe that'd still be ok
<youpi> i.e. things outside the remap want to know the actual system things
<youpi> while things inside want to know the remapped things
<youpi> and you need that to avoid getting fooled by the user remapping
<braunr> for credentials, i think it works because the client provides
  rights, so it would provide rights to the remapped translators in this
  case
<braunr> this would need to be generalized
<youpi> I believe it's already general
<braunr> well no
<braunr> procfs for example will always talk to the "true" proc server
<youpi> sure
<youpi> that's what I want from the outside
<youpi> if the user, from the inside, wants another view, he'll have to
  start another procfs
<youpi> his own one
<braunr> ok
<youpi> attached to the remapping

IRC, freenode, #hurd, 2013-01-29

<youpi> ok, the remap translator was too easy
<youpi> just took fakeroot.c
<youpi> added if (!strcmp("bin/foo", filename))         filename =
  "bin/bash"; in 
<youpi> netfs_S_dir_lookup
<youpi> and it just works

dir lookup.

<youpi> ok, remap does indeed take my own pfinet
<youpi> good :)
<youpi> pfinet's tun seems to be working too
<youpi> it's however not really flexible, it has to show up in /dev/tunx
<youpi> I'll have a look at fixing that
<youpi> yep, works fine

IRC, freenode, #hurd, 2013-02-01

<youpi> braunr: as I expected, simply passing FS_RETRY_REAUTH does the
  remapping trick

IRC, freenode, #hurd, 2013-02-12

<braunr>
  http://darnassus.sceen.net/~hurd-web/community/gsoc/project_ideas/server_overriding/
<braunr> youpi: isn't that your remap translator ?
<youpi> completely
<youpi> remap being (5)

IRC, freenode, #hurd, 2013-02-25

<youpi> I'm just having an issue with getcwd getting in the sky
<youpi> I wonder whether libc might need patching to understand it's in
  some sort of chroot
<youpi> or perhaps remap fixed into avoiding .. of / being odd
<youpi> erf, it's actually an explicit error
<youpi> libc just doesn't want to have a ".." / being different from CRDIR
<youpi> let me just comment out that :)
<youpi> way better :)
<youpi> yep, just works fine

IRC, freenode, #hurd, 2013-03-16

<braunr> youpi: is the /bin/remap --help output correct ?

fsysopts

Doesn't support fsysopts.