Blaisorblade wrote:
> On Tuesday 05 April 2005 19:01, Jan Rychter wrote:
> > > Blaisorblade wrote:
> > > > On Sunday 03 April 2005 13:48, Jan Rychter wrote:
> > > > > I used UML successfully in the past and recently I've tried to revive
> > > > > my UML installation. Unfortunately, I've run into a range of problems
> > > > > (updates required for newer host kernels, non-working hostfs, etc).
> > > > >
> > > > > I am now at a stage where I have linux-2.4.27 patched with
> > > > > uml-2.4.27-bs2-pre7.patch and a host 2.6.10 patched with
> > > > > skas3-v8-rc5.
> > > > >
> > > > > The issue I have is that hostfs doesn't work they way I want it to.
> > > > > What I get is all hostfs files owned by root:root.
> > > > >
> > > > > The semantics I'd like to see are:
> > > > >
> > > > > -- uid:gid mapping of 1:1 between the host and uml,
> > > > >
> > > > > -- normal permission checks done on the host for requests done by
> > > > > uml.
> > > >
> > > > This is the normal situation, but there are some rare exception (i.e.
> > > > when you boot from hostfs as rootfs, instead of using one UBD). So,
> > > > please describe exactly your setup (UML command line and boot messages,
> > > > mount command, and so on).
> > > >
> > > > Are you using hostfs as your rootfs? From what you say it seems not,
> > > > but the behaviour you see has been explicitly coded for that case.
> > >
> > > No, I am not using it as my rootfs. The semantics that I'd like to
> > > achieve are similar to an NFS-mounted /home filesystem, except that only
> > > the user running the UML should have access to his files.
> > >
> > > I mount the hostfs via the fstab entry:
> > >
> > > none /home hostfs /home,rw 0 0
> > >
> > > ... which gets me all files owned by root. What I'd like to have is
> > > either a full /home or a /home/jwr (my $HOME). As user 501 (my uid)
> > > inside UML I should be able to access files on the host beloging to uid
> > > 501, assuming that UML runs under uid 501.
> >
> > So, is this mapping of ownership of all files to root in hostfs a bug or
> > a feature?
> It's a bug. And I'm seeing some others one here, in the local tree.
> > Is there any other information that I should supply?
> Try with 2.4.27-bs1. And tell me if you see any difference.
I tried, and that version works just fine. It does exactly what I want
it to do.
However, I can't use it for real work, as it seems there is a strange
problem with TCP connections going over the TAP interface. A good way to
reproduce this is to run 'xlsfonts' on the client machine. It is
sometimes unable to get the full list of fonts from the server and hangs
forever.
Example strace output (filtered, let me know if you want the full thing,
which is quite long):
execve("/usr/X11R6/bin/xlsfonts", ["xlsfonts"], [/* 40 vars */]) = 0
[...]
socket(PF_INET, SOCK_STREAM, IPPROTO_IP) = 3
setsockopt(3, SOL_TCP, TCP_NODELAY, [1], 4) = 0
setsockopt(3, SOL_SOCKET, SO_KEEPALIVE, [1], 4) = 0
connect(3, {sin_family=AF_INET, sin_port=htons(6000),
sin_addr=inet_addr("10.200.254.1")}}, 16) = 0
getsockname(3, {sin_family=AF_INET, sin_port=htons(1038),
sin_addr=inet_addr("10.200.254.2")}}, [16]) = 0
getpeername(3, {sin_family=AF_INET, sin_port=htons(6000),
sin_addr=inet_addr("10.200.254.1")}}, [16]) = 0
fcntl64(3, F_SETFD, FD_CLOEXEC) = 0
[...]
writev(3, [{"l\0\v\0\0\0\0\0\0\0\0\0", 12}], 1) = 12
fcntl64(3, F_GETFL) = 0x2 (flags O_RDWR)
fcntl64(3, F_SETFL, O_RDWR|O_NONBLOCK) = 0
read(3, "\1\0\v\0\0\0m\0", 8) = 8
[...]
readv(3, [{"l-medium-r-normal--0-0-0-0-p-0-i"..., 66516}, {"", 0}], 2) = 15928
readv(3, [{"[EMAIL PROTECTED]"..., 50588}, {"", 0}], 2) = 15928
readv(3, 0xbffff770, 2) = -1 EAGAIN (Resource temporarily
unavailable)
select(4, [3], NULL, NULL, NULL) = 1 (in [3])
readv(3, [{"-bold-r-normal--0-0-0-0-p-0-koi8"..., 34660}, {"", 0}], 2) = 15928
readv(3, 0xbffff770, 2) = -1 EAGAIN (Resource temporarily
unavailable)
select(4, [3], NULL, NULL, NULL) = 1 (in [3])
readv(3, [{"8-ru7-urw-urw palladio l-medium-"..., 18732}, {"", 0}], 2) = 15928
readv(3, 0xbffff770, 2) = -1 EAGAIN (Resource temporarily
unavailable)
select(4, [3], NULL, NULL, NULL <unfinished ...>
--- SIGINT (Interrupt) ---
+++ killed by SIGINT +++
--J.
-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
_______________________________________________
User-mode-linux-user mailing list
User-mode-linux-user@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-user
