Linuxulator MFC and VMware
Marcel Moolenaar
I tracked the problem down to the socket ioctls. The LINUX_SIOCGIFADDR
ioctl fails because the ifname translation fails. A simple patch
that disables the name translation makes the problem go away, but the
only result it has is that VMware complains about an old vnet driver.
I'm not sure that's something that's broken on my machine or is still
something that needs to be fixed, so if someone can test the following
patch, I'd be grateful:
Index: linux_ioctl.c
===================================================================
RCS file: /home/ncvs/src/sys/compat/linux/linux_ioctl.c,v
retrieving revision 1.55.2.5
diff -u -r1.55.2.5 linux_ioctl.c
--- linux_ioctl.c 5 Nov 2001 19:08:22 -0000 1.55.2.5
+++ linux_ioctl.c 7 Nov 2001 00:57:56 -0000
@@ -1526,6 +1526,7 @@
case LINUX_SIOCGIFCONF:
case LINUX_SIOCGPGRP:
case LINUX_SIOCSPGRP:
+ case LINUX_SIOCGIFADDR:
/* these ioctls don't take an interface name */
#ifdef DEBUG
printf(__FUNCTION__ "(): ioctl %d\n",
@@ -1534,7 +1535,6 @@
break;
case LINUX_SIOCGIFFLAGS:
- case LINUX_SIOCGIFADDR:
case LINUX_SIOCGIFDSTADDR:
case LINUX_SIOCGIFBRDADDR:
case LINUX_SIOCGIFNETMASK:
Dag-Erling: did you test your changes with VMware on -current?
--
Marcel Moolenaar USPA: A-39004 mar...@xcllnt.net
To Unsubscribe: send mail to majo...@FreeBSD.org
with "unsubscribe freebsd-emulation" in the body of the message
K.Sumitani
I tried Dag-Erling's patch, but unfortunately, it doesn't work.
VMWare says 'Could not get address for /dev/vmnet1....'; same
message without patch.
I tried to set VMWare's Log-Level to Debug but no additional
information was logged.
Regards,
Koshin Sumitani
ksum...@mui.biglobe.ne.jp
Marcel Moolenaar
> Dag-Erling Smorgrav <d...@ofug.org> writes:
> > I'll have a patch ready later tonight.
>
> Attached.
>
> This is a hack; it's linux_ioctl()'s responsibility to check that the
> fd is a socket before calling linux_ioctl_socket(), and there should
> be a separate function for non-socket GIFADDR/SIFADDR/PRIVATE ioctls.
> I'll see if I can come up with something better when I'm back from
> Brighton; in the meantime, use either Marcel's patch or this one if
> they work and don't break anything else (I think Marcel's patch breaks
> ifconfig).
Bummer. I forgot to test this today at work (that's where I have the
only -stable machine I can "develop" on). I'll do it tomorrow...
If someone has tested this in the mean time; please report when it
works, not only when it doesn't work.
--
Marcel Moolenaar USPA: A-39004 mar...@xcllnt.net
To Unsubscribe: send mail to majo...@FreeBSD.org
Mark Santcroos
> Dag-Erling Smorgrav <d...@ofug.org> writes:
>
> ...and vmmon won't build on a post-KSE2 -CURRENT anyway. I believe
> this makes a good argument for moving the rtc and vmmon modules into
> the base system.
I posted a patch for this on current@ sometime ago, but no ports
committer seems to have interest to make the change.
(And the maintainer never respond to my emails)
Mark
--
Mark Santcroos RIPE Network Coordination Centre
http://www.ripe.net/home/mark/ New Projects Group/TTM
Marcel Moolenaar
>
> I posted a patch for this on current@ sometime ago, but no ports
> committer seems to have interest to make the change.
> (And the maintainer never respond to my emails)
I tried the patch, but it's stale now:
In file included from /usr/ports/emulators/vmware2/work/vmware-distrib/vmmon-only/freebsd/hostif.c:59:
/sys/sys/mutex.h:243: #error LOCK_DEBUG not defined, include <sys/lock.h> before <sys/mutex.h>
*** Error code 1
Apply the following diff *after* applying the patch:
--- hostif.c.~ Thu Nov 8 01:03:05 2001
+++ hostif.c Thu Nov 8 01:03:36 2001
@@ -55,10 +55,9 @@
#include <vm/vm.h>
#include <vm/vm_param.h>
+#include <sys/lock.h>
#if __FreeBSD_version >= 500018
#include <sys/mutex.h>
-#else
-#include <sys/lock.h>
#endif
#include <vm/pmap.h>
#include <vm/vm_map.h>
Mark Santcroos
I think it becomes time to replace the maintainer btw.
Mark
On Thu, Nov 08, 2001 at 01:05:53AM -0800, Marcel Moolenaar wrote:
> On Thu, Nov 08, 2001 at 07:49:30AM +0100, Mark Santcroos wrote:
> >
> > I posted a patch for this on current@ sometime ago, but no ports
> > committer seems to have interest to make the change.
> > (And the maintainer never respond to my emails)
>
> I tried the patch, but it's stale now:
>
> In file included from /usr/ports/emulators/vmware2/work/vmware-distrib/vmmon-only/freebsd/hostif.c:59:
> /sys/sys/mutex.h:243: #error LOCK_DEBUG not defined, include <sys/lock.h> before <sys/mutex.h>
> *** Error code 1
>
> Apply the following diff *after* applying the patch:
>
> --- hostif.c.~ Thu Nov 8 01:03:05 2001
> +++ hostif.c Thu Nov 8 01:03:36 2001
> @@ -55,10 +55,9 @@
>
> #include <vm/vm.h>
> #include <vm/vm_param.h>
> +#include <sys/lock.h>
> #if __FreeBSD_version >= 500018
> #include <sys/mutex.h>
> -#else
> -#include <sys/lock.h>
> #endif
> #include <vm/pmap.h>
> #include <vm/vm_map.h>
>
> --
> Marcel Moolenaar USPA: A-39004 mar...@xcllnt.net
>
> To Unsubscribe: send mail to majo...@FreeBSD.org
> with "unsubscribe freebsd-emulation" in the body of the message
--
Mark Santcroos RIPE Network Coordination Centre
http://www.ripe.net/home/mark/ New Projects Group/TTM
To Unsubscribe: send mail to majo...@FreeBSD.org
Marcel Moolenaar
> On Wed, Nov 07, 2001 at 04:58:41PM +0100, Dag-Erling Smorgrav wrote:
> > Dag-Erling Smorgrav <d...@ofug.org> writes:
> >
> > Attached.
> >
> > This is a hack; it's linux_ioctl()'s responsibility to check that the
> > fd is a socket before calling linux_ioctl_socket(), and there should
> > be a separate function for non-socket GIFADDR/SIFADDR/PRIVATE ioctls.
> > I'll see if I can come up with something better when I'm back from
> > Brighton; in the meantime, use either Marcel's patch or this one if
> > they work and don't break anything else (I think Marcel's patch breaks
> > ifconfig).
>
> Bummer. I forgot to test this today at work (that's where I have the
> only -stable machine I can "develop" on). I'll do it tomorrow...
I tried this on -current just now and it fails with the same popup I
got with my patch on -stable:
"Encountered an error while initializing the ethernet address.
You probably have an old vnet driver. Try installing a newer version
Failed to configure ethernet0."
I suspect another socket ioctl. I'll take a look at it tomorrow.
--
Marcel Moolenaar USPA: A-39004 mar...@xcllnt.net
To Unsubscribe: send mail to majo...@FreeBSD.org