Hi, On Tue, 29 May 2007, Clemens Kolbitsch wrote:
> Johannes Schindelin wrote: > > Hi, > > > > On Tue, 29 May 2007, Clemens Kolbitsch wrote: > > > > > > > Clemens Kolbitsch wrote: > > > > > > > > > > my virtual device inside qemu (a pci device) is listening for > > > > socket-(tcp)-connections. > > > > > > > > however, accept() always fails (code works fine if not executed inside > > > > the qemu-process)... now i'm wondering if qemu interferes somehow... > > > > > > > > is that possible? > > > > > > > > frustrated.... *gg* > > > > > > > > > > > ok.... i found an answer to it. > > > > > > obviously, qemu has to handle interrupt-signals (e.g. from its > > > client-os). this interrupts accepting connections. > > > > > > very bad for me, but at least a logical reason :-/ > > > > > > > I guess that you use the Slirp device. This device is masqueraded to the > > outside, so unless you use something different, like VLAN or TAP, you can > > only make it work using a tunnel. > > > > > no... i think you misunderstand... > > I'm trying to accept a socket inside the qemu-emulation-process. not inside > the simulated operating system. Ah, sorry. I cannot think of anything which could trigger that, except what you suggested (interrupts are in the way), or maybe timing problems. Sorry, Dscho