Re: [PATCH] hurd: Implement device memory mapping

[Samuel Thibault]

Hello,

Joan Lledó, le sam. 12 mars 2022 12:22:39 +0100, a ecrit:
> El 23/1/22 a les 18:44, Samuel Thibault ha escrit:
> > With a normal hurd system, i.e. no bootstrap pci-arbiter, and using
> > netdde, which is thus supposed to eventually use the translator sitting
> > on /servers/bus/pci
> 
> I still don't know how to reproduce this, if I run a normal Hurd with an
> e1000 network card, is it not already using netdde?

Yes.

> All seems to work fine for me.

Without a bootstrap pci-arbiter?

Is netdde using your modified libpciaccess?

I don't see how that can be working with your patch: netdde is a
privileged process, and thus get_privileged_ports will succeed, and thus
device_open will be used and fail.

Samuel