On Tue, Jan 09, 2018 at 10:00:25PM +0200, Mika Westerberg wrote:
> On Tue, Jan 09, 2018 at 03:25:58PM +0000, Lorenzo Pieralisi wrote:
> > On Tue, Jan 09, 2018 at 03:42:21PM +0100, Koen Vandeputte wrote:
> > > The subordinate value indicates the highest bus number which can be
> > > reached downstream though a certain device.
> > >
> > > Commit a20c7f36bd3d ("PCI: Do not allocate more buses than available in
> > > parent")
> > > ensures that downstream devices cannot assign busnumbers higher than the
> > > upstream device subordinate number, which was indeed illogical.
> > >
> > > By default, dw_pcie_setup_rc() inits the Root Complex subordinate to a
> > > value of 0x01.
> > >
> > > Due to this combined with above commit, enumeration stops digging deeper
> > > downstream as soon as bus num 0x01 has been assigned, which is always
> > > the case for a bridge device.
> > >
> > > This results in all devices behind a bridge bus to remain undetected, as
> > > these would be connected to bus 0x02 or higher.
> > >
> > > Fix this by initializing the RC to a subordinate value of 0xff, meaning
> > > that all busses [0x00-0xff] are reachable through this RC.
> >
> > This is not a correct description of the problem. AFAICS all busses
> > are reachable through this RC _regardless_ of whatever subordinate
> > bus number value you programme into it.
>
> Type 1 (the ones directed to the other side of the bridge) configuration
> transactions are not forwarded if the bus number in the transaction is
> not included in secondary and subordinate numbers of the root bridge. I
> think the description here is pretty accurate as the bridges (anything
> higher than bus number 1) below are effectively hidden.
That's correct - but that's not what this patch is fixing which is
what the commit log should describe.
See:
https://marc.info/?l=linux-arm-kernel&m=151540153522730&w=2
