On Mon, Jul 01, 2019 at 07:55:03AM +0200, Cédric Le Goater wrote: > On 01/07/2019 07:07, David Gibson wrote: > > On Fri, Jun 21, 2019 at 05:05:45PM +0200, Cédric Le Goater wrote: > >> On 21/06/2019 16:52, Greg Kurz wrote: > >>> As indicated in the function header, this "hcall is only supported for > >>> LISNs that have the ESB hcall flag set to 1 when returned from hcall() > >>> H_INT_GET_SOURCE_INFO". We only set that flag for LSIs actually. > >>> > >>> Check that in h_int_esb(). > >> > >> Indeed. H_INT_ESB should work on any IRQ, but I think it's better > >> to check that the HCALL is only used with the IRQ requiring it. > > > > I'm not so convinced. It seems to me that specifically limiting this > > to certain things limits our options if we ever need future > > workarounds for problems with ESB mapping. > > > > In addition using H_INT_ESB for non-LSIs could be useful for minimal > > guests (e.g. kvm-unit-tests) where mapping memory might be awkward. > > Ah yes. This is true. There is no real reason for enforcing this > restriction in QEMU as H_INT_ESB should work on any irq. We can > keep it that way I guess. It would be a small deviation from PAPR.
True, but "unexpectedly working" is a form of variation exceedingly unlikely to break anything. -- David Gibson | I'll have my music baroque, and my code david AT gibson.dropbear.id.au | minimalist, thank you. NOT _the_ _other_ | _way_ _around_! http://www.ozlabs.org/~dgibson
signature.asc
Description: PGP signature