On Thu, Jun 13, 2019 at 11:37:45AM +1000, Alexey Kardashevskiy wrote: > > > On 12/06/2019 16:11, David Gibson wrote: > > On Thu, Jun 06, 2019 at 02:09:19PM +1000, Alexey Kardashevskiy wrote: > >> The pseries guests do not normally allocate PCI resouces and rely on > >> the system firmware doing so. Furthermore at least at some point in > >> the past the pseries guests won't even be allowed to change BARs, probably > >> it is still the case for phyp. So since the initial commit we have [1] > >> which prevents resource reallocation. > >> > >> This is not a problem until we want specific BAR alignments, for example, > >> PAGE_SIZE==64k to make sure we can still map MMIO BARs directly. For > >> the boot time devices we handle this in SLOF [2] but since QEMU's RTAS > >> does not allocate BARs, the guest does this instead and does not align > >> BARs even if Linux is given pci=resource_alignment=16@pci:0:0 as > >> PCI_PROBE_ONLY makes Linux ignore alignment requests. > >> > >> ARM folks added a dial to control PCI_PROBE_ONLY via the device tree [3]. > >> This makes use of the dial to advertise to the guest that we can handle > >> BAR reassignments. > >> > >> We do not remove the flag from [1] as pseries guests are still supported > >> under phyp so having that removed may cause problems. > >> > >> [1] > >> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/arch/powerpc/platforms/pseries/setup.c?h=v5.1#n773 > >> [2] > >> https://git.qemu.org/?p=SLOF.git;a=blob;f=board-qemu/slof/pci-phb.fs;h=06729bcf77a0d4e900c527adcd9befe2a269f65d;hb=HEAD#l338 > >> [3] > >> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=f81c11af > >> Signed-off-by: Alexey Kardashevskiy <a...@ozlabs.ru> > > > > Changing a guest visible property, that could have a big effect on how > > the guest behaves, without a machine version change seems... unwise. > > > As a general rule - sure, not good. In this particular case QEMU has > always been able to cope with BAR reallocations.
That's not really the point. What I'm worried about is some old kernel version running on a guest in the wild having a bug here and the supposedly compatible qemu change breaking it. > What could probably > make sense is having it as a machine option (pci-probe-only=off by > default) in case if we find some old kernel which cannot handle > "linux,pci-probe-only" but I seriously doubt we'll find such a broken > kernel - I do remove the probe-only switch from guest kernels on a > regular basis last 7 or so years when debugging. Yeah, doing it when debugging isn't really the same as exercising it in production environments. -- 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