[qemu-xen master] spapr: don't initialize PATB entry if max-cpu-compat < power9

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view

[qemu-xen master] spapr: don't initialize PATB entry if max-cpu-compat < power9

commit 97d17551b57bfbcae5371295cf48850076ce6caa
Author:     Laurent Vivier <[hidden email]>
AuthorDate: Thu Dec 14 19:09:48 2017 +0100
Commit:     Michael Roth <[hidden email]>
CommitDate: Sun Feb 4 23:38:26 2018 -0600

    spapr: don't initialize PATB entry if max-cpu-compat < power9
    if KVM is enabled and KVM capabilities MMU radix is available,
    the partition table entry (patb_entry) for the radix mode is
    initialized by default in ppc_spapr_reset().
    It's a problem if we want to migrate the guest to a POWER8 host
    while the kernel is not started to set the value to the one
    expected for a POWER8 CPU.
    The "-machine max-cpu-compat=power8" should allow to migrate
    a POWER9 KVM host to a POWER8 KVM host, but because patb_entry
    is set, the destination QEMU tries to enable radix mode on the
    POWER8 host. This fails and cancels the migration:
        Process table config unsupported by the host
        error while loading state for instance 0x0 of device 'spapr'
        load of migration failed: Invalid argument
    This patch doesn't set the PATB entry if the user provides
    a CPU compatibility mode that doesn't support radix mode.
    Signed-off-by: Laurent Vivier <[hidden email]>
    Signed-off-by: David Gibson <[hidden email]>
    (cherry picked from commit 1481fe5fcfeb7fcf3c1ebb9d8c0432e3e0188ccf)
    Signed-off-by: Michael Roth <[hidden email]>
 hw/ppc/spapr.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/hw/ppc/spapr.c b/hw/ppc/spapr.c
index 1ac7eb0..66d0ed5 100644
--- a/hw/ppc/spapr.c
+++ b/hw/ppc/spapr.c
@@ -1440,7 +1440,10 @@ static void ppc_spapr_reset(void)
     /* Check for unknown sysbus devices */
     foreach_dynamic_sysbus_device(find_unknown_sysbus_device, NULL);
-    if (kvm_enabled() && kvmppc_has_cap_mmu_radix()) {
+    first_ppc_cpu = POWERPC_CPU(first_cpu);
+    if (kvm_enabled() && kvmppc_has_cap_mmu_radix() &&
+        ppc_check_compat(first_ppc_cpu, CPU_POWERPC_LOGICAL_3_00, 0,
+                         spapr->max_compat_pvr)) {
         /* If using KVM with radix mode available, VCPUs can be started
          * without a HPT because KVM will start them in radix mode.
          * Set the GR bit in PATB so that we know there is no HPT. */
@@ -1499,7 +1502,6 @@ static void ppc_spapr_reset(void)
     /* Set up the entry state */
-    first_ppc_cpu = POWERPC_CPU(first_cpu);
     first_ppc_cpu->env.gpr[3] = fdt_addr;
     first_ppc_cpu->env.gpr[5] = 0;
     first_cpu->halted = 0;
generated by git-patchbot for /home/xen/git/qemu-xen.git#master

Xen-changelog mailing list
[hidden email]