qemu-kvm/SOURCES/kvm-target-s390x-kvm-Honor-...

107 lines
3.8 KiB
Diff

From 8d1a60069cddcc69ef1a6f50f2b55343de348b57 Mon Sep 17 00:00:00 2001
From: Janis Schoetterl-Glausch <scgl@linux.ibm.com>
Date: Fri, 6 May 2022 17:39:56 +0200
Subject: [PATCH 2/3] target/s390x: kvm: Honor storage keys during emulation
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
RH-Author: Cédric Le Goater <None>
RH-MergeRequest: 220: s390x: Fix skey test in kvm_unit_test
RH-Bugzilla: 2124757
RH-Acked-by: Thomas Huth <thuth@redhat.com>
RH-Acked-by: David Hildenbrand <david@redhat.com>
RH-Acked-by: Cornelia Huck <cohuck@redhat.com>
RH-Commit: [2/2] 980dbb4eba8d2f1da7cf4113230d0a6483cffc4f
Storage key controlled protection is currently not honored when
emulating instructions.
If available, enable key protection for the MEM_OP ioctl, thereby
enabling it for the s390_cpu_virt_mem_* functions, when using kvm.
As a result, the emulation of the following instructions honors storage
keys:
* CLP
The Synch I/O CLP command would need special handling in order
to support storage keys, but is currently not supported.
* CHSC
Performing commands asynchronously would require special
handling, but commands are currently always synchronous.
* STSI
* TSCH
Must (and does) not change channel if terminated due to
protection.
* MSCH
Suppressed on protection, works because fetching instruction.
* SSCH
Suppressed on protection, works because fetching instruction.
* STSCH
* STCRW
Suppressed on protection, this works because no partial store is
possible, because the operand cannot span multiple pages.
* PCISTB
* MPCIFC
* STPCIFC
Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=2124757
Signed-off-by: Janis Schoetterl-Glausch <scgl@linux.ibm.com>
Message-Id: <20220506153956.2217601-3-scgl@linux.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
(cherry picked from commit 54354861d21b69ec0781f43e67b8d4f6edad7e3f)
Signed-off-by: Cédric Le Goater <clg@redhat.com>
---
target/s390x/kvm/kvm.c | 9 +++++++++
1 file changed, 9 insertions(+)
diff --git a/target/s390x/kvm/kvm.c b/target/s390x/kvm/kvm.c
index c52434985b..ba04997da1 100644
--- a/target/s390x/kvm/kvm.c
+++ b/target/s390x/kvm/kvm.c
@@ -152,12 +152,15 @@ const KVMCapabilityInfo kvm_arch_required_capabilities[] = {
static int cap_sync_regs;
static int cap_async_pf;
static int cap_mem_op;
+static int cap_mem_op_extension;
static int cap_s390_irq;
static int cap_ri;
static int cap_hpage_1m;
static int cap_vcpu_resets;
static int cap_protected;
+static bool mem_op_storage_key_support;
+
static int active_cmma;
static int kvm_s390_query_mem_limit(uint64_t *memory_limit)
@@ -355,6 +358,8 @@ int kvm_arch_init(MachineState *ms, KVMState *s)
cap_sync_regs = kvm_check_extension(s, KVM_CAP_SYNC_REGS);
cap_async_pf = kvm_check_extension(s, KVM_CAP_ASYNC_PF);
cap_mem_op = kvm_check_extension(s, KVM_CAP_S390_MEM_OP);
+ cap_mem_op_extension = kvm_check_extension(s, KVM_CAP_S390_MEM_OP_EXTENSION);
+ mem_op_storage_key_support = cap_mem_op_extension > 0;
cap_s390_irq = kvm_check_extension(s, KVM_CAP_S390_INJECT_IRQ);
cap_vcpu_resets = kvm_check_extension(s, KVM_CAP_S390_VCPU_RESETS);
cap_protected = kvm_check_extension(s, KVM_CAP_S390_PROTECTED);
@@ -843,6 +848,7 @@ int kvm_s390_mem_op(S390CPU *cpu, vaddr addr, uint8_t ar, void *hostbuf,
: KVM_S390_MEMOP_LOGICAL_READ,
.buf = (uint64_t)hostbuf,
.ar = ar,
+ .key = (cpu->env.psw.mask & PSW_MASK_KEY) >> PSW_SHIFT_KEY,
};
int ret;
@@ -852,6 +858,9 @@ int kvm_s390_mem_op(S390CPU *cpu, vaddr addr, uint8_t ar, void *hostbuf,
if (!hostbuf) {
mem_op.flags |= KVM_S390_MEMOP_F_CHECK_ONLY;
}
+ if (mem_op_storage_key_support) {
+ mem_op.flags |= KVM_S390_MEMOP_F_SKEY_PROTECTION;
+ }
ret = kvm_vcpu_ioctl(CPU(cpu), KVM_S390_MEM_OP, &mem_op);
if (ret < 0) {
--
2.35.3