Skip to contentSkip to conversation

target/s390x/tcg: Fix and improve the SACF instruction

PATCH In review Thomas Huth 4 participants 5 messages Dec 1, 2022 – Dec 2, 2022

lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 target/s390x/tcg: Fix and improve the SACF instruction

The SET ADDRESS SPACE CONTROL FAST instruction is not privileged, it can be used from problem space, too. Just the switching to the home address space is privileged and should still generate a privilege exception. This bug is e.g. causing programs like Java that use the "getcpu" vdso kernel function to crash (see https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=990417#26 ).

While we're at it, also check if DAT is not enabled. In that case the instruction is supposed to generate a special operation exception.

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/655

---
 target/s390x/tcg/insn-data.h.inc | 2 +-
 target/s390x/tcg/cc_helper.c     | 7 +++++++
 2 files changed, 8 insertions(+), 1 deletion(-)
target/s390x/tcg/insn-data.h.inc +1 -1
index 7e952bdfc8..54d4250c9f 100644
@@ -1365,7 +1365,7 @@
 /* SERVICE CALL LOGICAL PROCESSOR (PV hypercall) */
     F(0xb220, SERVC,   RRE,   Z,   r1_o, r2_o, 0, 0, servc, 0, IF_PRIV | IF_IO)
 /* SET ADDRESS SPACE CONTROL FAST */
-    F(0xb279, SACF,    S,     Z,   0, a2, 0, 0, sacf, 0, IF_PRIV)
+    C(0xb279, SACF,    S,     Z,   0, a2, 0, 0, sacf, 0)
 /* SET CLOCK */
     F(0xb204, SCK,     S,     Z,   0, m2_64a, 0, 0, sck, 0, IF_PRIV | IF_IO)
 /* SET CLOCK COMPARATOR */
target/s390x/tcg/cc_helper.c +7 -0
index b2e8d3d9f5..b36f8cdc8b 100644
@@ -487,6 +487,10 @@ void HELPER(sacf)(CPUS390XState *env, uint64_t a1)
 {
     HELPER_LOG("%s: %16" PRIx64 "\n", __func__, a1);
 
+    if (!(env->psw.mask & PSW_MASK_DAT)) {
+        tcg_s390_program_interrupt(env, PGM_SPECIAL_OP, GETPC());
+    }
+
     switch (a1 & 0xf00) {
     case 0x000:
         env->psw.mask &= ~PSW_MASK_ASC;
@@ -497,6 +501,9 @@ void HELPER(sacf)(CPUS390XState *env, uint64_t a1)
         env->psw.mask |= PSW_ASC_SECONDARY;
         break;
     case 0x300:
+        if ((env->psw.mask & PSW_MASK_PSTATE) != 0) {
+            tcg_s390_program_interrupt(env, PGM_PRIVILEGED, GETPC());
+        }
         env->psw.mask &= ~PSW_MASK_ASC;
         env->psw.mask |= PSW_ASC_HOME;
         break;
Signature · 1 line
2.31.1
Reviewed-by on v2 1/1
#1 lore raw
Quoted 6 lines · Thomas Huth

The SET ADDRESS SPACE CONTROL FAST instruction is not privileged, it can be used from problem space, too. Just the switching to the home address space is privileged and should still generate a privilege exception. This bug is e.g. causing programs like Java that use the "getcpu" vdso kernel function to crash (see https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=990417#26 ).

Quoted 6 lines

While we're at it, also check if DAT is not enabled. In that case the instruction is supposed to generate a special operation exception.

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/655

---

Looks ok, as far as it goes. We appear to be missing the check for CR0_SECONDARY, which is unpredictable for SACF but mandatory for SAC.

I'll give you

for fixing the incorrect IF_PRIV check, which by itself should be enough to fix the Java issue.

r~

#2 lore raw
Quoted 7 lines · Richard Henderson
Quoted 6 lines · Thomas Huth

The SET ADDRESS SPACE CONTROL FAST instruction is not privileged, it can be used from problem space, too. Just the switching to the home address space is privileged and should still generate a privilege exception. This bug is e.g. causing programs like Java that use the "getcpu" vdso kernel function to crash (see https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=990417#26 ).

Quoted 8 lines
Quoted 6 lines

While we're at it, also check if DAT is not enabled. In that case the instruction is supposed to generate a special operation exception.

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/655

---

Looks ok, as far as it goes.  We appear to be missing the check for CR0_SECONDARY, which is unpredictable for SACF but mandatory for SAC.

Yes, but if I got our sources right, we do not implement SAC yet. Looks like Linux does not use it, so nobody bothered to implement it yet. Since it should be very similar to SACF, it should be easy to add - I can try to come up with an additional patch for it later.

Quoted 5 lines

I'll give you

for fixing the incorrect IF_PRIV check, which by itself should be enough to fix the Java issue.

Thanks!

Thomas

PS: We might have a similar bug with the MVCP and MVCS instructions ... seems like they could be called from userspace in certain situations, too.

Reviewed-by on v2 1/1
#3 lore raw
In reply to Thomas Huth
Quoted 45 lines · Thomas Huth

The SET ADDRESS SPACE CONTROL FAST instruction is not privileged, it can be used from problem space, too. Just the switching to the home address space is privileged and should still generate a privilege exception. This bug is e.g. causing programs like Java that use the "getcpu" vdso kernel function to crash (see https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=990417#26 ).

While we're at it, also check if DAT is not enabled. In that case the instruction is supposed to generate a special operation exception.

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/655

---
  target/s390x/tcg/insn-data.h.inc | 2 +-
  target/s390x/tcg/cc_helper.c     | 7 +++++++
  2 files changed, 8 insertions(+), 1 deletion(-)
target/s390x/tcg/insn-data.h.inc +1 -1
index 7e952bdfc8..54d4250c9f 100644
@@ -1365,7 +1365,7 @@
  /* SERVICE CALL LOGICAL PROCESSOR (PV hypercall) */
      F(0xb220, SERVC,   RRE,   Z,   r1_o, r2_o, 0, 0, servc, 0, IF_PRIV | IF_IO)
  /* SET ADDRESS SPACE CONTROL FAST */
-    F(0xb279, SACF,    S,     Z,   0, a2, 0, 0, sacf, 0, IF_PRIV)
+    C(0xb279, SACF,    S,     Z,   0, a2, 0, 0, sacf, 0)
  /* SET CLOCK */
      F(0xb204, SCK,     S,     Z,   0, m2_64a, 0, 0, sck, 0, IF_PRIV | IF_IO)
  /* SET CLOCK COMPARATOR */
target/s390x/tcg/cc_helper.c +4 -0
index b2e8d3d9f5..b36f8cdc8b 100644
@@ -487,6 +487,10 @@ void HELPER(sacf)(CPUS390XState *env, uint64_t a1)
  {
      HELPER_LOG("%s: %16" PRIx64 "\n", __func__, a1);
 
+    if (!(env->psw.mask & PSW_MASK_DAT)) {
+        tcg_s390_program_interrupt(env, PGM_SPECIAL_OP, GETPC());
+    }
+
      switch (a1 & 0xf00) {
      case 0x000:
          env->psw.mask &= ~PSW_MASK_ASC;

target/s390x/tcg/cc_helper.c void HELPER(sacf)(CPUS390XState *env, uint64_t a1) · line 509

Quoted 9 lines
target/s390x/tcg/cc_helper.c +3 -0
@@ -497,6 +501,9 @@ void HELPER(sacf)(CPUS390XState *env, uint64_t a1)
          env->psw.mask |= PSW_ASC_SECONDARY;
          break;
      case 0x300:
+        if ((env->psw.mask & PSW_MASK_PSTATE) != 0) {
+            tcg_s390_program_interrupt(env, PGM_PRIVILEGED, GETPC());
+        }
          env->psw.mask &= ~PSW_MASK_ASC;
          env->psw.mask |= PSW_ASC_HOME;
          break;
Signature · 3 lines
Thanks,

David / dhildenb
Reviewed-by on v2 1/1
#4 lore raw
In reply to Thomas Huth
Quoted 14 lines · Thomas Huth

The SET ADDRESS SPACE CONTROL FAST instruction is not privileged, it can be
used from problem space, too. Just the switching to the home address space
is privileged and should still generate a privilege exception. This bug is
e.g. causing programs like Java that use the "getcpu" vdso kernel function
to crash (see
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=990417#26 ).

While we're at it, also check if DAT is not enabled. In that case the instruction is supposed to generate a special operation exception.

Quoted 6 lines

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/655

---
 target/s390x/tcg/insn-data.h.inc | 2 +-
 target/s390x/tcg/cc_helper.c     | 7 +++++++

 2 files changed, 8 insertions(+), 1 deletion(-)