Skip to contentSkip to conversation

target/s390x/tcg: Set STCK/STCKF condition code after the store

PATCH Applied Ilya Leoshkevich 2 participants 5 messages Jul 14 – Jul 31

lore
Share & tools
lore
#0 lore raw

v2 cover target/s390x/tcg: Set STCK/STCKF condition code after the store

Hi,

Ido reported an assertion failure after failing STCKF and provided a small assembly snippet that causes it.

This series fixes the issue and adds a test based on that snippet.

Best regards,
Ilya

Ilya Leoshkevich (2):
target/s390x/tcg: Set STCK/STCKF condition code after the store tests/tcg/s390x: Test STCKF condition code on a faulting store

 target/s390x/tcg/insn-data.h.inc |  4 +--
 target/s390x/tcg/translate.c     |  2 ++
 tests/tcg/s390x/Makefile.target  |  1 +
 tests/tcg/s390x/stckf.c          | 44 ++++++++++++++++++++++++++++++++
 4 files changed, 49 insertions(+), 2 deletions(-)
 create mode 100644 tests/tcg/s390x/stckf.c
Signature · 1 line
2.55.0
#1 lore raw

PATCH v2 1/2 target/s390x/tcg: Set STCK/STCKF condition code after the store

STORE CLOCK [FAST] to an inaccessible address aborts QEMU:

  $ qemu-s390x ./stckf
  ERROR:cc_helper.c:128:cc_calc_addu: assertion failed: (carry_out <= 1)

op_stck() sets the condition code with gen_op_movi_cc() before the output operand store, which is deferred to wout_m1_64(). Assigning a constant condition code discards the lazy CC values, so the optimizer drops the writes that produced them. When the store then raises an exception, the instruction is suppressed and s390x_restore_state_to_opc() reinstates the cc_op recorded at the start of STCK[F], but cc_src/cc_dst now hold stale values, so the next condition code evaluation reads garbage.

Fix by performing the store manually. The alternative of not discarding in gen_op_movi_cc() keeps the inputs live, but results in less optimal code.

Fixes 434c91a5f4ed ("target-s390: Convert STCK")
---
 target/s390x/tcg/insn-data.h.inc | 4 ++--
 target/s390x/tcg/translate.c     | 2 ++
 2 files changed, 4 insertions(+), 2 deletions(-)
target/s390x/tcg/insn-data.h.inc +2 -2
index 0d5392eac54..1ea72248a6a 100644
@@ -887,8 +887,8 @@
     C(0xe32f, STRVG,   RXY_a, Z,   la2, r1_o, new, m1_64, rev64, 0)
 
 /* STORE CLOCK */
-    F(0xb205, STCK,    S,     Z,   la2, 0, new, m1_64, stck, 0, IF_IO)
-    F(0xb27c, STCKF,   S,     SCF, la2, 0, new, m1_64, stck, 0, IF_IO)
+    F(0xb205, STCK,    S,     Z,   la2, 0, new, 0, stck, 0, IF_IO)
+    F(0xb27c, STCKF,   S,     SCF, la2, 0, new, 0, stck, 0, IF_IO)
 /* STORE CLOCK EXTENDED */
     F(0xb278, STCKE,   S,     Z,   0, a2, 0, 0, stcke, 0, IF_IO)
 
target/s390x/tcg/translate.c +2 -0
index 82165ac1ec0..1b6023168bb 100644
@@ -4108,7 +4108,9 @@ static DisasJumpType op_stap(DisasContext *s, DisasOps *o)
 static DisasJumpType op_stck(DisasContext *s, DisasOps *o)
 {
     gen_helper_stck(o->out, tcg_env);
+    tcg_gen_qemu_st_i64(o->out, o->addr1, get_mem_index(s), MO_BEUQ);
     /* ??? We don't implement clock states.  */
+    /* Set the CC after the store; a suppressed store must preserve it. */
     gen_op_movi_cc(s, 0);
     return DISAS_NEXT;
 }
Signature · 1 line
2.55.0
#2 lore raw

PATCH v2 2/2 tests/tcg/s390x: Test STCKF condition code on a faulting store

In reply to Ilya Leoshkevich

Add a small test to prevent regressions.

---
 tests/tcg/s390x/Makefile.target |  1 +
 tests/tcg/s390x/stckf.c         | 44 +++++++++++++++++++++++++++++++++
 2 files changed, 45 insertions(+)
 create mode 100644 tests/tcg/s390x/stckf.c
tests/tcg/s390x/Makefile.target +1 -0
index 0ca030ded01..6a5bf8b7813 100644
@@ -50,6 +50,7 @@ TESTS+=cvb
 TESTS+=ts
 TESTS+=ex-smc
 TESTS+=divide-to-integer
+TESTS+=stckf
 
 cdsg: CFLAGS+=-pthread
 cdsg: LDFLAGS+=-pthread
tests/tcg/s390x/stckf.c new file +44 -0

44 lines not on this page · raw message

Signature · 1 line
2.55.0
Reviewed-by on the v2 series
#3 lore raw
In reply to Ilya Leoshkevich
Quoted 5 lines · Ilya Leoshkevich

Hi,

Ido reported an assertion failure after failing STCKF and provided a small assembly snippet that causes it.

Quoted 8 lines

This series fixes the issue and adds a test based on that snippet.

Best regards,
Ilya

Ilya Leoshkevich (2):
target/s390x/tcg: Set STCK/STCKF condition code after the store tests/tcg/s390x: Test STCKF condition code on a faulting store

Quoted 8 lines
  target/s390x/tcg/insn-data.h.inc |  4 +--
  target/s390x/tcg/translate.c     |  2 ++
  tests/tcg/s390x/Makefile.target  |  1 +
  tests/tcg/s390x/stckf.c          | 44 ++++++++++++++++++++++++++++++++
  4 files changed, 49 insertions(+), 2 deletions(-)

create mode 100644 tests/tcg/s390x/stckf.c

Applied
#4 lore raw
In reply to Ilya Leoshkevich
Quoted 5 lines · Ilya Leoshkevich

Hi,

Ido reported an assertion failure after failing STCKF and provided a small assembly snippet that causes it.

Quoted 8 lines

This series fixes the issue and adds a test based on that snippet.

Best regards,
Ilya

Ilya Leoshkevich (2):
target/s390x/tcg: Set STCK/STCKF condition code after the store tests/tcg/s390x: Test STCKF condition code on a faulting store

Thanks! Applied for 11.2

Quoted 8 lines
  target/s390x/tcg/insn-data.h.inc |  4 +--
  target/s390x/tcg/translate.c     |  2 ++
  tests/tcg/s390x/Makefile.target  |  1 +
  tests/tcg/s390x/stckf.c          | 44 ++++++++++++++++++++++++++++++++
  4 files changed, 49 insertions(+), 2 deletions(-)

create mode 100644 tests/tcg/s390x/stckf.c