Skip to contentSkip to conversation

powerpc/64s: system call scv tabort fix for corrupt irq soft-mask state

PATCH Superseded Nicholas Piggin 4 participants 11 messages Sep 1, 2021 – Sep 3, 2021

Subsystems (2)

v2 of 3 Superseded Latest is v3 →

lore
Share & tools
lore

Nicholas Piggin

author reviewer
#0 lore raw

PATCH v2 1/2 powerpc/64s: system call scv tabort fix for corrupt irq soft-mask state

If a system call is made with a transaction active, the kernel immediately aborts it and returns. scv system calls disable irqs even earlier in their interrupt handler, and tabort_syscall does not fix this up.

This can result in irq soft-mask state being messed up on the next kernel entry, and crashing at BUG_ON(arch_irq_disabled_regs(regs)) in the kernel exit handlers, or possibly worse.

Fix this by having tabort_syscall setting irq soft-mask back to enabled (which requires MSR[EE] be disabled first).

Fixes 7fa95f9adaee7 ("powerpc/64s: system call support for scv/rfscv instructions")

---

Tested the wrong kernel before sending v1 and missed a bug, sorry.

 arch/powerpc/kernel/interrupt_64.S | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)
arch/powerpc/kernel/interrupt_64.S +7 -1
index d4212d2ff0b5..9c31d65b4851 100644
@@ -428,16 +428,22 @@ RESTART_TABLE(.Lsyscall_rst_start, .Lsyscall_rst_end, syscall_restart)
 #ifdef CONFIG_PPC_TRANSACTIONAL_MEM
 tabort_syscall:
 _ASM_NOKPROBE_SYMBOL(tabort_syscall)
-	/* Firstly we need to enable TM in the kernel */
+	/* We need to enable TM in the kernel, and disable EE (for scv) */
 	mfmsr	r10
 	li	r9, 1
 	rldimi	r10, r9, MSR_TM_LG, 63-MSR_TM_LG
+	LOAD_REG_IMMEDIATE(r9, MSR_EE)
+	andc	r10, r10, r9
 	mtmsrd	r10, 0
 
 	/* tabort, this dooms the transaction, nothing else */
 	li	r9, (TM_CAUSE_SYSCALL|TM_CAUSE_PERSISTENT)
 	TABORT(R9)
 
+	/* scv has disabled irqs so must re-enable. sc just remains enabled */
+	li	r9,IRQS_ENABLED
+	stb	r9,PACAIRQSOFTMASK(r13)
+
 	/*
 	 * Return directly to userspace. We have corrupted user register state,
 	 * but userspace will never see that register state. Execution will
Signature · 1 line
2.23.0
author reviewer
#1 lore raw

PATCH v2 2/2 selftests/powerpc: Add scv versions of the basic TM syscall tests

The basic TM vs syscall test code hard codes an sc instruction for the system call, which fails to cover scv even when the userspace libc has support for it.

Duplicate the tests with hard coded scv variants so both are tested when possible.

---
 .../selftests/powerpc/tm/tm-syscall-asm.S     | 46 +++++++++++++++++++
 .../testing/selftests/powerpc/tm/tm-syscall.c | 36 ++++++++++++---
 2 files changed, 75 insertions(+), 7 deletions(-)
tools/testing/selftests/powerpc/tm/tm-syscall-asm.S +46 -0
index bd1ca25febe4..849316831e6a 100644
@@ -2,6 +2,10 @@
 #include <ppc-asm.h>
 #include <asm/unistd.h>
 
+/* ppc-asm.h does not define r0 or r1 */
+#define r0 0
+#define r1 1
+
 	.text
 FUNC_START(getppid_tm_active)
 	tbegin.
@@ -26,3 +30,45 @@ FUNC_START(getppid_tm_suspended)
 1:
 	li	r3, -1
 	blr
+
+FUNC_START(getppid_scv_tm_active)
+	mflr	r0
+	std	r0,16(r1)
+	stdu	r1,-32(r1)
+	tbegin.
+	beq 1f
+	li	r0, __NR_getppid
+	scv	0
+	tend.
+	addi	r1,r1,32
+	ld	r0,16(r1)
+	mtlr	r0
+	blr
+1:
+	li	r3, -1
+	addi	r1,r1,32
+	ld	r0,16(r1)
+	mtlr	r0
+	blr
+
+FUNC_START(getppid_scv_tm_suspended)
+	mflr	r0
+	std	r0,16(r1)
+	stdu	r1,-32(r1)
+	tbegin.
+	beq 1f
+	li	r0, __NR_getppid
+	tsuspend.
+	scv	0
+	tresume.
+	tend.
+	addi	r1,r1,32
+	ld	r0,16(r1)
+	mtlr	r0
+	blr
+1:
+	li	r3, -1
+	addi	r1,r1,32
+	ld	r0,16(r1)
+	mtlr	r0
+	blr
tools/testing/selftests/powerpc/tm/tm-syscall.c +29 -7
index becb8207b432..9a822208680e 100644
@@ -19,24 +19,37 @@
 #include "utils.h"
 #include "tm.h"
 
+#ifndef PPC_FEATURE2_SCV
+#define PPC_FEATURE2_SCV               0x00100000 /* scv syscall */
+#endif
+
 extern int getppid_tm_active(void);
 extern int getppid_tm_suspended(void);
+extern int getppid_scv_tm_active(void);
+extern int getppid_scv_tm_suspended(void);
 
 unsigned retries = 0;
 
 #define TEST_DURATION 10 /* seconds */
 #define TM_RETRIES 100
 
-pid_t getppid_tm(bool suspend)
+pid_t getppid_tm(bool scv, bool suspend)
 {
 	int i;
 	pid_t pid;
 
 	for (i = 0; i < TM_RETRIES; i++) {
-		if (suspend)
-			pid = getppid_tm_suspended();
-		else
-			pid = getppid_tm_active();
+		if (suspend) {
+			if (scv)
+				pid = getppid_scv_tm_suspended();
+			else
+				pid = getppid_tm_suspended();
+		} else {
+			if (scv)
+				pid = getppid_scv_tm_active();
+			else
+				pid = getppid_tm_active();
+		}
 
 		if (pid >= 0)
 			return pid;
@@ -82,15 +95,24 @@ int tm_syscall(void)
 		 * Test a syscall within a suspended transaction and verify
 		 * that it succeeds.
 		 */
-		FAIL_IF(getppid_tm(true) == -1); /* Should succeed. */
+		FAIL_IF(getppid_tm(false, true) == -1); /* Should succeed. */
 
 		/*
 		 * Test a syscall within an active transaction and verify that
 		 * it fails with the correct failure code.
 		 */
-		FAIL_IF(getppid_tm(false) != -1);  /* Should fail... */
+		FAIL_IF(getppid_tm(false, false) != -1);  /* Should fail... */
 		FAIL_IF(!failure_is_persistent()); /* ...persistently... */
 		FAIL_IF(!failure_is_syscall());    /* ...with code syscall. */
+
+		/* Now do it all again with scv if it is available. */
+		if (have_hwcap2(PPC_FEATURE2_SCV)) {
+			FAIL_IF(getppid_tm(true, true) == -1); /* Should succeed. */
+			FAIL_IF(getppid_tm(true, false) != -1);  /* Should fail... */
+			FAIL_IF(!failure_is_persistent()); /* ...persistently... */
+			FAIL_IF(!failure_is_syscall());    /* ...with code syscall. */
+		}
+
 		gettimeofday(&now, 0);
 	}
 
Signature · 1 line
2.23.0
#2 lore raw

Re: [PATCH v2 2/2] selftests/powerpc: Add scv versions of the basic TM syscall tests

Le 01/09/2021 à 18:54, Nicholas Piggin a écrit :

Quoted 18 lines

The basic TM vs syscall test code hard codes an sc instruction for the system call, which fails to cover scv even when the userspace libc has support for it.

Duplicate the tests with hard coded scv variants so both are tested when possible.

---
  .../selftests/powerpc/tm/tm-syscall-asm.S     | 46 +++++++++++++++++++
  .../testing/selftests/powerpc/tm/tm-syscall.c | 36 ++++++++++++---
  2 files changed, 75 insertions(+), 7 deletions(-)
tools/testing/selftests/powerpc/tm/tm-syscall-asm.S +0 -0
index bd1ca25febe4..849316831e6a 100644

tools/testing/selftests/powerpc/tm/tm-syscall-asm.S line 8

Quoted 7 lines
tools/testing/selftests/powerpc/tm/tm-syscall-asm.S +4 -0
@@ -2,6 +2,10 @@
  #include <ppc-asm.h>
  #include <asm/unistd.h>
 
+/* ppc-asm.h does not define r0 or r1 */
+#define r0 0
+#define r1 1
+

See https://github.com/gcc-mirror/gcc/blob/master/gcc/config/rs6000/ppc-asm.h

It doesn't not define r1 but it defines r0.

And it defines 'sp' as register 1.

Quoted 124 lines
tools/testing/selftests/powerpc/tm/tm-syscall-asm.S +42 -0
  	.text
  FUNC_START(getppid_tm_active)
  	tbegin.
@@ -26,3 +30,45 @@ FUNC_START(getppid_tm_suspended)
  1:
  	li	r3, -1
  	blr
+
+FUNC_START(getppid_scv_tm_active)
+	mflr	r0
+	std	r0,16(r1)
+	stdu	r1,-32(r1)
+	tbegin.
+	beq 1f
+	li	r0, __NR_getppid
+	scv	0
+	tend.
+	addi	r1,r1,32
+	ld	r0,16(r1)
+	mtlr	r0
+	blr
+1:
+	li	r3, -1
+	addi	r1,r1,32
+	ld	r0,16(r1)
+	mtlr	r0
+	blr
+
+FUNC_START(getppid_scv_tm_suspended)
+	mflr	r0
+	std	r0,16(r1)
+	stdu	r1,-32(r1)
+	tbegin.
+	beq 1f
+	li	r0, __NR_getppid
+	tsuspend.
+	scv	0
+	tresume.
+	tend.
+	addi	r1,r1,32
+	ld	r0,16(r1)
+	mtlr	r0
+	blr
+1:
+	li	r3, -1
+	addi	r1,r1,32
+	ld	r0,16(r1)
+	mtlr	r0
+	blr
tools/testing/selftests/powerpc/tm/tm-syscall.c +29 -7
index becb8207b432..9a822208680e 100644
@@ -19,24 +19,37 @@
  #include "utils.h"
  #include "tm.h"
 
+#ifndef PPC_FEATURE2_SCV
+#define PPC_FEATURE2_SCV               0x00100000 /* scv syscall */
+#endif
+
  extern int getppid_tm_active(void);
  extern int getppid_tm_suspended(void);
+extern int getppid_scv_tm_active(void);
+extern int getppid_scv_tm_suspended(void);
 
  unsigned retries = 0;
 
  #define TEST_DURATION 10 /* seconds */
  #define TM_RETRIES 100
 
-pid_t getppid_tm(bool suspend)
+pid_t getppid_tm(bool scv, bool suspend)
  {
  	int i;
  	pid_t pid;
 
  	for (i = 0; i < TM_RETRIES; i++) {
-		if (suspend)
-			pid = getppid_tm_suspended();
-		else
-			pid = getppid_tm_active();
+		if (suspend) {
+			if (scv)
+				pid = getppid_scv_tm_suspended();
+			else
+				pid = getppid_tm_suspended();
+		} else {
+			if (scv)
+				pid = getppid_scv_tm_active();
+			else
+				pid = getppid_tm_active();
+		}
 
  		if (pid >= 0)
  			return pid;
@@ -82,15 +95,24 @@ int tm_syscall(void)
  		 * Test a syscall within a suspended transaction and verify
  		 * that it succeeds.
  		 */
-		FAIL_IF(getppid_tm(true) == -1); /* Should succeed. */
+		FAIL_IF(getppid_tm(false, true) == -1); /* Should succeed. */
 
  		/*
  		 * Test a syscall within an active transaction and verify that
  		 * it fails with the correct failure code.
  		 */
-		FAIL_IF(getppid_tm(false) != -1);  /* Should fail... */
+		FAIL_IF(getppid_tm(false, false) != -1);  /* Should fail... */
  		FAIL_IF(!failure_is_persistent()); /* ...persistently... */
  		FAIL_IF(!failure_is_syscall());    /* ...with code syscall. */
+
+		/* Now do it all again with scv if it is available. */
+		if (have_hwcap2(PPC_FEATURE2_SCV)) {
+			FAIL_IF(getppid_tm(true, true) == -1); /* Should succeed. */
+			FAIL_IF(getppid_tm(true, false) != -1);  /* Should fail... */
+			FAIL_IF(!failure_is_persistent()); /* ...persistently... */
+			FAIL_IF(!failure_is_syscall());    /* ...with code syscall. */
+		}
+
  		gettimeofday(&now, 0);
  	}
 
author reviewer
#3 lore raw

Re: [PATCH v2 2/2] selftests/powerpc: Add scv versions of the basic TM syscall tests

Excerpts from Christophe Leroy's message of September 2, 2021 3:15 am:

Quoted 24 lines

Le 01/09/2021 à 18:54, Nicholas Piggin a écrit :

Quoted 21 lines

The basic TM vs syscall test code hard codes an sc instruction for the system call, which fails to cover scv even when the userspace libc has support for it.

Duplicate the tests with hard coded scv variants so both are tested when possible.

---
  .../selftests/powerpc/tm/tm-syscall-asm.S     | 46 +++++++++++++++++++
  .../testing/selftests/powerpc/tm/tm-syscall.c | 36 ++++++++++++---
  2 files changed, 75 insertions(+), 7 deletions(-)
tools/testing/selftests/powerpc/tm/tm-syscall-asm.S +0 -0
index bd1ca25febe4..849316831e6a 100644
@@ -2,6 +2,10 @@
  #include <ppc-asm.h>
  #include <asm/unistd.h>
 
Quoted 8 lines

tools/testing/selftests/powerpc/tm/tm-syscall-asm.S line 8

Quoted 4 lines
tools/testing/selftests/powerpc/tm/tm-syscall-asm.S +4 -0
+/* ppc-asm.h does not define r0 or r1 */
+#define r0 0
+#define r1 1
+

See https://github.com/gcc-mirror/gcc/blob/master/gcc/config/rs6000/ppc-asm.h

It doesn't not define r1 but it defines r0.

Oops, I'll fix that.

Quoted 1 line

And it defines 'sp' as register 1.

Does userspace code typically use that? Kernel code AFAIKS does not.

Thanks,
Nick

#4 lore raw

Re: [PATCH v2 2/2] selftests/powerpc: Add scv versions of the basic TM syscall tests

Quoted 30 lines · Nicholas Piggin

Excerpts from Christophe Leroy's message of September 2, 2021 3:15 am:

Quoted 29 lines

Le 01/09/2021 à 18:54, Nicholas Piggin a écrit :

Quoted 18 lines

The basic TM vs syscall test code hard codes an sc instruction for the system call, which fails to cover scv even when the userspace libc has support for it.

Duplicate the tests with hard coded scv variants so both are tested when possible.

---
  .../selftests/powerpc/tm/tm-syscall-asm.S     | 46 +++++++++++++++++++
  .../testing/selftests/powerpc/tm/tm-syscall.c | 36 ++++++++++++---
  2 files changed, 75 insertions(+), 7 deletions(-)
tools/testing/selftests/powerpc/tm/tm-syscall-asm.S +0 -0
index bd1ca25febe4..849316831e6a 100644

tools/testing/selftests/powerpc/tm/tm-syscall-asm.S line 8

Quoted 7 lines
tools/testing/selftests/powerpc/tm/tm-syscall-asm.S +4 -0
@@ -2,6 +2,10 @@
  #include <ppc-asm.h>
  #include <asm/unistd.h>
 
+/* ppc-asm.h does not define r0 or r1 */
+#define r0 0
+#define r1 1
+

See https://github.com/gcc-mirror/gcc/blob/master/gcc/config/rs6000/ppc-asm.h

Quoted 7 lines
Quoted 1 line

It doesn't not define r1 but it defines r0.

Oops, I'll fix that.

Quoted 1 line

And it defines 'sp' as register 1.

Does userspace code typically use that? Kernel code AFAIKS does not.

Some does, but it's not used consistently IME.

I'd prefer you just use %r1.

cheers

#5 lore raw

Re: [PATCH v2 2/2] selftests/powerpc: Add scv versions of the basic TM syscall tests

In reply to Nicholas Piggin
Quoted 42 lines · Nicholas Piggin

The basic TM vs syscall test code hard codes an sc instruction for the system call, which fails to cover scv even when the userspace libc has support for it.

Duplicate the tests with hard coded scv variants so both are tested when possible.

---
 .../selftests/powerpc/tm/tm-syscall-asm.S     | 46 +++++++++++++++++++
 .../testing/selftests/powerpc/tm/tm-syscall.c | 36 ++++++++++++---
 2 files changed, 75 insertions(+), 7 deletions(-)
tools/testing/selftests/powerpc/tm/tm-syscall-asm.S +14 -0
index bd1ca25febe4..849316831e6a 100644
@@ -2,6 +2,10 @@
 #include <ppc-asm.h>
 #include <asm/unistd.h>
 
+/* ppc-asm.h does not define r0 or r1 */
+#define r0 0
+#define r1 1
+
 	.text
 FUNC_START(getppid_tm_active)
 	tbegin.
@@ -26,3 +30,45 @@ FUNC_START(getppid_tm_suspended)
 1:
 	li	r3, -1
 	blr
+
+FUNC_START(getppid_scv_tm_active)
+	mflr	r0
+	std	r0,16(r1)
+	stdu	r1,-32(r1)
+	tbegin.
+	beq 1f
+	li	r0, __NR_getppid
+	scv	0
+	tend.

tools/testing/selftests/powerpc/tm/tm-syscall-asm.S FUNC_START(getppid_tm_suspended) · line 52

Quoted 10 lines
tools/testing/selftests/powerpc/tm/tm-syscall-asm.S +10 -0
+	addi	r1,r1,32
+	ld	r0,16(r1)
+	mtlr	r0
+	blr
+1:
+	li	r3, -1
+	addi	r1,r1,32
+	ld	r0,16(r1)
+	mtlr	r0
+	blr

There's some macros in tools/testing/selftests/powerpc/include/basic_asm.h that can do some of this boiler plate stack setup/teardown.

Incremental diff below to use them, only build tested.

cheers

tools/testing/selftests/powerpc/tm/tm-syscall-asm.S +7 -24
index 849316831e6a..a73694daca71 100644
@@ -1,10 +1,5 @@
 /* SPDX-License-Identifier: GPL-2.0 */
-#include <ppc-asm.h>
-#include <asm/unistd.h>
-
-/* ppc-asm.h does not define r0 or r1 */
-#define r0 0
-#define r1 1
+#include <basic_asm.h>
 
 	.text
 FUNC_START(getppid_tm_active)
@@ -32,29 +27,21 @@ FUNC_START(getppid_tm_suspended)
 	blr
 
 FUNC_START(getppid_scv_tm_active)
-	mflr	r0
-	std	r0,16(r1)
-	stdu	r1,-32(r1)
+	PUSH_BASIC_STACK(0)
 	tbegin.
 	beq 1f
 	li	r0, __NR_getppid
 	scv	0
 	tend.
-	addi	r1,r1,32
-	ld	r0,16(r1)
-	mtlr	r0
+	POP_BASIC_STACK(0)
 	blr
 1:
 	li	r3, -1
-	addi	r1,r1,32
-	ld	r0,16(r1)
-	mtlr	r0
+	POP_BASIC_STACK(0)
 	blr
 
 FUNC_START(getppid_scv_tm_suspended)
-	mflr	r0
-	std	r0,16(r1)
-	stdu	r1,-32(r1)
+	PUSH_BASIC_STACK(0)
 	tbegin.
 	beq 1f
 	li	r0, __NR_getppid
@@ -62,13 +49,9 @@ FUNC_START(getppid_scv_tm_suspended)
 	scv	0
 	tresume.
 	tend.
-	addi	r1,r1,32
-	ld	r0,16(r1)
-	mtlr	r0
+	POP_BASIC_STACK(0)
 	blr
 1:
 	li	r3, -1
-	addi	r1,r1,32
-	ld	r0,16(r1)
-	mtlr	r0
+	POP_BASIC_STACK(0)
 	blr
#6 lore raw
In reply to Nicholas Piggin

Le 01/09/2021 à 18:54, Nicholas Piggin a écrit :

Quoted 27 lines

If a system call is made with a transaction active, the kernel immediately aborts it and returns. scv system calls disable irqs even earlier in their interrupt handler, and tabort_syscall does not fix this up.

This can result in irq soft-mask state being messed up on the next kernel entry, and crashing at BUG_ON(arch_irq_disabled_regs(regs)) in the kernel exit handlers, or possibly worse.

Fix this by having tabort_syscall setting irq soft-mask back to enabled (which requires MSR[EE] be disabled first).

Fixes 7fa95f9adaee7 ("powerpc/64s: system call support for scv/rfscv instructions")

---

Tested the wrong kernel before sending v1 and missed a bug, sorry.

  arch/powerpc/kernel/interrupt_64.S | 8 +++++++-
  1 file changed, 7 insertions(+), 1 deletion(-)
arch/powerpc/kernel/interrupt_64.S +0 -0
index d4212d2ff0b5..9c31d65b4851 100644

arch/powerpc/kernel/interrupt_64.S RESTART_TABLE(.Lsyscall_rst_start, .Lsyscall_rst_end, syscall_restart) · line 436

Quoted 10 lines
arch/powerpc/kernel/interrupt_64.S +3 -1
@@ -428,16 +428,22 @@ RESTART_TABLE(.Lsyscall_rst_start, .Lsyscall_rst_end, syscall_restart)
  #ifdef CONFIG_PPC_TRANSACTIONAL_MEM
  tabort_syscall:
  _ASM_NOKPROBE_SYMBOL(tabort_syscall)
-	/* Firstly we need to enable TM in the kernel */
+	/* We need to enable TM in the kernel, and disable EE (for scv) */
  	mfmsr	r10
  	li	r9, 1
  	rldimi	r10, r9, MSR_TM_LG, 63-MSR_TM_LG
+	LOAD_REG_IMMEDIATE(r9, MSR_EE)
+	andc	r10, r10, r9

Why not use 'rlwinm' to mask out MSR_EE ?

Something like

	rlwinm	r10, r10, 0, ~MSR_EE
Quoted 14 lines
arch/powerpc/kernel/interrupt_64.S +4 -0
  	mtmsrd	r10, 0
 
  	/* tabort, this dooms the transaction, nothing else */
  	li	r9, (TM_CAUSE_SYSCALL|TM_CAUSE_PERSISTENT)
  	TABORT(R9)
 
+	/* scv has disabled irqs so must re-enable. sc just remains enabled */
+	li	r9,IRQS_ENABLED
+	stb	r9,PACAIRQSOFTMASK(r13)
+
  	/*
  	 * Return directly to userspace. We have corrupted user register state,
  	 * but userspace will never see that register state. Execution will
author reviewer
#7 lore raw

Excerpts from Christophe Leroy's message of September 2, 2021 3:21 am:

Quoted 43 lines

Le 01/09/2021 à 18:54, Nicholas Piggin a écrit :

Quoted 27 lines

If a system call is made with a transaction active, the kernel immediately aborts it and returns. scv system calls disable irqs even earlier in their interrupt handler, and tabort_syscall does not fix this up.

This can result in irq soft-mask state being messed up on the next kernel entry, and crashing at BUG_ON(arch_irq_disabled_regs(regs)) in the kernel exit handlers, or possibly worse.

Fix this by having tabort_syscall setting irq soft-mask back to enabled (which requires MSR[EE] be disabled first).

Fixes 7fa95f9adaee7 ("powerpc/64s: system call support for scv/rfscv instructions")

---

Tested the wrong kernel before sending v1 and missed a bug, sorry.

  arch/powerpc/kernel/interrupt_64.S | 8 +++++++-
  1 file changed, 7 insertions(+), 1 deletion(-)
arch/powerpc/kernel/interrupt_64.S +0 -0
index d4212d2ff0b5..9c31d65b4851 100644

arch/powerpc/kernel/interrupt_64.S RESTART_TABLE(.Lsyscall_rst_start, .Lsyscall_rst_end, syscall_restart) · line 436

Quoted 10 lines
arch/powerpc/kernel/interrupt_64.S +3 -1
@@ -428,16 +428,22 @@ RESTART_TABLE(.Lsyscall_rst_start, .Lsyscall_rst_end, syscall_restart)
  #ifdef CONFIG_PPC_TRANSACTIONAL_MEM
  tabort_syscall:
  _ASM_NOKPROBE_SYMBOL(tabort_syscall)
-	/* Firstly we need to enable TM in the kernel */
+	/* We need to enable TM in the kernel, and disable EE (for scv) */
  	mfmsr	r10
  	li	r9, 1
  	rldimi	r10, r9, MSR_TM_LG, 63-MSR_TM_LG
+	LOAD_REG_IMMEDIATE(r9, MSR_EE)
+	andc	r10, r10, r9

Why not use 'rlwinm' to mask out MSR_EE ?

Quoted 3 lines

Something like

	rlwinm	r10, r10, 0, ~MSR_EE

Mainly because I'm bad at powerpc assembly. Why do you think I'm trying to change as much as possible to C?

Actually there should really be no need for mfmsr either, I wanted to rewrite the thing entirely as

	ld      r10,PACAKMSR(r13)
	LOAD_REG_IMMEDIATE(r9, MSR_TM)
	or	r10,r10,r9
	mtmsrd	r10

But I thought that's not a minimal bug fix.

Thanks,
Nick

Quoted 16 lines
Quoted 15 lines
arch/powerpc/kernel/interrupt_64.S +4 -0
  	mtmsrd	r10, 0
 
  	/* tabort, this dooms the transaction, nothing else */
  	li	r9, (TM_CAUSE_SYSCALL|TM_CAUSE_PERSISTENT)
  	TABORT(R9)
 
+	/* scv has disabled irqs so must re-enable. sc just remains enabled */
+	li	r9,IRQS_ENABLED
+	stb	r9,PACAIRQSOFTMASK(r13)
+
  	/*
  	 * Return directly to userspace. We have corrupted user register state,
  	 * but userspace will never see that register state. Execution will
#8 lore raw
Quoted 11 lines · Nicholas Piggin

Excerpts from Christophe Leroy's message of September 2, 2021 3:21 am:

Quoted 10 lines

arch/powerpc/kernel/interrupt_64.S RESTART_TABLE(.Lsyscall_rst_start, .Lsyscall_rst_end, syscall_restart) · line 436

Quoted 7 lines
arch/powerpc/kernel/interrupt_64.S +3 -1
-	/* Firstly we need to enable TM in the kernel */
+	/* We need to enable TM in the kernel, and disable EE (for scv) */
  	mfmsr	r10
  	li	r9, 1
  	rldimi	r10, r9, MSR_TM_LG, 63-MSR_TM_LG
+	LOAD_REG_IMMEDIATE(r9, MSR_EE)
+	andc	r10, r10, r9

Why not use 'rlwinm' to mask out MSR_EE ?

Quoted 6 lines
Quoted 3 lines

Something like

	rlwinm	r10, r10, 0, ~MSR_EE

Mainly because I'm bad at powerpc assembly. Why do you think I'm trying to change as much as possible to C?

The actual bit (bit 31, i.e. with value 1UL << 32) cannot be cleared with rlwinm (only the low 32 bits can). There are many ways to do it using two insns of course.

Quoted 9 lines

Actually there should really be no need for mfmsr either, I wanted to rewrite the thing entirely as

	ld      r10,PACAKMSR(r13)
	LOAD_REG_IMMEDIATE(r9, MSR_TM)
	or	r10,r10,r9
	mtmsrd	r10

But I thought that's not a minimal bug fix.

That (LOAD_REG_IMMEDIATE+or) can be done with just two insns, not three as written:
li r0,1
rldimi r10,r0,32,31
(you can write that last insns as
rldimi r10,r0,MSR_TM_LG,MSR_TM_LG-1
if you insist :-) )

It isn't like a few integer computational insns will kill you here of course, there are much more important cycles to shave off :-)

Segher

#9 lore raw
Quoted 14 lines · Segher Boessenkool
Quoted 13 lines · Nicholas Piggin

Excerpts from Christophe Leroy's message of September 2, 2021 3:21 am:

Quoted 12 lines

arch/powerpc/kernel/interrupt_64.S RESTART_TABLE(.Lsyscall_rst_start, .Lsyscall_rst_end, syscall_restart) · line 436

Quoted 7 lines
arch/powerpc/kernel/interrupt_64.S +3 -1
-	/* Firstly we need to enable TM in the kernel */
+	/* We need to enable TM in the kernel, and disable EE (for scv) */
  	mfmsr	r10
  	li	r9, 1
  	rldimi	r10, r9, MSR_TM_LG, 63-MSR_TM_LG
+	LOAD_REG_IMMEDIATE(r9, MSR_EE)
+	andc	r10, r10, r9

Why not use 'rlwinm' to mask out MSR_EE ?

Something like

Quoted 8 lines
Quoted 4 lines
Quoted 1 line
	rlwinm	r10, r10, 0, ~MSR_EE

Mainly because I'm bad at powerpc assembly. Why do you think I'm trying to change as much as possible to C?

The actual bit (bit 31, i.e. with value 1UL << 32) cannot be cleared with rlwinm (only the low 32 bits can). There are many ways to do it using two insns of course.

Wow I misread that, you want to clear MSR[EE] really, not MSR[TM].

You cannot use rlwinm and keep the high 32 bits of the target register intact. You either clear all to 0 or set them to a copy of the rotated value in the low 32 bits.

Segher

#10 lore raw

Le 03/09/2021 à 00:20, Segher Boessenkool a écrit :

Quoted 24 lines · Segher Boessenkool
Quoted 14 lines · Nicholas Piggin
Quoted 13 lines

Excerpts from Christophe Leroy's message of September 2, 2021 3:21 am:

Quoted 12 lines

arch/powerpc/kernel/interrupt_64.S RESTART_TABLE(.Lsyscall_rst_start, .Lsyscall_rst_end, syscall_restart) · line 436

Quoted 7 lines
arch/powerpc/kernel/interrupt_64.S +3 -1
-	/* Firstly we need to enable TM in the kernel */
+	/* We need to enable TM in the kernel, and disable EE (for scv) */
   	mfmsr	r10
   	li	r9, 1
   	rldimi	r10, r9, MSR_TM_LG, 63-MSR_TM_LG
+	LOAD_REG_IMMEDIATE(r9, MSR_EE)
+	andc	r10, r10, r9

Why not use 'rlwinm' to mask out MSR_EE ?

Something like

Quoted 8 lines
Quoted 4 lines
Quoted 1 line
	rlwinm	r10, r10, 0, ~MSR_EE

Mainly because I'm bad at powerpc assembly. Why do you think I'm trying to change as much as possible to C?

The actual bit (bit 31, i.e. with value 1UL << 32) cannot be cleared with rlwinm (only the low 32 bits can). There are many ways to do it using two insns of course.

Quoted 6 lines

Wow I misread that, you want to clear MSR[EE] really, not MSR[TM].

You cannot use rlwinm and keep the high 32 bits of the target register intact. You either clear all to 0 or set them to a copy of the rotated value in the low 32 bits.

Oops, my mistake. When I tested it in C to see what was generated by GCC I forgot the ~ so I got rlwinm r3,r3,0,16,16 and didn't realise it was different from rlwinm r3,r3,0,~(1<<15)

By the way it would be more explicit if objdump could display the mask instead of the mask boundaries. Is there a way to do that ?

Christophe