[PATCH v2 0/3] powerpc/tm: MSR oops printing and selftests

[PATCH v2 0/3] powerpc/tm: MSR oops printing and selftests

[Michael Neuling]

Here are 3 non-urgent TM patches.  These were left off my other post
of the urgent fixes.

v2:
  - Added invalid stack with signal delievery during tm selftest
  - Added selftest binaries to .gitignore
  - Improved signal return selftest based on mpe's feedback
  - Improved oops message printing based on mpe's feedback
  - Dropped get_tm_stackpointer() renaming patch in leui of more
     extensive rewrite later

Michael Neuling (3):
  powerpc: Print MSR TM bits in oops message
  selftests/powerpc: Add TM signal return test
  selftests/powerpc: Add TM signal with invalid stack test

 arch/powerpc/kernel/process.c                      | 51 ++++++++++++---
 tools/testing/selftests/powerpc/tm/.gitignore      |  2 +
 tools/testing/selftests/powerpc/tm/Makefile        |  2 +-
 .../selftests/powerpc/tm/tm-signal-msr-resv.c      | 71 +++++++++++++++++++++
 .../testing/selftests/powerpc/tm/tm-signal-stack.c | 73 ++++++++++++++++++++++
 5 files changed, 190 insertions(+), 9 deletions(-)
 create mode 100644 tools/testing/selftests/powerpc/tm/tm-signal-msr-resv.c
 create mode 100644 tools/testing/selftests/powerpc/tm/tm-signal-stack.c

-- 
2.5.0

[PATCH v2 1/3] powerpc: Print MSR TM bits in oops messages

[Michael Neuling]

Print MSR TM bits in oops messages.  This appends them to the end
like this:

    MSR: 8000000502823031 <SF,VEC,VSX,FP,ME,IR,DR,LE,TM[TE]>

You get the TM[] only if at least one TM MSR bit is set.  Inside the
TM[], E means Enabled (bit 32), S means Suspended (bit 33), and T
means Transactional (bit 34)

If no bits are set, you get no TM[] output.

Include rework of printbits() to handle this case.

Signed-off-by: Michael Neuling <mikey@neuling.org>
---
 arch/powerpc/kernel/process.c | 51 ++++++++++++++++++++++++++++++++++++-------
 1 file changed, 43 insertions(+), 8 deletions(-)

diff --git a/arch/powerpc/kernel/process.c b/arch/powerpc/kernel/process.c
index 646bf4d..f086e8b 100644
--- a/arch/powerpc/kernel/process.c
+++ b/arch/powerpc/kernel/process.c
@@ -952,10 +952,12 @@ static void show_instructions(struct pt_regs *regs)
 	printk("\n");
 }
 
-static struct regbit {
+struct regbit {
 	unsigned long bit;
 	const char *name;
-} msr_bits[] = {
+};
+
+static struct regbit msr_bits[] = {
 #if defined(CONFIG_PPC64) && !defined(CONFIG_BOOKE)
 	{MSR_SF,	"SF"},
 	{MSR_HV,	"HV"},
@@ -985,16 +987,49 @@ static struct regbit {
 	{0,		NULL}
 };
 
-static void printbits(unsigned long val, struct regbit *bits)
+static void print_bits(unsigned long val, struct regbit *bits, const char *sep)
 {
-	const char *sep = "";
+	const char *s = "";
 
-	printk("<");
 	for (; bits->bit; ++bits)
 		if (val & bits->bit) {
-			printk("%s%s", sep, bits->name);
-			sep = ",";
+			printk("%s%s", s, bits->name);
+			s = sep;
 		}
+}
+
+#ifdef CONFIG_PPC_TRANSACTIONAL_MEM
+static struct regbit msr_tm_bits[] = {
+	{MSR_TS_T,	"T"},
+	{MSR_TS_S,	"S"},
+	{MSR_TM,	"E"},
+	{0,		NULL}
+};
+
+static void print_tm_bits(unsigned long val)
+{
+/*
+ * This only prints something if at least one of the TM bit is set.
+ * Inside the TM[], the output means:
+ *   E: Enabled		(bit 32)
+ *   S: Suspended	(bit 33)
+ *   T: Transactional	(bit 34)
+ */
+	if (val & (MSR_TM | MSR_TS_S | MSR_TS_T)) {
+		printk(",TM[");
+		print_bits(val, msr_tm_bits, "");
+		printk("]");
+	}
+}
+#else
+static void print_tm_bits(unsigned long val) {}
+#endif
+
+static void print_msr_bits(unsigned long val)
+{
+	printk("<");
+	print_bits(val, msr_bits, ",");
+	print_tm_bits(val);
 	printk(">");
 }
 
@@ -1019,7 +1054,7 @@ void show_regs(struct pt_regs * regs)
 	printk("REGS: %p TRAP: %04lx   %s  (%s)\n",
 	       regs, regs->trap, print_tainted(), init_utsname()->release);
 	printk("MSR: "REG" ", regs->msr);
-	printbits(regs->msr, msr_bits);
+	print_msr_bits(regs->msr);
 	printk("  CR: %08lx  XER: %08lx\n", regs->ccr, regs->xer);
 	trap = TRAP(regs);
 	if ((regs->trap != 0xc00) && cpu_has_feature(CPU_FTR_CFAR))
-- 
2.5.0

[PATCH v2 2/3] selftests/powerpc: Add TM signal return test

[Michael Neuling]

Test the kernel's signal return code to ensure that it doesn't crash
when both the transactional and suspend MSR bits are set in the signal
context.

Signed-off-by: Michael Neuling <mikey@neuling.org>
---
 tools/testing/selftests/powerpc/tm/.gitignore      |  1 +
 tools/testing/selftests/powerpc/tm/Makefile        |  2 +-
 .../selftests/powerpc/tm/tm-signal-msr-resv.c      | 71 ++++++++++++++++++++++
 3 files changed, 73 insertions(+), 1 deletion(-)
 create mode 100644 tools/testing/selftests/powerpc/tm/tm-signal-msr-resv.c

diff --git a/tools/testing/selftests/powerpc/tm/.gitignore b/tools/testing/selftests/powerpc/tm/.gitignore
index 2699635d..61c318f 100644
--- a/tools/testing/selftests/powerpc/tm/.gitignore
+++ b/tools/testing/selftests/powerpc/tm/.gitignore
@@ -1,2 +1,3 @@
 tm-resched-dscr
 tm-syscall
+tm-signal-msr-resv
diff --git a/tools/testing/selftests/powerpc/tm/Makefile b/tools/testing/selftests/powerpc/tm/Makefile
index 4bea62a..0c28db7 100644
--- a/tools/testing/selftests/powerpc/tm/Makefile
+++ b/tools/testing/selftests/powerpc/tm/Makefile
@@ -1,4 +1,4 @@
-TEST_PROGS := tm-resched-dscr tm-syscall
+TEST_PROGS := tm-resched-dscr tm-syscall tm-signal-msr-resv
 
 all: $(TEST_PROGS)
 
diff --git a/tools/testing/selftests/powerpc/tm/tm-signal-msr-resv.c b/tools/testing/selftests/powerpc/tm/tm-signal-msr-resv.c
new file mode 100644
index 0000000..bdad348
--- /dev/null
+++ b/tools/testing/selftests/powerpc/tm/tm-signal-msr-resv.c
@@ -0,0 +1,71 @@
+/*
+ * Copyright 2015, Michael Neuling, IBM Corp.
+ * Licensed under GPLv2.
+ *
+ * Test the kernel's signal return code to ensure that it doesn't
+ * crash when both the transactional and suspend MSR bits are set in
+ * the signal context.
+ *
+ * For this test, we send ourselves a SIGUSR1.  In the SIGUSR1 handler
+ * we modify the signal context to set both MSR TM S and T bits (which
+ * is "reserved" by the PowerISA). When we return from the signal
+ * handler (implicit sigreturn), the kernel should detect reserved MSR
+ * value and send us with a SIGSEGV.
+ */
+
+#include <stdlib.h>
+#include <stdio.h>
+#include <signal.h>
+#include <unistd.h>
+
+#include "utils.h"
+
+int segv_expected = 0;
+
+void signal_segv(int signum)
+{
+	if (segv_expected && (signum == SIGSEGV))
+		_exit(0);
+	_exit(1);
+}
+
+void signal_usr1(int signum, siginfo_t *info, void *uc)
+{
+	ucontext_t *ucp = uc;
+
+	/* Link tm checkpointed context to normal context */
+	ucp->uc_link = ucp;
+	/* Set all TM bits so that the context is now invalid */
+#ifdef __powerpc64__
+	ucp->uc_mcontext.gp_regs[PT_MSR] |= (7ULL << 32);
+#else
+	ucp->uc_mcontext.regs->gpr[PT_MSR] |= (7ULL);
+#endif
+	/* Should segv on return becuase of invalid context */
+	segv_expected = 1;
+}
+
+int tm_signal_msr_resv()
+{
+	struct sigaction act;
+
+	act.sa_sigaction = signal_usr1;
+	sigemptyset(&act.sa_mask);
+	act.sa_flags = SA_SIGINFO;
+	if (sigaction(SIGUSR1, &act, NULL) < 0) {
+		perror("sigaction sigusr1");
+		exit(1);
+	}
+	if (signal(SIGSEGV, signal_segv) == SIG_ERR)
+		exit(1);
+
+	raise(SIGUSR1);
+
+	/* We shouldn't get here as we exit in the segv handler */
+	return 1;
+}
+
+int main(void)
+{
+	return test_harness(tm_signal_msr_resv, "tm_signal_msr_resv");
+}
-- 
2.5.0

[PATCH v2 3/3] selftests/powerpc: Add TM signal with invalid stack test

[Michael Neuling]

Test the kernels signal generation code to ensure it can handle an
invalid stack pointer when transactional.

Signed-off-by: Michael Neuling <mikey@neuling.org>
---
 tools/testing/selftests/powerpc/tm/.gitignore      |  1 +
 tools/testing/selftests/powerpc/tm/Makefile        |  2 +-
 .../testing/selftests/powerpc/tm/tm-signal-stack.c | 73 ++++++++++++++++++++++
 3 files changed, 75 insertions(+), 1 deletion(-)
 create mode 100644 tools/testing/selftests/powerpc/tm/tm-signal-stack.c

diff --git a/tools/testing/selftests/powerpc/tm/.gitignore b/tools/testing/selftests/powerpc/tm/.gitignore
index 61c318f..e666821 100644
--- a/tools/testing/selftests/powerpc/tm/.gitignore
+++ b/tools/testing/selftests/powerpc/tm/.gitignore
@@ -1,3 +1,4 @@
 tm-resched-dscr
 tm-syscall
 tm-signal-msr-resv
+tm-signal-stack
diff --git a/tools/testing/selftests/powerpc/tm/Makefile b/tools/testing/selftests/powerpc/tm/Makefile
index 0c28db7..0e45c7e 100644
--- a/tools/testing/selftests/powerpc/tm/Makefile
+++ b/tools/testing/selftests/powerpc/tm/Makefile
@@ -1,4 +1,4 @@
-TEST_PROGS := tm-resched-dscr tm-syscall tm-signal-msr-resv
+TEST_PROGS := tm-resched-dscr tm-syscall tm-signal-msr-resv tm-signal-stack
 
 all: $(TEST_PROGS)
 
diff --git a/tools/testing/selftests/powerpc/tm/tm-signal-stack.c b/tools/testing/selftests/powerpc/tm/tm-signal-stack.c
new file mode 100644
index 0000000..62f2a5f
--- /dev/null
+++ b/tools/testing/selftests/powerpc/tm/tm-signal-stack.c
@@ -0,0 +1,73 @@
+/*
+ * Copyright 2015, Michael Neuling, IBM Corp.
+ * Licensed under GPLv2.
+ *
+ * Test the kernel's signal delievery code to ensure that we don't
+ * trelaim twice in the kernel signal delivery code.  This can happen
+ * if we trigger a signal when in a transaction and the stack pointer
+ * is bogus.
+ *
+ * This test case registers a SEGV handler, sets the stack pointer
+ * (r1) to NULL, starts a transaction and then generates a SEGV.  The
+ * SEGV should be handled but we exit here as the stack pointer is
+ * invalid and hance we can't sigreturn.  We only need to check that
+ * this flow doesn't crash the kernel.
+ */
+
+#include <unistd.h>
+#include <sys/types.h>
+#include <sys/wait.h>
+#include <stdlib.h>
+#include <stdio.h>
+#include <signal.h>
+
+#include "utils.h"
+
+void signal_segv(int signum)
+{
+	/* This should never actually run since stack is foobar */
+	exit(1);
+}
+
+int tm_signal_stack()
+{
+	int pid;
+
+	pid = fork();
+	if (pid < 0)
+		exit(1);
+
+	if (pid) { /* Parent */
+		/*
+		 * It's likely the whole machine will crash here so if
+		 * the child ever exits, we are good.
+		 */
+		wait(NULL);
+		return 0;
+	}
+
+	/*
+	 * The flow here is:
+	 * 1) register a signal handler (so signal delievery occurs)
+	 * 2) make stack pointer (r1) = NULL
+	 * 3) start transaction
+	 * 4) cause segv
+	 */
+	if (signal(SIGSEGV, signal_segv) == SIG_ERR)
+		exit(1);
+	asm volatile("li 1, 0 ;"		/* stack ptr == NULL */
+		     "1:"
+		     ".long 0x7C00051D ;"	/* tbegin */
+		     "beq 1b ;"			/* retry forever */
+		     ".long 0x7C0005DD ; ;"	/* tsuspend */
+		     "ld 2, 0(1) ;"		/* trigger segv" */
+		     : : : "memory");
+
+	/* This should never get here due to above segv */
+	return 1;
+}
+
+int main(void)
+{
+	return test_harness(tm_signal_stack, "tm_signal_stack");
+}
-- 
2.5.0

Re: [PATCH v2 2/3] selftests/powerpc: Add TM signal return test

[Anshuman Khandual]

On 11/20/2015 09:45 AM, Michael Neuling wrote:
> Test the kernel's signal return code to ensure that it doesn't crash
> when both the transactional and suspend MSR bits are set in the signal
> context.
> 
> Signed-off-by: Michael Neuling <mikey@neuling.org>

Tested-by: Anshuman Khandual <khandual@linux.vnet.ibm.com>

Re: [PATCH v2 3/3] selftests/powerpc: Add TM signal with invalid stack test

[Anshuman Khandual]

On 11/20/2015 09:45 AM, Michael Neuling wrote:
> Test the kernels signal generation code to ensure it can handle an
> invalid stack pointer when transactional.
> 
> Signed-off-by: Michael Neuling <mikey@neuling.org>

Tested-by: Anshuman Khandual <khandual@linux.vnet.ibm.com>

Re: [v2,1/3] powerpc: Print MSR TM bits in oops messages

[Michael Ellerman]

On Fri, 2015-20-11 at 04:15:32 UTC, Michael Neuling wrote:
> Print MSR TM bits in oops messages.  This appends them to the end
> like this:
> 
>     MSR: 8000000502823031 <SF,VEC,VSX,FP,ME,IR,DR,LE,TM[TE]>
> 
> You get the TM[] only if at least one TM MSR bit is set.  Inside the
> TM[], E means Enabled (bit 32), S means Suspended (bit 33), and T
> means Transactional (bit 34)
> 
> If no bits are set, you get no TM[] output.
> 
> Include rework of printbits() to handle this case.
> 
> Signed-off-by: Michael Neuling <mikey@neuling.org>

Applied to powerpc next, thanks.

https://git.kernel.org/powerpc/c/801c0b2c4db3a33d56b3e192

cheers

Re: [v2,2/3] selftests/powerpc: Add TM signal return test

[Michael Ellerman]

On Fri, 2015-20-11 at 04:15:33 UTC, Michael Neuling wrote:
> Test the kernel's signal return code to ensure that it doesn't crash
> when both the transactional and suspend MSR bits are set in the signal
> context.
> 
> Signed-off-by: Michael Neuling <mikey@neuling.org>
> Tested-by: Anshuman Khandual <khandual@linux.vnet.ibm.com>

Applied to powerpc next, thanks.

https://git.kernel.org/powerpc/c/25007a69e852389985ee9823

cheers

Re: [v2, 3/3] selftests/powerpc: Add TM signal with invalid stack test

[Michael Ellerman]

On Fri, 2015-20-11 at 04:15:34 UTC, Michael Neuling wrote:
> Test the kernels signal generation code to ensure it can handle an
> invalid stack pointer when transactional.
> 
> Signed-off-by: Michael Neuling <mikey@neuling.org>
> Tested-by: Anshuman Khandual <khandual@linux.vnet.ibm.com>

Applied to powerpc next, thanks.

https://git.kernel.org/powerpc/c/a26f415bf71640f0141e5e94

cheers