# \[PATCH v4 0/2\] make sigaltstack() compatible with swapcontext()

- Subsystems: EXEC \& BINFMT API, ELF, KERNEL SELFTEST FRAMEWORK, MEMORY MANAGEMENT - CORE, SCHEDULER, THE REST
- Maintainers: Kees Cook, Shuah Khan, Shuah Khan, David Hildenbrand, Ingo Molnar, Peter Zijlstra, Juri Lelli, Vincent Guittot, Linus Torvalds
- Series: 2 patches touching 8 files, +187/-13
- Status: SUPERSEDED (Superseded by v6 posted 2016-03-07)
- Superseded by v6 (posted 2016-03-07).
- Revision v4 (6 revisions archived).
- List: linux-api
- 13 messages, 3 authors
- Page: https://ratatoskr.run/linux-api/2016/02/4697311/t

## \[PATCH v4 0/2\] make sigaltstack() compatible with swapcontext()

- From: Stas Sergeev \<hidden\>
- Date: 2016-02-29 21:30:00 
- Message: https://ratatoskr.run/linux-api/2016/02/4697311.md

```text
The following patches make it possible to use swapcontext()
in a sighandler that works on sigaltstack.
The approach is inspired by Andy Lutomirski's suggestion that
sigaltstack should disarm itself after saving into uc_stack:
https://lkml.org/lkml/2016/2/1/594

I add the SS_AUTODISARM flag that does exactly that.
On sighandler exit, the sigaltstack is restored from uc_stack.
Another possible name could be SS_ONESHOT, but, since it gets
always re-enabled, I choose SS_AUTODISARM.

[PATCH 1/2] sigaltstack: implement SS_AUTODISARM flag
	This patch implements SS_AUTODISARM flag
[PATCH 2/2] selftests: Add test for sigaltstack(SS_AUTODISARM)
	This patch adds the selftest code for new functionality

CC: <redacted>
CC: <redacted>
CC: Andy Lutomirski <redacted>
CC: Oleg Nesterov <redacted>
CC: Shuah Khan <redacted>

Diffstat:
 include/linux/sched.h                        |    8 +
 include/linux/signal.h                       |    4 
 include/uapi/linux/signal.h                  |    3 
 kernel/fork.c                                |    2 
 kernel/signal.c                              |   23 ++--
 tools/testing/selftests/Makefile             |    1 
 tools/testing/selftests/sigaltstack/Makefile |    8 +
 tools/testing/selftests/sigaltstack/sas.c    |  151 +++++++++++++++++++++++++++
 8 files changed, 187 insertions(+), 13 deletions(-)
```

## \[PATCH 1/2\] sigaltstack: implement SS\_AUTODISARM flag

- From: Stas Sergeev \<hidden\>
- Date: 2016-02-29 21:30:18 
- Message: https://ratatoskr.run/linux-api/2016/02/4697314.md

```text
This patch implements the SS_AUTODISARM flag that can be ORed with
SS_ONSTACK when forming ss_flags.
When this flag is set, sigaltstack will be disabled when entering
the signal handler; more precisely, after saving sas to uc_stack.
When leaving the signal handler, the sigaltstack is restored by
uc_stack.
When this flag is used, it is safe to switch from sighandler with
swapcontext(). Without this flag, the subsequent signal will corrupt
the state of the switched-away sighandler.

CC: Ingo Molnar <mingo@redhat.com>
CC: Peter Zijlstra <peterz@infradead.org>
CC: Richard Weinberger <richard@nod.at>
CC: Andrew Morton <akpm@linux-foundation.org>
CC: Oleg Nesterov <oleg@redhat.com>
CC: Tejun Heo <tj@kernel.org>
CC: Heinrich Schuchardt <redacted>
CC: Jason Low <redacted>
CC: Andrea Arcangeli <redacted>
CC: Frederic Weisbecker <redacted>
CC: Konstantin Khlebnikov <redacted>
CC: Josh Triplett <josh@joshtriplett.org>
CC: "Eric W. Biederman" <redacted>
CC: Aleksa Sarai <redacted>
CC: "Amanieu d'Antras" <redacted>
CC: Paul Moore <redacted>
CC: Sasha Levin <redacted>
CC: Palmer Dabbelt <palmer@dabbelt.com>
CC: Vladimir Davydov <redacted>
CC: <redacted>
CC: <redacted>
CC: Andy Lutomirski <luto@amacapital.net>

Signed-off-by: Stas Sergeev <redacted>
---
 include/linux/sched.h       |  8 ++++++++
 include/linux/signal.h      |  4 +++-
 include/uapi/linux/signal.h |  3 +++
 kernel/fork.c               |  2 +-
 kernel/signal.c             | 23 ++++++++++++-----------
 5 files changed, 27 insertions(+), 13 deletions(-)

diff --git a/include/linux/sched.h b/include/linux/sched.h
index a10494a..26201cd 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -1587,6 +1587,7 @@ struct task_struct {
 
 	unsigned long sas_ss_sp;
 	size_t sas_ss_size;
+	unsigned sas_ss_flags;
 
 	struct callback_head *task_works;
 
@@ -2573,6 +2574,13 @@ static inline int sas_ss_flags(unsigned long sp)
 	return on_sig_stack(sp) ? SS_ONSTACK : 0;
 }
 
+static inline void sas_ss_reset(struct task_struct *p)
+{
+	p->sas_ss_sp = 0;
+	p->sas_ss_size = 0;
+	p->sas_ss_flags = SS_DISABLE;
+}
+
 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
 {
 	if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
diff --git a/include/linux/signal.h b/include/linux/signal.h
index 92557bb..3fbe814 100644
--- a/include/linux/signal.h
+++ b/include/linux/signal.h
@@ -432,8 +432,10 @@ int __save_altstack(stack_t __user *, unsigned long);
 	stack_t __user *__uss = uss; \
 	struct task_struct *t = current; \
 	put_user_ex((void __user *)t->sas_ss_sp, &__uss->ss_sp); \
-	put_user_ex(sas_ss_flags(sp), &__uss->ss_flags); \
+	put_user_ex(t->sas_ss_flags, &__uss->ss_flags); \
 	put_user_ex(t->sas_ss_size, &__uss->ss_size); \
+	if (t->sas_ss_flags & SS_AUTODISARM) \
+		sas_ss_reset(t); \
 } while (0);
 
 #ifdef CONFIG_PROC_FS
diff --git a/include/uapi/linux/signal.h b/include/uapi/linux/signal.h
index e1bd50c2..4691bc5 100644
--- a/include/uapi/linux/signal.h
+++ b/include/uapi/linux/signal.h
@@ -6,5 +6,8 @@
 
 #define SS_ONSTACK	1
 #define SS_DISABLE	2
+#define SS_VALMASK	0xf
+/* bit-flags */
+#define SS_AUTODISARM	(1 << 4)	/* disable sas during sighandling */
 
 #endif /* _UAPI_LINUX_SIGNAL_H */
diff --git a/kernel/fork.c b/kernel/fork.c
index 2e391c7..68c8716 100644
--- a/kernel/fork.c
+++ b/kernel/fork.c
@@ -1483,7 +1483,7 @@ static struct task_struct *copy_process(unsigned long clone_flags,
 	 * sigaltstack should be cleared when sharing the same VM
 	 */
 	if ((clone_flags & (CLONE_VM|CLONE_VFORK)) == CLONE_VM)
-		p->sas_ss_sp = p->sas_ss_size = 0;
+		sas_ss_reset(p);
 
 	/*
 	 * Syscall tracing and stepping should be turned off in the
diff --git a/kernel/signal.c b/kernel/signal.c
index 0508544..fab6bea 100644
--- a/kernel/signal.c
+++ b/kernel/signal.c
@@ -3100,13 +3100,14 @@ do_sigaltstack (const stack_t __user *uss, stack_t __user *uoss, unsigned long s
 	if (uss) {
 		void __user *ss_sp;
 		size_t ss_size;
-		int ss_flags;
+		unsigned ss_flags;
+		unsigned ss_xflags;
 
 		error = -EFAULT;
 		if (!access_ok(VERIFY_READ, uss, sizeof(*uss)))
 			goto out;
 		error = __get_user(ss_sp, &uss->ss_sp) |
-			__get_user(ss_flags, &uss->ss_flags) |
+			__get_user(ss_xflags, &uss->ss_flags) |
 			__get_user(ss_size, &uss->ss_size);
 		if (error)
 			goto out;
@@ -3115,14 +3116,8 @@ do_sigaltstack (const stack_t __user *uss, stack_t __user *uoss, unsigned long s
 		if (on_sig_stack(sp))
 			goto out;
 
+		ss_flags = ss_xflags & SS_VALMASK;
 		error = -EINVAL;
-		/*
-		 * Note - this code used to test ss_flags incorrectly:
-		 *  	  old code may have been written using ss_flags==0
-		 *	  to mean ss_flags==SS_ONSTACK (as this was the only
-		 *	  way that worked) - this fix preserves that older
-		 *	  mechanism.
-		 */
 		if (ss_flags != SS_DISABLE && ss_flags != SS_ONSTACK && ss_flags != 0)
 			goto out;
 
@@ -3137,6 +3132,7 @@ do_sigaltstack (const stack_t __user *uss, stack_t __user *uoss, unsigned long s
 
 		current->sas_ss_sp = (unsigned long) ss_sp;
 		current->sas_ss_size = ss_size;
+		current->sas_ss_flags = ss_xflags;
 	}
 
 	error = 0;
@@ -3167,9 +3163,14 @@ int restore_altstack(const stack_t __user *uss)
 int __save_altstack(stack_t __user *uss, unsigned long sp)
 {
 	struct task_struct *t = current;
-	return  __put_user((void __user *)t->sas_ss_sp, &uss->ss_sp) |
-		__put_user(sas_ss_flags(sp), &uss->ss_flags) |
+	int err = __put_user((void __user *)t->sas_ss_sp, &uss->ss_sp) |
+		__put_user(t->sas_ss_flags, &uss->ss_flags) |
 		__put_user(t->sas_ss_size, &uss->ss_size);
+	if (err)
+		return err;
+	if (t->sas_ss_flags & SS_AUTODISARM)
+		sas_ss_reset(t);
+	return 0;
 }
 
 #ifdef CONFIG_COMPAT
-- 
2.7.2
```

## \[PATCH 2/2\] selftests: Add test for sigaltstack(SS\_AUTODISARM)

- From: Stas Sergeev \<hidden\>
- Date: 2016-02-29 21:30:20 
- Message: https://ratatoskr.run/linux-api/2016/02/4697312.md

```text
From: Stas Sergeev <redacted>

sigaltstack needs to be disabled before the signal handler can
safely use swapcontext().
This patch adds the SS_AUTODISARM flag.
This flag disables the sigaltstack when entering the signal handler.
When returning from signal handler, the sigaltstack is restored by
uc_stack.

CC: Shuah Khan <redacted>
CC: <redacted>
CC: <redacted>
CC: Andy Lutomirski <luto@amacapital.net>

Signed-off-by: Stas Sergeev <redacted>
---
 tools/testing/selftests/Makefile             |   1 +
 tools/testing/selftests/sigaltstack/Makefile |   8 ++
 tools/testing/selftests/sigaltstack/sas.c    | 151 +++++++++++++++++++++++++++
 3 files changed, 160 insertions(+)
 create mode 100644 tools/testing/selftests/sigaltstack/Makefile
 create mode 100644 tools/testing/selftests/sigaltstack/sas.c

diff --git a/tools/testing/selftests/Makefile b/tools/testing/selftests/Makefile
index b04afc3..ff9e5f2 100644
--- a/tools/testing/selftests/Makefile
+++ b/tools/testing/selftests/Makefile
@@ -19,6 +19,7 @@ TARGETS += powerpc
 TARGETS += pstore
 TARGETS += ptrace
 TARGETS += seccomp
+TARGETS += sigaltstack
 TARGETS += size
 TARGETS += static_keys
 TARGETS += sysctl
diff --git a/tools/testing/selftests/sigaltstack/Makefile b/tools/testing/selftests/sigaltstack/Makefile
new file mode 100644
index 0000000..56af56e
--- /dev/null
+++ b/tools/testing/selftests/sigaltstack/Makefile
@@ -0,0 +1,8 @@
+CFLAGS = -Wall
+BINARIES = sas
+all: $(BINARIES)
+
+include ../lib.mk
+
+clean:
+	rm -rf $(BINARIES)
diff --git a/tools/testing/selftests/sigaltstack/sas.c b/tools/testing/selftests/sigaltstack/sas.c
new file mode 100644
index 0000000..bba149a
--- /dev/null
+++ b/tools/testing/selftests/sigaltstack/sas.c
@@ -0,0 +1,151 @@
+/*
+ * Stas Sergeev <stsp@users.sourceforge.net>
+ *
+ * test sigaltstack(SS_AUTODISARM)
+ * If that succeeds, then swapcontext() can be used inside sighandler safely.
+ *
+ */
+
+#define _GNU_SOURCE
+#include <signal.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <sys/mman.h>
+#include <ucontext.h>
+#include <alloca.h>
+#include <string.h>
+#include <assert.h>
+
+#ifndef SS_AUTODISARM
+#define SS_AUTODISARM  (1 << 4)
+#endif
+
+static void *sstack, *ustack;
+static ucontext_t uc, sc;
+static const char *msg = "[OK]\tStack preserved";
+static const char *msg2 = "[FAIL]\tStack corrupted";
+struct stk_data {
+	char msg[128];
+	int flag;
+};
+
+void my_usr1(int sig, siginfo_t *si, void *u)
+{
+	char *aa;
+	int err;
+	stack_t stk;
+	struct stk_data *p;
+
+	register unsigned long sp asm("sp");
+
+	if (sp < (unsigned long)sstack ||
+			sp >= (unsigned long)sstack + SIGSTKSZ) {
+		printf("[FAIL]\tSP is not on sigaltstack\n");
+		exit(EXIT_FAILURE);
+	}
+	/* put some data on stack. other sighandler will try to overwrite it */
+	aa = alloca(1024);
+	assert(aa);
+	p = (struct stk_data *)(aa + 512);
+	strcpy(p->msg, msg);
+	p->flag = 1;
+	printf("[RUN]\tsignal USR1\n");
+	err = sigaltstack(NULL, &stk);
+	if (err) {
+		perror("[FAIL]\tsigaltstack()");
+		exit(EXIT_FAILURE);
+	}
+	if (stk.ss_flags != SS_DISABLE)
+		printf("[FAIL]\tss_flags=%i, should be SS_DISABLE\n",
+				stk.ss_flags);
+	else
+		printf("[OK]\tsigaltstack is disabled in sighandler\n");
+	swapcontext(&sc, &uc);
+	printf("%s\n", p->msg);
+	if (!p->flag) {
+		printf("[RUN]\tAborting\n");
+		exit(EXIT_FAILURE);
+	}
+}
+
+void my_usr2(int sig, siginfo_t *si, void *u)
+{
+	char *aa;
+	struct stk_data *p;
+
+	printf("[RUN]\tsignal USR2\n");
+	aa = alloca(1024);
+	/* dont run valgrind on this */
+	/* try to find the data stored by previous sighandler */
+	p = memmem(aa, 1024, msg, strlen(msg));
+	if (p) {
+		printf("[FAIL]\tsigaltstack re-used\n");
+		/* corrupt the data */
+		strcpy(p->msg, msg2);
+		/* tell other sighandler that his data is corrupted */
+		p->flag = 0;
+	}
+}
+
+static void switch_fn(void)
+{
+	printf("[RUN]\tswitched to user ctx\n");
+	raise(SIGUSR2);
+	setcontext(&sc);
+}
+
+int main(void)
+{
+	struct sigaction act;
+	stack_t stk;
+	int err;
+
+	sigemptyset(&act.sa_mask);
+	act.sa_flags = SA_ONSTACK | SA_SIGINFO;
+	act.sa_sigaction = my_usr1;
+	sigaction(SIGUSR1, &act, NULL);
+	act.sa_sigaction = my_usr2;
+	sigaction(SIGUSR2, &act, NULL);
+	sstack = mmap(NULL, SIGSTKSZ, PROT_READ | PROT_WRITE,
+		      MAP_PRIVATE | MAP_ANONYMOUS | MAP_STACK, -1, 0);
+	if (sstack == MAP_FAILED) {
+		perror("mmap()");
+		return EXIT_FAILURE;
+	}
+	stk.ss_sp = sstack;
+	stk.ss_size = SIGSTKSZ;
+	stk.ss_flags = SS_ONSTACK | SS_AUTODISARM;
+	err = sigaltstack(&stk, NULL);
+	if (err) {
+		perror("sigaltstack()");
+		return EXIT_FAILURE;
+	}
+
+	ustack = mmap(NULL, SIGSTKSZ, PROT_READ | PROT_WRITE,
+		      MAP_PRIVATE | MAP_ANONYMOUS | MAP_STACK, -1, 0);
+	if (ustack == MAP_FAILED) {
+		perror("mmap()");
+		return EXIT_FAILURE;
+	}
+	getcontext(&uc);
+	uc.uc_link = NULL;
+	uc.uc_stack.ss_sp = ustack;
+	uc.uc_stack.ss_size = SIGSTKSZ;
+	makecontext(&uc, switch_fn, 0);
+	raise(SIGUSR1);
+
+	err = sigaltstack(NULL, &stk);
+	if (err) {
+		perror("[FAIL]\tsigaltstack()");
+		exit(EXIT_FAILURE);
+	}
+	if (stk.ss_flags != 0) {
+		printf("[FAIL]\tss_flags=%i, should be 0\n",
+				stk.ss_flags);
+		exit(EXIT_FAILURE);
+	}
+	printf("[OK]\tsigaltstack is enabled after signal\n");
+
+	printf("[OK]\tTest passed\n");
+	return 0;
+}
-- 
2.7.2
```

## Re: \[PATCH 1/2\] sigaltstack: implement SS\_AUTODISARM flag

- From: Andy Lutomirski \<luto@amacapital.net\>
- Date: 2016-03-04 22:22:47 
- Message: https://ratatoskr.run/linux-api/2016/03/4701902.md

```text
On Mon, Feb 29, 2016 at 1:29 PM, Stas Sergeev [ref] wrote:
> This patch implements the SS_AUTODISARM flag that can be ORed with
> SS_ONSTACK when forming ss_flags.
> When this flag is set, sigaltstack will be disabled when entering
> the signal handler; more precisely, after saving sas to uc_stack.
> When leaving the signal handler, the sigaltstack is restored by
> uc_stack.
> When this flag is used, it is safe to switch from sighandler with
> swapcontext(). Without this flag, the subsequent signal will corrupt
> the state of the switched-away sighandler.
>

This looks reasonable to me with one exception: how does a user
program detect the presence of this feature?  Anyone else have any
thoughts?
```

## Re: \[PATCH 1/2\] sigaltstack: implement SS\_AUTODISARM flag

- From: Stas Sergeev \<hidden\>
- Date: 2016-03-05 07:39:48 
- Message: https://ratatoskr.run/linux-api/2016/03/4702069.md

```text
05.03.2016 01:22, Andy Lutomirski пишет:
> On Mon, Feb 29, 2016 at 1:29 PM, Stas Sergeev [ref] wrote:
>> This patch implements the SS_AUTODISARM flag that can be ORed with
>> SS_ONSTACK when forming ss_flags.
>> When this flag is set, sigaltstack will be disabled when entering
>> the signal handler; more precisely, after saving sas to uc_stack.
>> When leaving the signal handler, the sigaltstack is restored by
>> uc_stack.
>> When this flag is used, it is safe to switch from sighandler with
>> swapcontext(). Without this flag, the subsequent signal will corrupt
>> the state of the switched-away sighandler.
>>
> This looks reasonable to me with one exception: how does a user
> program detect the presence of this feature?
Compile-time detection:
#ifdef SS_AUTODISARM
  # I have this feature
  ...
#endif

Run-time detection:
int err = sigaltstack(SS_ONSTACK | SS_AUTODISARM);
if (err == EINVAL) {
  i_dont_have_this_feature = 1;
  err = sigaltstack(SS_ONSTACK);
}

Note: if you want to keep such detection for the future
additions, the mask can be enlarged to, say, ((1 << 24) - 1),
and whenever someone adds a new flag, he can lower the
mask by one bit.
But I think this would be an overkill in that particular case.
```

## Re: \[PATCH 1/2\] sigaltstack: implement SS\_AUTODISARM flag

- From: Stas Sergeev \<hidden\>
- Date: 2016-03-05 07:41:18 
- Message: https://ratatoskr.run/linux-api/2016/03/4702070.md

```text
05.03.2016 10:39, Stas Sergeev пишет:
> 05.03.2016 01:22, Andy Lutomirski пишет:
>> On Mon, Feb 29, 2016 at 1:29 PM, Stas Sergeev [ref] wrote:
>>> This patch implements the SS_AUTODISARM flag that can be ORed with
>>> SS_ONSTACK when forming ss_flags.
>>> When this flag is set, sigaltstack will be disabled when entering
>>> the signal handler; more precisely, after saving sas to uc_stack.
>>> When leaving the signal handler, the sigaltstack is restored by
>>> uc_stack.
>>> When this flag is used, it is safe to switch from sighandler with
>>> swapcontext(). Without this flag, the subsequent signal will corrupt
>>> the state of the switched-away sighandler.
>>>
>> This looks reasonable to me with one exception: how does a user
>> program detect the presence of this feature?
> Compile-time detection:
> #ifdef SS_AUTODISARM
>   # I have this feature
>   ...
> #endif
>
> Run-time detection:
> int err = sigaltstack(SS_ONSTACK | SS_AUTODISARM);
> if (err == EINVAL) {
I guess I mean here
if (err && errno == EINVAL)
```

## Re: \[PATCH 1/2\] sigaltstack: implement SS\_AUTODISARM flag

- From: Andy Lutomirski \<luto@amacapital.net\>
- Date: 2016-03-06 20:07:34 
- Message: https://ratatoskr.run/linux-api/2016/03/4702455.md

```text
On Mon, Feb 29, 2016 at 1:29 PM, Stas Sergeev [ref] wrote:
> This patch implements the SS_AUTODISARM flag that can be ORed with
> SS_ONSTACK when forming ss_flags.
> When this flag is set, sigaltstack will be disabled when entering
> the signal handler; more precisely, after saving sas to uc_stack.
> When leaving the signal handler, the sigaltstack is restored by
> uc_stack.
> When this flag is used, it is safe to switch from sighandler with
> swapcontext(). Without this flag, the subsequent signal will corrupt
> the state of the switched-away sighandler.
>
> CC: Ingo Molnar <redacted>
> CC: Peter Zijlstra <redacted>
> CC: Richard Weinberger <redacted>
> CC: Andrew Morton <redacted>
> CC: Oleg Nesterov <redacted>
> CC: Tejun Heo <redacted>
> CC: Heinrich Schuchardt <redacted>
> CC: Jason Low <redacted>
> CC: Andrea Arcangeli <redacted>
> CC: Frederic Weisbecker <redacted>
> CC: Konstantin Khlebnikov <redacted>
> CC: Josh Triplett <redacted>
> CC: "Eric W. Biederman" <redacted>
> CC: Aleksa Sarai <redacted>
> CC: "Amanieu d'Antras" <redacted>
> CC: Paul Moore <redacted>
> CC: Sasha Levin <redacted>
> CC: Palmer Dabbelt <redacted>
> CC: Vladimir Davydov <redacted>
> CC: <redacted>
> CC: <redacted>
> CC: Andy Lutomirski <redacted>
>
> Signed-off-by: Stas Sergeev <redacted>
> ---
>  include/linux/sched.h       |  8 ++++++++
>  include/linux/signal.h      |  4 +++-
>  include/uapi/linux/signal.h |  3 +++
>  kernel/fork.c               |  2 +-
>  kernel/signal.c             | 23 ++++++++++++-----------
>  5 files changed, 27 insertions(+), 13 deletions(-)
>
> diff --git a/include/linux/sched.h b/include/linux/sched.h
> index a10494a..26201cd 100644
> --- a/include/linux/sched.h
> +++ b/include/linux/sched.h
> @@ -1587,6 +1587,7 @@ struct task_struct {
>
>         unsigned long sas_ss_sp;
>         size_t sas_ss_size;
> +       unsigned sas_ss_flags;
>
>         struct callback_head *task_works;
>
> @@ -2573,6 +2574,13 @@ static inline int sas_ss_flags(unsigned long sp)
>         return on_sig_stack(sp) ? SS_ONSTACK : 0;
>  }
>
> +static inline void sas_ss_reset(struct task_struct *p)
> +{
> +       p->sas_ss_sp = 0;
> +       p->sas_ss_size = 0;
> +       p->sas_ss_flags = SS_DISABLE;
> +}
> +
>  static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
>  {
>         if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
> diff --git a/include/linux/signal.h b/include/linux/signal.h
> index 92557bb..3fbe814 100644
> --- a/include/linux/signal.h
> +++ b/include/linux/signal.h
> @@ -432,8 +432,10 @@ int __save_altstack(stack_t __user *, unsigned long);
>         stack_t __user *__uss = uss; \
>         struct task_struct *t = current; \
>         put_user_ex((void __user *)t->sas_ss_sp, &__uss->ss_sp); \
> -       put_user_ex(sas_ss_flags(sp), &__uss->ss_flags); \
> +       put_user_ex(t->sas_ss_flags, &__uss->ss_flags); \
>         put_user_ex(t->sas_ss_size, &__uss->ss_size); \
> +       if (t->sas_ss_flags & SS_AUTODISARM) \
> +               sas_ss_reset(t); \
>  } while (0);
>
>  #ifdef CONFIG_PROC_FS
> diff --git a/include/uapi/linux/signal.h b/include/uapi/linux/signal.h
> index e1bd50c2..4691bc5 100644
> --- a/include/uapi/linux/signal.h
> +++ b/include/uapi/linux/signal.h
> @@ -6,5 +6,8 @@
>
>  #define SS_ONSTACK     1
>  #define SS_DISABLE     2
> +#define SS_VALMASK     0xf

SS_MODE_MASK, perhaps?

> diff --git a/kernel/signal.c b/kernel/signal.c
> index 0508544..fab6bea 100644
> --- a/kernel/signal.c
> +++ b/kernel/signal.c
> @@ -3100,13 +3100,14 @@ do_sigaltstack (const stack_t __user *uss, stack_t __user *uoss, unsigned long s
>         if (uss) {
>                 void __user *ss_sp;
>                 size_t ss_size;
> -               int ss_flags;
> +               unsigned ss_flags;
> +               unsigned ss_xflags;
>
>                 error = -EFAULT;
>                 if (!access_ok(VERIFY_READ, uss, sizeof(*uss)))
>                         goto out;
>                 error = __get_user(ss_sp, &uss->ss_sp) |
> -                       __get_user(ss_flags, &uss->ss_flags) |
> +                       __get_user(ss_xflags, &uss->ss_flags) |
>                         __get_user(ss_size, &uss->ss_size);
>                 if (error)
>                         goto out;
> @@ -3115,14 +3116,8 @@ do_sigaltstack (const stack_t __user *uss, stack_t __user *uoss, unsigned long s
>                 if (on_sig_stack(sp))
>                         goto out;
>
> +               ss_flags = ss_xflags & SS_VALMASK;

This is poorly named.  How about "ss_mode = ss_flags & SS_MODE_MASK";

--Andy
```

## Re: \[PATCH v4 0/2\] make sigaltstack() compatible with swapcontext()

- From: Szabolcs Nagy \<hidden\>
- Date: 2016-03-06 20:07:39 
- Message: https://ratatoskr.run/linux-api/2016/03/4702456.md

```text
* Stas Sergeev [ref] [2016-03-01 00:29:03 +0300]:
> The following patches make it possible to use swapcontext()
> in a sighandler that works on sigaltstack.

i don't think that's possible, the (obsolete) userspace
*context functions cannot operate on kernel provided
ucontext_t structs, so they are not usable in signal
handlers.

at least not on most target archs.

so i don't understand how the tests can work portably.

> The approach is inspired by Andy Lutomirski's suggestion that
> sigaltstack should disarm itself after saving into uc_stack:
> https://lkml.org/lkml/2016/2/1/594
> 
> I add the SS_AUTODISARM flag that does exactly that.
> On sighandler exit, the sigaltstack is restored from uc_stack.
> Another possible name could be SS_ONESHOT, but, since it gets
> always re-enabled, I choose SS_AUTODISARM.
> 
> [PATCH 1/2] sigaltstack: implement SS_AUTODISARM flag
> 	This patch implements SS_AUTODISARM flag
> [PATCH 2/2] selftests: Add test for sigaltstack(SS_AUTODISARM)
> 	This patch adds the selftest code for new functionality
> 
> CC: <redacted>
> CC: <redacted>
> CC: Andy Lutomirski <redacted>
> CC: Oleg Nesterov <redacted>
> CC: Shuah Khan <redacted>
> 
> Diffstat:
>  include/linux/sched.h                        |    8 +
>  include/linux/signal.h                       |    4 
>  include/uapi/linux/signal.h                  |    3 
>  kernel/fork.c                                |    2 
>  kernel/signal.c                              |   23 ++--
>  tools/testing/selftests/Makefile             |    1 
>  tools/testing/selftests/sigaltstack/Makefile |    8 +
>  tools/testing/selftests/sigaltstack/sas.c    |  151 +++++++++++++++++++++++++++
>  8 files changed, 187 insertions(+), 13 deletions(-)
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-api" in
> the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
```

## Re: \[PATCH 1/2\] sigaltstack: implement SS\_AUTODISARM flag

- From: Andy Lutomirski \<luto@amacapital.net\>
- Date: 2016-03-06 20:10:50 
- Message: https://ratatoskr.run/linux-api/2016/03/4702457.md

```text
On Sun, Mar 6, 2016 at 12:07 PM, Andy Lutomirski [ref] wrote:
> On Mon, Feb 29, 2016 at 1:29 PM, Stas Sergeev [ref] wrote:
>> This patch implements the SS_AUTODISARM flag that can be ORed with
>> SS_ONSTACK when forming ss_flags.
>> When this flag is set, sigaltstack will be disabled when entering
>> the signal handler; more precisely, after saving sas to uc_stack.
>> When leaving the signal handler, the sigaltstack is restored by
>> uc_stack.
>> When this flag is used, it is safe to switch from sighandler with
>> swapcontext(). Without this flag, the subsequent signal will corrupt
>> the state of the switched-away sighandler.
>>
>> CC: Ingo Molnar <redacted>
>> CC: Peter Zijlstra <redacted>
>> CC: Richard Weinberger <redacted>
>> CC: Andrew Morton <redacted>
>> CC: Oleg Nesterov <redacted>
>> CC: Tejun Heo <redacted>
>> CC: Heinrich Schuchardt <redacted>
>> CC: Jason Low <redacted>
>> CC: Andrea Arcangeli <redacted>
>> CC: Frederic Weisbecker <redacted>
>> CC: Konstantin Khlebnikov <redacted>
>> CC: Josh Triplett <redacted>
>> CC: "Eric W. Biederman" <redacted>
>> CC: Aleksa Sarai <redacted>
>> CC: "Amanieu d'Antras" <redacted>
>> CC: Paul Moore <redacted>
>> CC: Sasha Levin <redacted>
>> CC: Palmer Dabbelt <redacted>
>> CC: Vladimir Davydov <redacted>
>> CC: <redacted>
>> CC: <redacted>
>> CC: Andy Lutomirski <redacted>
>>
>> Signed-off-by: Stas Sergeev <redacted>
>> ---
>>  include/linux/sched.h       |  8 ++++++++
>>  include/linux/signal.h      |  4 +++-
>>  include/uapi/linux/signal.h |  3 +++
>>  kernel/fork.c               |  2 +-
>>  kernel/signal.c             | 23 ++++++++++++-----------
>>  5 files changed, 27 insertions(+), 13 deletions(-)
>>
>> diff --git a/include/linux/sched.h b/include/linux/sched.h
>> index a10494a..26201cd 100644
>> --- a/include/linux/sched.h
>> +++ b/include/linux/sched.h
>> @@ -1587,6 +1587,7 @@ struct task_struct {
>>
>>         unsigned long sas_ss_sp;
>>         size_t sas_ss_size;
>> +       unsigned sas_ss_flags;
>>
>>         struct callback_head *task_works;
>>
>> @@ -2573,6 +2574,13 @@ static inline int sas_ss_flags(unsigned long sp)
>>         return on_sig_stack(sp) ? SS_ONSTACK : 0;
>>  }
>>
>> +static inline void sas_ss_reset(struct task_struct *p)
>> +{
>> +       p->sas_ss_sp = 0;
>> +       p->sas_ss_size = 0;
>> +       p->sas_ss_flags = SS_DISABLE;
>> +}
>> +
>>  static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
>>  {
>>         if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
>> diff --git a/include/linux/signal.h b/include/linux/signal.h
>> index 92557bb..3fbe814 100644
>> --- a/include/linux/signal.h
>> +++ b/include/linux/signal.h
>> @@ -432,8 +432,10 @@ int __save_altstack(stack_t __user *, unsigned long);
>>         stack_t __user *__uss = uss; \
>>         struct task_struct *t = current; \
>>         put_user_ex((void __user *)t->sas_ss_sp, &__uss->ss_sp); \
>> -       put_user_ex(sas_ss_flags(sp), &__uss->ss_flags); \
>> +       put_user_ex(t->sas_ss_flags, &__uss->ss_flags); \
>>         put_user_ex(t->sas_ss_size, &__uss->ss_size); \
>> +       if (t->sas_ss_flags & SS_AUTODISARM) \
>> +               sas_ss_reset(t); \
>>  } while (0);
>>
>>  #ifdef CONFIG_PROC_FS
>> diff --git a/include/uapi/linux/signal.h b/include/uapi/linux/signal.h
>> index e1bd50c2..4691bc5 100644
>> --- a/include/uapi/linux/signal.h
>> +++ b/include/uapi/linux/signal.h
>> @@ -6,5 +6,8 @@
>>
>>  #define SS_ONSTACK     1
>>  #define SS_DISABLE     2
>> +#define SS_VALMASK     0xf
>
> SS_MODE_MASK, perhaps?

Actually, let's invert that.

#define SS_AUTODISARM (1U << 31)
#define SS_FLAG_BITS SS_AUTODISARM

ss_mode = ss_flags & ~SS_FLAG_BITS;

this way flag bits that are currently undefined will continue to trigger EINVAL.

--Andy
```

## Re: \[PATCH 1/2\] sigaltstack: implement SS\_AUTODISARM flag

- From: Stas Sergeev \<hidden\>
- Date: 2016-03-06 20:17:48 
- Message: https://ratatoskr.run/linux-api/2016/03/4702458.md

```text
06.03.2016 23:10, Andy Lutomirski пишет:
> On Sun, Mar 6, 2016 at 12:07 PM, Andy Lutomirski [ref] wrote:
>> On Mon, Feb 29, 2016 at 1:29 PM, Stas Sergeev [ref] wrote:
>>> This patch implements the SS_AUTODISARM flag that can be ORed with
>>> SS_ONSTACK when forming ss_flags.
>>> When this flag is set, sigaltstack will be disabled when entering
>>> the signal handler; more precisely, after saving sas to uc_stack.
>>> When leaving the signal handler, the sigaltstack is restored by
>>> uc_stack.
>>> When this flag is used, it is safe to switch from sighandler with
>>> swapcontext(). Without this flag, the subsequent signal will corrupt
>>> the state of the switched-away sighandler.
>>>
>>> CC: Ingo Molnar <redacted>
>>> CC: Peter Zijlstra <redacted>
>>> CC: Richard Weinberger <redacted>
>>> CC: Andrew Morton <redacted>
>>> CC: Oleg Nesterov <redacted>
>>> CC: Tejun Heo <redacted>
>>> CC: Heinrich Schuchardt <redacted>
>>> CC: Jason Low <redacted>
>>> CC: Andrea Arcangeli <redacted>
>>> CC: Frederic Weisbecker <redacted>
>>> CC: Konstantin Khlebnikov <redacted>
>>> CC: Josh Triplett <redacted>
>>> CC: "Eric W. Biederman" <redacted>
>>> CC: Aleksa Sarai <redacted>
>>> CC: "Amanieu d'Antras" <redacted>
>>> CC: Paul Moore <redacted>
>>> CC: Sasha Levin <redacted>
>>> CC: Palmer Dabbelt <redacted>
>>> CC: Vladimir Davydov <redacted>
>>> CC: <redacted>
>>> CC: <redacted>
>>> CC: Andy Lutomirski <redacted>
>>>
>>> Signed-off-by: Stas Sergeev <redacted>
>>> ---
>>>   include/linux/sched.h       |  8 ++++++++
>>>   include/linux/signal.h      |  4 +++-
>>>   include/uapi/linux/signal.h |  3 +++
>>>   kernel/fork.c               |  2 +-
>>>   kernel/signal.c             | 23 ++++++++++++-----------
>>>   5 files changed, 27 insertions(+), 13 deletions(-)
>>>
>>> diff --git a/include/linux/sched.h b/include/linux/sched.h
>>> index a10494a..26201cd 100644
>>> --- a/include/linux/sched.h
>>> +++ b/include/linux/sched.h
>>> @@ -1587,6 +1587,7 @@ struct task_struct {
>>>
>>>          unsigned long sas_ss_sp;
>>>          size_t sas_ss_size;
>>> +       unsigned sas_ss_flags;
>>>
>>>          struct callback_head *task_works;
>>>
>>> @@ -2573,6 +2574,13 @@ static inline int sas_ss_flags(unsigned long sp)
>>>          return on_sig_stack(sp) ? SS_ONSTACK : 0;
>>>   }
>>>
>>> +static inline void sas_ss_reset(struct task_struct *p)
>>> +{
>>> +       p->sas_ss_sp = 0;
>>> +       p->sas_ss_size = 0;
>>> +       p->sas_ss_flags = SS_DISABLE;
>>> +}
>>> +
>>>   static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
>>>   {
>>>          if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
>>> diff --git a/include/linux/signal.h b/include/linux/signal.h
>>> index 92557bb..3fbe814 100644
>>> --- a/include/linux/signal.h
>>> +++ b/include/linux/signal.h
>>> @@ -432,8 +432,10 @@ int __save_altstack(stack_t __user *, unsigned long);
>>>          stack_t __user *__uss = uss; \
>>>          struct task_struct *t = current; \
>>>          put_user_ex((void __user *)t->sas_ss_sp, &__uss->ss_sp); \
>>> -       put_user_ex(sas_ss_flags(sp), &__uss->ss_flags); \
>>> +       put_user_ex(t->sas_ss_flags, &__uss->ss_flags); \
>>>          put_user_ex(t->sas_ss_size, &__uss->ss_size); \
>>> +       if (t->sas_ss_flags & SS_AUTODISARM) \
>>> +               sas_ss_reset(t); \
>>>   } while (0);
>>>
>>>   #ifdef CONFIG_PROC_FS
>>> diff --git a/include/uapi/linux/signal.h b/include/uapi/linux/signal.h
>>> index e1bd50c2..4691bc5 100644
>>> --- a/include/uapi/linux/signal.h
>>> +++ b/include/uapi/linux/signal.h
>>> @@ -6,5 +6,8 @@
>>>
>>>   #define SS_ONSTACK     1
>>>   #define SS_DISABLE     2
>>> +#define SS_VALMASK     0xf
>> SS_MODE_MASK, perhaps?
> Actually, let's invert that.
>
> #define SS_AUTODISARM (1U << 31)
> #define SS_FLAG_BITS SS_AUTODISARM
>
> ss_mode = ss_flags & ~SS_FLAG_BITS;
>
> this way flag bits that are currently undefined will continue to trigger EINVAL.
OK, will do.
Though I am still going to keep SS_AUTODISARM =(1 << 4) -
it doesn't matter from what side we add new flags, does it?
```

## Re: \[PATCH v4 0/2\] make sigaltstack() compatible with swapcontext()

- From: Stas Sergeev \<hidden\>
- Date: 2016-03-06 20:27:46 
- Message: https://ratatoskr.run/linux-api/2016/03/4702459.md

```text
06.03.2016 23:02, Szabolcs Nagy пишет:
> * Stas Sergeev [ref] [2016-03-01 00:29:03 +0300]:
>> The following patches make it possible to use swapcontext()
>> in a sighandler that works on sigaltstack.
> i don't think that's possible, the (obsolete) userspace
> *context functions cannot operate on kernel provided
> ucontext_t structs,
It doesn't have to.
It does the normal getcontext()/swapcontext() sequences,
as if it would be switching from any casual code place.
Kernel-provided context structs are completely irrelevant.
For more details please see the test-case I posted in that patch series.
It shows what can be done and how.

>   so they are not usable in signal
> handlers.
Please note the fact that swapcontext() also replaces the
signal mask as part of the context switching. This is a very
strong hint that it was intended to work with signal handlers.
And in fact it perfectly does.
What I want is only to make it friendly to sigaltstack.
I am not making it friendly to signal handlers in general,
because it already is.

> at least not on most target archs.
>
> so i don't understand how the tests can work portably.
If you apply my patch series, you can immediately test
it on any arch you want, thanks to the test-case my patch
adds to the kernel. Please let me know on what arch it fails.
```

## Re: \[PATCH v4 0/2\] make sigaltstack() compatible with swapcontext()

- From: Szabolcs Nagy \<hidden\>
- Date: 2016-03-06 20:27:55 
- Message: https://ratatoskr.run/linux-api/2016/03/4702460.md

```text
* Szabolcs Nagy [ref] [2016-03-06 21:02:08 +0100]:
> * Stas Sergeev [ref] [2016-03-01 00:29:03 +0300]:
> > The following patches make it possible to use swapcontext()
> > in a sighandler that works on sigaltstack.
> 
> i don't think that's possible, the (obsolete) userspace
> *context functions cannot operate on kernel provided
> ucontext_t structs, so they are not usable in signal
> handlers.
> 
> at least not on most target archs.
> 
> so i don't understand how the tests can work portably.
> 

i see the test does not use the kernel ucontext_t
and only synchronous context switch is done

sorry for the noise.
```

## Re: \[PATCH 1/2\] sigaltstack: implement SS\_AUTODISARM flag

- From: Andy Lutomirski \<luto@amacapital.net\>
- Date: 2016-03-06 20:55:48 
- Message: https://ratatoskr.run/linux-api/2016/03/4702467.md

```text
On Sun, Mar 6, 2016 at 12:17 PM, Stas Sergeev [ref] wrote:
> 06.03.2016 23:10, Andy Lutomirski пишет:
>
>> On Sun, Mar 6, 2016 at 12:07 PM, Andy Lutomirski [ref]
>> wrote:
>>>
>>> On Mon, Feb 29, 2016 at 1:29 PM, Stas Sergeev [ref] wrote:
>>>>
>>>> This patch implements the SS_AUTODISARM flag that can be ORed with
>>>> SS_ONSTACK when forming ss_flags.
>>>> When this flag is set, sigaltstack will be disabled when entering
>>>> the signal handler; more precisely, after saving sas to uc_stack.
>>>> When leaving the signal handler, the sigaltstack is restored by
>>>> uc_stack.
>>>> When this flag is used, it is safe to switch from sighandler with
>>>> swapcontext(). Without this flag, the subsequent signal will corrupt
>>>> the state of the switched-away sighandler.
>>>>
>>>> CC: Ingo Molnar <redacted>
>>>> CC: Peter Zijlstra <redacted>
>>>> CC: Richard Weinberger <redacted>
>>>> CC: Andrew Morton <redacted>
>>>> CC: Oleg Nesterov <redacted>
>>>> CC: Tejun Heo <redacted>
>>>> CC: Heinrich Schuchardt <redacted>
>>>> CC: Jason Low <redacted>
>>>> CC: Andrea Arcangeli <redacted>
>>>> CC: Frederic Weisbecker <redacted>
>>>> CC: Konstantin Khlebnikov <redacted>
>>>> CC: Josh Triplett <redacted>
>>>> CC: "Eric W. Biederman" <redacted>
>>>> CC: Aleksa Sarai <redacted>
>>>> CC: "Amanieu d'Antras" <redacted>
>>>> CC: Paul Moore <redacted>
>>>> CC: Sasha Levin <redacted>
>>>> CC: Palmer Dabbelt <redacted>
>>>> CC: Vladimir Davydov <redacted>
>>>> CC: <redacted>
>>>> CC: <redacted>
>>>> CC: Andy Lutomirski <redacted>
>>>>
>>>> Signed-off-by: Stas Sergeev <redacted>
>>>> ---
>>>>   include/linux/sched.h       |  8 ++++++++
>>>>   include/linux/signal.h      |  4 +++-
>>>>   include/uapi/linux/signal.h |  3 +++
>>>>   kernel/fork.c               |  2 +-
>>>>   kernel/signal.c             | 23 ++++++++++++-----------
>>>>   5 files changed, 27 insertions(+), 13 deletions(-)
>>>>
>>>> diff --git a/include/linux/sched.h b/include/linux/sched.h
>>>> index a10494a..26201cd 100644
>>>> --- a/include/linux/sched.h
>>>> +++ b/include/linux/sched.h
>>>> @@ -1587,6 +1587,7 @@ struct task_struct {
>>>>
>>>>          unsigned long sas_ss_sp;
>>>>          size_t sas_ss_size;
>>>> +       unsigned sas_ss_flags;
>>>>
>>>>          struct callback_head *task_works;
>>>>
>>>> @@ -2573,6 +2574,13 @@ static inline int sas_ss_flags(unsigned long sp)
>>>>          return on_sig_stack(sp) ? SS_ONSTACK : 0;
>>>>   }
>>>>
>>>> +static inline void sas_ss_reset(struct task_struct *p)
>>>> +{
>>>> +       p->sas_ss_sp = 0;
>>>> +       p->sas_ss_size = 0;
>>>> +       p->sas_ss_flags = SS_DISABLE;
>>>> +}
>>>> +
>>>>   static inline unsigned long sigsp(unsigned long sp, struct ksignal
>>>> *ksig)
>>>>   {
>>>>          if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && !
>>>> sas_ss_flags(sp))
>>>> diff --git a/include/linux/signal.h b/include/linux/signal.h
>>>> index 92557bb..3fbe814 100644
>>>> --- a/include/linux/signal.h
>>>> +++ b/include/linux/signal.h
>>>> @@ -432,8 +432,10 @@ int __save_altstack(stack_t __user *, unsigned
>>>> long);
>>>>          stack_t __user *__uss = uss; \
>>>>          struct task_struct *t = current; \
>>>>          put_user_ex((void __user *)t->sas_ss_sp, &__uss->ss_sp); \
>>>> -       put_user_ex(sas_ss_flags(sp), &__uss->ss_flags); \
>>>> +       put_user_ex(t->sas_ss_flags, &__uss->ss_flags); \
>>>>          put_user_ex(t->sas_ss_size, &__uss->ss_size); \
>>>> +       if (t->sas_ss_flags & SS_AUTODISARM) \
>>>> +               sas_ss_reset(t); \
>>>>   } while (0);
>>>>
>>>>   #ifdef CONFIG_PROC_FS
>>>> diff --git a/include/uapi/linux/signal.h b/include/uapi/linux/signal.h
>>>> index e1bd50c2..4691bc5 100644
>>>> --- a/include/uapi/linux/signal.h
>>>> +++ b/include/uapi/linux/signal.h
>>>> @@ -6,5 +6,8 @@
>>>>
>>>>   #define SS_ONSTACK     1
>>>>   #define SS_DISABLE     2
>>>> +#define SS_VALMASK     0xf
>>>
>>> SS_MODE_MASK, perhaps?
>>
>> Actually, let's invert that.
>>
>> #define SS_AUTODISARM (1U << 31)
>> #define SS_FLAG_BITS SS_AUTODISARM
>>
>> ss_mode = ss_flags & ~SS_FLAG_BITS;
>>
>> this way flag bits that are currently undefined will continue to trigger
>> EINVAL.
>
> OK, will do.
> Though I am still going to keep SS_AUTODISARM =(1 << 4) -
> it doesn't matter from what side we add new flags, does it?

Adding them starting with a high bit means that more low bits are
available for use as mode numbers.  Admittedly, this barely matters.

--Andy

-- 
Andy Lutomirski
AMA Capital Management, LLC
```