Re: [PATCH v16 5/6] signal: define the SA_UNSUPPORTED bit in sa_flags

On Sat, Nov 14, 2020 at 5:53 AM Eric W. Biederman [off-list ref] wrote:

Peter Collingbourne [off-list ref] writes:

quoted

Define a sa_flags bit, SA_UNSUPPORTED, which will never be supported
in the uapi. The purpose of this flag bit is to allow userspace to
distinguish an old kernel that does not clear unknown sa_flags bits
from a kernel that supports every flag bit.

In other words, if userspace does something like:

  act.sa_flags |= SA_UNSUPPORTED;
  sigaction(SIGSEGV, &act, 0);
  sigaction(SIGSEGV, 0, &oldact);

and finds that SA_UNSUPPORTED remains set in oldact.sa_flags, it means
that the kernel cannot be trusted to have cleared unknown flag bits
from sa_flags, so no assumptions about flag bit support can be made.

Signed-off-by: Peter Collingbourne <redacted>
Reviewed-by: Dave Martin <Dave.Martin@arm.com>
Link: https://linux-review.googlesource.com/id/Ic2501ad150a3a79c1cf27fb8c99be342e9dffbcb
---
v11:
- clarify the commit message

 include/uapi/asm-generic/signal-defs.h | 7 +++++++
 kernel/signal.c                        | 6 ++++++
 2 files changed, 13 insertions(+)

diff --git a/include/uapi/asm-generic/signal-defs.h b/include/uapi/asm-generic/signal-defs.h
index 493953fe319b..0126ebda4d31 100644
--- a/include/uapi/asm-generic/signal-defs.h
+++ b/include/uapi/asm-generic/signal-defs.h

@@ -14,6 +14,12 @@
  * SA_RESTART flag to get restarting signals (which were the default long ago)
  * SA_NODEFER prevents the current signal from being masked in the handler.
  * SA_RESETHAND clears the handler when the signal is delivered.
+ * SA_UNSUPPORTED is a flag bit that will never be supported. Kernels from
+ * before the introduction of SA_UNSUPPORTED did not clear unknown bits from
+ * sa_flags when read using the oldact argument to sigaction and rt_sigaction,
+ * so this bit allows flag bit support to be detected from userspace while
+ * allowing an old kernel to be distinguished from a kernel that supports every
+ * flag bit.
  *
  * SA_ONESHOT and SA_NOMASK are the historical Linux names for the Single
  * Unix names RESETHAND and NODEFER respectively.

@@ -42,6 +48,7 @@
 #ifndef SA_RESETHAND
 #define SA_RESETHAND 0x80000000
 #endif
+#define SA_UNSUPPORTED       0x00000400

Why this value and why not in numerical order with the other flags?

At the very least not being in order with the other bits makes it
a little easier to overlook it and define something at that position.

The value is because this is the first bit that isn't already taken by
an architecture-specific flag bit. It seems okay to move it into
numerical order.

The taken flag bits are listed in the comment that I added in patch 3.
Do you think there would be a more prominent way to document them?
Maybe we can replace that comment with inline, in-order comments along
the lines of:

#ifndef SA_NOCLDSTOP
#define SA_NOCLDSTOP   0x00000001
#endif
#ifndef SA_NOCLDWAIT
#define SA_NOCLDWAIT   0x00000002
#endif
#ifndef SA_SIGINFO
#define SA_SIGINFO     0x00000004
#endif
/* 0x00000008 has arch-specific definition */
/* 0x00000010 has arch-specific definition */

etc.

And then this patch would add the new bit in the right place.

Peter

Eric

quoted

 #define SA_NOMASK    SA_NODEFER
 #define SA_ONESHOT   SA_RESETHAND

diff --git a/kernel/signal.c b/kernel/signal.c
index 8f5bd12ee41b..8f34819e80de 100644
--- a/kernel/signal.c
+++ b/kernel/signal.c

@@ -3985,6 +3985,12 @@ int do_sigaction(int sig, struct k_sigaction *act, struct k_sigaction *oact)
      if (oact)
              *oact = *k;

+     /*
+      * Make sure that we never accidentally claim to support SA_UNSUPPORTED,
+      * e.g. by having an architecture use the bit in their uapi.
+      */
+     BUILD_BUG_ON(UAPI_SA_FLAGS & SA_UNSUPPORTED);
+
      /*
       * Clear unknown flag bits in order to allow userspace to detect missing
       * support for flag bits and to allow the kernel to use non-uapi bits