arm64: compat: Avoid sending SIGILL for unallocated syscall numbers
authorWill Deacon <will.deacon@arm.com>
Thu, 3 Jan 2019 17:45:07 +0000 (17:45 +0000)
committerWill Deacon <will.deacon@arm.com>
Fri, 4 Jan 2019 14:18:01 +0000 (14:18 +0000)
commit169113ece0f29ebe884a6cfcf57c1ace04d8a36a
tree6f5dfd2cb72ed264b96097768e29c2c2f341b481
parent9966a05c7b80f075f2bc7e48dbb108d3f2927234
arm64: compat: Avoid sending SIGILL for unallocated syscall numbers

The ARM Linux kernel handles the EABI syscall numbers as follows:

  0           - NR_SYSCALLS-1 : Invoke syscall via syscall table
  NR_SYSCALLS - 0xeffff : -ENOSYS (to be allocated in future)
  0xf0000     - 0xf07ff : Private syscall or -ENOSYS if not allocated
  > 0xf07ff : SIGILL

Our compat code gets this wrong and ends up sending SIGILL in response
to all syscalls greater than NR_SYSCALLS which have a value greater
than 0x7ff in the bottom 16 bits.

Fix this by defining the end of the ARM private syscall region and
checking the syscall number against that directly. Update the comment
while we're at it.

Cc: <stable@vger.kernel.org>
Cc: Dave Martin <Dave.Martin@arm.com>
Reported-by: Pi-Hsun Shih <pihsun@chromium.org>
Signed-off-by: Will Deacon <will.deacon@arm.com>
arch/arm64/include/asm/unistd.h
arch/arm64/kernel/sys_compat.c