On Tue, Jul 18, 2017 at 03:41:41PM +0200, Stefan Liebler wrote:
On 07/18/2017 12:20 PM, Dmitry V. Levin wrote:
On Mon, Jun 19, 2017 at 03:10:57PM +0200, Stefan Liebler wrote:
[...]
Mark Wielaard has spotted [1] a regression that I missed during review.
After this change, this test case fails to compile with the following
diagnostics:
$ gcc -c -xc -o/dev/null - <<'EOF'
#include <asm/ptrace.h>
#include <sys/ptrace.h>
EOF
Is this order required?
No, the reversed order also works.
I've tried it on x86_64 RHEL 7.3:
In file included from /usr/include/asm/ptrace.h:5:0,
from <stdin>:1:
/usr/include/sys/ptrace.h:74:4: error: expected identifier before
numeric constant
PTRACE_GETREGS = 12,
^
Interesting. So it means that it didn't work with some old linux headers.
However, it used to work with recent linux headers and glibc-2.25.
In file included from <stdin>:1:0:
/usr/include/sys/ptrace.h:93:3: error: expected identifier before numeric constant
PTRACE_SINGLEBLOCK = 12,
^
In file included from <stdin>:2:0:
/usr/include/sys/ptrace.h:158:0: warning: "PTRACE_PEEKUSR_AREA" redefined
#define PTRACE_PEEKUSR_AREA PTRACE_PEEKUSR_AREA
^
In file included from <stdin>:1:0:
/usr/include/asm/ptrace.h:392:0: note: this is the location of the previous definition
#define PTRACE_PEEKUSR_AREA 0x5000
^
In file included from <stdin>:2:0:
/usr/include/sys/ptrace.h:161:0: warning: "PTRACE_POKEUSR_AREA" redefined
#define PTRACE_POKEUSR_AREA PTRACE_POKEUSR_AREA
^
In file included from <stdin>:1:0:
/usr/include/asm/ptrace.h:393:0: note: this is the location of the previous definition
#define PTRACE_POKEUSR_AREA 0x5001
^
In file included from <stdin>:2:0:
/usr/include/sys/ptrace.h:164:0: warning: "PTRACE_GET_LAST_BREAK" redefined
#define PTRACE_GET_LAST_BREAK PTRACE_GET_LAST_BREAK
^
In file included from <stdin>:1:0:
/usr/include/asm/ptrace.h:398:0: note: this is the location of the previous definition
#define PTRACE_GET_LAST_BREAK 0x5006
^
In file included from <stdin>:2:0:
/usr/include/sys/ptrace.h:167:0: warning: "PTRACE_ENABLE_TE" redefined
#define PTRACE_ENABLE_TE PTRACE_ENABLE_TE
^
In file included from <stdin>:1:0:
/usr/include/asm/ptrace.h:401:0: note: this is the location of the previous definition
#define PTRACE_ENABLE_TE 0x5009
^
In file included from <stdin>:2:0:
/usr/include/sys/ptrace.h:170:0: warning: "PTRACE_DISABLE_TE" redefined
#define PTRACE_DISABLE_TE PTRACE_DISABLE_TE
^
In file included from <stdin>:1:0:
/usr/include/asm/ptrace.h:402:0: note: this is the location of the previous definition
#define PTRACE_DISABLE_TE 0x5010
^
In file included from <stdin>:2:0:
/usr/include/sys/ptrace.h:173:0: warning: "PTRACE_TE_ABORT_RAND" redefined
#define PTRACE_TE_ABORT_RAND PTRACE_TE_ABORT_RAND
^
In file included from <stdin>:1:0:
/usr/include/asm/ptrace.h:403:0: note: this is the location of the previous definition
#define PTRACE_TE_ABORT_RAND 0x5011
^
The following change fixes this and similar compilation issues that arise
when sys/ptrace.h is included after linux/ptrace.h:
--- a/sysdeps/unix/sysv/linux/s390/sys/ptrace.h
+++ b/sysdeps/unix/sysv/linux/s390/sys/ptrace.h
@@ -24,24 +24,61 @@
#include <bits/types.h>
__BEGIN_DECLS
-#ifdef _LINUX_PTRACE_H
+#if defined _LINUX_PTRACE_H || defined _S390_PTRACE_H
Shall we also add "|| defined _UAPI_LINUX_PTRACE_H" here?
Isn't _UAPI prefix automatically stripped during linux headers export
procedure?