[PATCH RFC] debugger: detect presence of ptrace type debuggers

[PATCH RFC] debugger: detect presence of ptrace type debuggers

[Bart Trojanowski]

According to the internet [1], this is a more reliable way of detecting if
one is under the spell of a debugger.  Should work for ptrace, strace,
gdb, etc.  Basically anything that uses the ptrace() syscall.

[1] http://vx.netlux.org/lib/vsc04.html
---
 debugger.c |    5 +++++
 1 files changed, 5 insertions(+), 0 deletions(-)

diff --git a/debugger.c b/debugger.c
index e8b9378..f486032 100644
--- a/debugger.c
+++ b/debugger.c
@@ -22,6 +22,8 @@
 
 #include <libgen.h>
 
+#include <sys/ptrace.h>
+
 #if HAVE_VALGRIND
 #include <valgrind.h>
 #else
@@ -36,6 +38,9 @@ debugger_is_active (void)
     if (RUNNING_ON_VALGRIND)
 	return TRUE;
 
+    if (ptrace(PTRACE_TRACEME, 0, 1, 0) < 0)
+	return TRUE;
+
     sprintf (buf, "/proc/%d/exe", getppid ());
     if (readlink (buf, buf, sizeof (buf)) != -1 &&
 	strncmp (basename (buf), "gdb", 3) == 0)
-- 
1.6.4.4.2.gc2f148

Re: [PATCH RFC] debugger: detect presence of ptrace type debuggers

[Carl Worth]

[-- Attachment #1: Type: text/plain, Size: 476 bytes --]

On Fri, 27 Nov 2009 22:11:04 -0500, Bart Trojanowski <bart@jukie.net> wrote:
> According to the internet [1], this is a more reliable way of detecting if
> one is under the spell of a debugger.  Should work for ptrace, strace,
> gdb, etc.  Basically anything that uses the ptrace() syscall.

Doesn't avoid having side effects in the non-debugger case:

	$ ./notmuch new
	No new mail---and that's not much.
	
	[1]+  Stopped                 ./notmuch new
	$

Not pushed.

-Carl

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]