Am Donnerstag 21 Juli 2011, 11:38:11 schrieb Geert Uytterhoeven:

On Thu, Jul 21, 2011 at 11:18, Richard Weinberger [off-list ref] wrote:

quoted

Linux can have pids up to 4*1024*1024.
To handle such huge numbers pid_buf needs to be larger.

Reported-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Richard Weinberger <richard@nod.at>
---
 arch/um/drivers/harddog_user.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/arch/um/drivers/harddog_user.c

b/arch/um/drivers/harddog_user.c index b56f8e0..84dce3f 100644

--- a/arch/um/drivers/harddog_user.c
+++ b/arch/um/drivers/harddog_user.c

@@ -32,7 +32,7 @@ int start_watchdog(int *in_fd_ret, int *out_fd_ret,

char *sock) {
       struct dog_data data;
       int in_fds[2], out_fds[2], pid, n, err;
-       char pid_buf[sizeof("nnnnn\0")], c;
+       char pid_buf[sizeof("nnnnnnn\0")], c;

Why not make it handle the full 32-bit? That's just a few bytes extra
on the stack...

True.
The sizeof("nnnnnnn\0") crap will anyway go away.
ASAP I'll submit a patch which removes all sizeof("nnnnnnn\0") and replaces
sprintf() with snprintf().

Thanks,
//richard