Re: [Jack-Devel] [PATCH] jackdbus: Fix sigsegv handling segfaulting by itself
Hi David,
David Henningsson <[hidden]> writes:
> In my case, the siginfo code was -6, causing the sigsegv handling
> to crash on its own. This patch adds a range check for siginfo code.
>
> BugLink: https://bugs.launchpad.net/bugs/983835
> Signed-off-by: David Henningsson <[hidden]>
> ---
> dbus/sigsegv.c | 8 +++++++-
> 1 file changed, 7 insertions(+), 1 deletion(-)
>
> diff --git a/dbus/sigsegv.c b/dbus/sigsegv.c
> index 03b1183..8a87dc8 100644
> --- a/dbus/sigsegv.c
> +++ b/dbus/sigsegv.c
> @@ -60,6 +60,7 @@ static void signal_segv(int signum, siginfo_t* info, void*ptr) {
> static const char *si_codes[3] = {"", "SEGV_MAPERR", "SEGV_ACCERR"};
>
> size_t i;
> + const char *si_code_str;
> ucontext_t *ucontext = (ucontext_t*)ptr;
>
> #if defined(SIGSEGV_STACK_X86) || defined(SIGSEGV_STACK_IA64)
> @@ -94,9 +95,14 @@ static void signal_segv(int signum, siginfo_t* info, void*ptr) {
> jack_error("Unknown bad signal catched!");
> }
>
> + if (info->si_code >= 0 && info->si_code < 3)
> + si_code_str = si_codes[info->si_code];
> + else
> + si_code_str = "unknown";
> +
> jack_error("info.si_signo = %d", signum);
> jack_error("info.si_errno = %d", info->si_errno);
> - jack_error("info.si_code = %d (%s)", info->si_code, si_codes[info->si_code]);
> + jack_error("info.si_code = %d (%s)", info->si_code, si_code_str);
> jack_error("info.si_addr = %p", info->si_addr);
> #if !defined(__alpha__) && !defined(__ia64__) && !defined(__FreeBSD_kernel__) && !defined(__arm__) && !defined(__hppa__) && !defined(__sh__)
> for(i = 0; i < NGREG; i++)
I've applied the patch and pushed to the jack2 master branch. Thank you
for your contribution!
The si_code decoding is still suboptimal because it cannot handle
negative si_codes that are valid. si_codes are non-continuous and
the signal specific si_code values are reused for different
signals. Thus, implementing si_code to string descriptions map with
directly indexed array is not good idea in the first place.
FTR, a2jmidid, ladish and lash are also affected by this bug.
--
Nedko Arnaudov <GnuPG KeyID: 5D1B58ED>
1334710093.1482_0.ltw:2,a <87fwc19883.fsf at arnaudov dot name>