* [PATCH] pseudo: Fix problem where pseudo could kill a container init
@ 2016-09-22 21:37 Mark Hatle
0 siblings, 0 replies; only message in thread
From: Mark Hatle @ 2016-09-22 21:37 UTC (permalink / raw)
To: openembedded-core
In a heavily loaded container, the child process might not started
before the parent process had terminated. The child process attempts to
signal the parent with SIGUSR1. If the parent had terminated, the
parent becomes PID 1, which is generally init. When it signaled pid 1,
it caused the docker mini-init to terminate.
This doesn't happen in a traditional system, as systemd/sysvinit is
protected to only root users can signal it.
[YOCTO #10324]
Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
---
.../files/0001-Don-t-send-SIGUSR1-to-init.patch | 48 ++++++++++++++++++++++
meta/recipes-devtools/pseudo/pseudo_1.8.1.bb | 1 +
meta/recipes-devtools/pseudo/pseudo_git.bb | 2 +-
3 files changed, 50 insertions(+), 1 deletion(-)
create mode 100644 meta/recipes-devtools/pseudo/files/0001-Don-t-send-SIGUSR1-to-init.patch
diff --git a/meta/recipes-devtools/pseudo/files/0001-Don-t-send-SIGUSR1-to-init.patch b/meta/recipes-devtools/pseudo/files/0001-Don-t-send-SIGUSR1-to-init.patch
new file mode 100644
index 0000000..6c694ce
--- /dev/null
+++ b/meta/recipes-devtools/pseudo/files/0001-Don-t-send-SIGUSR1-to-init.patch
@@ -0,0 +1,48 @@
+From befc6dbd6469d428c9e0830dbe51bdf7ac39d9ae Mon Sep 17 00:00:00 2001
+From: Seebs <seebs@seebs.net>
+Date: Thu, 22 Sep 2016 14:35:04 -0500
+Subject: [PATCH] Don't send SIGUSR1 to init.
+
+If the parent exits due to child process being slow, getppid() will return
+1, and we'll send SIGUSR1 to init, which can break things like dumbinit
+which aren't adequately protected against non-root processes sending them
+signals.
+
+Signed-off-by: Seebs <seebs@seebs.net>
+
+Upstream-Status: Backport (commit befc6dbd6469d428c9e0830dbe51bdf7ac39d9ae)
+
+[YOCTO #10324]
+
+This resolves an issue where a docker container running builds would die
+due to it's 'mini init' being signaled by pseudo.
+
+Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
+---
+ pseudo_server.c | 11 ++++++++---
+
+diff --git a/pseudo_server.c b/pseudo_server.c
+index 65102dd..8731d20 100644
+--- a/pseudo_server.c
++++ b/pseudo_server.c
+@@ -358,9 +358,14 @@ pseudo_server_start(int daemonize) {
+ signal(SIGTERM, quit_now);
+ /* tell parent process to stop waiting */
+ if (daemonize) {
+- pseudo_diag("Setup complete, sending SIGUSR1 to pid %d.\n",
+- getppid());
+- kill(getppid(), SIGUSR1);
++ pid_t ppid = getppid();
++ if (ppid == 1) {
++ pseudo_diag("Setup complete, but parent is init, not sending SIGUSR1.\n");
++ } else {
++ pseudo_diag("Setup complete, sending SIGUSR1 to pid %d.\n",
++ ppid);
++ kill(ppid, SIGUSR1);
++ }
+ }
+ pseudo_server_loop();
+ return 0;
+--
+2.5.5
+
diff --git a/meta/recipes-devtools/pseudo/pseudo_1.8.1.bb b/meta/recipes-devtools/pseudo/pseudo_1.8.1.bb
index f45912a..cec2149 100644
--- a/meta/recipes-devtools/pseudo/pseudo_1.8.1.bb
+++ b/meta/recipes-devtools/pseudo/pseudo_1.8.1.bb
@@ -6,6 +6,7 @@ SRC_URI = "http://downloads.yoctoproject.org/releases/pseudo/${BPN}-${PV}.tar.bz
file://fallback-group \
file://moreretries.patch \
file://Fix-xattr-performance.patch \
+ file://0001-Don-t-send-SIGUSR1-to-init.patch \
"
SRC_URI[md5sum] = "ee38e4fb62ff88ad067b1a5a3825bac7"
diff --git a/meta/recipes-devtools/pseudo/pseudo_git.bb b/meta/recipes-devtools/pseudo/pseudo_git.bb
index 8735cf4..8110b1a 100644
--- a/meta/recipes-devtools/pseudo/pseudo_git.bb
+++ b/meta/recipes-devtools/pseudo/pseudo_git.bb
@@ -1,6 +1,6 @@
require pseudo.inc
-SRCREV = "f4b1c752186f4d08f1fadb0ea10ebcde9b0ea251"
+SRCREV = "befc6dbd6469d428c9e0830dbe51bdf7ac39d9ae"
PV = "1.8.1+git${SRCPV}"
DEFAULT_PREFERENCE = "-1"
--
2.5.5
^ permalink raw reply related [flat|nested] only message in thread
only message in thread, other threads:[~2016-09-22 21:37 UTC | newest]
Thread overview: (only message) (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2016-09-22 21:37 [PATCH] pseudo: Fix problem where pseudo could kill a container init Mark Hatle
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox