Skip to contentSkip to conversation

sysvinit: start .sh scripts correctly

PATCH NAK [email protected] 3 participants 6 messages Feb 26, 2013 – Feb 27, 2013

lore
Share & tools
lore
#0 lore raw

cover sysvinit: start .sh scripts correctly

From: Chen Qi <[email protected]>

The following changes since commit e96d541c8edccbb69602eace900ce9293ee12d22:

tiny-init: Mount devtmpfs manually (2013-02-22 06:39:07 -0800)

are available in the git repository at:

git://git.pokylinux.org/poky-contrib ChenQi/sysv-consistent
http://git.pokylinux.org/cgit.cgi/poky-contrib/log/?h=ChenQi/sysv-consistent

Chen Qi (1):
sysvinit: start .sh scripts correctly

 meta/recipes-core/sysvinit/sysvinit/rc |   16 +---------------
 1 file changed, 1 insertion(+), 15 deletions(-)
Signature · 1 line
1.7.9.5
#1 lore raw

PATCH 1/1 sysvinit: start .sh scripts correctly

From: Chen Qi <[email protected]>

Previously, scripts which end with '.sh' were sourced, so the arguments like 'start' and 'stop' were just ignored.

This resulted in some init scripts not being able to start correctly. For example, sourcing hwclock.sh in busybox actually does nothing. It should be invoked as 'hwclock.sh start' or 'hwclock.sh stop'.

This patch fixes this issue.

[YOCTO #3612]

---
 meta/recipes-core/sysvinit/sysvinit/rc |   16 +---------------
 1 file changed, 1 insertion(+), 15 deletions(-)
meta/recipes-core/sysvinit/sysvinit/rc +1 -15
index 44bc9bf..50951da 100755
@@ -41,21 +41,7 @@ startup_progress() {
 startup() {
   # Handle verbosity
   [ "$VERBOSE" = very ] && echo "INIT: Running $@..."
-
-  case "$1" in
-	*.sh)
-		# Source shell script for speed.
-		(
-			trap - INT QUIT TSTP
-			scriptname=$1
-			shift
-			. $scriptname
-		)
-		;;
-	*)
-		"$@"
-		;;
-  esac
+  "$@"
   startup_progress
 }
 
Signature · 1 line
1.7.9.5
#2 lore raw
Quoted 23 lines · On 02/25/2013 05:39 PM, [email protected] wrote:

From: Chen Qi <[email protected]>

Previously, scripts which end with '.sh' were sourced, so the arguments like 'start' and 'stop' were just ignored.

This resulted in some init scripts not being able to start correctly. For example, sourcing hwclock.sh in busybox actually does nothing. It should be invoked as 'hwclock.sh start' or 'hwclock.sh stop'.

This patch fixes this issue.

[YOCTO #3612]

---
  meta/recipes-core/sysvinit/sysvinit/rc |   16 +---------------
  1 file changed, 1 insertion(+), 15 deletions(-)
meta/recipes-core/sysvinit/sysvinit/rc +0 -0
index 44bc9bf..50951da 100755

meta/recipes-core/sysvinit/sysvinit/rc startup_progress() { · line 49

Quoted 9 lines
meta/recipes-core/sysvinit/sysvinit/rc +0 -6
@@ -41,21 +41,7 @@ startup_progress() {
  startup() {
    # Handle verbosity
    [ "$VERBOSE" = very ] && echo "INIT: Running $@..."
-
-  case "$1" in
-	*.sh)
-		# Source shell script for speed.
-		(
-			trap - INT QUIT TSTP

Are you sure you don't want the trap still? I realize this reset it to default..

Sau!

Quoted 14 lines
meta/recipes-core/sysvinit/sysvinit/rc +1 -9
-			scriptname=$1
-			shift
-			. $scriptname
-		)
-		;;
-	*)
-		"$@"
-		;;
-  esac
+  "$@"
    startup_progress
  }
 
author
#3 lore raw
Quoted 19 lines · Saul Wold
Quoted 18 lines · On 02/25/2013 05:39 PM, [email protected] wrote:

From: Chen Qi <[email protected]>

Previously, scripts which end with '.sh' were sourced, so the arguments like 'start' and 'stop' were just ignored.

This resulted in some init scripts not being able to start correctly. For example, sourcing hwclock.sh in busybox actually does nothing. It should be invoked as 'hwclock.sh start' or 'hwclock.sh stop'.

This patch fixes this issue.

[YOCTO #3612]

---
  meta/recipes-core/sysvinit/sysvinit/rc |   16 +---------------
  1 file changed, 1 insertion(+), 15 deletions(-)
Quoted 17 lines
Quoted 5 lines
meta/recipes-core/sysvinit/sysvinit/rc +0 -0
index 44bc9bf..50951da 100755

meta/recipes-core/sysvinit/sysvinit/rc startup_progress() { · line 49

Quoted 9 lines
meta/recipes-core/sysvinit/sysvinit/rc +0 -6
@@ -41,21 +41,7 @@ startup_progress() {
  startup() {
    # Handle verbosity
    [ "$VERBOSE" = very ] && echo "INIT: Running $@..."
-
-  case "$1" in
-    *.sh)
-        # Source shell script for speed.
-        (
-            trap - INT QUIT TSTP

Are you sure you don't want the trap still? I realize this reset it to default..

Yes.

Code snippet from rc:
# Ignore CTRL-C only in this shell, so we can interrupt subprocesses. trap ":" INT QUIT TSTP

So these signals are ignored in rc but passed on to subprocesses. And the reset is not needed.

If in rc we use
trap "" INIT QUIT TSTP
then the reset is needed.

Best Regards,
Chen Qi

Quoted 18 lines

Sau!

Quoted 14 lines
meta/recipes-core/sysvinit/sysvinit/rc +1 -9
-            scriptname=$1
-            shift
-            . $scriptname
-        )
-        ;;
-    *)
-        "$@"
-        ;;
-  esac
+  "$@"
    startup_progress
  }
 
#4 lore raw
In reply to [email protected]
Quoted 8 lines · On 26 February 2013 01:39, <[email protected]> wrote:
meta/recipes-core/sysvinit/sysvinit/rc +0 -8
-
-  case "$1" in
-       *.sh)
-               # Source shell script for speed.
-               (
-                       trap - INT QUIT TSTP
-                       scriptname=$1
-                       shift

meta/recipes-core/sysvinit/sysvinit/rc startup_progress() { · line 46

Quoted 10 lines
meta/recipes-core/sysvinit/sysvinit/rc +1 -7
-                       . $scriptname
-               )
-               ;;
-       *)
-               "$@"
-               ;;
-  esac
+  "$@"
   startup_progress
 }

NACK.

By "fix" you mean "remove the tested and proven optimisation"? The "if .sh use ." test was designed to speed up booting by not forking a new bash, and it's been demonstrated to have a noticeable difference on slower hardware.

You can pass arguments to "." as this little test demonstrates:

$ cat service.sh
echo My arguments are "$@"
$ . service.sh foo bar
My arguments are foo bar

The "shift" command shows that passing the arguments to the script was the intention, and a few lines of micro-test demonstrate that it *should* work:

$ cat rc.sh
startup() {
    scriptname=$1
    shift
    . $scriptname
}
startup ./service.sh start
$ busybox sh ./rc.sh
My arguments are start

So, something else is going wrong.

Ross

author
#5 lore raw
Quoted 47 lines · Ross
Quoted 8 lines · On 26 February 2013 01:39, <[email protected]> wrote:
meta/recipes-core/sysvinit/sysvinit/rc +0 -8
-
-  case "$1" in
-       *.sh)
-               # Source shell script for speed.
-               (
-                       trap - INT QUIT TSTP
-                       scriptname=$1
-                       shift

meta/recipes-core/sysvinit/sysvinit/rc startup_progress() { · line 46

Quoted 10 lines
meta/recipes-core/sysvinit/sysvinit/rc +1 -7
-                       . $scriptname
-               )
-               ;;
-       *)
-               "$@"
-               ;;
-  esac
+  "$@"
    startup_progress
  }

NACK.

By "fix" you mean "remove the tested and proven optimisation"? The "if .sh use ." test was designed to speed up booting by not forking a new bash, and it's been demonstrated to have a noticeable difference on slower hardware.

You can pass arguments to "." as this little test demonstrates:

$ cat service.sh
echo My arguments are "$@"
$ . service.sh foo bar
My arguments are foo bar

The "shift" command shows that passing the arguments to the script was the intention, and a few lines of micro-test demonstrate that it *should* work:

$ cat rc.sh
startup() {
     scriptname=$1
     shift
     . $scriptname
}
startup ./service.sh start
$ busybox sh ./rc.sh
My arguments are start
Quoted 5 lines

So, something else is going wrong.

Ross

Yes, you're right.
I absolutely made a mistake.

Thank you for pointing it out.

Best Regards,
Chen Qi