On Tue, 2011-12-13 at 12:26 +0000, Richard Purdie wrote:
> On Tue, 2011-12-13 at 08:58 +0200, Lauri Hintsala wrote:
> > # Set the system clock from hardware clock
> > -# If the timestamp is 1 day or more recent than the current time,
> > +# If the timestamp is more recent than the current time,
> > # use the timestamp instead.
> > /etc/init.d/hwclock.sh start
> > if test -e /etc/timestamp
> > then
> > - SYSTEMDATE=`date -u +%2m%2d%2H%2M%4Y`
> > - read TIMESTAMP < /etc/timestamp
> > - NEEDUPDATE=`expr \( $TIMESTAMP \> $SYSTEMDATE + 10000 \)`
> > - if [ $NEEDUPDATE -eq 1 ]; then
> > - date -u $TIMESTAMP
> > + SYSTEMDATE=`date -u +%4Y%2m%2d`
> > + TIMESTAMP=`cat /etc/timestamp | awk '{ print substr($0,9,4)
> > substr($0,1,4); }'`
> > + NEEDUPDATE=`expr \( $TIMESTAMP \> $SYSTEMDATE \)`
> > + if [ $NEEDUPDATE -eq 1 ]; then
> > + date -u `cat /etc/timestamp`
> > /etc/init.d/hwclock.sh stop
> > fi
> > fi
>
>
> For reference, the code in the boot process is trying not to cause
> fork/exec calls. This is why it does:
>
> read TIMESTAMP < /etc/timestamp
>
> since this is faster than forking to run cat. Could we fix this in a
> different way to avoid the fork/execs?
For the same reason it would probably be nice to replace that call to
"expr" (which was in the old version too) with a shell expansion.
p.
_______________________________________________
Openembedded-core mailing list
[email protected]
http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-core
