[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: [libvirt] [PATCH] Make QEMU text monitor parsing more robust



On Wed, Dec 09, 2009 at 10:46:58AM +0100, Daniel Veillard wrote:
> On Tue, Dec 08, 2009 at 06:08:04PM +0000, Daniel P. Berrange wrote:
> > The QEMU 0.10.0 release (and possibly other 0.10.x) has a bug where
> > it sometimes/often forgets to display the initial monitor greeting
> > line, soley printing a (qemu).  This in turn confuses the text
> > console parsing because it has a '(qemu)' it is not expecting. The
> > confusion results in a negative malloc. Bad things follow.
> > 
> > This re-writes the text console handling to be more robust. The key
> > idea is that it should only look for a (qemu), once it has seen the
> > original command echo'd back. This ensures it'll skip the bogus stray
> > (qemu) with broken QEMUs.
> > 
> > * src/qemu/qemu_monitor.c: Add some (disabled) debug code
> > * src/qemu/qemu_monitor_text.c: Re-write way command replies
> >   are detected
> > ---
> >  src/qemu/qemu_monitor.c      |   28 +++++++++++
> >  src/qemu/qemu_monitor_text.c |  102 +++++++++++++++++++++++++++--------------
> >  2 files changed, 95 insertions(+), 35 deletions(-)
> > 
> > diff --git a/src/qemu/qemu_monitor.c b/src/qemu/qemu_monitor.c
> > index abb763c..103cf28 100644
> > --- a/src/qemu/qemu_monitor.c
> > +++ b/src/qemu/qemu_monitor.c
> > @@ -39,6 +39,8 @@
> >  
> >  #define VIR_FROM_THIS VIR_FROM_QEMU
> >  
> > +#define QEMU_DEBUG_RAW_IO 0
> > +
> >  struct _qemuMonitor {
> >      virMutex lock;
> >      virCond notify;
> > @@ -163,6 +165,24 @@ char *qemuMonitorEscapeShell(const char *in)
> >  }
> >  
> >  
> > +#if QEMU_DEBUG_RAW_IO
> > +#include <c-ctype.h>
> > +static char * qemuMonitorEscapeNonPrintable(const char *text)
> > +{
> > +    int i;
> > +    virBuffer buf = VIR_BUFFER_INITIALIZER;
> > +    for (i = 0 ; text[i] != '\0' ; i++) {
> > +        if (c_isprint(text[i]) ||
> 
> Hum I just hope c_isprint behaviour doesn't depend on the user locale,
> otherwise I would just replace it with an explicit value test on the
> ASCII range, not a big deal since it's just for debug though ...

isprint() is locale sensitive

c_isprint() is fixed to the C locale


Either way, this code is never intended to be enabled in GIT - its just
there to make it easier to debug in the future, since I've added &
deleted this code in my working copy too many times now :-)

Regards,
Daniel
-- 
|: Red Hat, Engineering, London   -o-   http://people.redhat.com/berrange/ :|
|: http://libvirt.org  -o-  http://virt-manager.org  -o-  http://ovirt.org :|
|: http://autobuild.org       -o-         http://search.cpan.org/~danberr/ :|
|: GnuPG: 7D3B9505  -o-  F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 :|


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]