[Qemu-devel] [PATCH] flush stdout after printing usage()

[Qemu-devel] [PATCH] flush stdout after printing usage()

[Riku Voipio]

testcase:

qemu-arm|grep cpu

Without fflush() getting output from piped qemu is a bit random.

Signed-off-by: Riku Voipio <riku.voipio@iki.fi>
---
 bsd-user/main.c    |    1 +
 darwin-user/main.c |    1 +
 linux-user/main.c  |    1 +
 3 files changed, 3 insertions(+), 0 deletions(-)

diff --git a/bsd-user/main.c b/bsd-user/main.c
index 636f1dc..f3cb3b1 100644
--- a/bsd-user/main.c
+++ b/bsd-user/main.c
@@ -351,6 +351,7 @@ static void usage(void)
            interp_prefix,
            x86_stack_size,
            DEBUG_LOGFILE);
+    fflush(stdout);
     _exit(1);
 }
 
diff --git a/darwin-user/main.c b/darwin-user/main.c
index 3edad73..44e8f7c 100644
--- a/darwin-user/main.c
+++ b/darwin-user/main.c
@@ -758,6 +758,7 @@ void usage(void)
            interp_prefix,
            stack_size,
            DEBUG_LOGFILE);
+    fflush(stdout);
     _exit(1);
 }
 
diff --git a/linux-user/main.c b/linux-user/main.c
index 3418ca6..9b807e5 100644
--- a/linux-user/main.c
+++ b/linux-user/main.c
@@ -2216,6 +2216,7 @@ static void usage(void)
            interp_prefix,
            x86_stack_size,
            DEBUG_LOGFILE);
+    fflush(stdout);
     _exit(1);
 }
 
-- 
1.5.6.5


-- 
"rm -rf" only sounds scary if you don't have backups

Re: [Qemu-devel] [PATCH] flush stdout after printing usage()

[Anthony Liguori]

Riku Voipio wrote:
> testcase:
>
> qemu-arm|grep cpu
>
> Without fflush() getting output from piped qemu is a bit random.
>   

It should get flushed upon exit(). Perhaps the problem is that we're 
using _exit() instead of exit()?

Regards,

Anthony Liguori

> Signed-off-by: Riku Voipio <riku.voipio@iki.fi>
> ---
>  bsd-user/main.c    |    1 +
>  darwin-user/main.c |    1 +
>  linux-user/main.c  |    1 +
>  3 files changed, 3 insertions(+), 0 deletions(-)
>
> diff --git a/bsd-user/main.c b/bsd-user/main.c
> index 636f1dc..f3cb3b1 100644
> --- a/bsd-user/main.c
> +++ b/bsd-user/main.c
> @@ -351,6 +351,7 @@ static void usage(void)
>             interp_prefix,
>             x86_stack_size,
>             DEBUG_LOGFILE);
> +    fflush(stdout);
>      _exit(1);
>  }
>  
> diff --git a/darwin-user/main.c b/darwin-user/main.c
> index 3edad73..44e8f7c 100644
> --- a/darwin-user/main.c
> +++ b/darwin-user/main.c
> @@ -758,6 +758,7 @@ void usage(void)
>             interp_prefix,
>             stack_size,
>             DEBUG_LOGFILE);
> +    fflush(stdout);
>      _exit(1);
>  }
>  
> diff --git a/linux-user/main.c b/linux-user/main.c
> index 3418ca6..9b807e5 100644
> --- a/linux-user/main.c
> +++ b/linux-user/main.c
> @@ -2216,6 +2216,7 @@ static void usage(void)
>             interp_prefix,
>             x86_stack_size,
>             DEBUG_LOGFILE);
> +    fflush(stdout);
>      _exit(1);
>  }
>  
>   

Re: [Qemu-devel] [PATCH] flush stdout after printing usage()

[Riku Voipio]

On Wed, Jan 28, 2009 at 02:10:42PM -0600, Anthony Liguori wrote:
> Riku Voipio wrote:
> >testcase:
> >
> >qemu-arm|grep cpu
> >
> >Without fflush() getting output from piped qemu is a bit random.

> It should get flushed upon exit(). Perhaps the problem is that we're 
> using _exit() instead of exit()?

Well yes in my testing, replacing _exit() with exit() fixes it too.
I just presumed _exit() was there for a purpose.

> Regards,
> 
> Anthony Liguori
> 
> >Signed-off-by: Riku Voipio <riku.voipio@iki.fi>
> >---
> > bsd-user/main.c    |    1 +
> > darwin-user/main.c |    1 +
> > linux-user/main.c  |    1 +
> > 3 files changed, 3 insertions(+), 0 deletions(-)
> >
> >diff --git a/bsd-user/main.c b/bsd-user/main.c
> >index 636f1dc..f3cb3b1 100644
> >--- a/bsd-user/main.c
> >+++ b/bsd-user/main.c
> >@@ -351,6 +351,7 @@ static void usage(void)
> >            interp_prefix,
> >            x86_stack_size,
> >            DEBUG_LOGFILE);
> >+    fflush(stdout);
> >     _exit(1);
> > }
> > 
> >diff --git a/darwin-user/main.c b/darwin-user/main.c
> >index 3edad73..44e8f7c 100644
> >--- a/darwin-user/main.c
> >+++ b/darwin-user/main.c
> >@@ -758,6 +758,7 @@ void usage(void)
> >            interp_prefix,
> >            stack_size,
> >            DEBUG_LOGFILE);
> >+    fflush(stdout);
> >     _exit(1);
> > }
> > 
> >diff --git a/linux-user/main.c b/linux-user/main.c
> >index 3418ca6..9b807e5 100644
> >--- a/linux-user/main.c
> >+++ b/linux-user/main.c
> >@@ -2216,6 +2216,7 @@ static void usage(void)
> >            interp_prefix,
> >            x86_stack_size,
> >            DEBUG_LOGFILE);
> >+    fflush(stdout);
> >     _exit(1);
> > }
> > 
> >  
> 
> 

-- 
"rm -rf" only sounds scary if you don't have backups

Re: [Qemu-devel] [PATCH] flush stdout after printing usage()

[Lionel Landwerlin]

Le mercredi 28 janvier 2009 à 23:06 +0200, Riku Voipio a écrit :
> On Wed, Jan 28, 2009 at 02:10:42PM -0600, Anthony Liguori wrote:
> > Riku Voipio wrote:
> > >testcase:
> > >
> > >qemu-arm|grep cpu
> > >
> > >Without fflush() getting output from piped qemu is a bit random.
> 
> > It should get flushed upon exit(). Perhaps the problem is that we're 
> > using _exit() instead of exit()?
> 

_exit does not trigger callbacks registred with at_exit.

from the man page :

       The function _exit() is like exit(3), but does not call  any  functions
       registered  with  atexit(3) or on_exit(3).  Whether it flushes standard
       I/O buffers and removes temporary  files  created  with  tmpfile(3)  is
       implementation-dependent.   On  the other hand, _exit() does close open
       file descriptors, and this may cause  an  unknown  delay,  waiting  for
       pending  output to finish.  If the delay is undesired, it may be useful
       to call functions like tcflush(3) before calling _exit().  Whether  any
       pending  I/O  is  canceled,  and which pending I/O may be canceled upon
       _exit(), is implementation-dependent.

Re: [Qemu-devel] [PATCH] flush stdout after printing usage()

[Anthony Liguori]

Lionel Landwerlin wrote:
> Le mercredi 28 janvier 2009 à 23:06 +0200, Riku Voipio a écrit :
>   
>> On Wed, Jan 28, 2009 at 02:10:42PM -0600, Anthony Liguori wrote:
>>     
>>> Riku Voipio wrote:
>>>       
>>>> testcase:
>>>>
>>>> qemu-arm|grep cpu
>>>>
>>>> Without fflush() getting output from piped qemu is a bit random.
>>>>         
>>> It should get flushed upon exit(). Perhaps the problem is that we're 
>>> using _exit() instead of exit()?
>>>       
>
> _exit does not trigger callbacks registred with at_exit.
>   

Yes.  The question is why is it being used?  What at_exit handler are we 
trying to avoid.

Regards,

Anthony Liguori

> from the man page :
>
>        The function _exit() is like exit(3), but does not call  any  functions
>        registered  with  atexit(3) or on_exit(3).  Whether it flushes standard
>        I/O buffers and removes temporary  files  created  with  tmpfile(3)  is
>        implementation-dependent.   On  the other hand, _exit() does close open
>        file descriptors, and this may cause  an  unknown  delay,  waiting  for
>        pending  output to finish.  If the delay is undesired, it may be useful
>        to call functions like tcflush(3) before calling _exit().  Whether  any
>        pending  I/O  is  canceled,  and which pending I/O may be canceled upon
>        _exit(), is implementation-dependent.
>
>
>
>
>
>   

Re: [Qemu-devel] [PATCH] flush stdout after printing usage()

[Riku Voipio]

On Wed, Jan 28, 2009 at 04:00:09PM -0600, Anthony Liguori wrote:
> Lionel Landwerlin wrote:
>>> On Wed, Jan 28, 2009 at 02:10:42PM -0600, Anthony Liguori wrote:
>>>> Riku Voipio wrote:

>>>>> Without fflush() getting output from piped qemu is a bit random.
>>>>>         
>>>> It should get flushed upon exit(). Perhaps the problem is that 
>>>> we're using _exit() instead of exit()?
>>>>       

>> _exit does not trigger callbacks registred with at_exit.

> Yes.  The question is why is it being used?  What at_exit handler are we  
> trying to avoid.

Historic reasons? Just to "be safe" since system qemu uses atexit handlers?
I don't see any atexit handlers registered for linux-user (or bsd-user)
targets for current svn qemu. Definetly none registered at command line parsing
stage.