Re: [Xenomai-core] [patch 2/4] RTDM support for select-like service.

[Jan Kiszka <jan.kiszka@domain.hid>]

[-- Attachment #1: Type: text/plain, Size: 2613 bytes --]

Gilles Chanteperdrix wrote:
> Jan Kiszka wrote:
>  > Gilles Chanteperdrix wrote:
>  > > Jan Kiszka wrote:
>  > >  > Gilles Chanteperdrix wrote:
>  > >  > > stats:
>  > >  > >  include/rtdm/rtdm.h        |    4 ++
>  > >  > >  include/rtdm/rtdm_driver.h |   30 +++++++++++++++-
>  > >  > >  ksrc/skins/rtdm/core.c     |   84 +++++++++++++++++++++++++++++++--------------
>  > >  > >  ksrc/skins/rtdm/device.c   |   10 +++++
>  > >  > >  ksrc/skins/rtdm/drvlib.c   |   76 ++++++++++++++++++++++++++++++++++++----
>  > >  > >  5 files changed, 170 insertions(+), 34 deletions(-)
>  > >  > > 
>  > >  > 
>  > >  > /me thinks that going for inline patches on this list is overdue...
>  > >  > 
>  > >  > > @@ -1148,8 +1208,8 @@ int rtdm_sem_timeddown(rtdm_sem_t *sem, 
>  > >  > >  
>  > >  > >  	if (testbits(sem->synch_base.status, RTDM_SYNCH_DELETED))
>  > >  > >  		err = -EIDRM;
>  > >  > > -	else if (sem->value > 0)
>  > >  > > -		sem->value--;
>  > >  > > +	else if (sem->value > 0 && !--sem->value)
>  > >  > > +			xnselect_signal(&sem->select_block, 0);
>  > >  > >  	else if (timeout < 0) /* non-blocking mode */
>  > >  > >  		err = -EWOULDBLOCK;
>  > >  > >  	else {
>  > >  > 
>  > >  > Uuuh, this doesn't look equivalent (for --sem->value < 0).
>  > >  > 
>  > >  > Otherwise it's OK with me.
>  > > 
>  > > After testing, it appears that the "owner" of a file descriptor changed from 
>  > > current->mm 
>  > > to
>  > > container_of(xnshadow_ppd_get(__rtdm_muxid), struct rtdm_process, ppd)
>  > 
>  > Yes, indeed, forgot about it. We have a full-blown owner structure now
>  > to track comm and pid.
>  > 
>  > > 
>  > > So, I now need a ppd hash lookup in __rtdm_context_get.
>  > > 
>  > 
>  > You could add mm to rtdm_process. This means another indirection and
>  > makes things worse (more cache misses in the worst case...).
> 
> Would not it be simpler to put a pointer to the task_struct ? After all,
> it already has a pid, comm and mm, and a file descriptor will not
> survive a task_struct thanks to automatic closing of file descriptors.

Hmm, hmm, hmmmm... Sounds reasonable, should be safe.

> 
>  Could you
>  > live without the check until we have per-process fd tabled, or was it
>  > essential for the select thing?
> 
> An application which I ported to Xenomai (and which uses the select
> call) closes all file descriptors in a for loop. The purpose of this
> loop is, I guess, to avoid leaving a file descriptor opened that was
> passed through exec.

OK.

So, will you change rtdm_process too? Thanks.

Jan


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 254 bytes --]