[ivtv-devel] [PATCH] saa717x.c for after 0.8.0

Kyuma Ohta whatisthis at jcom.home.ne.jp
Fri Nov 3 01:22:57 CET 2006


Hi,

Now,I tested this saa717x patch without semaphore over 4days
with SVN 3495, any accident has not happened.
Parhaps,this accident is because of uncomplete initalize of saa7173.

Sorry for missing.

K.Ohta.
--

On Sat, 28 Oct 2006 02:13:10 +0900
Kyuma Ohta <whatisthis at jcom.home.ne.jp> wrote:

> On Fri, 27 Oct 2006 11:19:55 +0200
> Hans Verkuil <hverkuil at xs4all.nl> wrote:
> 
> > On Thursday 26 October 2006 10:57, Kyuma Ohta wrote:
> > > Hello;)
> > >
> > > On Thu, 26 Oct 2006 00:42:09 +0200
> > >
> > > Hans Verkuil <hverkuil at xs4all.nl> wrote:
> > > > Thank you for the patch, but I'm afraid that it needs some more
> > > > work before I'll apply it:
> > > >
> > > > On Wednesday 25 October 2006 19:51, Kyuma Ohta wrote:
> > > > > Hi,
> > > > >   I made patch for saa717x(7173/7174) driver after 0.8.0.
> > > > >  (which includes one line patch for able to build ivtv driver on
> > > > > 2.6.18 and above.) *Changes:
> > > > >      * Apply after Vanilla 2.6.18 : replace linux/version.h to
> > > > > linux/autoconf.h * Fix endian multibytes value on writing tables.
> > > > >     * Fix writing algorythm.
> > > >
> > > > Is the schedule_timeout really needed? And are the retries needed
> > > > at all?
> > >
> > > Parhaps,schedule_timeout is not needed.
> > >
> > > And I think,retrying sequence has needed to send commands via
> > > I2C-Bus,because some slave devices with one bus, and slave devices
> > > running asyncronous. So, I'm afraid to conflic data/command via
> > > I2C-Bus if kernel driver (or master device) has not supported 
> > > exclusion writing.
> > 
> > It's the ivtv driver itself that is responsible for ensuring that you do 
> > not get i2c bus conflicts. So there is no need to bother with that in 
> > the i2c driver itself. The saa7115 module is a good template to see how 
> > an i2c module should work.
> 
> Hmm..I re-think not need to use semaphore from that precedent.
> 
> > 
> > >
> > > > >     * Fix multibytes reading algorythm.
> > > > >     * Using semaphore to write value tables correctry.
> > > >
> > > > Why would you need a semaphore for that? And the semaphore should
> > > > certainly not be a driver global, it should be part of
> > > > saa717x_state instead, or am I missing something?
> > >
> > > Ah,I missed writing standards of driver...sorry...
> > >
> > > > The locking should also be done inside a reg_write function, and
> > > > not duplicated all over.
> > >
> > > When checking initialization of SAA7173 (on CX23416-STVLP
> > > :cardtype=12 ), someone was throwing IOCTL via I2C  occasionally ,
> > > initial sequence of SAA7173 was interrupted another device command
> > > anytime by IOCTL routine.
> > >
> > > So,I thought needs selialization collectry to use semaphore .
> > > If you don't think that,parhaps initialization and running is almost
> > > okay without semaphore in this driver.
> > 
> > Which ioctl is called? That should not happen and it indicates an ivtv 
> > error.
> 
> Sorry,I didn't log when these "accidents" had occured,sorry...
> But,I remember these accidents had occured when 
> Initialization sequence was imcompleted yet.
> I'll restart to test with SVN HEAD any sequences again 
> (parhaps afternoon of next Monday),with logging,without semaphore,
> maybe accedents will be occured. 
> 
> > 
> > Thanks,
> > 
> > 	Hans
> 
>  Thanks and Sorry
> 
> K.Ohta
> 
> _______________________________________________
> ivtv-devel mailing list
> ivtv-devel at ivtvdriver.org
> http://ivtvdriver.org/mailman/listinfo/ivtv-devel



More information about the ivtv-devel mailing list