Fix for breakage caused by kfifo backport

Fix for breakage caused by kfifo backport

[Andy Walls]

Mauro,

At 

http://linuxtv.org/hg/~awalls/cx23885-ir2

I have a change checked in to fix the v4l-dvb compilation breakage for
kernels less than 2.6.33 cause by the kfifo API change.  I have fixed
both the cx23885 and meye driver so they compile again for older
kernels.

All the changes in this repo are OK to PULL as is, even though I haven't
finished all the changes for the TeVii S470 IR  (I was planning on a
PULL request late this evening EST).  You can also just cherry pick the
one that fixes the kfifo problem if you want.

[I was unaware of the timing of the backport, but since it was stopping
me from working, I fixed it as I thought appropriate.  Please feel free
to contact me on any backport changes that have my fingerprints all over
it, with which you would like help.  I'd like to help minimize the
impact to users, testers, and developers, who may not have the bleeding
edge kernel - or at least the impact to me ;) ]


Regards,
Andy

Re: Fix for breakage caused by kfifo backport

[Douglas Schilling Landgraf]

Hello Andy,

On 01/15/2010 10:00 AM, Andy Walls wrote:
> Mauro,
>
> At
>
> http://linuxtv.org/hg/~awalls/cx23885-ir2
>
> I have a change checked in to fix the v4l-dvb compilation breakage for
> kernels less than 2.6.33 cause by the kfifo API change.  I have fixed
> both the cx23885 and meye driver so they compile again for older
> kernels.
>
> All the changes in this repo are OK to PULL as is, even though I haven't
> finished all the changes for the TeVii S470 IR  (I was planning on a
> PULL request late this evening EST).  You can also just cherry pick the
> one that fixes the kfifo problem if you want.
>
> [I was unaware of the timing of the backport, but since it was stopping
> me from working, I fixed it as I thought appropriate.

Ouch, I was expecting to send the pull request for these changes before.
I have created the exactly same patches. Anyway, good that we have these 
backports done.

>    Please feel free
> to contact me on any backport changes that have my fingerprints all over
> it, with which you would like help.  I'd like to help minimize the
> impact to users, testers, and developers, who may not have the bleeding
> edge kernel - or at least the impact to me ;) ]
>
>    

Cheers,
Douglas

Re: Fix for breakage caused by kfifo backport

[Mauro Carvalho Chehab]

Andy Walls wrote:
> Mauro,
> 
> At 
> 
> http://linuxtv.org/hg/~awalls/cx23885-ir2
> 
> I have a change checked in to fix the v4l-dvb compilation breakage for
> kernels less than 2.6.33 cause by the kfifo API change.  I have fixed
> both the cx23885 and meye driver so they compile again for older
> kernels.

As patches that do backports aren't applied upstream, they can't change any
line at the upstream code. However, your patch is changing two comments:

+#if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 33)
+		kfifo_reset(state->rx_kfifo);
+#else
 		unsigned long flags;
 
 		spin_lock_irqsave(&state->rx_kfifo_lock, flags);
 		kfifo_reset(&state->rx_kfifo);
-		/* reset tx_fifo too if there is one... */
 		spin_unlock_irqrestore(&state->rx_kfifo_lock, flags);
+#endif

@@ -977,6 +1000,7 @@
 	o->interrupt_enable = p->interrupt_enable;
 	o->enable = p->enable;
 	if (p->enable) {
+		/* reset tx_fifo here */
 		if (p->interrupt_enable)
 			irqenable_tx(dev, IRQEN_TSE);
 		control_tx_enable(dev, p->enable);
@@ -1256,8 +1280,15 @@


This means that upstream and your -hg will be different. Please, don't do that.
If you want to touch on comments or at the upstream code, please send a separate
patch.

> 
> All the changes in this repo are OK to PULL as is, even though I haven't
> finished all the changes for the TeVii S470 IR  (I was planning on a
> PULL request late this evening EST).  You can also just cherry pick the
> one that fixes the kfifo problem if you want.

Yet, the series contains that issue I've already pointed:
+struct cx23885_ir_input {
...
+       char                    name[48];
+       char                    phys[48];

> 
> [I was unaware of the timing of the backport, but since it was stopping
> me from working, I fixed it as I thought appropriate.  Please feel free
> to contact me on any backport changes that have my fingerprints all over
> it, with which you would like help.  I'd like to help minimize the
> impact to users, testers, and developers, who may not have the bleeding
> edge kernel - or at least the impact to me ;) ]

I do the backports when I'm about to prepare patches to submit upstream, generally
after doing a large patch merge.

That's said, maintainng both upstream and the backports are consuming a lot
of my time. I'll probably pass the task of keeping the -hg tree with the
backports to somebody else receiving and applying patches directly on my
-git tree.

Douglas already offered to do it, so I'll likely pass him the task to maintain
the -hg tree soon. Then, all patches there will be simply a backport of what
we'll have on -git.

Cheers,
Mauro.

Re: Fix for breakage caused by kfifo backport

[Andy Walls]

On Fri, 2010-01-15 at 17:47 -0200, Mauro Carvalho Chehab wrote:
> Andy Walls wrote:
> > Mauro,
> > 
> > At 
> > 
> > http://linuxtv.org/hg/~awalls/cx23885-ir2
> > 
> > I have a change checked in to fix the v4l-dvb compilation breakage for
> > kernels less than 2.6.33 cause by the kfifo API change.  I have fixed
> > both the cx23885 and meye driver so they compile again for older
> > kernels.
> 
> As patches that do backports aren't applied upstream, they can't change any
> line at the upstream code. However, your patch is changing two comments:


> This means that upstream and your -hg will be different. Please, don't do that.
> If you want to touch on comments or at the upstream code, please send a separate
> patch.

Ooops.  OK, I didn't consider that.  Thanks.  I guess Douglas' change
will be the fix then.


> > All the changes in this repo are OK to PULL as is, even though I haven't
> > finished all the changes for the TeVii S470 IR  (I was planning on a
> > PULL request late this evening EST).  You can also just cherry pick the
> > one that fixes the kfifo problem if you want.
> 
> Yet, the series contains that issue I've already pointed:
> +struct cx23885_ir_input {
> ...
> +       char                    name[48];
> +       char                    phys[48];
> 

Mauro Carvalho Chehab wrote in another thread:
> >> Why are you creating a name[] and phys[] chars here? It should be using the names already
> >> defined at struct input_dev.

-ENOMEM

>From linux/include/linux/input.h:

        struct input_dev {
                const char *name;
                const char *phys;
        [...]

My previous full response is here in case it got lost:

http://article.gmane.org/gmane.linux.drivers.video-input-infrastructure/14745

Please advise if this is still an issue, or if there is some other
storage that you were thinking about.

Regards,
Andy