[PATCH] dvb-core: fix initialization of feeds list in demux filter

[PATCH] dvb-core: fix initialization of feeds list in demux filter

[Francesco Lavra]

A DVB demultiplexer device can be used to set up either a PES filter or
a section filter. In the former case, the ts field of the feed union of
struct dmxdev_filter is used, in the latter case the sec field of the
same union is used.
The ts field is a struct list_head, and is currently initialized in the
open() method of the demux device. When for a given demuxer a section
filter is set up, the sec field is played with, thus if a PES filter
needs to be set up after that the ts field will be corrupted, causing a
kernel oops.
This fix moves the list head initialization to
dvb_dmxdev_pes_filter_set(), so that the ts field is properly
initialized every time a PES filter is set up.

Signed-off-by: Francesco Lavra <francescolavra@interfree.it>
Cc: stable <stable@kernel.org>
---

--- a/drivers/media/dvb/dvb-core/dmxdev.c	2010-02-07 13:19:18.000000000 +0100
+++ b/drivers/media/dvb/dvb-core/dmxdev.c	2010-02-07 13:23:39.000000000 +0100
@@ -761,7 +761,6 @@ static int dvb_demux_open(struct inode *
 	dvb_ringbuffer_init(&dmxdevfilter->buffer, NULL, 8192);
 	dmxdevfilter->type = DMXDEV_TYPE_NONE;
 	dvb_dmxdev_filter_state_set(dmxdevfilter, DMXDEV_STATE_ALLOCATED);
-	INIT_LIST_HEAD(&dmxdevfilter->feed.ts);
 	init_timer(&dmxdevfilter->timer);
 
 	dvbdev->users++;
@@ -887,6 +886,7 @@ static int dvb_dmxdev_pes_filter_set(str
 	dmxdevfilter->type = DMXDEV_TYPE_PES;
 	memcpy(&dmxdevfilter->params, params,
 	       sizeof(struct dmx_pes_filter_params));
+	INIT_LIST_HEAD(&dmxdevfilter->feed.ts);
 
 	dvb_dmxdev_filter_state_set(dmxdevfilter, DMXDEV_STATE_SET);
 

Re: [PATCH] dvb-core: fix initialization of feeds list in demux filter

[Andy Walls]

On Sun, 2010-02-07 at 13:49 +0100, Francesco Lavra wrote:
> A DVB demultiplexer device can be used to set up either a PES filter or
> a section filter. In the former case, the ts field of the feed union of
> struct dmxdev_filter is used, in the latter case the sec field of the
> same union is used.
> The ts field is a struct list_head, and is currently initialized in the
> open() method of the demux device. When for a given demuxer a section
> filter is set up, the sec field is played with, thus if a PES filter
> needs to be set up after that the ts field will be corrupted, causing a
> kernel oops.
> This fix moves the list head initialization to
> dvb_dmxdev_pes_filter_set(), so that the ts field is properly
> initialized every time a PES filter is set up.
> 
> Signed-off-by: Francesco Lavra <francescolavra@interfree.it>
> Cc: stable <stable@kernel.org>


Reviewed-by: Andy Walls <awalls@radix.net>

This certainly helps.

It catches the case of setting a Section filter with DMX_SET_FILTER and
then a PES filter with DMX_SET_PES_FILTER on the same file descriptor.
This is exactly what alevt does.  See vbi.c:vbi_dvb_open() and
vbi.c:dvb_get_table() in the alevt source code.


Regards,
Andy

> ---
> 
> --- a/drivers/media/dvb/dvb-core/dmxdev.c	2010-02-07 13:19:18.000000000 +0100
> +++ b/drivers/media/dvb/dvb-core/dmxdev.c	2010-02-07 13:23:39.000000000 +0100
> @@ -761,7 +761,6 @@ static int dvb_demux_open(struct inode *
>  	dvb_ringbuffer_init(&dmxdevfilter->buffer, NULL, 8192);
>  	dmxdevfilter->type = DMXDEV_TYPE_NONE;
>  	dvb_dmxdev_filter_state_set(dmxdevfilter, DMXDEV_STATE_ALLOCATED);
> -	INIT_LIST_HEAD(&dmxdevfilter->feed.ts);
>  	init_timer(&dmxdevfilter->timer);
>  
>  	dvbdev->users++;
> @@ -887,6 +886,7 @@ static int dvb_dmxdev_pes_filter_set(str
>  	dmxdevfilter->type = DMXDEV_TYPE_PES;
>  	memcpy(&dmxdevfilter->params, params,
>  	       sizeof(struct dmx_pes_filter_params));
> +	INIT_LIST_HEAD(&dmxdevfilter->feed.ts);
>  
>  	dvb_dmxdev_filter_state_set(dmxdevfilter, DMXDEV_STATE_SET);
>  
> 
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-media" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> 

Re: [PATCH] dvb-core: fix initialization of feeds list in demux filter

[hermann pitton]

Hi!

Am Sonntag, den 07.02.2010, 10:13 -0500 schrieb Andy Walls:
> On Sun, 2010-02-07 at 13:49 +0100, Francesco Lavra wrote:
> > A DVB demultiplexer device can be used to set up either a PES filter or
> > a section filter. In the former case, the ts field of the feed union of
> > struct dmxdev_filter is used, in the latter case the sec field of the
> > same union is used.
> > The ts field is a struct list_head, and is currently initialized in the
> > open() method of the demux device. When for a given demuxer a section
> > filter is set up, the sec field is played with, thus if a PES filter
> > needs to be set up after that the ts field will be corrupted, causing a
> > kernel oops.
> > This fix moves the list head initialization to
> > dvb_dmxdev_pes_filter_set(), so that the ts field is properly
> > initialized every time a PES filter is set up.
> > 
> > Signed-off-by: Francesco Lavra <francescolavra@interfree.it>
> > Cc: stable <stable@kernel.org>
> 
> 
> Reviewed-by: Andy Walls <awalls@radix.net>
Tested-by: hermann pitton <hermann-pitton@arcor.de>

> 
> This certainly helps.

Yes, as already confirmed by Uwe.

> It catches the case of setting a Section filter with DMX_SET_FILTER and
> then a PES filter with DMX_SET_PES_FILTER on the same file descriptor.
> This is exactly what alevt does.  See vbi.c:vbi_dvb_open() and
> vbi.c:dvb_get_table() in the alevt source code.
> 
> 
> Regards,
> Andy
> 
> > ---
> > 
> > --- a/drivers/media/dvb/dvb-core/dmxdev.c	2010-02-07 13:19:18.000000000 +0100
> > +++ b/drivers/media/dvb/dvb-core/dmxdev.c	2010-02-07 13:23:39.000000000 +0100
> > @@ -761,7 +761,6 @@ static int dvb_demux_open(struct inode *
> >  	dvb_ringbuffer_init(&dmxdevfilter->buffer, NULL, 8192);
> >  	dmxdevfilter->type = DMXDEV_TYPE_NONE;
> >  	dvb_dmxdev_filter_state_set(dmxdevfilter, DMXDEV_STATE_ALLOCATED);
> > -	INIT_LIST_HEAD(&dmxdevfilter->feed.ts);
> >  	init_timer(&dmxdevfilter->timer);
> >  
> >  	dvbdev->users++;
> > @@ -887,6 +886,7 @@ static int dvb_dmxdev_pes_filter_set(str
> >  	dmxdevfilter->type = DMXDEV_TYPE_PES;
> >  	memcpy(&dmxdevfilter->params, params,
> >  	       sizeof(struct dmx_pes_filter_params));
> > +	INIT_LIST_HEAD(&dmxdevfilter->feed.ts);
> >  
> >  	dvb_dmxdev_filter_state_set(dmxdevfilter, DMXDEV_STATE_SET);
> >  
> > 

Many thanks to Francesco and all taking it up with this one.

Cheers,
Hermann

Re: [PATCH] dvb-core: fix initialization of feeds list in demux filter

[Mauro Carvalho Chehab]

Francesco Lavra wrote:
> A DVB demultiplexer device can be used to set up either a PES filter or
> a section filter. In the former case, the ts field of the feed union of
> struct dmxdev_filter is used, in the latter case the sec field of the
> same union is used.
> The ts field is a struct list_head, and is currently initialized in the
> open() method of the demux device. When for a given demuxer a section
> filter is set up, the sec field is played with, thus if a PES filter
> needs to be set up after that the ts field will be corrupted, causing a
> kernel oops.
> This fix moves the list head initialization to
> dvb_dmxdev_pes_filter_set(), so that the ts field is properly
> initialized every time a PES filter is set up.
> 
> Signed-off-by: Francesco Lavra <francescolavra@interfree.it>
> Cc: stable <stable@kernel.org>
> ---
> 
> --- a/drivers/media/dvb/dvb-core/dmxdev.c	2010-02-07 13:19:18.000000000 +0100
> +++ b/drivers/media/dvb/dvb-core/dmxdev.c	2010-02-07 13:23:39.000000000 +0100
> @@ -761,7 +761,6 @@ static int dvb_demux_open(struct inode *
>  	dvb_ringbuffer_init(&dmxdevfilter->buffer, NULL, 8192);
>  	dmxdevfilter->type = DMXDEV_TYPE_NONE;
>  	dvb_dmxdev_filter_state_set(dmxdevfilter, DMXDEV_STATE_ALLOCATED);
> -	INIT_LIST_HEAD(&dmxdevfilter->feed.ts);
>  	init_timer(&dmxdevfilter->timer);
>  
>  	dvbdev->users++;
> @@ -887,6 +886,7 @@ static int dvb_dmxdev_pes_filter_set(str
>  	dmxdevfilter->type = DMXDEV_TYPE_PES;
>  	memcpy(&dmxdevfilter->params, params,
>  	       sizeof(struct dmx_pes_filter_params));
> +	INIT_LIST_HEAD(&dmxdevfilter->feed.ts);
>  
>  	dvb_dmxdev_filter_state_set(dmxdevfilter, DMXDEV_STATE_SET);
>  

Good catch, but it seems better to initialize both the mutex and the list head
at dvb_dmx_dev_init. Please test if the following patch fixes the issue. If so, please
sign.

Cheers,
Mauro

diff --git a/drivers/media/dvb/dvb-core/dmxdev.c b/drivers/media/dvb/dvb-core/dmxdev.c
index c37790a..be9bef1 100644
--- a/drivers/media/dvb/dvb-core/dmxdev.c
+++ b/drivers/media/dvb/dvb-core/dmxdev.c
@@ -755,13 +755,11 @@ static int dvb_demux_open(struct inode *inode, struct file *file)
 	}
 
 	dmxdevfilter = &dmxdev->filter[i];
-	mutex_init(&dmxdevfilter->mutex);
 	file->private_data = dmxdevfilter;
 
 	dvb_ringbuffer_init(&dmxdevfilter->buffer, NULL, 8192);
 	dmxdevfilter->type = DMXDEV_TYPE_NONE;
 	dvb_dmxdev_filter_state_set(dmxdevfilter, DMXDEV_STATE_ALLOCATED);
-	INIT_LIST_HEAD(&dmxdevfilter->feed.ts);
 	init_timer(&dmxdevfilter->timer);
 
 	dvbdev->users++;
@@ -1239,6 +1237,8 @@ int dvb_dmxdev_init(struct dmxdev *dmxdev, struct dvb_adapter *dvb_adapter)
 		dmxdev->filter[i].buffer.data = NULL;
 		dvb_dmxdev_filter_state_set(&dmxdev->filter[i],
 					    DMXDEV_STATE_FREE);
+		mutex_init(&dmxdev->filter[i].mutex);
+		INIT_LIST_HEAD(&dmxdev->filter[i].feed.ts);
 	}
 
 	dvb_register_device(dvb_adapter, &dmxdev->dvbdev, &dvbdev_demux, dmxdev,



Cheers,
Mauro

Re: [PATCH] dvb-core: fix initialization of feeds list in demux filter

[Andreas Oberritter]

Hello Mauro,

Mauro Carvalho Chehab wrote:
> Good catch, but it seems better to initialize both the mutex and the list head
> at dvb_dmx_dev_init. Please test if the following patch fixes the issue. If so, please
> sign.

please apply Francesco's original patch. Yours won't work, because
"feed" is a union. It must be initialized each time DMX_SET_PES_FILTER
gets called, because the memory might have been overwritten by a
previous call to DMX_SET_FILTER, which uses "feed.sec".

Regards,
Andreas

Re: [PATCH] dvb-core: fix initialization of feeds list in demux filter (Was: Videotext application crashes the kernel due to DVB-demux patch)

[Chicken Shack]

Am Montag, den 08.02.2010, 13:24 +0100 schrieb Andreas Oberritter:
> Hello Mauro,
> 
> Mauro Carvalho Chehab wrote:
> > Good catch, but it seems better to initialize both the mutex and the list head
> > at dvb_dmx_dev_init. Please test if the following patch fixes the issue. If so, please
> > sign.
> 
> please apply Francesco's original patch. Yours won't work, because
> "feed" is a union. It must be initialized each time DMX_SET_PES_FILTER
> gets called, because the memory might have been overwritten by a
> previous call to DMX_SET_FILTER, which uses "feed.sec".
> 
> Regards,
> Andreas

Now if I were a cynical or ranter or another kind of dumb primitive
persona non grata I would just add "Lol" or stuff like that and turn
myself away.

But this is no fun here.

It's nothing but a big proof that one Brazilian person in Mr. Torvalds
"dream team of untouchables" needs to be URGENTLY replaced by another
real capable person.

NO IDEA ABOUT DVB ISSUES, BUT DVB MAINTAINER!

This is a SCANDAL, not fun! This is SCANDALOUS!

1. When you start to complain then this moron asks you why you are so
late.
2. Instead of organizing or really trying to win people with the
necessary skills he starts dumbest silliest thinkable flame wars with
people who are not 100 % conform with him. Thus he does not stop the
bleeding and runaway processes of real interested persons who are
urgently needed. On the contrary he accelerates those processes.
3. When the situation was in balance and no help was in sight I decided
to organize help from far outside the list.
I was lucky finding Francesco Lavra. If there hadn't been him, endless
effectless ranting would still be the way to go, and the kernel
regression of a stable kernel would persist (!).
In this situation our Brazilian counterproductive seat farter starts to
demotivate you by: "It's too late to write compat levels now, we're too
late in the release circle and blablablablahhhh Blubblubblubblub.....
Blather blather blather blather.........

4. Before the decisive patch was there our completely incapable
Brazilian spurked out some completely ineffective and silly
pseudo-patches who for each did not even touch the problem.
Thus he did de facto nothing than stealing my time, attacking my humble
nerves. The rule of our Brazilian: Even if I am a moron knowing
absolutely nothing I must PRETEND some brainless activism. So show is
everything, while manpower is being chased away.
Competitive manpower is dangerous for the Brazilian pretender.

5. And when the decisive patch is there he wastes it for a reason that
only he knows. Thanks to Andreas we all know now.

That is exactly what he did with Markus Rechbergers contributions:
He broke them by transforming them in a definitely unqualified manner.
Until Markus ran away. With the effect that the Empia stuff is being
hosted elsewhere now.
Mauro Carvalho Chehab's behaviour and policy is so goddamn dumb,
ineffective, destructive, .....
It is a secure way to put a project finally over the edge which has got
real structural problems ....

I want him to be replaced as soon as possible by a real fair, qualified
and capable person. Every second that one tolerates Mauro Carvalho
Chehab as a common V4l / DVB maintainer is a second of pain and
catastrophe.

I vote for Andy Walls, Devin Heitmueller. There are a couple of capable
candidates here who can really do that job with a better output for the
whole Linux community.
There is ABSOLUTELY NO necessity to continue with Mauro Carvalho Chehab.

I cced all relevant persons. I want Chehab away from here! It's enough
now. My tolerance is at zero. PERIOD!

CS

P. S.: If the RedHat company can afford people like MCC then there is
something wrong with the internal structures of RedHat.

Re: [PATCH] dvb-core: fix initialization of feeds list in demux filter (Was: Videotext application crashes the kernel due to DVB-demux patch)

[Linus Torvalds]

On Mon, 8 Feb 2010, Chicken Shack wrote:
> 
> This is a SCANDAL, not fun! This is SCANDALOUS!

I agree that this whole thread has been totally inappropriate from 
beginning to end. 

On all sides, btw. I would suggest that somebody who can't even use his 
own name should be the last to complain about other peoples behavior. I 
have suspicions that the reason you're not using your own name and instead 
go by "Chicken Shack" is that you're one of the DVB people that get 
ignored because everybody knows you always just argue mindlessly, so now 
you use made-up accounts just to not be immediately ignored.

And if that is the case, you should take a hard look at your _own_ 
behavior, and ask yourself why you get ignored on multiple accounts. When 
people ignore your bug reports, maybe it's because they know that the pain 
of interacting with you is simply NOT WORTH IT?

Anyway, the amount of just this kind of pure _crap_ in the whole DVB 
development environment is scary. There are no other areas of Linux kernel 
development where we see this kind of personal and _pointless_ invective.

I realize that I probably see the absolute sewer of the whole discussion, 
since people seem to Cc me only when things are going badly, but _still_. 
I don't see that kind of childish behavior anywhere else in kernel 
development.

Guys, get a f*cking grip already.

Here's the rule:

 - if somebody reports a regression, IT MUST BE FIXED. Not "discussed" for 
   two weeks. Not talked around. If we have a bisected commit that starts 
   oopsing with an existing setup that used to work, there is no 
   discussion. That commit absolutely _has_ to be reverted or fixed. No 
   ifs, buts, or maybes about it.

   Anybody who argues anything else is simply totally wrong. And 
   discussing various semantic changes or asking people to use other 
   programs instead of the one that causes the problem is totally 
   irrelevant until the oops has been fixed.

   An oops is not acceptable in _any_ situation, and saying that the user 
   program is doing something wrong is totally ludicrous. So is breaking a 
   program that used to work.

The fix, btw, for those that haven't seen it, seems to be here:

	http://patchwork.kernel.org/patch/77615/

and people should look at that fix, the explanation, and feel embarrassed 
about th pure crazy invective that has been going on in this thread. It 
was a simple bug, nothing more. It was not worth the intense level of 
craziness seen in this thread.

Btw, there's a very real reason I haven't replied to this thread before: I 
don't like the DVB development process. I've seen the crazyness before, 
and I don't know where it comes from. Some people blame Mauro, but the 
thing is, this whole problem with the DVB development tone used to be 
_worse_. 

It's likely some deep social reason. I have a few suspicions, but they are 
just guesses. For example, I suspect it's partly because DVB gets a much 
less wide development audience - there are clearly some people in the DVB 
driver area that are totally unable to see anything but their own small 
area, and they are too focused on just some detail, and then they get 
upset when others care about other details or about the big picture.

I admit that I used to think that it was because a lot of the people 
involved were Germans, and that the whole mindless bickering was somehow 
cultural. A _lot_ of the worst invective comes from exactly places like 
'gmx.de', which seems to be one of the big mail servers in Germany. But 
there are lots of _good_ developers with that address too, so I suspect my 
"Oh, it's the Germans" explanation was just a funny/sad prejudice.

Why can't DVB people be sane? Please, guys?

And btw, don't in any way think that I think that the problem has just 
been that people have had a hard time just admitting that that commit 
that caused the oops need to be fixed. One of the reasons I think people 
had a hard time fixing the regression was that the original bug report, 
and all the invective was all just totally hiding the real issue.

Instead of making a nice bugzilla and talking about the technical problem, 
the whole beginning of this thread from Chicken Shack has been all about 
shouting at people.

I'm perfectly happy with shouting at people (I'm doing it right now), but 
damn, this thread has just been totally incoherent and stupid. Put some 
real _technical_ reasoning in your crazy rants. 

So damn it, stop cc'ing me on DVB issues. Chicken Shack - I very much 
mean you. I'm not interested in crazy bickering. If you Cc me, include 
proper technical details, and keep it at that level.

And Mauro &co: get me that fix, and stop discussing anything else until 
the actual oops has been fixed.

Everybody: start acting like adults, guys.

			Linus

Re: [PATCH] dvb-core: fix initialization of feeds list in demux filter (Was: Videotext application crashes the kernel due to DVB-demux patch)

[Mauro Carvalho Chehab]

Linus Torvalds wrote:

>  - if somebody reports a regression, IT MUST BE FIXED. Not "discussed" for 
>    two weeks. Not talked around. If we have a bisected commit that starts 
>    oopsing with an existing setup that used to work, there is no 
>    discussion. That commit absolutely _has_ to be reverted or fixed. No 
>    ifs, buts, or maybes about it.
>    Anybody who argues anything else is simply totally wrong. And 
>    discussing various semantic changes or asking people to use other 
>    programs instead of the one that causes the problem is totally 
>    irrelevant until the oops has been fixed.
> 
>    An oops is not acceptable in _any_ situation, and saying that the user 
>    program is doing something wrong is totally ludicrous. So is breaking a 
>    program that used to work.
> 
> The fix, btw, for those that haven't seen it, seems to be here:
> 
> 	http://patchwork.kernel.org/patch/77615/

Yes, this patch actually fixed the OOPS, although it were a report From Chicken
saying that a previous patch got fixed it (http://patchwork.kernel.org/patch/76071/).

I submitted you both fixes, plus a third one (http://patchwork.kernel.org/patch/76083/)
for a potential usage of vmalloc during interrupt time.


Cheers,
Mauro

Re: [PATCH] dvb-core: fix initialization of feeds list in demux filter (Was: Videotext application crashes the kernel due to DVB-demux patch)

[Chicken Shack]

Am Montag, den 08.02.2010, 08:14 -0800 schrieb Linus Torvalds:
> 
> On Mon, 8 Feb 2010, Chicken Shack wrote:
> > 
> > This is a SCANDAL, not fun! This is SCANDALOUS!
> 
> I agree that this whole thread has been totally inappropriate from 
> beginning to end. 

Yes, as a matter of frustration and the way some people around here deal
with kernel regressions (shrug shoulders, endless discussions etc.).

> On all sides, btw. I would suggest that somebody who can't even use his 
> own name should be the last to complain about other peoples behavior.

Interesting point of view. Would be valid to critically analyze the
gesture behind. "Arrogance" would be the most harmless word behind it.
Sharper evaluations I better avoid here!

>  I 
> have suspicions that the reason you're not using your own name and instead 
> go by "Chicken Shack" is that you're one of the DVB people that get 
> ignored because everybody knows you always just argue mindlessly, so now 
> you use made-up accounts just to not be immediately ignored.

Richard Stallman told me:

"Banning you as a punishment seems foolishly harsh, too."

It's you, Linus Torvalds who makes those foolishly harsh punishments
possible. They are very equivalent to a fallback in the middle ages when
a term called "bourgeois rights" wasn't even utopic.

> And if that is the case, you should take a hard look at your _own_ 
> behavior, and ask yourself why you get ignored on multiple accounts. When 
> people ignore your bug reports, maybe it's because they know that the pain 
> of interacting with you is simply NOT WORTH IT?

Well thanks for the flowers, from somebody who goes like "I see that
there is an issue, but I do not comprehend or know the what and why."

Very helpful, very intelligent, isn't it? And how constructive!?

> Anyway, the amount of just this kind of pure _crap_ in the whole DVB 
> development environment is scary. There are no other areas of Linux kernel 
> development where we see this kind of personal and _pointless_ invective.

It was not not pointless at all. It was a kernel regression, and I found
the kernel patch responsive for that.
The way this regression is dealt with here is the only thing that is
"pointless".

> I realize that I probably see the absolute sewer of the whole discussion, 
> since people seem to Cc me only when things are going badly, but _still_. 
> I don't see that kind of childish behavior anywhere else in kernel 
> development.
> 
> Guys, get a f*cking grip already.
> 
> Here's the rule:
> 
>  - if somebody reports a regression, IT MUST BE FIXED. Not "discussed" for 
>    two weeks. Not talked around. If we have a bisected commit that starts 
>    oopsing with an existing setup that used to work, there is no 
>    discussion. That commit absolutely _has_ to be reverted or fixed. No 
>    ifs, buts, or maybes about it.

Go tell that to guys like Devin Heitmueller. Not me!

I really do not know for what kind of ideas or ideals people like him do
struggle for when they start hacking here.
But the "standard" answers you get when you come up with that first rule
above, are:
1. Sorry, I am doing this only in my spare time, without getting paid.
2. Variation of 1.: Sorry, I am short in time.
3. You cannot expect anything unless you pay with a cheque.

I am really wondering for what goddamn reason they keep on repeating
this continuously.
But when it's personally THEM to cause regressions, to fuck up the code,
to break peripheral compatibilities, THEN, YES, THEN you do not hear any
excuses for being short in time, or for being paid.

I call that behaviour asocial, irresponsible, inacceptable.
How would you call it, Linus Torvalds?

Read Heitmueller's contribution and you know what this kiddy thinks and
does not think.

>    Anybody who argues anything else is simply totally wrong. And 
>    discussing various semantic changes or asking people to use other 
>    programs instead of the one that causes the problem is totally 
>    irrelevant until the oops has been fixed.

Here we are! Fully accepted!

>    An oops is not acceptable in _any_ situation, and saying that the user 
>    program is doing something wrong is totally ludicrous. So is breaking a 
>    program that used to work.
> 
> The fix, btw, for those that haven't seen it, seems to be here:
> 
> 	http://patchwork.kernel.org/patch/77615/
> 
> and people should look at that fix, the explanation, and feel embarrassed 
> about th pure crazy invective that has been going on in this thread. It 
> was a simple bug, nothing more. It was not worth the intense level of 
> craziness seen in this thread.

Yes, that is really a shame.

> Btw, there's a very real reason I haven't replied to this thread before: I 
> don't like the DVB development process. I've seen the crazyness before, 
> and I don't know where it comes from.

I am breathless. I simply cannot believe it!
Ok! You like strict guidelines?

Here they are:

1. There is one guy for instance who, from time to time, really manages
to develop something valid: for instance a "Mantis" driver.
But behind the guidelines this man is doing nothing but pure agitation
of the worst kind: throwing stones and preparing flame wars. Insane!

2. Mauro maybe well in V4l terms, his coding experience I would call
mediocre, but as a maintainer of it all hemixes his hands in affairs
that he has no idea about. And that's DVB.

There should be a second maintainer for DVB only like it was
traditionally when this whole stuff was nearly completely in German
hands (I am talking about the convergence aera).

Mauro as the only one for all - that is highly insane.
It simply does not work!

>  Some people blame Mauro, but the 
> thing is, this whole problem with the DVB development tone used to be 
> _worse_. 

When was it _worse_? What do you mean?

> It's likely some deep social reason. I have a few suspicions, but they are 
> just guesses. For example, I suspect it's partly because DVB gets a much 
> less wide development audience - there are clearly some people in the DVB 
> driver area that are totally unable to see anything but their own small 
> area, and they are too focused on just some detail, and then they get 
> upset when others care about other details or about the big picture.

Sounds warm, but isn't hot at all. One possible reason, not more.

> I admit that I used to think that it was because a lot of the people 
> involved were Germans, and that the whole mindless bickering was somehow 
> cultural. A _lot_ of the worst invective comes from exactly places like 
> 'gmx.de', which seems to be one of the big mail servers in Germany. But 
> there are lots of _good_ developers with that address too, so I suspect my 
> "Oh, it's the Germans" explanation was just a funny/sad prejudice.

Yes it is. Guess it's _your_ personal mind a point of view which is
short-minded here. And if especially you regard "Mister Mantis" from
India you should be clear that "mindless bickering" is not something
specifically German out of _cultural_ terms.

I do not say either that a _social competence below zero_ is something
typical appearing very often within Swedish minorities who are growing
up in some very provincial angle of Finland, or do I say that, little
boy?


> Why can't DVB people be sane? Please, guys?
> 
> And btw, don't in any way think that I think that the problem has just 
> been that people have had a hard time just admitting that that commit 
> that caused the oops need to be fixed. One of the reasons I think people 
> had a hard time fixing the regression was that the original bug report, 
> and all the invective was all just totally hiding the real issue.

Pure nonsense.
I was reporting the symptoms of the crash - but noone listened.
I was again reporting the crash plus the patch that caused it - this
time a little louder, because noone listened at the first time. And then
it was Andy Walls asking me to send in a complete syslog.
So Andy Walls was nearly the only one here reacting as expected.
Picking up the issue, at least trying to perceive the technical problem.

There were, among my personal entries, two user requests:

Emil Meyer and Thomas Voegtle.
Emil Meyer's was completely ignored.

> Instead of making a nice bugzilla and talking about the technical problem, 
> the whole beginning of this thread from Chicken Shack has been all about 
> shouting at people.

That has perhaps to do with my kind of approach:
I directly mailed Andreas Oberritter, the author of the patch causing
the regression in 2.6.32.
His response was one of these typical asocial and insane behaviour types
of disgusting and polluted persons:
_Promises are made not to be kept_
He went. I am short in time. If I do not get a notice from me during the
next 2 weeks please mail me again.

5 mails followed - unanswered. Too lazy and not willing to take the
responsibility - that's the insane state of mind behind.

Hope you comprehend now how foolish and nasty your "scapegoat theory_
is, Mister Torvalds!!!

> I'm perfectly happy with shouting at people (I'm doing it right now), but 
> damn, this thread has just been totally incoherent and stupid. Put some 
> real _technical_ reasoning in your crazy rants. 

What a holy wish. Personal consequences would be more efficient, as you
CANNOT Reduce every issue on a technical one!

When will you get that, Mister Torvalds?

> So damn it, stop cc'ing me on DVB issues. Chicken Shack - I very much 
> mean you. I'm not interested in crazy bickering. If you Cc me, include 
> proper technical details, and keep it at that level.

Is that all??

> And Mauro &co: get me that fix, and stop discussing anything else until 
> the actual oops has been fixed.
> 
> Everybody: start acting like adults, guys.
> 
> 			Linus
> --
> To unsubscribe from this list: send the line "unsubscribe linux-media" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

CS

P. S.: One last hint for Mister Linus Torvalds, the man with the social
competence below zero:

There are people sampled around here who really think that hacking free
software (drivers, applications, anything else) is adequate to freely
producing crap, rubbish, doing everything they ever want to without
taking responsibility if it breaks or damages hardware or software.
The pure dumbness and thickness, the imbearable primitivity and
superficiality in their brains CANNOT BE evaluated via expressions like
_kiddish_ or _childish_ or similar.

The dumbness and thickness, the destructivity, the primitivity and the
superficiality of them mixes up the term "freedom" with a french term
called "Laissez-Faire" which means something like "Do what you want,
without responsiblities, without limits, without laws, without rules."

"And if it does harm - shrug your shoulders!"

Just to make that one quite clear, Linus Torvalds:
This gesture has got NOTHING to do with _acting kiddish_.
And people with that gesture I do hate like I hate the pest and cholera.

No matter if those fuckers claim to have produced 1000 or 0 drivers, no
matter if those motherfuckers claim to have produced 1000 or 0
regressions:
People with that gesture and attitude should immediately fuck off from
here without return ticket. No project needs them! PERIOD!

Re: [PATCH] dvb-core: fix initialization of feeds list in demux filter (Was: Videotext application crashes the kernel due to DVB-demux patch)

[Linus Torvalds]

On Mon, 8 Feb 2010, Chicken Shack wrote:
> ...

Please stop emailing me. I'm simply not interested in your political 
troubles with the DVB people.

I see with my own eyes why people have issues with your way of 
communication, and quite frankly, I'm not interested in discussing 
name-calling with somebody who is anonymous.

		Linus

Re: [PATCH] dvb-core: fix initialization of feeds list in demux filter (Was: Videotext application crashes the kernel due to DVB-demux patch)

[Chicken Shack]

Am Montag, den 08.02.2010, 11:09 -0800 schrieb Linus Torvalds:
> 
> On Mon, 8 Feb 2010, Chicken Shack wrote:
> > ...
> 
> Please stop emailing me. I'm simply not interested in your political 
> troubles with the DVB people.
> 
> I see with my own eyes why people have issues with your way of 
> communication, and quite frankly, I'm not interested in discussing 
> name-calling with somebody who is anonymous.

C R A P ! ! ! !

> 
> 		Linus
> --
> To unsubscribe from this list: send the line "unsubscribe linux-media" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

I admit I completely overestimated you.

HOW POOR!

SHAME ON YOU!

Re: [PATCH] dvb-core: fix initialization of feeds list in demux filter (Was: Videotext application crashes the kernel due to DVB-demux patch)

[Rudy Zijlstra]

Chicken Shack wrote:
ok anonymous chicken

you've managed a distinct first: i've installed a block filter on your 
email address.

Rudy

Re: [PATCH] dvb-core: fix initialization of feeds list in demux filter (Was: Videotext application crashes the kernel due to DVB-demux patch)

[Chicken Shack]

Am Dienstag, den 09.02.2010, 17:09 -0500 schrieb Richard Stallman:
> "Banning you as a punishment seems foolishly harsh, too."
> 
> The word "too" is there because, in my previous paragraph, I said it
> seemed you had been overly harsh in your criticism.  Everyone makes
> mistakes.  Being busy is not malicious.
> 
Yes. Sometimes I lose my temper. If you ask me how to avoid it - I do
not know. I'll keep on trying.

The more important question is:
Is there a way to lift the ban?

NOT to at least show a way is malicious.

It's not that my self-confidence depends on such an idiotic punishment.
But lifting the ban would make things easier (-> stigmatization effects
by people not concerned with the causes of the ban f. ex.).

Thanks

CS

Re: [PATCH] dvb-core: fix initialization of feeds list in demux filter (Was: Videotext application crashes the kernel due to DVB-demux patch)

[hermann pitton]

Am Montag, den 08.02.2010, 08:14 -0800 schrieb Linus Torvalds:
> 
> On Mon, 8 Feb 2010, Chicken Shack wrote:
> > 
> > This is a SCANDAL, not fun! This is SCANDALOUS!
> 
> I agree that this whole thread has been totally inappropriate from 
> beginning to end. 

The initial problem was, how to find such software producing that oops
at all.

Uwe/Chicken only later friendly distributed it to us.

Is there some alevt 1.7.0 (dvb-t/dvb-s :) anywhere else already?

Since only then, we have been able to discover, that it is valid bug
that needs investigation.

Else I do agree with all your findings, you had him blacklisted on LKML
already too, but v4l and dvb people are also somehow forced to work
together, because of the hybrid devices everywhere now.

Some don't like it and he is always on top of it again.

We make progress in so far.

Hermann



 

Re: [PATCH] dvb-core: fix initialization of feeds list in demux filter (Was: Videotext application crashes the kernel due to DVB-demux patch)

[Chicken Shack]

Am Dienstag, den 09.02.2010, 01:53 +0100 schrieb hermann pitton:
> Am Montag, den 08.02.2010, 08:14 -0800 schrieb Linus Torvalds:
> > 
> > On Mon, 8 Feb 2010, Chicken Shack wrote:
> > > 
> > > This is a SCANDAL, not fun! This is SCANDALOUS!
> > 
> > I agree that this whole thread has been totally inappropriate from 
> > beginning to end. 
> 
> The initial problem was, how to find such software producing that oops
> at all.

Pure crap! Because, as I stated already, you are too lazy to read the
context.

Proofs:

1. http://www.spinics.net/lists/linux-media/msg15182.html

This guy was compeletely ignored, although he announced a kernel bug /
security risk. Neither is he banned somewhere, nor are there other ifs
or whatevers to justify just ignoring him.
Also the utmost insane Torvalds-scapegoat-theory dowes not work here at
all. This guy reported a problem, and noone except me listened: Period!

2. http://www.spinics.net/lists/linux-media/msg15356.html

This one the bums here could not ignore at all, because the noise around
was already so loud that ignoring was no longer possible.


MOST IMPORTANT: NONE of them had my overworked version.
And I do not know to which one they were referring to.
There is one at:

1. http://pluto.blackbone-ev.de/v1/AleVT%20mit%20DVB-T.html

and there is one at:


2. http://packages.debian.org/sid/alevt


BOTH are DVB compatible. So with both YOU CAN identify the kernel
security risk / the oops.

So as long as you do not know the facts, better shut up and stay quiet,
stupid!

There's no need to comment your spit-licking that follows.
Everyone knows who you moron are by evaluating your stupid rant, the sum
of self-invented lies.

CS

Re: [PATCH] dvb-core: fix initialization of feeds list in demux filter (Was: Videotext application crashes the kernel due to DVB-demux patch)

[hermann pitton]

Am Dienstag, den 09.02.2010, 09:38 +0100 schrieb Chicken Shack:
> Am Dienstag, den 09.02.2010, 01:53 +0100 schrieb hermann pitton:
> > Am Montag, den 08.02.2010, 08:14 -0800 schrieb Linus Torvalds:
> > > 
> > > On Mon, 8 Feb 2010, Chicken Shack wrote:
> > > > 
> > > > This is a SCANDAL, not fun! This is SCANDALOUS!
> > > 
> > > I agree that this whole thread has been totally inappropriate from 
> > > beginning to end. 
> > 
> > The initial problem was, how to find such software producing that oops
> > at all.
> 
> Pure crap! Because, as I stated already, you are too lazy to read the
> context.
> 
> Proofs:
> 
> 1. http://www.spinics.net/lists/linux-media/msg15182.html
> 
> This guy was compeletely ignored, although he announced a kernel bug /
> security risk. Neither is he banned somewhere, nor are there other ifs
> or whatevers to justify just ignoring him.
> Also the utmost insane Torvalds-scapegoat-theory dowes not work here at
> all. This guy reported a problem, and noone except me listened: Period!
> 
> 2. http://www.spinics.net/lists/linux-media/msg15356.html
> 
> This one the bums here could not ignore at all, because the noise around
> was already so loud that ignoring was no longer possible.
> 
> 
> MOST IMPORTANT: NONE of them had my overworked version.
> And I do not know to which one they were referring to.
> There is one at:
> 
> 1. http://pluto.blackbone-ev.de/v1/AleVT%20mit%20DVB-T.html
> 
> and there is one at:
> 
> 
> 2. http://packages.debian.org/sid/alevt
> 
> 
> BOTH are DVB compatible. So with both YOU CAN identify the kernel
> security risk / the oops.
> 
> So as long as you do not know the facts, better shut up and stay quiet,
> stupid!
> 
> There's no need to comment your spit-licking that follows.
> Everyone knows who you moron are by evaluating your stupid rant, the sum
> of self-invented lies.
> 
> CS
> 

For the record.

Uwe is right, the bug is already at the above link

pluto.blackbone-ev.de

coming with its own headers.

It is for sure not in any Fedora and he asked only for DVB-S testers, IIRC.

I'm sorry to have tested anything at all in such a crappy way ...

Cheers,
Hermann

Re: [PATCH] dvb-core: fix initialization of feeds list in demux filter (Was: Videotext application crashes the kernel due to DVB-demux patch)

[Devin Heitmueller]

Hello Mr. Shack,

On Mon, Feb 8, 2010 at 8:43 AM, Chicken Shack <chicken.shack@gmx.de> wrote:
> I vote for Andy Walls, Devin Heitmueller. There are a couple of capable
> candidates here who can really do that job with a better output for the
> whole Linux community.
> There is ABSOLUTELY NO necessity to continue with Mauro Carvalho Chehab.

I want to take this opportunity to thank you for your nomination and
vote.  It means so much to me to see this nomination coming from such
a dedicated contributor to the project, given your years of service
and hundreds of patches to the LinuxTV project.

<sarcasm mode off>

I've quietly sat back and watched this thread over the last few days.
I've watched your abusive and condescending attitude toward the
volunteers who have spent their limited time trying to help resolve
the issue you reported.  You seem to have forgotten that you are
dealing with a group of volunteers, and that they have absolutely *no*
obligation whatsoever to help you in making your application work.

As a user, I can appreciate your frustration:  your application used
to work and now it doesn't.  But *demanding* that developers stop what
they are doing to fix your problem is not particularly productive.
Unless you are cutting them a check, they have no obligation to do
what you want.

Was there a regression?  Yup.  Shit happens.  Let's focus on what
needs to be done to fix it.  While you interpreted the extended thread
of discussion of various options as counter-productive, this is
actually how development works.  Somebody reports a problem.
Developers discuss the various potential approaches available to deal
with the problem.  Other developers point out why those approaches
aren't appropriate or could cause other issues.

"When you are dumb, you surround yourself with smart people.  When you
are smart, you surround yourself with smart people who disagree with
you."

In this case, while not doubting the reported problem was valid, the
notion of rolling back functionality that made it into a stable kernel
in order to address the problem is something that isn't desirable.
The best case is when a fix can be made while preserving compatibility
with both usage patterns.  When that isn't possible, then hard
decisions need to be made.

And as someone who has both fixed bugs in the DVB core as well as
having introduced them, I can appreciate how hard it can be to
understand that piece of code.  It's an inherently complicated
problem, involving multiple threads of execution and concurrency, as
well as a large number of applications that use the API in different
ways.

With regards to your concerns about Mauro being the maintainer, what
can I say?  Does he make mistakes?  Sure.  Does he sometimes do things
I disagree with?  Yup.  Would I want his job?  Absolutely not.  It's a
thankless job and people only notice what you do when you make a
mistake.  He has to evaluate patches not just in terms of how they
effect whatever tuner the developer submitted them for, but also for
all the other products that use the same code.  When people submit
patches that effect application behavior, he has put forth his best
effort to ensure that it doesn't break other applications.  This is a
nontrivial exercise, and we should be thankful that Mauro has been
willing to do it for as long as he has (most maintainers of large
codebases experience "burn out" after a relatively short period of
time).

One of the things that is so empowering about open source is the
ability to fix your own problems if it matters enough to you.  You
have the full source code to all the components, and if some issue is
critical to you then you have everything you need to fix your own
problems.  You are only limited by your own commitment to learn how to
program.

Now I'm going to go back to working on my driver code.

Cheers,

Devin

-- 
Devin J. Heitmueller - Kernel Labs
http://www.kernellabs.com

Re: [PATCH] dvb-core: fix initialization of feeds list in demux filter (Was: Videotext application crashes the kernel due to DVB-demux patch)

[Alain Kalker]

Op maandag 08-02-2010 om 14:43 uur [tijdzone +0100], schreef Chicken
Shack:
> Now if I were a cynical or ranter or another kind of dumb primitive
> persona non grata I would just add "Lol" or stuff like that and turn
> myself away.
> 
> But this is no fun here.
> 
> It's nothing but a big proof that one Brazilian person in Mr. Torvalds
> "dream team of untouchables" needs to be URGENTLY replaced by another
> real capable person.
> 
> NO IDEA ABOUT DVB ISSUES, BUT DVB MAINTAINER!

I would like to SINCERELY urge you to cut the ad hominems. If you feel
you have arguments furthering your position, please let them stand for
themselves. No need to involve someones nationality, religion, star-sign
or whatever to make your point.

I think people on this list have more pressing issues to deal with (like
bisecting hard-to-find kernel oopses or digging up datasheets on
undocumented mixer/PLLs and the like) than listening to rants about
Brazilians not being team players.

Please, just the facts, ma'am. Trolls can feed elsewhere.

Sincerely,

Alain