Hi,
On 30.12.2015 02:09, Ronald S. Bultje wrote:
> On Tue, Dec 29, 2015 at 6:54 PM, Andreas Cadhalpun <
> andreas.cadhal...@googlemail.com> wrote:
>> I prefer that input for such performance-sensitive dsp code gets sanitized,
>> so that overflows are (nearly) impossible.
>
>
> I don't think it m
On Tue, Dec 29, 2015 at 7:13 PM, Rostislav Pehlivanov
wrote:
> I agree with BBB (Ronald), this project claims to have the fastest decoders
> available and we shouldn't sacrifice performance. Since basically the only
> input of the decoder goes through the entropy decoding system occasionally
> che
I agree with BBB (Ronald), this project claims to have the fastest decoders
available and we shouldn't sacrifice performance. Since basically the only
input of the decoder goes through the entropy decoding system occasionally
checking for overflows in a few performance sensitive parts there is wort
On Tue, Dec 29, 2015 at 5:09 PM, Ronald S. Bultje wrote:
> Hi,
>
> On Tue, Dec 29, 2015 at 6:54 PM, Andreas Cadhalpun <
> andreas.cadhal...@googlemail.com> wrote:
>
>> Hi,
>>
>> On 29.12.2015 19:46, Ronald S. Bultje wrote:
>> > On Tue, Dec 29, 2015 at 11:49 AM, Andreas Cadhalpun <
>> > andreas.cad
Hi,
On Tue, Dec 29, 2015 at 6:54 PM, Andreas Cadhalpun <
andreas.cadhal...@googlemail.com> wrote:
> Hi,
>
> On 29.12.2015 19:46, Ronald S. Bultje wrote:
> > On Tue, Dec 29, 2015 at 11:49 AM, Andreas Cadhalpun <
> > andreas.cadhal...@googlemail.com> wrote:
> >> Do you have a sample causing overflo
On 29.12.2015 22:34, Rostislav Pehlivanov wrote:
>> This is an infinite loop that can hang the decoder.
> Will try to get the upstream to accept a fix.
Thanks.
>> overflows in the entropy decoding system
> The assertions should be able to catch them, so they should be fine.
Are you talking about
Hi,
On 29.12.2015 19:46, Ronald S. Bultje wrote:
> On Tue, Dec 29, 2015 at 11:49 AM, Andreas Cadhalpun <
> andreas.cadhal...@googlemail.com> wrote:
>> Do you have a sample causing overflows in the vp9 decoder?
>
>
> Nope, but I'm going to assume they're not hard to create, just use a few
> high
Hi,
Thanks for the feedback.
>This is an infinite loop that can hang the decoder.
Will try to get the upstream to accept a fix.
>overflows in the entropy decoding system
The assertions should be able to catch them, so they should be fine.
They'll trigger if there's a desync with the bitstream. I
On Tue, Dec 29, 2015 at 10:42 AM, Ronald S. Bultje wrote:
> Hi,
>
> On Tue, Dec 29, 2015 at 12:25 PM, Michael Niedermayer <
> mich...@niedermayer.cc> wrote:
>
>> On Tue, Dec 29, 2015 at 02:12:44AM +, Rostislav Pehlivanov wrote:
>> > +#define DAALAENT_PVQ_NOVAL(-2147483647-1)
>>
>> why is t
Hi,
On Tue, Dec 29, 2015 at 11:49 AM, Andreas Cadhalpun <
andreas.cadhal...@googlemail.com> wrote:
> On 29.12.2015 17:32, Ronald S. Bultje wrote:
> > On Tue, Dec 29, 2015 at 11:29 AM, Andreas Cadhalpun <
> > andreas.cadhal...@googlemail.com> wrote:
> >> On 29.12.2015 17:15, Ronald S. Bultje wrote
Hi,
On Tue, Dec 29, 2015 at 12:25 PM, Michael Niedermayer <
mich...@niedermayer.cc> wrote:
> On Tue, Dec 29, 2015 at 02:12:44AM +, Rostislav Pehlivanov wrote:
> > +#define DAALAENT_PVQ_NOVAL(-2147483647-1)
>
> why is the 1 not already subtracted in the written value ?
I think it's MIN_I
On Tue, Dec 29, 2015 at 02:12:44AM +, Rostislav Pehlivanov wrote:
> Hmm, kinda odd.
> I've attached a v2 of the RFC Patch which does this:
> 1. Fixed some typos.
> 2. Fixed some missing newlines at the end of daalatab*
> 3. Moves the pix_fmt finding function inside the main decoder file.
[...]
On 29.12.2015 17:32, Ronald S. Bultje wrote:
> On Tue, Dec 29, 2015 at 11:29 AM, Andreas Cadhalpun <
> andreas.cadhal...@googlemail.com> wrote:
>> On 29.12.2015 17:15, Ronald S. Bultje wrote:
>>> In that case, you likely need 64bit integers for 15bit math precision
>> (look at
>>> vp9 code), or daa
Hi,
On Tue, Dec 29, 2015 at 11:29 AM, Andreas Cadhalpun <
andreas.cadhal...@googlemail.com> wrote:
> Hi,
>
> On 29.12.2015 17:15, Ronald S. Bultje wrote:
> > On Tue, Dec 29, 2015 at 11:07 AM, Andreas Cadhalpun <
> > andreas.cadhal...@googlemail.com> wrote:
> >
> >> On 29.12.2015 16:58, Ronald S.
Hi,
On 29.12.2015 17:15, Ronald S. Bultje wrote:
> On Tue, Dec 29, 2015 at 11:07 AM, Andreas Cadhalpun <
> andreas.cadhal...@googlemail.com> wrote:
>
>> On 29.12.2015 16:58, Ronald S. Bultje wrote:
>>> On Tue, Dec 29, 2015 at 10:55 AM, Andreas Cadhalpun <
>>> andreas.cadhal...@googlemail.com> wro
Hi,
On Tue, Dec 29, 2015 at 11:07 AM, Andreas Cadhalpun <
andreas.cadhal...@googlemail.com> wrote:
> On 29.12.2015 16:58, Ronald S. Bultje wrote:
> > Hi,
> >
> > On Tue, Dec 29, 2015 at 10:55 AM, Andreas Cadhalpun <
> > andreas.cadhal...@googlemail.com> wrote:
> >
> >>> +static av_always_inline v
On 29.12.2015 16:58, Ronald S. Bultje wrote:
> Hi,
>
> On Tue, Dec 29, 2015 at 10:55 AM, Andreas Cadhalpun <
> andreas.cadhal...@googlemail.com> wrote:
>
>>> +static av_always_inline void idct_1D_8(pixel *x, int xstride, const
>> pixel y[16])
>>> +{
>>> +int t1h, t4h, t6h, t0 = y[0], t1 = y[1
Hi,
On Tue, Dec 29, 2015 at 10:55 AM, Andreas Cadhalpun <
andreas.cadhal...@googlemail.com> wrote:
> > +static av_always_inline void idct_1D_8(pixel *x, int xstride, const
> pixel y[16])
> > +{
> > +int t1h, t4h, t6h, t0 = y[0], t1 = y[1], t2 = y[2], t3 = y[3], t4 =
> y[4];
> > +int t5 =
On 29.12.2015 03:12, Rostislav Pehlivanov wrote:
> Hmm, kinda odd.
> I've attached a v2 of the RFC Patch which does this:
> 1. Fixed some typos.
> 2. Fixed some missing newlines at the end of daalatab*
> 3. Moves the pix_fmt finding function inside the main decoder file.
>
> Try building now.
Fir
On 12/28/2015 11:12 PM, Rostislav Pehlivanov wrote:
> diff --git a/libavcodec/daalatab.h b/libavcodec/daalatab.h
> new file mode 100644
> index 000..6a4eb34
> --- /dev/null
> +++ b/libavcodec/daalatab.h
> @@ -0,0 +1,85 @@
> +/*
> + * Daala video decoder
> + *
> + * Copyright (C) 2015 Rostislav
On Tue, Dec 29, 2015 at 02:12:44AM +, Rostislav Pehlivanov wrote:
> Hmm, kinda odd.
> I've attached a v2 of the RFC Patch which does this:
> 1. Fixed some typos.
> 2. Fixed some missing newlines at the end of daalatab*
> 3. Moves the pix_fmt finding function inside the main decoder file.
>
> T
21 matches
Mail list logo
