[FFmpeg-devel] [libav-devel] [PATCH] libopenjpegdec: check existence of image component data

[Andreas Cadhalpun]

On 01.06.2015 15:30, Michael Niedermayer wrote:
> On Mon, Jun 01, 2015 at 12:27:22AM +0200, Andreas Cadhalpun wrote:
>> On 31.05.2015 11:11, Luca Barbato wrote:
>>> On 30/05/15 20:00, Andreas Cadhalpun wrote:
>>>> +    memset(&ctx->event_mgr, 0, sizeof(opj_event_mgr_t));
>>>
>>> usually the size is sizeof(*actualpoint)
>>
>> Yes that's better, patch updated.
>> I just copied this from libopenjpegenc.c, so it should be fixed
>> there too. Additional patch for that attached.
>>
>>>> +    ctx->event_mgr.info_handler    = info_callback;
>>>> +    ctx->event_mgr.error_handler   = error_callback;
>>>> +    ctx->event_mgr.warning_handler = warning_callback;
>>>> +    opj_set_event_mgr((opj_common_ptr) dec, &ctx->event_mgr, avctx);
>>>
>>> The rest seems fine, not sure info_callback should be DEBUG or VERBOSE,
>>> I'll go with your pick if nobody has other opinions.
>>
>> I just picked what was used by the encoder.
>>
>> Best regards,
>> Andreas
>>
> 
>>  libopenjpegdec.c |   22 +++++++++++++++++++++-
>>  1 file changed, 21 insertions(+), 1 deletion(-)
>> 98b2b77bfa56fb844103c32628de8d78fb16d29a  0001-libopenjpegdec-register-logging-callback-functions.patch
>> From 79490bdcae3ac44c0383136064f3e267d98c5ec9 Mon Sep 17 00:00:00 2001
>> From: Andreas Cadhalpun <Andreas.Cadhalpun at googlemail.com>
>> Date: Sat, 30 May 2015 19:40:22 +0200
>> Subject: [PATCH 1/2] libopenjpegdec: register logging callback functions
> 
> LGTM

Pushed.

>>  libopenjpegenc.c |    2 +-
>>  1 file changed, 1 insertion(+), 1 deletion(-)
>> 2cf81129db2c06575289a01e795bced18cf980eb  0002-libopenjpegenc-use-variable-instead-of-type-for-size.patch
>> From 822f2eb3e4c47eed74f4ede5fa52a1112d5cb506 Mon Sep 17 00:00:00 2001
>> From: Andreas Cadhalpun <Andreas.Cadhalpun at googlemail.com>
>> Date: Mon, 1 Jun 2015 00:16:29 +0200
>> Subject: [PATCH 2/2] libopenjpegenc: use variable instead of type for sizeof
> 
> LGTM

Pushed.

> not sure if the memset isnt redundant though as the whole context
> should be cleared on allocation but it should do no harm

I think the memset is currently necessary, because this happens in
encode_frame and not in encode_init. (I'm not sure why, though.)

Best regards,
Andreas