[FFmpeg-devel] [PATCH] SIPR: kill variable-length arrays

[Måns Rullgård]

Vitor Sessak <vitor1001 at gmail.com> writes:

> Mans Rullgard wrote:
>> Two of these are in fact constant size, so use the constant instead of
>> a variable in the declarations.  The remaining one is small enough
>> that always using the maximum size is acceptable.
>> ---
>>  libavcodec/sipr.c |    8 +++++---
>>  1 files changed, 5 insertions(+), 3 deletions(-)
>> diff --git a/libavcodec/sipr.c b/libavcodec/sipr.c
>> index 7bfa3ab..e84e353 100644
>> --- a/libavcodec/sipr.c
>> +++ b/libavcodec/sipr.c
>> @@ -46,6 +46,8 @@
>>  /**  Subframe size for all modes except 16k */
>>  #define SUBFR_SIZE           48
>>  +#define MAX_SUBFRAME_COUNT   5
>> +
>>  #include "siprdata.h"
>>   typedef enum {
>> @@ -233,8 +235,8 @@ static void decode_parameters(SiprParameters* parms, GetBitContext *pgb,
>>  static void lsp2lpc_sipr(const double *lsp, float *Az)
>>  {
>>      int lp_half_order = LP_FILTER_ORDER >> 1;
>> -    double buf[lp_half_order + 1];
>> -    double pa[lp_half_order + 1];
>> +    double buf[(LP_FILTER_ORDER >> 1) + 1];
>> +    double pa[(LP_FILTER_ORDER >> 1) + 1];
>>      double *qa = buf + 1;
>>      int i,j;
>>  @@ -409,7 +411,7 @@ static void decode_frame(SiprContext *ctx,
>> SiprParameters *params,
>>  {
>>      int i, j;
>>      int frame_size = ctx->m.subframe_count * SUBFR_SIZE;
>> -    float Az[LP_FILTER_ORDER * ctx->m.subframe_count];
>> +    float Az[LP_FILTER_ORDER * MAX_SUBFRAME_COUNT];
>>      float *excitation;
>>      float ir_buf[SUBFR_SIZE + LP_FILTER_ORDER];
>>      float lsf_new[LP_FILTER_ORDER];
>
> Patch ok.

Applied.

-- 
M?ns Rullg?rd
mans at mansr.com