[pulseaudio-discuss] [PATCH 3/3] tests: Fix mix-test on big-endian systems
Felipe Sateler
fsateler at debian.org
Fri Aug 29 18:11:01 PDT 2014
On Wed, Aug 13, 2014 at 3:50 AM, Peter Meerwald <pmeerw at pmeerw.net> wrote:
>
>> the mix test code never worked on big-endian systems
>>
>> patch saves a lot of duplicate test and uses more logical naming
>>
>> see
>> http://lists.freedesktop.org/archives/pulseaudio-discuss/2014-August/021035.html
>>
>> Signed-off-by: Peter Meerwald <pmeerw at pmeerw.net>
>> Reported-by: Felipe Sateler <fsateler at debian.org>
>> ---
>> src/tests/mix-test.c | 229 ++++++++++++---------------------------------------
>> 1 file changed, 54 insertions(+), 175 deletions(-)
>>
>> diff --git a/src/tests/mix-test.c b/src/tests/mix-test.c
>> index c0cd559..d5bae13 100644
>> --- a/src/tests/mix-test.c
>> +++ b/src/tests/mix-test.c
>> @@ -56,79 +56,37 @@ static const uint8_t ulaw_result[3][10] = {
>> { 0x00, 0xff, 0xff, 0x80, 0x91, 0x31, 0x00, 0xe9, 0x12, 0x13 },
>> };
>>
>> -/* PA_SAMPLE_S16LE */
>> -static const uint16_t s16le_result[3][10] = {
>> +static const uint16_t s16ne_result[3][10] = {
>> { 0x0000, 0xffff, 0x7fff, 0x8000, 0x9fff, 0x3fff, 0x0001, 0xf000, 0x0020, 0x0021 },
>> { 0x0000, 0xffff, 0x7332, 0x8ccd, 0xa998, 0x3998, 0x0000, 0xf199, 0x001c, 0x001d },
>> { 0x0000, 0xfffe, 0x7fff, 0x8000, 0x8000, 0x7997, 0x0001, 0xe199, 0x003c, 0x003e },
>> };
>>
>> -/* PA_SAMPLE_S16BE */
>> -static const uint16_t s16be_result[3][10] = {
>> -{ 0x0000, 0xffff, 0x7fff, 0x8000, 0x9fff, 0x3fff, 0x0001, 0xf000, 0x0020, 0x0021 },
>> -{ 0x0000, 0xffff, 0x8bff, 0x7300, 0xa8ff, 0x52ff, 0xe600, 0xd700, 0xcc1c, 0xb31d },
>> -{ 0x0000, 0xfeff, 0x0aff, 0xf300, 0x47ff, 0x91fe, 0xe601, 0xc701, 0xcc3c, 0xb33e },
>> -};
>> -
>> -/* PA_SAMPLE_FLOAT32LE */
>> -static const float float32le_result[3][10] = {
>> -{ 0.000000, -1.000000, 1.000000, 4711.000000, 0.222000, 0.330000, -0.300000, 99.000000, -0.555000, -0.123000 },
>> -{ 0.000000, -0.899987, 0.899987, 4239.837402, 0.199797, 0.296996, -0.269996, 89.098679, -0.499493, -0.110698 },
>> -{ 0.000000, -1.899987, 1.899987, 8950.837891, 0.421797, 0.626996, -0.569996, 188.098679, -1.054493, -0.233698 },
>> -};
>> -
>> -/* PA_SAMPLE_FLOAT32BE */
>> -static const float float32be_result[3][10] = {
>> +static const float float32ne_result[3][10] = {
>> { 0.000000, -1.000000, 1.000000, 4711.000000, 0.222000, 0.330000, -0.300000, 99.000000, -0.555000, -0.123000 },
>> { 0.000000, -0.899987, 0.899987, 4239.837402, 0.199797, 0.296996, -0.269996, 89.098679, -0.499493, -0.110698 },
>> { 0.000000, -1.899987, 1.899987, 8950.837891, 0.421797, 0.626996, -0.569996, 188.098679, -1.054493, -0.233698 },
>> };
>>
>> -/* PA_SAMPLE_S32LE */
>> -static const uint32_t s32le_result[3][10] = {
>> +static const uint32_t s32ne_result[3][10] = {
>> { 0x00000001, 0xffff0002, 0x7fff0003, 0x80000004, 0x9fff0005, 0x3fff0006, 0x00010007, 0xf0000008, 0x00200009, 0x0021000a },
>> { 0x00000000, 0xffff199b, 0x7332199c, 0x8ccd0003, 0xa998d99e, 0x3998999f, 0x0000e66c, 0xf199a007, 0x001cccc8, 0x001db32e },
>> { 0x00000001, 0xfffe199d, 0x7fffffff, 0x80000000, 0x80000000, 0x799799a5, 0x0001e673, 0xe199a00f, 0x003cccd1, 0x003eb338 },
>> };
>>
>> -/* PA_SAMPLE_S32BE */
>> -static const uint32_t s32be_result[3][10] = {
>> -{ 0x00000001, 0xffff0002, 0x7fff0003, 0x80000004, 0x9fff0005, 0x3fff0006, 0x00010007, 0xf0000008, 0x00200009, 0x0021000a },
>> -{ 0x0066e600, 0x65b2cd01, 0xf117b402, 0x73989903, 0x0ee48004, 0xb8496705, 0xe6ca4c06, 0xd7303307, 0xccb21908, 0xb3190009 },
>> -{ 0x0066e601, 0x64b2ce03, 0x7017b505, 0xf3989907, 0xade38109, 0xf748680b, 0xe6cb4c0d, 0xc731330f, 0xccd21911, 0xb33a0013 },
>> -};
>> -
>> -/* PA_SAMPLE_S24LE */
>> -static const uint8_t s24le_result[3][30] = {
>> -{ 0x00, 0x00, 0x01, 0xff, 0xff, 0x02, 0x7f, 0xff, 0x03, 0x80, 0x00, 0x04, 0x9f, 0xff, 0x05, 0x3f, 0xff, 0x06, 0x01, 0x00, 0x07, 0xf0, 0x00, 0x08, 0x20, 0x00, 0x09, 0x21, 0x00, 0x0a },
>> -{ 0x66, 0xe6, 0x00, 0x31, 0xb3, 0x02, 0x23, 0x99, 0x03, 0x0b, 0x9a, 0x03, 0x0c, 0x66, 0x05, 0x1c, 0x4c, 0x06, 0xca, 0x4c, 0x06, 0x07, 0x34, 0x07, 0xb2, 0x19, 0x08, 0x19, 0x00, 0x09 },
>> -{ 0x66, 0xe6, 0x01, 0x30, 0xb3, 0x05, 0xa2, 0x98, 0x07, 0x8b, 0x9a, 0x07, 0xab, 0x65, 0x0b, 0x5b, 0x4b, 0x0d, 0xcb, 0x4c, 0x0d, 0xf7, 0x34, 0x0f, 0xd2, 0x19, 0x11, 0x3a, 0x00, 0x13 },
>> -};
>> -
>> -/* PA_SAMPLE_S24BE */
>> +/* attention: result is in BE, not NE! */
>> static const uint8_t s24be_result[3][30] = {
>> -{ 0x00, 0x00, 0x01, 0xff, 0xff, 0x02, 0x7f, 0xff, 0x03, 0x80, 0x00, 0x04, 0x9f, 0xff, 0x05, 0x3f, 0xff, 0x06, 0x01, 0x00, 0x07,
>> - 0xf0, 0x00, 0x08, 0x20, 0x00, 0x09, 0x21, 0x00, 0x0a },
>> -{ 0x00, 0x00, 0x00, 0xff, 0xff, 0x1b, 0x73, 0x32, 0x1c, 0x8c, 0xcd, 0x03, 0xa9, 0x98, 0xde, 0x39, 0x98, 0x9f, 0x00, 0xe6, 0x6c,
>> - 0xf1, 0x99, 0xa7, 0x1c, 0xcc, 0xc8, 0x1d, 0xb3, 0x2e },
>> -{ 0x00, 0x00, 0x01, 0xff, 0xfe, 0x1d, 0x7f, 0xff, 0xff, 0x80, 0x00, 0x00, 0x80, 0x00, 0x00, 0x79, 0x97, 0xa5, 0x01, 0xe6, 0x73,
>> - 0xe1, 0x99, 0xaf, 0x3c, 0xcc, 0xd1, 0x3e, 0xb3, 0x38 },
>> +{ 0x00, 0x00, 0x01, 0xff, 0xff, 0x02, 0x7f, 0xff, 0x03, 0x80, 0x00, 0x04, 0x9f, 0xff, 0x05, 0x3f, 0xff, 0x06, 0x01, 0x00, 0x07, 0xf0, 0x00, 0x08, 0x20, 0x00, 0x09, 0x21, 0x00, 0x0a },
>> +{ 0x00, 0x00, 0x00, 0xff, 0xff, 0x1b, 0x73, 0x32, 0x1c, 0x8c, 0xcd, 0x03, 0xa9, 0x98, 0xde, 0x39, 0x98, 0x9f, 0x00, 0xe6, 0x6c, 0xf1, 0x99, 0xa7, 0x1c, 0xcc, 0xc8, 0x1d, 0xb3, 0x2e },
>> +{ 0x00, 0x00, 0x01, 0xff, 0xfe, 0x1d, 0x7f, 0xff, 0xff, 0x80, 0x00, 0x00, 0x80, 0x00, 0x00, 0x79, 0x97, 0xa5, 0x01, 0xe6, 0x73, 0xe1, 0x99, 0xaf, 0x3c, 0xcc, 0xd1, 0x3e, 0xb3, 0x38 },
>> };
>>
>> -/* PA_SAMPLE_S24_32LE */
>> -static const uint32_t s24_32le_result[3][10] = {
>> +static const uint32_t s24_32ne_result[3][10] = {
>> { 0x00000001, 0xffff0002, 0x7fff0003, 0x80000004, 0x9fff0005, 0x3fff0006, 0x00010007, 0xf0000008, 0x00200009, 0x0021000a },
>> { 0x00000000, 0x00ff199b, 0x00ff199c, 0x00000003, 0x00ff199e, 0x00ff199f, 0x0000e66c, 0x00000007, 0x001cccc8, 0x001db32e },
>> { 0x00000001, 0x00fe199d, 0x00fe199f, 0x00000007, 0x00fe19a3, 0x00fe19a5, 0x0001e673, 0x0000000f, 0x003cccd1, 0x003eb338 },
>> };
>>
>> -/* PA_SAMPLE_S24_32BE */
>> -static const uint32_t s24_32be_result[3][10] = {
>> -{ 0x00000001, 0xffff0002, 0x7fff0003, 0x80000004, 0x9fff0005, 0x3fff0006, 0x00010007, 0xf0000008, 0x00200009, 0x0021000a },
>> -{ 0x00000000, 0x65e60000, 0xf1e50000, 0x73000000, 0x0ee60000, 0xb8e50000, 0xe6000000, 0xd7000000, 0xcc1c0000, 0xb31d0000 },
>> -{ 0x00000000, 0x64e60100, 0x70e50100, 0xf3000000, 0xade50100, 0xf7e40100, 0xe6010000, 0xc7010000, 0xcc3c0000, 0xb33e0000 },
>> -};
>> -
>> static void compare_block(const pa_sample_spec *ss, const pa_memchunk *chunk, int iter) {
>> void *d;
>> unsigned i;
>> @@ -172,49 +130,27 @@ static void compare_block(const pa_sample_spec *ss, const pa_memchunk *chunk, in
>> break;
>> }
>>
>> - case PA_SAMPLE_S16LE: {
>> - const uint16_t *v = s16le_result[iter];
>> - uint16_t *u = d;
>> -
>> - for (i = 0; i < chunk->length / pa_frame_size(ss); i++) {
>> - fail_unless(*u == *v, NULL);
>> - ++u;
>> - ++v;
>> - }
>> - break;
>> - }
>> -
>> - case PA_SAMPLE_S16BE: {
>> - const uint16_t *v = s16be_result[iter];
>> + case PA_SAMPLE_S16NE:
>> + case PA_SAMPLE_S16RE: {
>> + const uint16_t *v = s16ne_result[iter];
>> uint16_t *u = d;
>>
>> for (i = 0; i < chunk->length / pa_frame_size(ss); i++) {
>> - fail_unless(*u == *v, NULL);
>> - ++u;
>> - ++v;
>> - }
>> - break;
>> - }
>> -
>> - case PA_SAMPLE_FLOAT32LE: {
>> - const float *v = float32le_result[iter];
>> - float *u = d;
>> -
>> - for (i = 0; i < chunk->length / pa_frame_size(ss); i++) {
>> - float uu = ss->format == PA_SAMPLE_FLOAT32NE ? *u : PA_FLOAT32_SWAP(*u);
>> - fail_unless(fabsf(uu - *v) <= 1e-6f, NULL);
>> + uint16_t uu = PA_MAYBE_UINT16_SWAP(ss->format != PA_SAMPLE_S16NE, *u);
>> + fail_unless(uu == *v, NULL);
>> ++u;
>> ++v;
>> }
>> break;
>> }
>>
>> - case PA_SAMPLE_FLOAT32BE: {
>> - const float *v = float32be_result[iter];
>> + case PA_SAMPLE_FLOAT32NE:
>> + case PA_SAMPLE_FLOAT32RE: {
>> + const float *v = float32ne_result[iter];
>> float *u = d;
>>
>> for (i = 0; i < chunk->length / pa_frame_size(ss); i++) {
>> - float uu = ss->format == PA_SAMPLE_FLOAT32NE ? *u : PA_FLOAT32_SWAP(*u);
>> + float uu = PA_MAYBE_FLOAT32_SWAP(ss->format != PA_SAMPLE_FLOAT32NE, *u);
>> fail_unless(fabsf(uu - *v) <= 1e-6f, NULL);
>> ++u;
>> ++v;
>> @@ -222,77 +158,42 @@ static void compare_block(const pa_sample_spec *ss, const pa_memchunk *chunk, in
>> break;
>> }
>>
>> - case PA_SAMPLE_S32LE: {
>> - const uint32_t *v = s32le_result[iter];
>> - uint32_t *u = d;
>> -
>> - for (i = 0; i < chunk->length / pa_frame_size(ss); i++) {
>> - fail_unless(*u == *v, NULL);
>> - ++u;
>> - ++v;
>> - }
>> - break;
>> - }
>> -
>> - case PA_SAMPLE_S32BE: {
>> - const uint32_t *v = s32be_result[iter];
>> - uint32_t *u = d;
>> -
>> - for (i = 0; i < chunk->length / pa_frame_size(ss); i++) {
>> - fail_unless(*u == *v, NULL);
>> - ++u;
>> - ++v;
>> - }
>> - break;
>> - }
>> -
>> - case PA_SAMPLE_S24_32LE: {
>> - const uint32_t *v = s24_32le_result[iter];
>> + case PA_SAMPLE_S32NE:
>> + case PA_SAMPLE_S32RE: {
>> + const uint32_t *v = s32ne_result[iter];
>> uint32_t *u = d;
>>
>> for (i = 0; i < chunk->length / pa_frame_size(ss); i++) {
>> - fail_unless(*u == *v, NULL);
>> + uint32_t uu = PA_MAYBE_UINT32_SWAP(ss->format != PA_SAMPLE_S32NE, *u);
>> + fail_unless(uu == *v, NULL);
>> ++u;
>> ++v;
>> }
>> break;
>> }
>>
>> - case PA_SAMPLE_S24_32BE: {
>> - const uint32_t *v = s24_32be_result[iter];
>> + case PA_SAMPLE_S24_32NE:
>> + case PA_SAMPLE_S24_32RE: {
>> + const uint32_t *v = s24_32ne_result[iter];
>> uint32_t *u = d;
>>
>> for (i = 0; i < chunk->length / pa_frame_size(ss); i++) {
>> - fail_unless(*u == *v, NULL);
>> + uint32_t uu = PA_MAYBE_UINT32_SWAP(ss->format != PA_SAMPLE_S24_32NE, *u);
>> + fail_unless(uu == *v, NULL);
>> ++u;
>> ++v;
>> }
>> break;
>> }
>>
>> - case PA_SAMPLE_S24LE: {
>> - const uint8_t *v = s24le_result[iter];
>> - uint8_t *u = d;
>> -
>> - for (i = 0; i < chunk->length / pa_frame_size(ss); i++) {
>> - fail_unless(*u == *v, NULL);
>> - fail_unless(*(u+1) == *(v+1), NULL);
>> - fail_unless(*(u+2) == *(v+2), NULL);
>> -
>> - u += 3;
>> - v += 3;
>> - }
>> - break;
>> - }
>> -
>> - case PA_SAMPLE_S24BE: {
>> + case PA_SAMPLE_S24NE:
>> + case PA_SAMPLE_S24RE: {
>> const uint8_t *v = s24be_result[iter];
>> uint8_t *u = d;
>>
>> for (i = 0; i < chunk->length / pa_frame_size(ss); i++) {
>> - fail_unless(*u == *v, NULL);
>> - fail_unless(*(u+1) == *(v+1), NULL);
>> - fail_unless(*(u+2) == *(v+2), NULL);
>> + uint32_t uu = ss->format == PA_SAMPLE_S24LE ? PA_READ24LE(u) : PA_READ24BE(u);
>> + fail_unless(uu == PA_READ24BE(v), NULL);
>>
>> u += 3;
>> v += 3;
>> @@ -320,23 +221,18 @@ static pa_memblock* generate_block(pa_mempool *pool, const pa_sample_spec *ss) {
>> case PA_SAMPLE_U8:
>> case PA_SAMPLE_ULAW:
>> case PA_SAMPLE_ALAW: {
>> - static const uint8_t u8_samples[] = {
>> - 0x00, 0xFF, 0x7F, 0x80, 0x9f,
>> - 0x3f, 0x01, 0xF0, 0x20, 0x21
>> - };
>> -
>> - memcpy(d, u8_samples, sizeof(u8_samples));
>> + memcpy(d, u8_result[0], sizeof(u8_result[0]));
>> break;
>> }
>>
>> case PA_SAMPLE_S16NE:
>> case PA_SAMPLE_S16RE: {
>> - static const uint16_t u16_samples[] = {
>> - 0x0000, 0xFFFF, 0x7FFF, 0x8000, 0x9fff,
>> - 0x3fff, 0x0001, 0xF000, 0x0020, 0x0021
>> - };
>> -
>> - memcpy(d, u16_samples, sizeof(u16_samples));
>> + if (ss->format == PA_SAMPLE_S16RE) {
>> + uint16_t *u = d;
>> + for (i = 0; i < 10; i++)
>> + u[i] = PA_UINT16_SWAP(s16ne_result[0][i]);
>> + } else
>> + memcpy(d, s16ne_result[0], sizeof(s16ne_result[0]));
>> break;
>> }
>>
>> @@ -344,48 +240,33 @@ static pa_memblock* generate_block(pa_mempool *pool, const pa_sample_spec *ss) {
>> case PA_SAMPLE_S24_32RE:
>> case PA_SAMPLE_S32NE:
>> case PA_SAMPLE_S32RE: {
>> - static const uint32_t u32_samples[] = {
>> - 0x00000001, 0xFFFF0002, 0x7FFF0003, 0x80000004, 0x9fff0005,
>> - 0x3fff0006, 0x00010007, 0xF0000008, 0x00200009, 0x0021000A
>> - };
>> -
>> - memcpy(d, u32_samples, sizeof(u32_samples));
>> + if (ss->format == PA_SAMPLE_S24_32RE || ss->format == PA_SAMPLE_S32RE) {
>> + uint32_t *u = d;
>> + for (i = 0; i < 10; i++)
>> + u[i] = PA_UINT32_SWAP(s32ne_result[0][i]);
>> + } else
>> + memcpy(d, s32ne_result[0], sizeof(s32ne_result[0]));
>> break;
>> }
>>
>> case PA_SAMPLE_S24NE:
>> - case PA_SAMPLE_S24RE: {
>> - /* Need to be on a byte array because they are not aligned */
>> - static const uint8_t u24_samples[] = {
>> - 0x00, 0x00, 0x01,
>> - 0xFF, 0xFF, 0x02,
>> - 0x7F, 0xFF, 0x03,
>> - 0x80, 0x00, 0x04,
>> - 0x9f, 0xff, 0x05,
>> - 0x3f, 0xff, 0x06,
>> - 0x01, 0x00, 0x07,
>> - 0xF0, 0x00, 0x08,
>> - 0x20, 0x00, 0x09,
>> - 0x21, 0x00, 0x0A
>> - };
>> -
>> - memcpy(d, u24_samples, sizeof(u24_samples));
>> + case PA_SAMPLE_S24RE:
>> + if (ss->format == PA_SAMPLE_S24LE) {
>> + uint8_t *u = d;
>> + for (i = 0; i < 30; i += 3)
>> + PA_WRITE24LE(&u[i], PA_READ24BE(&s24be_result[0][i]));
>> + } else
>> + memcpy(d, s24be_result[0], sizeof(s24be_result[0]));
>> break;
>> - }
>>
>> case PA_SAMPLE_FLOAT32NE:
>> case PA_SAMPLE_FLOAT32RE: {
>> - float *u = d;
>> - static const float float_samples[] = {
>> - 0.0f, -1.0f, 1.0f, 4711.0f, 0.222f,
>> - 0.33f, -.3f, 99.0f, -0.555f, -.123f
>> - };
>> -
>> if (ss->format == PA_SAMPLE_FLOAT32RE) {
>> + float *u = d;
>> for (i = 0; i < 10; i++)
>> - u[i] = PA_FLOAT32_SWAP(float_samples[i]);
>> + u[i] = PA_FLOAT32_SWAP(float32ne_result[0][i]);
>> } else
>> - memcpy(d, float_samples, sizeof(float_samples));
>> + memcpy(d, float32ne_result[0], sizeof(float32ne_result[0]));
>>
>> break;
>> }
>> @@ -427,8 +308,6 @@ START_TEST (mix_test) {
>> i.length = pa_memblock_get_length(i.memblock);
>> i.index = 0;
>>
>> - compare_block(&a, &i, 0);
>> -
>> /* Make a copy */
>> j = i;
>> pa_memblock_ref(j.memblock);
>>
>
> pushed (old test code was definitely broken, what can go wrong?)
It appears that the new code is not really correct either: builds are
failing on the 3 i386 builders.
https://buildd.debian.org/status/logs.php?pkg=pulseaudio
tests/mix-test.c:154:F:mix:mix_test:0: Assertion 'fabsf(uu - *v) <=
1e-6f' failed
--
Saludos,
Felipe Sateler
More information about the pulseaudio-discuss
mailing list