[pulseaudio-discuss] [PATCH 2/2] endianmacros: Replace borked PA_FLOAT32_SWAP() with PA_READ_FLOAT32RE() / PA_WRITE_FLOAT32RE()
David Henningsson
david.henningsson at canonical.com
Tue Sep 2 23:32:52 PDT 2014
I was considering whether "const void *p" should instead be a "const
float *p" (in the function prototypes), but maybe not - a reversed float
cannot be considered a float anymore, due to the NaN problems you
describe below.
OTOH, that leaves us with a few float pointers in the callers to these
functions, which should perhaps be converted to void* too, but then you
can't index them...
So hey, it fixes a bug, and I assume you have run the relevant tests and
they succeeded. So while it's slightly inconsistent to mix float* and
void*, this one is acked too.
On 2014-09-03 02:25, Peter Meerwald wrote:
> From: Peter Meerwald <pmeerw at debian>
>
> building PA with -O0 leads to test failure in mix-test on i386
>
> issue reported by Felipe, see
> http://lists.freedesktop.org/archives/pulseaudio-discuss/2014-August/021406.html
>
> the problem is the value 0xbeffbd7f: when byte-swapped it becomes 0x7fbdffbe and according
> to IEEE-754 represents a signalling NaN (starting with s111 1111 10, see http://en.wikipedia.org/wiki/NaN)
>
> when this value is assigned to a floating point register, it becomes 0x7ffdffbe, representing
> a quiet NaN (starting with s111 1111 11) -- a signalling NaN is turned into a quiet NaN!
>
> so PA_FLOAT32_SWAP(PA_FLOAT32_SWAP(x)) != x for certain values, uhuh!
>
> the following test code can be used; due to volatile, it will always demonstrate the issue;
> without volatile, it depends on the optimization level (i386, 32-bit, gcc 4.9):
>
> // snip
>
> static inline float PA_FLOAT32_SWAP(float x) {
> union {
> float f;
> uint32_t u;
> } t;
>
> t.f = x;
> t.u = bswap_32(t.u);
> return t.f;
> }
>
> int main() {
> unsigned x = 0xbeffbd7f;
> volatile float f = PA_FLOAT32_SWAP(*(float *)&x);
> printf("%08x %08x %08x %f\n", 0xbeffbd7f, *(unsigned *)&f, bswap_32(*(unsigned *)&f), f);
> }
> // snip
>
> the problem goes away with optimization when no temporary floating point registers are used
>
> the proposed solution is to avoid passing swapped floating point data in a
> float; this is done with new functions PA_READ_FLOAT32RE() and PA_WRITE_FLOAT32RE()
> which use uint32_t to dereference a pointer and byte-swap the data, hence no temporary
> float variable is used
>
> also delete PA_FLOAT32_TO_LE()/_BE(), not used
>
> Signed-off-by: Peter Meerwald <pmeerw at pmeerw.net>
> Reported-by: Felipe Sateler <fsateler at debian.org>
> ---
> src/pulsecore/endianmacros.h | 23 ++++++++++++-----------
> src/pulsecore/mix.c | 11 ++++-------
> src/pulsecore/sample-util.c | 4 ++--
> src/pulsecore/sconv-s16le.c | 22 ++++++++--------------
> src/pulsecore/svolume_c.c | 4 ++--
> src/tests/mix-test.c | 4 ++--
> src/tests/resampler-test.c | 4 ++--
> 7 files changed, 32 insertions(+), 40 deletions(-)
>
> diff --git a/src/pulsecore/endianmacros.h b/src/pulsecore/endianmacros.h
> index 4822fc4..2e73136 100644
> --- a/src/pulsecore/endianmacros.h
> +++ b/src/pulsecore/endianmacros.h
> @@ -71,25 +71,32 @@ static inline void PA_WRITE24LE(uint8_t *p, uint32_t u) {
> p[0] = (uint8_t) u;
> }
>
> -static inline float PA_FLOAT32_SWAP(float x) {
> +static inline float PA_READ_FLOAT32RE(const void *p) {
> union {
> float f;
> uint32_t u;
> } t;
>
> - t.f = x;
> - t.u = PA_UINT32_SWAP(t.u);
> + t.u = PA_UINT32_SWAP(*(uint32_t *) p);
> return t.f;
> }
>
> +static inline void PA_WRITE_FLOAT32RE(void *p, float x) {
> + union {
> + float f;
> + uint32_t u;
> + } t;
> +
> + t.f = x;
> + *(uint32_t *) p = PA_UINT32_SWAP(t.u);
> +}
> +
> #define PA_MAYBE_INT16_SWAP(c,x) ((c) ? PA_INT16_SWAP(x) : (x))
> #define PA_MAYBE_UINT16_SWAP(c,x) ((c) ? PA_UINT16_SWAP(x) : (x))
>
> #define PA_MAYBE_INT32_SWAP(c,x) ((c) ? PA_INT32_SWAP(x) : (x))
> #define PA_MAYBE_UINT32_SWAP(c,x) ((c) ? PA_UINT32_SWAP(x) : (x))
>
> -#define PA_MAYBE_FLOAT32_SWAP(c,x) ((c) ? PA_FLOAT32_SWAP(x) : (x))
> -
> #ifdef WORDS_BIGENDIAN
> #define PA_INT16_FROM_LE(x) PA_INT16_SWAP(x)
> #define PA_INT16_FROM_BE(x) ((int16_t)(x))
> @@ -115,9 +122,6 @@ static inline float PA_FLOAT32_SWAP(float x) {
> #define PA_UINT32_TO_LE(x) PA_UINT32_SWAP(x)
> #define PA_UINT32_TO_BE(x) ((uint32_t)(x))
>
> - #define PA_FLOAT32_TO_LE(x) PA_FLOAT32_SWAP(x)
> - #define PA_FLOAT32_TO_BE(x) ((float) (x))
> -
> #define PA_READ24NE(x) PA_READ24BE(x)
> #define PA_WRITE24NE(x,y) PA_WRITE24BE((x),(y))
>
> @@ -148,9 +152,6 @@ static inline float PA_FLOAT32_SWAP(float x) {
> #define PA_UINT32_TO_LE(x) ((uint32_t)(x))
> #define PA_UINT32_TO_BE(x) PA_UINT32_SWAP(x)
>
> - #define PA_FLOAT32_TO_LE(x) ((float) (x))
> - #define PA_FLOAT32_TO_BE(x) PA_FLOAT32_SWAP(x)
> -
> #define PA_READ24NE(x) PA_READ24LE(x)
> #define PA_WRITE24NE(x,y) PA_WRITE24LE((x),(y))
>
> diff --git a/src/pulsecore/mix.c b/src/pulsecore/mix.c
> index 10567d6..03c71f0 100644
> --- a/src/pulsecore/mix.c
> +++ b/src/pulsecore/mix.c
> @@ -576,17 +576,14 @@ static void pa_mix_float32re_c(pa_mix_info streams[], unsigned nstreams, unsigne
>
> for (i = 0; i < nstreams; i++) {
> pa_mix_info *m = streams + i;
> - float v, cv = m->linear[channel].f;
> + float cv = m->linear[channel].f;
>
> - if (PA_LIKELY(cv > 0)) {
> - v = PA_FLOAT32_SWAP(*(float*) m->ptr);
> - v *= cv;
> - sum += v;
> - }
> + if (PA_LIKELY(cv > 0))
> + sum += PA_READ_FLOAT32RE(m->ptr) * cv;
> m->ptr = (uint8_t*) m->ptr + sizeof(float);
> }
>
> - *data = PA_FLOAT32_SWAP(sum);
> + PA_WRITE_FLOAT32RE(data, sum);
>
> if (PA_UNLIKELY(++channel >= channels))
> channel = 0;
> diff --git a/src/pulsecore/sample-util.c b/src/pulsecore/sample-util.c
> index dc5c1fd..539f3b8 100644
> --- a/src/pulsecore/sample-util.c
> +++ b/src/pulsecore/sample-util.c
> @@ -296,9 +296,9 @@ void pa_sample_clamp(pa_sample_format_t format, void *dst, size_t dstr, const vo
> for (; n > 0; n--) {
> float f;
>
> - f = PA_FLOAT32_SWAP(*s);
> + f = PA_READ_FLOAT32RE(s);
> f = PA_CLAMP_UNLIKELY(f, -1.0f, 1.0f);
> - *d = PA_FLOAT32_SWAP(f);
> + PA_WRITE_FLOAT32RE(d, f);
>
> s = (const float*) ((const uint8_t*) s + sstr);
> d = (float*) ((uint8_t*) d + dstr);
> diff --git a/src/pulsecore/sconv-s16le.c b/src/pulsecore/sconv-s16le.c
> index 45cf972..0c1d16c 100644
> --- a/src/pulsecore/sconv-s16le.c
> +++ b/src/pulsecore/sconv-s16le.c
> @@ -157,8 +157,7 @@ void pa_sconv_s16le_to_float32re(unsigned n, const int16_t *a, float *b) {
> for (; n > 0; n--) {
> int16_t s = *(a++);
> float k = INT16_FROM(s) * (1.0f / (1 << 15));
> - k = PA_FLOAT32_SWAP(k);
> - *(b++) = k;
> + PA_WRITE_FLOAT32RE(b++, k);
> }
> }
>
> @@ -169,8 +168,7 @@ void pa_sconv_s32le_to_float32re(unsigned n, const int32_t *a, float *b) {
> for (; n > 0; n--) {
> int32_t s = *(a++);
> float k = INT32_FROM(s) * (1.0f / (1U << 31));
> - k = PA_FLOAT32_SWAP(k);
> - *(b++) = k;
> + PA_WRITE_FLOAT32RE(b++, k);
> }
> }
>
> @@ -180,8 +178,7 @@ void pa_sconv_s16le_from_float32re(unsigned n, const float *a, int16_t *b) {
>
> for (; n > 0; n--) {
> int16_t s;
> - float v = *(a++);
> - v = PA_FLOAT32_SWAP(v) * (1 << 15);
> + float v = PA_READ_FLOAT32RE(a++) * (1 << 15);
> s = (int16_t) PA_CLAMP_UNLIKELY(lrintf(v), -0x8000, 0x7FFF);
> *(b++) = INT16_TO(s);
> }
> @@ -193,8 +190,7 @@ void pa_sconv_s32le_from_float32re(unsigned n, const float *a, int32_t *b) {
>
> for (; n > 0; n--) {
> int32_t s;
> - float v = *(a++);
> - v = PA_FLOAT32_SWAP(v) * (1U << 31);
> + float v = PA_READ_FLOAT32RE(a++) * (1U << 31);
> s = (int32_t) PA_CLAMP_UNLIKELY(llrintf(v), -0x80000000LL, 0x7FFFFFFFLL);
> *(b++) = INT32_TO(s);
> }
> @@ -325,7 +321,7 @@ void pa_sconv_s24le_to_float32re(unsigned n, const uint8_t *a, float *b) {
> for (; n > 0; n--) {
> int32_t s = READ24(a) << 8;
> float k = s * (1.0f / (1U << 31));
> - *b = PA_FLOAT32_SWAP(k);
> + PA_WRITE_FLOAT32RE(b, k);
> a += 3;
> b++;
> }
> @@ -337,8 +333,7 @@ void pa_sconv_s24le_from_float32re(unsigned n, const float *a, uint8_t *b) {
>
> for (; n > 0; n--) {
> int32_t s;
> - float v = *a;
> - v = PA_FLOAT32_SWAP(v) * (1U << 31);
> + float v = PA_READ_FLOAT32RE(a) * (1U << 31);
> s = (int32_t) PA_CLAMP_UNLIKELY(llrint(v), -0x80000000LL, 0x7FFFFFFFLL);
> WRITE24(b, ((uint32_t) s) >> 8);
> a++;
> @@ -411,7 +406,7 @@ void pa_sconv_s24_32le_to_float32re(unsigned n, const uint32_t *a, float *b) {
> for (; n > 0; n--) {
> int32_t s = (int32_t) (UINT32_FROM(*a) << 8);
> float k = s * (1.0f / (1U << 31));
> - *b = PA_FLOAT32_SWAP(k);
> + PA_WRITE_FLOAT32RE(b, k);
> a++;
> b++;
> }
> @@ -437,8 +432,7 @@ void pa_sconv_s24_32le_from_float32re(unsigned n, const float *a, uint32_t *b) {
>
> for (; n > 0; n--) {
> int32_t s;
> - float v = *a;
> - v = PA_FLOAT32_SWAP(v) * (1U << 31);
> + float v = PA_READ_FLOAT32RE(a) * (1U << 31);
> s = (int32_t) PA_CLAMP_UNLIKELY(llrint(v), -0x80000000LL, 0x7FFFFFFFLL);
> *b = UINT32_TO(((uint32_t) s) >> 8);
> a++;
> diff --git a/src/pulsecore/svolume_c.c b/src/pulsecore/svolume_c.c
> index ad2df18..2e2db12 100644
> --- a/src/pulsecore/svolume_c.c
> +++ b/src/pulsecore/svolume_c.c
> @@ -125,9 +125,9 @@ static void pa_volume_float32re_c(float *samples, const float *volumes, unsigned
> for (channel = 0; length; length--) {
> float t;
>
> - t = PA_FLOAT32_SWAP(*samples);
> + t = PA_READ_FLOAT32RE(samples);
> t *= volumes[channel];
> - *samples++ = PA_FLOAT32_SWAP(t);
> + PA_WRITE_FLOAT32RE(samples++, t);
>
> if (PA_UNLIKELY(++channel >= channels))
> channel = 0;
> diff --git a/src/tests/mix-test.c b/src/tests/mix-test.c
> index d5bae13..e95607d 100644
> --- a/src/tests/mix-test.c
> +++ b/src/tests/mix-test.c
> @@ -150,7 +150,7 @@ static void compare_block(const pa_sample_spec *ss, const pa_memchunk *chunk, in
> float *u = d;
>
> for (i = 0; i < chunk->length / pa_frame_size(ss); i++) {
> - float uu = PA_MAYBE_FLOAT32_SWAP(ss->format != PA_SAMPLE_FLOAT32NE, *u);
> + float uu = ss->format == PA_SAMPLE_FLOAT32NE ? *u : PA_READ_FLOAT32RE(u);
> fail_unless(fabsf(uu - *v) <= 1e-6f, NULL);
> ++u;
> ++v;
> @@ -264,7 +264,7 @@ static pa_memblock* generate_block(pa_mempool *pool, const pa_sample_spec *ss) {
> if (ss->format == PA_SAMPLE_FLOAT32RE) {
> float *u = d;
> for (i = 0; i < 10; i++)
> - u[i] = PA_FLOAT32_SWAP(float32ne_result[0][i]);
> + PA_WRITE_FLOAT32RE(&u[i], float32ne_result[0][i]);
> } else
> memcpy(d, float32ne_result[0], sizeof(float32ne_result[0]));
>
> diff --git a/src/tests/resampler-test.c b/src/tests/resampler-test.c
> index 60345af..f547770 100644
> --- a/src/tests/resampler-test.c
> +++ b/src/tests/resampler-test.c
> @@ -98,7 +98,7 @@ static void dump_block(const char *label, const pa_sample_spec *ss, const pa_mem
> float *u = d;
>
> for (i = 0; i < chunk->length / pa_frame_size(ss); i++) {
> - printf("%4.3g ", ss->format == PA_SAMPLE_FLOAT32NE ? *u : PA_FLOAT32_SWAP(*u));
> + printf("%4.3g ", ss->format == PA_SAMPLE_FLOAT32NE ? *u : PA_READ_FLOAT32RE(u));
> u++;
> }
>
> @@ -222,7 +222,7 @@ static pa_memblock* generate_block(pa_mempool *pool, const pa_sample_spec *ss) {
>
> if (ss->format == PA_SAMPLE_FLOAT32RE)
> for (i = 0; i < 10; i++)
> - u[i] = PA_FLOAT32_SWAP(u[i]);
> + PA_WRITE_FLOAT32RE(&u[i], u[i]);
>
> break;
> }
>
--
David Henningsson, Canonical Ltd.
https://launchpad.net/~diwic
More information about the pulseaudio-discuss
mailing list