diff --git a/libavcodec/mips/acelp_filters_mips.c b/libavcodec/mips/acelp_filters_mips.c index be686c287a..498660b448 100644 --- a/libavcodec/mips/acelp_filters_mips.c +++ b/libavcodec/mips/acelp_filters_mips.c @@ -75,7 +75,7 @@ static void ff_acelp_interpolatef_mips(float *out, const float *in, float v = 0; for (i = 0; i < filter_length;i++) { - __asm__ __volatile__ ( + __asm__ volatile ( "lwc1 %[in_val_p], 0(%[p_in_p]) \n\t" "lwc1 %[fc_val_p], 0(%[p_filter_coeffs_p]) \n\t" "lwc1 %[in_val_m], 0(%[p_in_m]) \n\t" @@ -108,7 +108,7 @@ static void ff_acelp_apply_order_2_transfer_function_mips(float *out, const floa * loop is unrolled eight times */ - __asm__ __volatile__ ( + __asm__ volatile ( "lwc1 $f0, 0(%[mem]) \n\t" "blez %[n], ff_acelp_apply_order_2_transfer_function_end%= \n\t" "lwc1 $f1, 4(%[mem]) \n\t" diff --git a/libavcodec/mips/acelp_vectors_mips.c b/libavcodec/mips/acelp_vectors_mips.c index d62b37798c..6f9390f225 100644 --- a/libavcodec/mips/acelp_vectors_mips.c +++ b/libavcodec/mips/acelp_vectors_mips.c @@ -61,7 +61,7 @@ static void ff_weighted_vector_sumf_mips( const float *a_end = in_a + length; /* loop unrolled two times */ - __asm__ __volatile__ ( + __asm__ volatile ( "blez %[length], ff_weighted_vector_sumf_end%= \n\t" "ff_weighted_vector_sumf_madd%=: \n\t" diff --git a/libavcodec/mips/amrwbdec_mips.c b/libavcodec/mips/amrwbdec_mips.c index ad08b63095..4bfbb8c6b9 100644 --- a/libavcodec/mips/amrwbdec_mips.c +++ b/libavcodec/mips/amrwbdec_mips.c @@ -70,7 +70,7 @@ void hb_fir_filter_mips(float *out, const float fir_coef[HB_FIR_SIZE + 1], * inner loop is entirely unrolled and instructions are scheduled * to minimize pipeline stall */ - __asm__ __volatile__( + __asm__ volatile( "mtc1 $zero, %[output] \n\t" "lwc1 $f0, 0(%[p_data]) \n\t" "lwc1 $f1, 0(%[fir_coef]) \n\t" diff --git a/libavcodec/mips/celp_filters_mips.c b/libavcodec/mips/celp_filters_mips.c index a31b81db17..06e0127b19 100644 --- a/libavcodec/mips/celp_filters_mips.c +++ b/libavcodec/mips/celp_filters_mips.c @@ -88,7 +88,7 @@ static void ff_celp_lp_synthesis_filterf_mips(float *out, out2 = in[2]; out3 = in[3]; - __asm__ __volatile__( + __asm__ volatile( "lwc1 $f2, 8(%[filter_coeffs]) \n\t" "lwc1 $f1, 4(%[filter_coeffs]) \n\t" "lwc1 $f0, 0(%[filter_coeffs]) \n\t" @@ -113,7 +113,7 @@ static void ff_celp_lp_synthesis_filterf_mips(float *out, ); for (i = 5; i <= filter_length; i += 2) { - __asm__ __volatile__( + __asm__ volatile( "lwc1 %[old_out3], -20(%[p_out]) \n\t" "lwc1 $f5, 16(%[p_filter_coeffs]) \n\t" "addiu %[p_out], -8 \n\t" @@ -141,7 +141,7 @@ static void ff_celp_lp_synthesis_filterf_mips(float *out, FFSWAP(float, old_out0, old_out2); } - __asm__ __volatile__( + __asm__ volatile( "nmsub.s %[out3], %[out3], %[a], %[out2] \n\t" "nmsub.s %[out2], %[out2], %[a], %[out1] \n\t" "nmsub.s %[out3], %[out3], %[b], %[out1] \n\t" @@ -176,7 +176,7 @@ static void ff_celp_lp_synthesis_filterf_mips(float *out, p_out = &out[n]; out_val = in[n]; for (i = 1; i <= filter_length; i++) { - __asm__ __volatile__( + __asm__ volatile( "lwc1 %[fc_val], 0(%[p_filter_coeffs]) \n\t" "lwc1 %[out_val_i], -4(%[p_out]) \n\t" "addiu %[p_filter_coeffs], 4 \n\t" @@ -219,7 +219,7 @@ static void ff_celp_lp_zero_synthesis_filterf_mips(float *out, * outer loop is unrolled eight times so there is less memory access * inner loop is unrolled two times */ - __asm__ __volatile__( + __asm__ volatile( "filt_lp_inner%=: \n\t" "lwc1 %[fc_val], 0(%[p_filter_coeffs]) \n\t" "lwc1 $f7, 6*4(%[p_in]) \n\t" diff --git a/libavcodec/mips/celp_math_mips.c b/libavcodec/mips/celp_math_mips.c index 0af4171bca..6ab1823d77 100644 --- a/libavcodec/mips/celp_math_mips.c +++ b/libavcodec/mips/celp_math_mips.c @@ -59,7 +59,7 @@ static float ff_dot_productf_mips(const float* a, const float* b, float sum; const float* a_end = a + length; - __asm__ __volatile__ ( + __asm__ volatile ( "mtc1 $zero, %[sum] \n\t" "blez %[length], ff_dot_productf_end%= \n\t" "ff_dot_productf_madd%=: \n\t" diff --git a/libavcodec/mips/compute_antialias_fixed.h b/libavcodec/mips/compute_antialias_fixed.h index 0c0059cecd..528411fff2 100644 --- a/libavcodec/mips/compute_antialias_fixed.h +++ b/libavcodec/mips/compute_antialias_fixed.h @@ -83,7 +83,7 @@ static void compute_antialias_mips_fixed(MPADecodeContext *s, /** * instructions are scheduled to minimize pipeline stall. */ - __asm__ __volatile__ ( + __asm__ volatile ( "lw %[tmp0], -1*4(%[ptr]) \n\t" "lw %[tmp1], 0*4(%[ptr]) \n\t" "lw %[temp_reg1], 0*4(%[csa]) \n\t" diff --git a/libavcodec/mips/compute_antialias_float.h b/libavcodec/mips/compute_antialias_float.h index 0abdef4fb2..3a3f5d6dde 100644 --- a/libavcodec/mips/compute_antialias_float.h +++ b/libavcodec/mips/compute_antialias_float.h @@ -82,7 +82,7 @@ static void compute_antialias_mips_float(MPADecodeContext *s, * instructions are scheduled to minimize pipeline stall. */ - __asm__ __volatile__ ( + __asm__ volatile ( "compute_antialias_float_loop%=: \t\n" "lwc1 %[in1], -1*4(%[ptr]) \t\n" "lwc1 %[in2], 0(%[csa]) \t\n" diff --git a/libavcodec/mips/lsp_mips.h b/libavcodec/mips/lsp_mips.h index f875392099..749770625c 100644 --- a/libavcodec/mips/lsp_mips.h +++ b/libavcodec/mips/lsp_mips.h @@ -70,7 +70,7 @@ static av_always_inline void ff_lsp2polyf_mips(const double *lsp, double *f, int double tmp, f_j_2, f_j_1, f_j; double val = lsp[2*i]; - __asm__ __volatile__( + __asm__ volatile( "move %[p_f], %[p_fi] \n\t" "add.d %[val], %[val], %[val] \n\t" "addiu %[p_fi], 8 \n\t" diff --git a/libavcodec/mips/mpegaudiodsp_mips_fixed.c b/libavcodec/mips/mpegaudiodsp_mips_fixed.c index 259fb856bb..d1b02775d4 100644 --- a/libavcodec/mips/mpegaudiodsp_mips_fixed.c +++ b/libavcodec/mips/mpegaudiodsp_mips_fixed.c @@ -84,7 +84,7 @@ static void ff_mpadsp_apply_window_mips_fixed(int32_t *synth_buf, int32_t *windo * use of round_sample function from the original code is eliminated, * changed with appropriate assembly instructions. */ - __asm__ __volatile__ ( + __asm__ volatile ( "mthi $zero \n\t" "mtlo %[temp1] \n\t" "lw %[w_asm], 0(%[w]) \n\t" @@ -175,7 +175,7 @@ static void ff_mpadsp_apply_window_mips_fixed(int32_t *synth_buf, int32_t *windo access per two sample */ for(j = 1; j < 16; j++) { - __asm__ __volatile__ ( + __asm__ volatile ( "mthi $0, $ac1 \n\t" "mtlo $0, $ac1 \n\t" "mthi $0 \n\t" @@ -299,7 +299,7 @@ static void ff_mpadsp_apply_window_mips_fixed(int32_t *synth_buf, int32_t *windo p = synth_buf + 32; - __asm__ __volatile__ ( + __asm__ volatile ( "mthi $0 \n\t" "mtlo %[temp1] \n\t" "lw %[w_asm], 32*4(%[w]) \n\t" @@ -372,7 +372,7 @@ static void imdct36_mips_fixed(int *out, int *buf, int *in, int *win) * in order to eliminate unnecessary readings and writings in array */ - __asm__ __volatile__ ( + __asm__ volatile ( "lw %[t1], 17*4(%[in]) \n\t" "lw %[t2], 16*4(%[in]) \n\t" "lw %[t3], 15*4(%[in]) \n\t" @@ -469,7 +469,7 @@ static void imdct36_mips_fixed(int *out, int *buf, int *in, int *win) * "sub %[t0], %[temp_reg1],%[t0] \n\t" */ - __asm__ __volatile__ ( + __asm__ volatile ( "lw %[t7], 4*4(%[in1]) \n\t" "lw %[t8], 8*4(%[in1]) \n\t" "lw %[t6], 16*4(%[in1]) \n\t" @@ -585,7 +585,7 @@ static void imdct36_mips_fixed(int *out, int *buf, int *in, int *win) * */ - __asm__ __volatile__ ( + __asm__ volatile ( "lw %[t2], 1*4(%[tmp]) \n\t" "lw %[t3], 3*4(%[tmp]) \n\t" "lw %[t0], 0*4(%[tmp]) \n\t" diff --git a/libavcodec/mips/mpegaudiodsp_mips_float.c b/libavcodec/mips/mpegaudiodsp_mips_float.c index 2516c5a760..00a7f20c1e 100644 --- a/libavcodec/mips/mpegaudiodsp_mips_float.c +++ b/libavcodec/mips/mpegaudiodsp_mips_float.c @@ -79,7 +79,7 @@ static void ff_mpadsp_apply_window_mips_float(float *synth_buf, float *window, * changed with appropriate assembly instructions. */ - __asm__ __volatile__ ( + __asm__ volatile ( "lwc1 %[sum], 0(%[dither_state]) \t\n" "sll %[t_sample], %[incr1], 5 \t\n" "sub %[t_sample], %[t_sample], %[incr1] \n\t" @@ -288,7 +288,7 @@ static void ff_dct32_mips_float(float *out, const float *tab) /** * instructions are scheduled to minimize pipeline stall. */ - __asm__ __volatile__ ( + __asm__ volatile ( "lwc1 %[fTmp1], 0*4(%[tab]) \n\t" "lwc1 %[fTmp2], 31*4(%[tab]) \n\t" "lwc1 %[fTmp3], 15*4(%[tab]) \n\t" @@ -350,7 +350,7 @@ static void ff_dct32_mips_float(float *out, const float *tab) : [tab] "r" (tab) ); - __asm__ __volatile__ ( + __asm__ volatile ( "lwc1 %[fTmp1], 3*4(%[tab]) \n\t" "lwc1 %[fTmp2], 28*4(%[tab]) \n\t" "lwc1 %[fTmp3], 12*4(%[tab]) \n\t" @@ -412,7 +412,7 @@ static void ff_dct32_mips_float(float *out, const float *tab) : [tab] "r" (tab) ); - __asm__ __volatile__ ( + __asm__ volatile ( "li.s %[fTmp1], 0.54119610014619698439 \n\t" "sub.s %[fTmp2], %[val0], %[val3] \n\t" "add.s %[val0], %[val0], %[val3] \n\t" @@ -436,7 +436,7 @@ static void ff_dct32_mips_float(float *out, const float *tab) : ); - __asm__ __volatile__ ( + __asm__ volatile ( "sub.s %[fTmp2], %[val16], %[val19] \n\t" "add.s %[val16], %[val16], %[val19] \n\t" "sub.s %[fTmp3], %[val23], %[val20] \n\t" @@ -457,7 +457,7 @@ static void ff_dct32_mips_float(float *out, const float *tab) : [fTmp1] "f" (fTmp1) ); - __asm__ __volatile__ ( + __asm__ volatile ( "lwc1 %[fTmp1], 1*4(%[tab]) \n\t" "lwc1 %[fTmp2], 30*4(%[tab]) \n\t" "lwc1 %[fTmp3], 14*4(%[tab]) \n\t" @@ -519,7 +519,7 @@ static void ff_dct32_mips_float(float *out, const float *tab) : [tab] "r" (tab) ); - __asm__ __volatile__ ( + __asm__ volatile ( "lwc1 %[fTmp1], 2*4(%[tab]) \n\t" "lwc1 %[fTmp2], 29*4(%[tab]) \n\t" "lwc1 %[fTmp3], 13*4(%[tab]) \n\t" @@ -581,7 +581,7 @@ static void ff_dct32_mips_float(float *out, const float *tab) : [tab] "r" (tab) ); - __asm__ __volatile__ ( + __asm__ volatile ( "li.s %[fTmp1], 1.30656296487637652785 \n\t" "sub.s %[fTmp2], %[val1], %[val2] \n\t" "add.s %[val1], %[val1], %[val2] \n\t" @@ -605,7 +605,7 @@ static void ff_dct32_mips_float(float *out, const float *tab) : ); - __asm__ __volatile__ ( + __asm__ volatile ( "sub.s %[fTmp2], %[val17], %[val18] \n\t" "add.s %[val17], %[val17], %[val18] \n\t" "sub.s %[fTmp3], %[val22], %[val21] \n\t" @@ -626,7 +626,7 @@ static void ff_dct32_mips_float(float *out, const float *tab) : [fTmp1] "f" (fTmp1) ); - __asm__ __volatile__ ( + __asm__ volatile ( "li.s %[fTmp1], 0.70710678118654752439 \n\t" "sub.s %[fTmp2], %[val0], %[val1] \n\t" "add.s %[val0], %[val0], %[val1] \n\t" @@ -663,7 +663,7 @@ static void ff_dct32_mips_float(float *out, const float *tab) : [out] "r" (out) ); - __asm__ __volatile__ ( + __asm__ volatile ( "sub.s %[fTmp2], %[val8], %[val9] \n\t" "add.s %[val8], %[val8], %[val9] \n\t" "sub.s %[fTmp3], %[val11], %[val10] \n\t" @@ -704,7 +704,7 @@ static void ff_dct32_mips_float(float *out, const float *tab) : [fTmp1] "f" (fTmp1), [out] "r" (out) ); - __asm__ __volatile__ ( + __asm__ volatile ( "sub.s %[fTmp2], %[val16], %[val17] \n\t" "add.s %[val16], %[val16], %[val17] \n\t" "sub.s %[fTmp3], %[val19], %[val18] \n\t" @@ -730,7 +730,7 @@ static void ff_dct32_mips_float(float *out, const float *tab) : [fTmp1] "f" (fTmp1) ); - __asm__ __volatile__ ( + __asm__ volatile ( "sub.s %[fTmp2], %[val24], %[val25] \n\t" "add.s %[val24], %[val24], %[val25] \n\t" "sub.s %[fTmp3], %[val27], %[val26] \n\t" @@ -799,7 +799,7 @@ static void imdct36_mips_float(float *out, float *buf, float *in, float *win) */ /* loop 1 and 2 */ - __asm__ __volatile__ ( + __asm__ volatile ( "lwc1 %[in1], 17*4(%[in]) \t\n" "lwc1 %[in2], 16*4(%[in]) \t\n" "lwc1 %[in3], 15*4(%[in]) \t\n" @@ -871,7 +871,7 @@ static void imdct36_mips_float(float *out, float *buf, float *in, float *win) ); /* loop 3 */ - __asm__ __volatile__ ( + __asm__ volatile ( "li.s %[c1], 0.5 \t\n" "lwc1 %[in1], 8*4(%[in]) \t\n" "lwc1 %[in2], 16*4(%[in]) \t\n" @@ -1002,7 +1002,7 @@ static void imdct36_mips_float(float *out, float *buf, float *in, float *win) ); /* loop 4 */ - __asm__ __volatile__ ( + __asm__ volatile ( "lwc1 %[in1], 2*4(%[tmp]) \t\n" "lwc1 %[in2], 0(%[tmp]) \t\n" "lwc1 %[in3], 3*4(%[tmp]) \t\n" diff --git a/libavutil/mips/libm_mips.h b/libavutil/mips/libm_mips.h index 9cc87b75fa..8853bbc751 100644 --- a/libavutil/mips/libm_mips.h +++ b/libavutil/mips/libm_mips.h @@ -57,7 +57,7 @@ static av_always_inline av_const long int lrintf_mips(float x) { register int ret_int; - __asm__ __volatile__ ( + __asm__ volatile ( "cvt.w.s %[x], %[x] \n\t" "mfc1 %[ret_int], %[x] \n\t"