On Wed, 03 Jan 2007 10:06:37 +0100 Sebastian Schäfer <[EMAIL PROTECTED]> wrote:
[stuff removed] > I applied the changes mentioned above and stumbled over one very curious > thing: > The following is the output of gcc when compiling the plugin: > > --------------------- > gcc -DHAVE_CONFIG_H -I. -I. -I. -I/usr/local/include -g -O2 -Wall -O6 > -fomit-frame-pointer -fstrength-reduce -funroll-loops -ffast-math -fPIC > -DPIC -march=i686 -MT mbeq_119700_la-mbeq_119700.lo -MD -MP > -MF .deps/mbeq_119700_la-mbeq_119700.Tpo -c mbeq_119700.c -fPIC -DPIC > -o .libs/mbeq_119700_la-mbeq_119700.o > mbeq_119700.c: In function 'runMChMBEq': > mbeq_119700.c:3893: warning: operation on 'comp_rp_ptr_L' may be > undefined > mbeq_119700.c:3894: warning: operation on 'comp_ip_ptr_L' may be > undefined > mbeq_119700.c:3896: warning: operation on 'comp_rp_ptr_R' may be > undefined > mbeq_119700.c:3897: warning: operation on 'comp_ip_ptr_R' may be > undefined > mbeq_119700.c:3899: warning: operation on 'comp_rp_ptr_C' may be > undefined > mbeq_119700.c:3900: warning: operation on 'comp_ip_ptr_C' may be > undefined > mbeq_119700.c:3902: warning: operation on 'comp_rp_ptr_S' may be > undefined > mbeq_119700.c:3903: warning: operation on 'comp_ip_ptr_S' may be > undefined > mbeq_119700.c:3905: warning: operation on 'comp_rp_ptr_LS' may be > undefined > mbeq_119700.c:3906: warning: operation on 'comp_ip_ptr_LS' may be > undefined > mbeq_119700.c:3908: warning: operation on 'comp_rp_ptr_RS' may be > undefined > mbeq_119700.c:3909: warning: operation on 'comp_ip_ptr_RS' may be > undefined > mbeq_119700.c: In function 'runMChMBEq': > mbeq_119700.c:3893: warning: operation on 'comp_rp_ptr_L' may be > undefined > mbeq_119700.c:3894: warning: operation on 'comp_ip_ptr_L' may be > undefined > mbeq_119700.c:3896: warning: operation on 'comp_rp_ptr_R' may be > undefined > mbeq_119700.c:3897: warning: operation on 'comp_ip_ptr_R' may be > undefined > mbeq_119700.c:3899: warning: operation on 'comp_rp_ptr_C' may be > undefined > mbeq_119700.c:3900: warning: operation on 'comp_ip_ptr_C' may be > undefined > mbeq_119700.c:3902: warning: operation on 'comp_rp_ptr_S' may be > undefined > mbeq_119700.c:3903: warning: operation on 'comp_ip_ptr_S' may be > undefined > mbeq_119700.c:3905: warning: operation on 'comp_rp_ptr_LS' may be > undefined > mbeq_119700.c:3906: warning: operation on 'comp_ip_ptr_LS' may be > undefined > mbeq_119700.c:3908: warning: operation on 'comp_rp_ptr_RS' may be > undefined > mbeq_119700.c:3909: warning: operation on 'comp_ip_ptr_RS' may be > undefined > ------------------ > > As you can see, it complains _twice_ about errors/warnings in the same > function! I don't think this is right and as the segfault occurs in > exactly the function gcc complains about I think this is somehow > connected... > I do pay attention to such warnings, but I think in this case they are bogus. Let's have a look at 'comp_rp_ptr_L', for example - please have a look at output of 'grep': " [243] 11:16 [EMAIL PROTECTED]:/ibm/home/sergei> grep -n comp_rp_ptr_L swh-plugins/swh-plugins-0.4.11/mbeq_119700.c 3554: fftw_real *comp_rp_ptr_L = comp_L + 1; 3579: *comp_rp_ptr_L++ = *comp_rp_ptr_L * cos_times_gain; [244] 11:16 [EMAIL PROTECTED]:/ibm/home/sergei> grep -n comp_L swh-plugins/swh-plugins-0.4.11/mbeq_119700.c 254: fftw_real *comp_L; 466: free((*plugin_data).comp_L); 771: CALLOC_AND_CHECK(fftw_real *, (*plugin_data).comp_L, fft_length, sizeof(fftw_real), __LINE__) 788: (*plugin_data).plan_rc_L = fftwf_plan_r2r_1d(fft_length, (*plugin_data).real_L, (*plugin_data).comp_L, FFTW_R2HC, FFTW_MEASURE); 789: (*plugin_data).plan_cr_L = fftwf_plan_r2r_1d(fft_length, (*plugin_data).comp_L, (*plugin_data).real_L, FFTW_HC2R, FFTW_MEASURE); 1607: fftw_real *comp_L = (*plugin_data).comp_L; 3542: rfftw_one(plan_rc, real_L, comp_L); 3552: //fprintf(stderr, "just after direct FFT: *(comp_L + input_buffer_pos)=%G\n", (LADSPA_Data)(*(comp_L + input_buffer_pos))); 3554: fftw_real *comp_rp_ptr_L = comp_L + 1; 3556: // in comp_L array - DC is left intact 3558: fftw_real *comp_ip_ptr_L = comp_L + fft_length - 1; 3559: // the above is imaginary part pointer - end of comp_L array 3592: //fprintf(stderr, "just after multiplying by gain: *(comp_L + input_buffer_pos)=%G\n", (LADSPA_Data)(*(comp_L + input_buffer_pos))); 3601: rfftw_one(plan_cr, comp_L, real_L); [245] 11:16 [EMAIL PROTECTED]:/ibm/home/sergei> ". The 3554: fftw_real *comp_rp_ptr_L = comp_L + 1; line says that at the moment 'comp_rp_ptr_L' is created it gets its value from 'comp_L'. 'comp_L' gets initialized by 771: CALLOC_AND_CHECK(fftw_real *, (*plugin_data).comp_L, fft_length, sizeof(fftw_real), __LINE__) and '(*plugin_data).comp_L' is copied later into the 'comp_L' in question: 1607: fftw_real *comp_L = (*plugin_data).comp_L; So, as I see it, the chain of defined value passing is: 771: CALLOC_AND_CHECK(fftw_real *, (*plugin_data).comp_L, fft_length, sizeof(fftw_real), __LINE__) -> 1607: fftw_real *comp_L = (*plugin_data).comp_L; -> 3554: fftw_real *comp_rp_ptr_L = comp_L + 1; -> 3579: *comp_rp_ptr_L++ = *comp_rp_ptr_L * cos_times_gain; ... A stray thought - try '-O0' in gcc, i.e. no optimization - I recently fixed a non-crash problem using '-O0'. Regards, Sergei. -- Visit my http://appsfromscratch.berlios.de/ open source project. ------------------------------------------------------------------------- Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys - and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV _______________________________________________ Alsa-user mailing list Alsa-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/alsa-user
