On Tue, 18 Aug 2015 11:51:18 +0200, Alexandru Costache wrote:

Hi,

I was reproducing the crash that Valentin Corfu reported earlier with his test.

Valgrind and gdb show that expand_linear_s16() is somehow corrupting the area used for memory management by libc for rate->pareas[0].addr

More exactly, the line that seems to corrupt the heap in

static void expand_linear_s16() is:

*dst = (old_sample * old_weight + new_sample * new_weight) >> 16;

As Valgrind says:

==11161== Invalid write of size 2 ==11161== at 0x40A8A82: linear_expand_s16 (pcm_rate_linear.c:179) ==11161== by 0x40A85A9: linear_convert (pcm_rate_linear.c:320) ==11161== by 0x40A6915: do_convert (pcm_rate.c:537) ==11161== by 0x40A6C7D: snd_pcm_rate_write_areas1 (pcm_rate.c:550) ==11161== by 0x40A6C7D: snd_pcm_rate_commit_area (pcm_rate.c:749) ==11161== by 0x40A7014: snd_pcm_rate_drain (pcm_rate.c:1105) ==11161== by 0x4084511: snd_pcm_drain (pcm.c:1122) ==11161== by 0x8049288: main (in /root/pcm_min_daisuke) ==11161== Address 0x43d6258 is 0 bytes after a block of size 86,016 alloc'd ==11161== at 0x40261B1: malloc (in /usr/lib/valgrind/vgpreload_memcheck-x86-linux.so) ==11161== by 0x4099801: snd_pcm_mmap (pcm_mmap.c:425) ==11161== by 0x4093CA7: sndrv_pcm_hw_params (pcm_params.c:2366) ==11161== by 0x409A59C: snd1_pcm_generic_hw_params (pcm_generic.c:104) ==11161== by 0x4091E7F: snd1_pcm_hw_params_slave (pcm_params.c:2250) ==11161== by 0x40A54AC: snd_pcm_rate_hw_params (pcm_rate.c:246) ==11161== by 0x4093AE3: sndrv_pcm_hw_params (pcm_params.c:2326) ==11161== by 0x40ABC37: snd_pcm_plug_hw_params (pcm_plug.c:1045) ==11161== by 0x4093AE3: sndrv_pcm_hw_params (pcm_params.c:2326) ==11161== by 0x4084281: snd_pcm_hw_params (pcm.c:830) ==11161== by 0x8048F84: main (in /root/pcm_min_daisuke) ==11161== ==11161== Invalid free() / delete / delete[] / realloc() ==11161== at 0x402726D: free (in /usr/lib/valgrind/vgpreload_memcheck-x86-linux.so) ==11161== by 0x40A53C0: snd_pcm_rate_hw_free (pcm_rate.c:344) ==11161== by 0x4083CDA: snd_pcm_hw_free (pcm.c:853) ==11161== by 0x40A9E8D: snd_pcm_plug_hw_free (pcm_plug.c:1061) ==11161== by 0x4083CDA: snd_pcm_hw_free (pcm.c:853) ==11161== by 0x80492AC: main (in /root/pcm_min_daisuke) ==11161== Address 0xe705004b is not stack'd, malloc'd or (recently) free'd

I'm not very familiar with alsa in general and was wondering if someone who knows more on it can drop a look on this one?

This is very much dependent on the setup. Please clarify your setup and the exact code you're using.

Takashi