Dear Maintainer,
the original staden-io-lib seems to be built with
libhtscodecs2 in version 1.2.2-1 [1].
The rebuild seems to be using libhtscodecs2 in version 1.3.0-4.
Unfortunately htscodecs upstream integrated this commit [3],
which renamed e.g. function encode_names to tok3_encode_names.
It added a compatibility function encode_names, but does not
declare it in the header file.
Therefore in the build of staden-io-lib the function encode_names
is implicitly declared - therefore seems to default to
a return type of int [4] - and therefore the returned pointer
gets truncated to the lower 4 bytes [5].
This get through unnoticed until the pointer is freed and
produces a segfault there [6].
I guess the right thing would be to have the short function name
in the header file of htscodecs, therefore solve this issue for
other packages using htscodecs.
A short term solution might be the patch in [7] which makes
the build and tests succeed without "implicit declarations".
Kind regards,
Bernhard
[1]
https://buildd.debian.org/status/fetch.php?pkg=staden-io-lib&arch=amd64&ver=1.14.14%2Bdfsg-1%2Bb1&stamp=1663567926&raw=0
[2] https://tracker.debian.org/pkg/htscodecs
[3]
https://github.com/jkbonfield/htscodecs/commit/6211b208d2bd21e93f3f62c0cd0d8c43546f98b5
[4]
cram_io.c: In function ‘cram_compress_by_method’:
cram_io.c:2420:23: warning: implicit declaration of function
‘encode_names’; did you mean ‘tok3_encode_names’?
[-Wimplicit-function-declaration]
2420 | uint8_t *cp = encode_names(in, in_size, lev, strat,
&out_len, NULL);
| ^~~~~~~~~~~~
| tok3_encode_names
cram_io.c:2420:23: warning: initialization of ‘uint8_t *’ {aka ‘unsigned
char *’} from ‘int’ makes pointer from integer without a cast [-Wint-conversion]
[5]
(rr) bt
#0 tok3_encode_names (blk=<optimized out>, len=<optimized out>, level=<optimized out>,
use_arith=<optimized out>, out_len=<optimized out>, last_start_p=<optimized out>) at
./htscodecs/tokenise_name3.c:1540
#1 0x00007f281aa34454 in cram_compress_by_method (s=s@entry=0x55e9af757e50, in=0x7f281000e930
"s0", in_size=9, out_size=out_size@entry=0x7f2819611730, method=method@entry=TOK3,
level=<optimized out>, strat=0, content_id=<optimized out>) at ./io_lib/cram_io.c:2420
#2 0x00007f281aa38ebe in cram_compress_block (fd=fd@entry=0x55e9af71f930,
s=s@entry=0x55e9af757e50, b=0x55e9af758830, metrics=0x55e9af731f50,
method=65794, level=level@entry=5) at ./io_lib/cram_io.c:2562
#3 0x00007f281aa22d75 in cram_compress_slice (s=0x55e9af757e50, c=<optimized
out>, fd=0x55e9af71f930) at ./io_lib/cram_encode.c:951
#4 cram_encode_slice (fd=fd@entry=0x55e9af71f930,
c=c@entry=0x55e9af73c270, h=h@entry=0x55e9af73c520, s=0x55e9af757e50) at
./io_lib/cram_encode.c:1219
#5 0x00007f281aa27881 in cram_encode_container (fd=<optimized out>,
c=<optimized out>) at ./io_lib/cram_encode.c:2119
#6 0x00007f281aa33fb0 in cram_flush_thread (arg=0x55e9af4d0500) at
./io_lib/cram_io.c:4328
#7 0x00007f281aa44272 in t_pool_worker (arg=0x55e9af737460) at
./io_lib/thread_pool.c:434
#8 0x00007f281a68784a in start_thread (arg=<optimized out>) at
./nptl/pthread_create.c:442
#9 0x00007f281a70a530 in clone () at
../sysdeps/unix/sysv/linux/x86_64/clone.S:100
(rr) print out
$6 = (uint8_t *) 0x7f2810010320 "\t"
(rr) finish
Run till exit from #0 tok3_encode_names (blk=<optimized out>, len=<optimized out>,
level=<optimized out>, use_arith=<optimized out>, out_len=<optimized out>,
last_start_p=<optimized out>) at ./htscodecs/tokenise_name3.c:1540
0x00007f281aa34454 in cram_compress_by_method (s=s@entry=0x55e9af757e50, in=0x7f281000e930
"s0", in_size=9, out_size=out_size@entry=0x7f2819611730, method=method@entry=TOK3,
level=<optimized out>, strat=0, content_id=<optimized out>) at ./io_lib/cram_io.c:2420
2420 uint8_t *cp = encode_names(in, in_size, lev, strat,
&out_len, NULL);
Value returned is $8 = (uint8_t *) 0x7f2810010320 "\t"
(rr) stepi
2421 *out_size = out_len;
(rr) print cp
$7 = (uint8_t *) 0x10010320 <error: Cannot access memory at address
0x10010320>
(rr) list
2416 int out_len;
2417 int lev = level;
2418 if (method == NAME_TOK3 && lev > 3)
2419 lev = 3;
2420 uint8_t *cp = encode_names(in, in_size, lev, strat,
&out_len, NULL);
2421 *out_size = out_len;
2422 return (char *)cp;
2423 }
2424
2425 case RAW:
[6]
Thread 2 received signal SIGSEGV, Segmentation fault.
0x00007f281a69788a in __GI___libc_free (mem=0x10010320) at
./malloc/malloc.c:3363
3363 ./malloc/malloc.c: Datei oder Verzeichnis nicht gefunden.
(rr) bt
#0 0x00007f281a69788a in __GI___libc_free (mem=0x10010320) at
./malloc/malloc.c:3363
#1 0x00007f281aa39238 in cram_compress_block (fd=fd@entry=0x55e9af71f930,
s=s@entry=0x55e9af757e50, b=0x55e9af758830, metrics=0x55e9af731f50,
method=65794, level=level@entry=5) at ./io_lib/cram_io.c:2575
#2 0x00007f281aa22d75 in cram_compress_slice (s=0x55e9af757e50, c=<optimized
out>, fd=0x55e9af71f930) at ./io_lib/cram_encode.c:951
#3 cram_encode_slice (fd=fd@entry=0x55e9af71f930,
c=c@entry=0x55e9af73c270, h=h@entry=0x55e9af73c520, s=0x55e9af757e50) at
./io_lib/cram_encode.c:1219
#4 0x00007f281aa27881 in cram_encode_container (fd=<optimized out>,
c=<optimized out>) at ./io_lib/cram_encode.c:2119
#5 0x00007f281aa33fb0 in cram_flush_thread (arg=0x55e9af4d0500) at
./io_lib/cram_io.c:4328
#6 0x00007f281aa44272 in t_pool_worker (arg=0x55e9af737460) at
./io_lib/thread_pool.c:434
#7 0x00007f281a68784a in start_thread (arg=<optimized out>) at
./nptl/pthread_create.c:442
#8 0x00007f281a70a530 in clone () at
../sysdeps/unix/sysv/linux/x86_64/clone.S:100
(rr) print mem
$10 = (void *) 0x10010320
(rr) up
#1 0x00007f281aa39238 in cram_compress_block (fd=fd@entry=0x55e9af71f930,
s=s@entry=0x55e9af757e50, b=0x55e9af758830, metrics=0x55e9af731f50,
method=65794, level=level@entry=5) at ./io_lib/cram_io.c:2575
2575 free(c);
(rr) list
2570 method_best = m;
2571 if (c_best)
2572 free(c_best);
2573 c_best = c;
2574 } else if (c) {
2575 free(c);
2576 } else {
2577 sz[m] = b->uncomp_size*2+1000; //
arbitrarily worse than raw
2578 }
2579
[7]
--- staden-io-lib-1.14.14+dfsg.orig/io_lib/cram_codecs.c
+++ staden-io-lib-1.14.14+dfsg/io_lib/cram_codecs.c
@@ -1970,10 +1970,10 @@ static int cram_xrle_decode_expand_char(
int nb = var_get_u64(len_dat, len_dat+len_sz, &out_sz);
if (!(b->data = malloc(out_sz)))
return -1;
- rle_decode(lit_dat, lit_sz,
- len_dat+nb, len_sz-nb,
- rle_syms, rle_nsyms,
- b->data, &out_sz);
+ hts_rle_decode(lit_dat, lit_sz,
+ len_dat+nb, len_sz-nb,
+ rle_syms, rle_nsyms,
+ b->data, &out_sz);
b->uncomp_size = out_sz;
return 0;
@@ -2139,10 +2139,10 @@ int cram_xrle_encode_flush(cram_codec *c
int nb = var_put_u64(out_len, NULL, c->e_xrle.to_flush_size);
- out_lit = rle_encode((uint8_t *)c->e_xrle.to_flush, c->e_xrle.to_flush_size,
- out_len+nb, &out_len_size,
- rle_syms, &rle_nsyms,
- NULL, &out_lit_size);
+ out_lit = hts_rle_encode((uint8_t *)c->e_xrle.to_flush,
c->e_xrle.to_flush_size,
+ out_len+nb, &out_len_size,
+ rle_syms, &rle_nsyms,
+ NULL, &out_lit_size);
out_len_size += nb;
--- staden-io-lib-1.14.14+dfsg.orig/io_lib/cram_io.c
+++ staden-io-lib-1.14.14+dfsg/io_lib/cram_io.c
@@ -2162,7 +2162,7 @@ int cram_uncompress_block(cram_block *b)
case NAME_TOK3: {
uint32_t out_len;
- uint8_t *cp = decode_names(b->data, b->comp_size, &out_len);
+ uint8_t *cp = tok3_decode_names(b->data, b->comp_size, &out_len);
b->orig_method = NAME_TOK3;
b->method = RAW;
free(b->data);
@@ -2417,7 +2417,7 @@ static char *cram_compress_by_method(cra
int lev = level;
if (method == NAME_TOK3 && lev > 3)
lev = 3;
- uint8_t *cp = encode_names(in, in_size, lev, strat, &out_len, NULL);
+ uint8_t *cp = tok3_encode_names(in, in_size, lev, strat, &out_len,
NULL);
*out_size = out_len;
return (char *)cp;
}
