tlsf memory manager status weirdness

[lazedo]

Description

TLSF mem manager accounting for wrong values

was trying to debug a crash and this popped up.

using a lot of small allocs until pkg_alloc returns NULL (when exhausted)
when NULL is returned the memory used+overhead exceeds the heap allocated.

Troubleshooting

Reproduction

use frequent small memory allocs until pool is exausted (returns NULL) , check tlsf status.

static void do_test_mem()
{
    void* chunk;
    tlsf_t pool = NULL;
    size_t total = 16 * 1024 * 1024;
    size_t half = 8 * 1024 * 1024;
    size_t sz = 6;
    int x;
    void* mem = malloc(total);
    char* mem2 = tlsf_cast(char*, tlsf_cast(ptrdiff_t, mem) + half);
    memset(mem2, 'X', half);
    assert(*mem2 == 'X');
    pool = tlsf_create_with_pool(mem, half);
    do { chunk = tlsf_malloc(pool, sz); } while(chunk != NULL);
    tlsf_status(pool);
    for(x=0; x < half; x++) {
        assert(*mem2 == 'X');
        mem2++;
    }
    free(mem);
}

heap size= **8388592**
used= **6285144**, used+overhead=**12578696**, free=**18446744073705361512**, fragments=0
max used (+overhead)=12578696, max fragments=1
Free blocks matrix ('.': none, 'X': between 2^X and (2^(X+1)-1) free blocks, X=A..Z, A=0, B=1, ...)
> first-level: 32 block list arrays between 2^fl and 2^(fl+1) bytes (fl=8..39)
v second-level: 32 block lists between 2^fl+sl*2^(fl-5) and 2^fl+(sl+1)*2^(fl-5)-1 bytes (sl=0..31)
 0|................................|
 1|................................|
 2|................................|
 3|................................|
 4|................................|
 5|................................|
 6|................................|
 7|................................|
 8|................................|
 9|................................|
10|................................|
11|................................|
12|................................|
13|................................|
14|................................|
15|................................|
16|................................|
17|................................|
18|................................|
19|................................|
20|................................|
21|................................|
22|................................|
23|................................|
24|................................|
25|................................|
26|................................|
27|................................|
28|................................|
29|................................|
30|................................|
31|................................|

Log Messages

identical to reproduce step when calling mem debug on a running instance

Possible Solutions

Additional Information

  * **Kamailio Version** - built from master

• Operating System:

CentOS7 and alpine

comments

heap size= 8388592 shouldn't it be 8388608 ?
used+overhead=12578696 misleading
free=18446744073705361512 misleading, looks like it has a negative value
the map doesn't look ok (or i'm not understanding what it should show), its the same output before the alloc and after.

[miconda]

The original developer of the code is no longer active, I can try to look over it as I get some time during next days, if nobody else does it meanwhile.

@lazedo - as we help with this one, can you take care of #976 ? It is an old one related to the kazoo module that you maintain -- check if it is still something to fix the docs or if already done, close it.

[github-actions]

This issue is stale because it has been open 6 weeks with no activity. Remove stale label or comment or this will be closed in 2 weeks.