fedab40576
With 64 bytes heap and 1 byte allocation on a big heap, we get: 0 1 2 3 4 5 6 7 | h | h | b | b | c | 1 | s | f | where - h: chunk0 header - b: buckets in chunk0 - c: chunk header for the first allocation - 1: chunk mem - s: solo free header - f: end marker / footer max_chunkid() was returning h->end_chunk - min_chunk_size(h), which is 5 because min_chunk_size() on a big heap is 2. This works if you don't have the solo free header at 6 and the heap is like: 0 1 2 3 4 5 6 | h | h | b | b | c | 1 | f | max_chunkid() in this case gives you 6 - 2 = 4, which is the right chunkid for the last chunk header. This commit replaces max_chunkid() with h->end_chunk and "<=" (less than or equal to) with "<" (less than), so that it always compares against the end maker chunkid, but the code won't touch the end maker itself. Signed-off-by: Yasushi SHOJI <yashi@spacecubics.com> |
||
|---|---|---|
| .. | ||
| CMakeLists.txt | ||
| Kconfig | ||
| Kconfig.cbprintf | ||
| assert.c | ||
| base64.c | ||
| bitarray.c | ||
| cbprintf.c | ||
| cbprintf_complete.c | ||
| cbprintf_nano.c | ||
| cbprintf_packaged.c | ||
| crc7_sw.c | ||
| crc8_sw.c | ||
| crc16_sw.c | ||
| crc32_sw.c | ||
| crc32c_sw.c | ||
| dec.c | ||
| fdtable.c | ||
| heap-validate.c | ||
| heap.c | ||
| heap.h | ||
| hex.c | ||
| json.c | ||
| mpsc_pbuf.c | ||
| mutex.c | ||
| notify.c | ||
| onoff.c | ||
| p4wq.c | ||
| printk.c | ||
| rb.c | ||
| reboot.c | ||
| ring_buffer.c | ||
| sem.c | ||
| thread_entry.c | ||
| timeutil.c | ||
| user_work.c | ||