diff options
| author | Heiko Carstens <heiko.carstens@de.ibm.com> | 2016-12-27 15:36:34 +0100 |
|---|---|---|
| committer | Martin Schwidefsky <schwidefsky@de.ibm.com> | 2017-01-16 07:27:54 +0100 |
| commit | 00de54c803d5526a791112e653614656365c5578 (patch) | |
| tree | 44a28936630b093bf582abf04f976b1062b35627 /arch/s390/mm | |
| parent | a2ce2a956814b52c062d17422ebbf87ce4f0158d (diff) | |
| download | linux-00de54c803d5526a791112e653614656365c5578.tar.gz linux-00de54c803d5526a791112e653614656365c5578.tar.xz | |
s390/mem_detect: fix memory type of first block
Fix a long-standing but currently irrelevant bug: the memory detection
code performs a tprot instruction on address zero to figure out if the
first memory chunk is readable or writable. Due to low address
protection the result is "read-only". If the memory detection code
would actually care, it would have to ignore the first memory
increment, but it adds the memory increment to writable memory anyway.
If memblock debugging is enabled this leads to an extra rather
surprising call which registers memory. To avoid this get rid of the
first misleading tprot call and simply assume that the first memory
increment is writable. Otherwise we wouldn't have reached the memory
detection code anyway.
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Diffstat (limited to 'arch/s390/mm')
| -rw-r--r-- | arch/s390/mm/mem_detect.c | 3 |
1 files changed, 2 insertions, 1 deletions
diff --git a/arch/s390/mm/mem_detect.c b/arch/s390/mm/mem_detect.c index 0389d8d24358..185e70d3e153 100644 --- a/arch/s390/mm/mem_detect.c +++ b/arch/s390/mm/mem_detect.c @@ -41,7 +41,8 @@ void __init detect_memory_memblock(void) memblock_set_bottom_up(true); do { size = 0; - type = tprot(addr); + /* assume lowcore is writable */ + type = addr ? tprot(addr) : CHUNK_READ_WRITE; do { size += rzm; if (max_physmem_end && addr + size >= max_physmem_end) |