On 3/26/20 3:10 PM, Michal Hocko wrote:
On Wed 25-03-20 08:49:14, Aneesh Kumar K.V wrote:
Fixes the below crash
BUG: Kernel NULL pointer dereference on read at 0x00000000
Faulting instruction address: 0xc000000000c3447c
Oops: Kernel access of bad area, sig: 11 [#1]
LE PAGE_SIZE=64K MMU=Hash SMP NR_CPUS=2048 NUMA pSeries
CPU: 11 PID: 7519 Comm: lt-ndctl Not tainted 5.6.0-rc7-autotest #1
...
NIP [c000000000c3447c] vmemmap_populated+0x98/0xc0
LR [c000000000088354] vmemmap_free+0x144/0x320
Call Trace:
section_deactivate+0x220/0x240
It would be great to match this to the specific source code.
The crash is due to NULL dereference at
test_bit(idx, ms->usage->subsection_map); due to ms->usage = NULL;
that is explained in later part of the commit.
__remove_pages+0x118/0x170
arch_remove_memory+0x3c/0x150
memunmap_pages+0x1cc/0x2f0
devm_action_release+0x30/0x50
release_nodes+0x2f8/0x3e0
device_release_driver_internal+0x168/0x270
unbind_store+0x130/0x170
drv_attr_store+0x44/0x60
sysfs_kf_write+0x68/0x80
kernfs_fop_write+0x100/0x290
__vfs_write+0x3c/0x70
vfs_write+0xcc/0x240
ksys_write+0x7c/0x140
system_call+0x5c/0x68
With commit: d41e2f3bd546 ("mm/hotplug: fix hot remove failure in SPARSEMEM|!VMEMMAP case")
section_mem_map is set to NULL after depopulate_section_mem(). This
was done so that pfn_page() can work correctly with kernel config that disables
SPARSEMEM_VMEMMAP. With that config pfn_to_page does
__section_mem_map_addr(__sec) + __pfn;
where
static inline struct page *__section_mem_map_addr(struct mem_section *section)
{
unsigned long map = section->section_mem_map;
map &= SECTION_MAP_MASK;
return (struct page *)map;
}
Now with SPASEMEM_VMEMAP enabled, mem_section->usage->subsection_map is used to
check the pfn validity (pfn_valid()). Since section_deactivate release
mem_section->usage if a section is fully deactivated, pfn_valid() check after
a subsection_deactivate cause a kernel crash.
static inline int pfn_valid(unsigned long pfn)
{
...
return early_section(ms) || pfn_section_valid(ms, pfn);
}
where
static inline int pfn_section_valid(struct mem_section *ms, unsigned long pfn)
{
int idx = subsection_map_index(pfn);
return test_bit(idx, ms->usage->subsection_map);
}
Avoid this by clearing SECTION_HAS_MEM_MAP when mem_section->usage is freed.
I am sorry, I haven't noticed that during the review of the commit
mentioned above. This is all subtle as hell, I have to say.
Why do we have to free usage before deactivaing section memmap? Now that
we have a late section_mem_map reset shouldn't we tear down the usage in
the same branch?
We still need to make the section invalid before we call into
depopulate_section_memmap(). Because architecture like powerpc can share
vmemmap area across sections (16MB mapping of vmemmap area) and we use
vmemmap_popluated() to make that decision.
Fixes: d41e2f3bd546 ("mm/hotplug: fix hot remove failure in SPARSEMEM|!VMEMMAP case")
Cc: Baoquan He <bhe@xxxxxxxxxx>
Reported-by: Sachin Sant <sachinp@xxxxxxxxxxxxxxxxxx>
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@xxxxxxxxxxxxx>
---
mm/sparse.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/mm/sparse.c b/mm/sparse.c
index aadb7298dcef..3012d1f3771a 100644
--- a/mm/sparse.c
+++ b/mm/sparse.c
@@ -781,6 +781,8 @@ static void section_deactivate(unsigned long pfn, unsigned long nr_pages,
ms->usage = NULL;
}
memmap = sparse_decode_mem_map(ms->section_mem_map, section_nr);
+ /* Mark the section invalid */
+ ms->section_mem_map &= ~SECTION_HAS_MEM_MAP;
Btw. this comment is not really helping at all.
That is marking the section invalid so that
static inline int valid_section(struct mem_section *section)
{
return (section && (section->section_mem_map & SECTION_HAS_MEM_MAP));
}
returns false.
/*
* section->usage is gone and VMEMMAP's pfn_valid depens
* on it (see pfn_section_valid)
*/
}
if (section_is_early && memmap)
--
2.25.1
