On some platforms cold reboot may not zero out DRAM and
as a consequence, the next kernel boot with a new pmem layout
change may end up using stale superblocks. This can cause
altmap to run out of memory and consequently the pmem drivers
may fail to probe.
failed to free all reserved pages
[trimmed]
Call trace:
memunmap_pages+0x2f8/0x358
memremap_pages+0x590/0x728
devm_memremap_pages+0x30/0x88
pmem_attach_disk+0x178/0x488
nd_pmem_probe+0x60/0xe0
nvdimm_bus_probe+0xec/0x208
really_probe+0xc0/0x2e8
__driver_probe_device+0xc4/0x168
driver_probe_device+0x48/0x218
__device_attach_driver+0x130/0x160
bus_for_each_drv+0x94/0xf0
__device_attach+0xfc/0x1a0
device_initial_probe+0x18/0x28
bus_probe_device+0x40/0xa0
device_add+0x270/0x498
nd_async_device_register+0x20/0x70
async_run_entry_fn+0x30/0x158
process_scheduled_works+0x248/0x4b0
worker_thread+0x284/0x3a0
kthread+0xe0/0x180
ret_from_fork+0x10/0x20
---[ end trace 0000000000000000 ]---
nd_pmem: probe of pfnX.1 failed with error -12
Fix this by discarding superblock if there is a mismatch b/w
dataoffset value from the preserved superblock and the
dataoffset from the resource.
While this fix is not robust but it can be combined with userspace
namespace destruction and recreation to remove all the stale pmems
metadata and reconfigure the fresh data using ndctl
Signed-off-by: Shyam Saini <[email protected]>
---
drivers/nvdimm/pfn_devs.c | 12 +++++++++---
1 file changed, 9 insertions(+), 3 deletions(-)
diff --git a/drivers/nvdimm/pfn_devs.c b/drivers/nvdimm/pfn_devs.c
index 8fa9c16aba7ea..667115ce88e40 100644
--- a/drivers/nvdimm/pfn_devs.c
+++ b/drivers/nvdimm/pfn_devs.c
@@ -439,12 +439,12 @@ static bool nd_supported_alignment(unsigned long align)
*/
int nd_pfn_validate(struct nd_pfn *nd_pfn, const char *sig)
{
- u64 checksum, offset;
+ u64 checksum, offset, offset_res;
struct resource *res;
enum nd_pfn_mode mode;
- resource_size_t res_size;
+ resource_size_t res_start, res_size;
struct nd_namespace_io *nsio;
- unsigned long align, start_pad, end_trunc;
+ unsigned long align, start_pad, end_trunc, npfns;
struct nd_pfn_sb *pfn_sb = nd_pfn->pfn_sb;
struct nd_namespace_common *ndns = nd_pfn->ndns;
const uuid_t *parent_uuid = nd_dev_to_uuid(&ndns->dev);
@@ -577,6 +577,12 @@ int nd_pfn_validate(struct nd_pfn *nd_pfn, const char *sig)
nsio = to_nd_namespace_io(&ndns->dev);
res = &nsio->res;
res_size = resource_size(res);
+ res_start = res->start;
+ npfns = PHYS_PFN(res_size - SZ_8K);
+ offset_res = ALIGN(res_start + SZ_8K + (MAX_STRUCT_PAGE_SIZE * npfns),
align) - res_start;
+ if ((mode == PFN_MODE_PMEM) && offset_res != offset)
+ return -ENODEV;
+
if (offset >= res_size) {
dev_err(&nd_pfn->dev, "pfn array size exceeds capacity of %s\n",
dev_name(&ndns->dev));
--
2.43.0
