From: Aneesh Kumar K.V <[email protected]>

Commit c1f45d86a522d568aef541dbbc066ccac262b4c3 upstream.

nvdimm core use nd_pfn_validate when looking for devdax or fsdax namespace. In 
this
case device resources are allocated against nd_namespace_io dev. In-order to
allow remap of range in nd_pfn_clear_memmap_error(), move the device memmap
area clearing while initializing pfn namespace. With this device
resource are allocated against nd_pfn and we can use nd_pfn->dev for remapping.

This also avoids calling nd_pfn_clear_mmap_errors twice. Once while probing the
namespace and second while initializing a pfn namespace.

Signed-off-by: Aneesh Kumar K.V <[email protected]>
Link: 
https://lore.kernel.org/r/[email protected]
Signed-off-by: Dan Williams <[email protected]>
---
 drivers/nvdimm/pfn_devs.c |    8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/drivers/nvdimm/pfn_devs.c b/drivers/nvdimm/pfn_devs.c
index 8c5b13567f55..6e5b042f453e 100644
--- a/drivers/nvdimm/pfn_devs.c
+++ b/drivers/nvdimm/pfn_devs.c
@@ -591,7 +591,7 @@ int nd_pfn_validate(struct nd_pfn *nd_pfn, const char *sig)
                return -EOPNOTSUPP;
        }
 
-       return nd_pfn_clear_memmap_errors(nd_pfn);
+       return 0;
 }
 EXPORT_SYMBOL(nd_pfn_validate);
 
@@ -729,6 +729,8 @@ static int nd_pfn_init(struct nd_pfn *nd_pfn)
                sig = PFN_SIG;
 
        rc = nd_pfn_validate(nd_pfn, sig);
+       if (rc == 0)
+               return nd_pfn_clear_memmap_errors(nd_pfn);
        if (rc != -ENODEV)
                return rc;
 
@@ -796,6 +798,10 @@ static int nd_pfn_init(struct nd_pfn *nd_pfn)
        checksum = nd_sb_checksum((struct nd_gen_sb *) pfn_sb);
        pfn_sb->checksum = cpu_to_le64(checksum);
 
+       rc = nd_pfn_clear_memmap_errors(nd_pfn);
+       if (rc)
+               return rc;
+
        return nvdimm_write_bytes(ndns, SZ_4K, pfn_sb, sizeof(*pfn_sb), 0);
 }
 
_______________________________________________
Linux-nvdimm mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to