On 2017年10月31日 15:15, Nikolay Borisov wrote:
>
>
> On 31.10.2017 06:03, Qu Wenruo wrote:
>> For key type BTRFS_UUID_KEY_SUBVOL or BTRFS_UUID_KEY_RECEIVED_SUBVOL the
>> key objectid and key offset are just half of the UUID.
>>
>> However we just print the key as %llu, which is converted from little
>> endian, not byte order for UUID, nor the traditional 36 bytes human
>> readable uuid format.
>>
>> Although true engineer can easily convert it in their brain, but to
>> make it easier for search, output the result UUID using the 36 chars format.
>>
>> Cc: Misono Tomohiro <misono.tomohiro@xxxxxxxxxxxxxx>
>> Signed-off-by: Qu Wenruo <wqu@xxxxxxxx>
>> ---
>> Inspired by UUID related work from Misono.
>> ---
>> print-tree.c | 17 ++++++++++++++---
>> 1 file changed, 14 insertions(+), 3 deletions(-)
>>
>> diff --git a/print-tree.c b/print-tree.c
>> index 3c585e31f1fc..687f871db302 100644
>> --- a/print-tree.c
>> +++ b/print-tree.c
>> @@ -803,14 +803,25 @@ void btrfs_print_key(struct btrfs_disk_key *disk_key)
>> }
>> }
>>
>> -static void print_uuid_item(struct extent_buffer *l, unsigned long offset,
>> - u32 item_size)
>> +static void print_uuid_item(struct extent_buffer *l, int slot,
>> + unsigned long offset, u32 item_size)
>> {
>> + struct btrfs_key key;
>> + char uuid_str[BTRFS_UUID_UNPARSED_SIZE];
>> + u8 uuid[BTRFS_UUID_SIZE];
>> +
>> + /* Reassemble the uuid from key.objecitd and key.offset */
>> + btrfs_item_key_to_cpu(l, &key, slot);
>> + put_unaligned_le64(key.objectid, uuid);
>> + put_unaligned_le64(key.offset, uuid + sizeof(u64));
>
> I don't think this will work on a BE system. Because
> btrfs_item_key_to_cpu take the LE representation on-disk and turns it
> into a cpu representation which might very well be BE. And then you
> essentially reverse it by using put_unaligned_le64 for x86 it works fine
> due to it being a LE system.
I know this can be tricky, let's assume the following case:
UUID: 0x0123456789abcdef0123456789abcdef (byte order, no endian)
Low bit high bit
Key objectid: 0x0123456789abcdef (LE on-disk)
0xefcdab8967452301 (u64) <- CPU key.objectid
Low bit hight bit
key.offset: 0x123456789abcdef (LE on-disk)
0xefcdab8967452301 (u64) <- CPU key.offset
put_unaligned_le64 will convert CPU key.objectid/offset to LE on-disk
again, so we get
Low bit high bit
0x01 23456789abcd ef (LE on-disk)
uuid[0] ... uuid[7]
And that's what we need.
We did the LE->native and native->LE, so the result is not changed at all.
And we just need byte order, so the result is correct.
Just like what kernel did.
Thanks,
Qu
>
>
>> + uuid_unparse(uuid, uuid_str);
>> +
>> if (item_size & (sizeof(u64) - 1)) {
>> printf("btrfs: uuid item with illegal size %lu!\n",
>> (unsigned long)item_size);
>> return;
>> }
>> + printf("\t\tuuid %s\n", uuid_str);
>> while (item_size) {
>> __le64 subvol_id;
>>
>> @@ -1297,7 +1308,7 @@ void btrfs_print_leaf(struct btrfs_root *root, struct extent_buffer *eb)
>> break;
>> case BTRFS_UUID_KEY_SUBVOL:
>> case BTRFS_UUID_KEY_RECEIVED_SUBVOL:
>> - print_uuid_item(eb, btrfs_item_ptr_offset(eb, i),
>> + print_uuid_item(eb, i, btrfs_item_ptr_offset(eb, i),
>> btrfs_item_size_nr(eb, i));
>> break;
>> case BTRFS_STRING_ITEM_KEY: {
>>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
> the body of a message to majordomo@xxxxxxxxxxxxxxx
> More majordomo info at http://vger.kernel.org/majordomo-info.html
>
Attachment:
signature.asc
Description: OpenPGP digital signature
