Re: [PATCH v2 01/14] btrfs: don't try to delete RAID stripe-extents if we don't need to

[Johannes Thumshirn <Johannes.Thumshirn@wdc.com>]

On 09.01.25 16:27, Johannes Thumshirn wrote:
> On 09.01.25 16:15, Filipe Manana wrote:
>> On Thu, Jan 9, 2025 at 2:39 PM Johannes Thumshirn
>> <Johannes.Thumshirn@wdc.com> wrote:
>>>
>>> On 09.01.25 13:35, Filipe Manana wrote:
>>>>> diff --git a/fs/btrfs/tests/raid-stripe-tree-tests.c b/fs/btrfs/tests/raid-stripe-tree-tests.c
>>>>> index 30f17eb7b6a8a1dfa9f66ed5508da42a70db1fa3..f060c04c7f76357e6d2c6ba78a8ba981e35645bd 100644
>>>>> --- a/fs/btrfs/tests/raid-stripe-tree-tests.c
>>>>> +++ b/fs/btrfs/tests/raid-stripe-tree-tests.c
>>>>> @@ -478,8 +478,9 @@ static int run_test(test_func_t test, u32 sectorsize, u32 nodesize)
>>>>>                    ret = PTR_ERR(root);
>>>>>                    goto out;
>>>>>            }
>>>>> -       btrfs_set_super_compat_ro_flags(root->fs_info->super_copy,
>>>>> +       btrfs_set_super_incompat_flags(root->fs_info->super_copy,
>>>>>                                            BTRFS_FEATURE_INCOMPAT_RAID_STRIPE_TREE);
>>>> This hunk seems unrelated to the rest of the patch, could be fixed in
>>>> a different patch in case it actually solves any problem (probably
>>>> not, but it's an incompat feature so it should be changed anyway).
>>>
>>> I'll make it a separate patch. RST is an incompat feature not a compat one.
>>>
>>> With this patch btrfs_delete_raid_extent() starts checking the incompat
>>> bit so it is fixing a 'problem'.
>>
>> Yes, but for that all that's needed is this call:
>>
>> btrfs_set_fs_incompat(root->fs_info, RAID_STRIPE_TREE);
>>
>> Right?
>>
>> Replacing the btrfs_set_super_compat_ro_flags() call with a call to
>> btrfs_set_super_incompat_flags() shouldn't be needed for this patch.
>> That's what I was referring to.
>>
> 
> Ah now I see the problem. I used btrfs_set_super_incompat_flags()
> instead of btrfs_set_fs_incompat() *facepalm*
> 

But when using btrfs_set_fs_incompat() we get the annoying btrfs_info() 
call about setting the flag. Which in case of a selftest is pointless.

Also btrfs_set_fs_incompat() calls btrfs_set_super_incompat_flags() 
internally, so I think using btrfs_set_super_incompat_flags() here is 
the way to go.