Re: [PATCH] man2/fallocate.2: add doc for FALLOC_FL_WRITE_ZEROES

> FALLOC_FL_WRITE_ZEROES itself is relatively flexible. It does not
> mandate that the filesystem must implement zeroing via unmap write
> zeroes. That depends on the specific implementation of each filesystem.
> Currently, ext4, xfs, and block devices (bdev) support this operation
> through that approach, since writing zeros directly on devices that do
> not support unmap zeroes can be very slow and thus of little practical
> value.
> 
> Therefore, I would suggest referring to the description in, for
> example, fallocate(1):
> 
> https://man7.org/linux/man-pages/man1/fallocate.1.html
> 
>   "Zeroing is done within the filesystem. The
>    filesystem may use a hardware-accelerated zeroing command or may
>    submit regular writes. The behavior depends on the filesystem design
>    and the available hardware."
> 
> What do you think?
> 

You are right. It is up to the FSs to implement however it wants.
I will use this framing instead. Thanks.

>> +This approach means that the specified range will be physically zeroed out
>> +on the device (including partial blocks at the either end of the range).
>> +.P
>> +No other flags may be specified in
>> +.I mode
>> +in conjunction with
>> +.BR FALLOC_FL_WRITE_ZEROES .
>> +.P
>> +Not all filesystems support
>> +.BR FALLOC_FL_WRITE_ZEROES ;
>> +if a filesystem doesn't support the operation, an error is returned.
>> +The operation is supported only in ext4 (since Linux 6.17).
>                                       ^^^^
> ext4 (only for extent-based files)
> 

Sounds good.

--
Pankaj