Re: [PATCH v4 1/5] fs, block: refactor enum rw_hint
From: Kanchan Joshi <hidden>
Date: 2024-08-27 05:12:55
Also in:
linux-f2fs-devel, linux-fsdevel, linux-nvme, linux-scsi
From: Kanchan Joshi <hidden>
Date: 2024-08-27 05:12:55
Also in:
linux-f2fs-devel, linux-fsdevel, linux-nvme, linux-scsi
On 8/26/2024 11:14 PM, Bart Van Assche wrote:
On 8/26/24 10:06 AM, Kanchan Joshi wrote:quoted
Change i_write_hint (in inode), bi_write_hint (in bio) and write_hint (in request) to use u8 data-type rather than this enum.That sounds fishy to me. Why to increase the size of this enum? Why to reduce the ability of the compiler to perform type checking? I think this needs to be motivated clearly in the patch description.
Since inode/bio/request stopped using this, the __packed annotation did not seem to serve much purpose. But sure, I can retain the size/checks on the renamed enum (rw_life_hint) too. Motivation for keeping u8 in inode/bio/request is to represent another hint type. This is similar to ioprio where multiple io priority classes/values are expressed within an int type.