On Fri, Feb 14, 2025 at 10:07:23AM +0800, Ming Lei wrote: > On Thu, Feb 13, 2025 at 07:41:46AM +0100, Christoph Hellwig wrote: > > On Sun, Feb 09, 2025 at 08:20:25PM +0800, Ming Lei wrote: > > > For each hctx, its debugfs path is fixed, which can be queried from > > > its parent dentry and hctx queue num, so it isn't necessary to cache > > > it in hctx structure because it isn't used in fast path. > > > > It's not needed, but it's also kinda convenient. What is the argument > > that speaks for recalculating the path? > > q->debugfs_lock will be not necessary if these cached entries are removed, > please see the last patch. Please document the reasoning (and tradeoff if there are any) in the commit log. No one will see your cover letter when bisecting a bug down to this commit in the future.
