Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.33 as permitted sender) client-ip=23.128.96.33;
Date:   Fri, 29 Sep 2023 13:59:30 +0100
From:   Matthew Wilcox <willy@infradead.org>
To:     Mateusz Guzik <mjguzik@gmail.com>
Cc:     brauner@kernel.org, viro@zeniv.linux.org.uk,
        linux-kernel@vger.kernel.org, linux-fsdevel@vger.kernel.org,
        torvalds@linux-foundation.org
Subject: Re: [PATCH v3] vfs: avoid delegating to task_work when cleaning up
 failed open
Message-ID: <ZRbKMmRm8i+/E88f@casper.infradead.org>
References: <20230928102516.186008-1-mjguzik@gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20230928102516.186008-1-mjguzik@gmail.com>
Precedence: bulk

On Thu, Sep 28, 2023 at 12:25:16PM +0200, Mateusz Guzik wrote:
> Below is my rebased patch + rewritten commit message with updated bench
> results. I decided to stick to fput_badopen name because with your patch
> it legitimately has to unref. Naming that "release_empty_file" or
> whatever would be rather misleading imho.

Do we still need fput_badopen()?  Couldn't we just make this part of
regular fput() at this point?  ie:

+++ b/fs/file_table.c
@@ -435,6 +435,10 @@ void fput(struct file *file)
        if (atomic_long_dec_and_test(&file->f_count)) {
                struct task_struct *task = current;
 
+               if (!(file->f_mode & FMODE_OPENED)) {
+                       file_free(file);
+                       return;
+               }
                if (likely(!in_interrupt() && !(task->flags & PF_KTHREAD))) {
                        init_task_work(&file->f_rcuhead, ____fput);
                        if (!task_work_add(task, &file->f_rcuhead, TWA_RESUME))