Re: [PATCH v3 05/13] fanotify: classify events that hold a file identifier

From: Jan Kara <jack@suse.cz>
To: Amir Goldstein <amir73il@gmail.com>
Cc: Jan Kara <jack@suse.cz>,
	Matthew Bobrowski <mbobrowski@mbobrowski.org>,
	linux-fsdevel@vger.kernel.org
Subject: Re: [PATCH v3 05/13] fanotify: classify events that hold a file identifier
Date: Wed, 28 Nov 2018 16:44:50 +0100	[thread overview]
Message-ID: <20181128154450.GK15604@quack2.suse.cz> (raw)
In-Reply-To: <20181128153312.GJ15604@quack2.suse.cz>

Sorry, sent previous reply too early.

On Wed 28-11-18 16:33:12, Jan Kara wrote:
> @@ -38,6 +50,21 @@ struct fanotify_event {
>  	struct pid *pid;
>  };
>  
> +/*
> + * Bit 0 of info.flags is set when event has fid information.
> + * event->info shares the same union address with event->path, so this helps
> + * us tell if event has fid or path.
> + */
> +#define __FANOTIFY_FID		1UL
> +#define FANOTIFY_HAS_FID(event)	((event)->info.flags & __FANOTIFY_FID)

Why isn't this an inline function? And then the name wouldn't have to be
all caps...

Also I'd prefer if we just enlarged struct fanotify_event by 8 bytes to
allow simple 'flags' field and info about possible embedded fid, rather
than playing these bit tricks. They save space but make code more subtle
and I don't file struct fanotify_event size so critical.

								Honza
-- 
Jan Kara <jack@suse.com>
SUSE Labs, CR