[Top] [All Lists]

Re: [PATCH V3] Make inode64 a remountable option

To: xfs@xxxxxxxxxxx
Subject: Re: [PATCH V3] Make inode64 a remountable option
From: Christoph Hellwig <hch@xxxxxxxxxxxxx>
Date: Fri, 17 Aug 2012 13:34:49 -0400
In-reply-to: <20120817144950.GA17092@xxxxxxxxxxxxxxxxxxxxxxxx>
References: <1345171178-10447-1-git-send-email-cmaiolino@xxxxxxxxxx> <20120817122437.GC2502@xxxxxxxxxxxxx> <20120817144950.GA17092@xxxxxxxxxxxxxxxxxxxxxxxx>
User-agent: Mutt/1.5.21 (2010-09-15)
On Fri, Aug 17, 2012 at 11:49:50AM -0300, Carlos Maiolino wrote:
> I don't think there is any lock protection around m_flags, I did a search on 
> the
> code and couldn't find anything protecting it. At a first glance though, I 
> don't
> think there is a need to protect it once this flag is managed only during 
> super
> operations - mount/umount/remount -
> Also, I *think* the sb->s_umount rw_semaphore is enough for protection, once 
> it
> protects the whole mount/umount operation, but I'm 100% sure of it.

Seems like rw_semaphore takes care of it, and we already have a trace
when setting XFS_MOUNT_FS_SHUTDOWN.  I guess it's fine as is for now,
but I'd love to have comment explaining why it's safe.

<Prev in Thread] Current Thread [Next in Thread>