Date: Wed, 13 Apr 2005 18:38:42 -0700 From: Alfred Perlstein <alfred@freebsd.org> To: Jeff Roberson <jroberson@chesapeake.net> Cc: cvs-all@FreeBSD.org Subject: Re: cvs commit: src/sys/coda coda_vnops.c src/sys/compat/linux linux_getcwd.c src/sys/fs/devfs devfs_vnops.c src/sys/fs/hpfs hpfs_vnops.c src/sys/fs/msdosfs msdosfs_lookup.c src/sys/fs/nwfs nwfs_vnops Message-ID: <20050414013842.GH60345@elvis.mu.org> In-Reply-To: <20050413182958.L28571@mail.chesapeake.net> References: <200504131059.j3DAxAUf039007@repoman.freebsd.org> <20050413195741.GE60345@elvis.mu.org> <20050413182958.L28571@mail.chesapeake.net>
next in thread | previous in thread | raw e-mail | index | archive | help
* Jeff Roberson <jroberson@chesapeake.net> [050413 15:31] wrote: > On Wed, 13 Apr 2005, Alfred Perlstein wrote: > > > * Jeff Roberson <jeff@FreeBSD.org> [050413 03:59] wrote: > > > jeff 2005-04-13 10:59:10 UTC > > > > > > FreeBSD src repository > > > > > > Modified files: > > > sys/coda coda_vnops.c > > > sys/compat/linux linux_getcwd.c > > > Log: > > > - Change all filesystems and vfs_cache to relock the dvp once the child is > > > locked in the ISDOTDOT case. Se vfs_lookup.c r1.79 for details. > > > > Sorry to bumble into your way, but can you explain this? I thought > > the lock order was always dvp->vp, doing a vp->dvp lock is a reversal > > and can cause deadlock. I guess I'm missing something, can you explain > > please? > > For ISDOTDOT dvp is really the child of vp. For example > > a/b/c > > b/.. is a, and a must be locked before b. So b must be unlocked before we > can lock ..(a), and then b must be locked again. Ok thanks! -- - Alfred Perlstein - email: bright@mu.org cell: 408-480-4684
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20050414013842.GH60345>