Messages in this thread | | | Date | Tue, 11 Nov 2008 07:37:54 +0100 | From | Alain Knaff <> | Subject | Re: [PATCH] VFS: lseek(fd, 0, SEEK_CUR) race condition |
| |
Andrew Morton wrote: [...] > OK, I think that a concurrent lseek(fd, 0, SEEK_CUR) is a sufficiently > sane operation that this is worth doing. As you point out, there is no > other way of userspace doing what is effectively a read-only operation > - userspace would be entitled to wonder "ytf did the kernel rewrite the > file offset for that?". > > > Do the below additions look OK?
Yes, these additions look ok to me. Thanks for your quick and easy reply.
> From: Andrew Morton <akpm@linux-foundation.org> > > - fix coding-style > - fix default_llseek() as well
Wow, that one was missing an unlock_kernel()... I'm wondering why this hadn't let to some major SNAFU...
> - add comments
Thanks,
Alain
| |