On Jan 29, 2008 9:37 PM, Andrew Lunn <[EMAIL PROTECTED]> wrote: > On Thu, Jan 24, 2008 at 12:33:14PM +0100, ?yvind Harboe wrote: > > Allow filing system to access files during close(). Without > > this fix, any file access causes a deadlock during > > close(). > > It might stop a deadlock, but it opens up a race condition. Imaging a > close() and a dup() at the same time.
From what you write above, I draw the conclusion that you think, like me, that file system access during close() should be allowed. Correct? If we can agree on what correct behavior is, then I can get to work on making a fix that passes muster. -- Øyvind Harboe http://www.zylin.com - eCos ARM & FPGA developer kit
