Re: [patch] convert aio event reap to use atomic-op instead of spin_lock

[Jeff Moyer <jmoyer@redhat.com>]

==> On Thu, 12 Apr 2007 00:29:28 -0700, "Ken Chen" <kenchen@google.com> said:

Ken> On 4/11/07, Ken Chen <kenchen@google.com> wrote:
Ken> > On 4/11/07, Zach Brown <zach.brown@oracle.com> wrote:
Ken> > > First, I'll NAK this and all AIO patches until the patch
Ken> description > > says that it's been run through the regression
Ken> tests that we've started > > collecting in autotest.  They're
Ken> trivial to run, never fear:
Ken> >
Ken> > OK.  I will run those regression tests.

Ken> Unfortunately, the aio_dio_bugs test in autotest has bug in it
Ken> :-( We need stress test the "test code".

Ken> on stock 2.6.21-rc6 kernel:

Ken> [rock-me-baby]$ cd autotest/tests/aio_dio_bugs/src
Ken> [rock-me-baby]$ make [rock-me-baby]$
Ken> ./aio-free-ring-with-bogus-nr-pages
Ken> aio-free-ring-with-bogus-nr-pages: Error: io_setup returned -22,
Ken> expected -ENOMEM

Ken> hmm???  The problem is that the test code forgot to initialized
Ken> ctx variable and in the kernel, sys_io_setup returns EINVAL if
Ken> user address contain none-zero value.

Zach had already sent a fix for that.  I wonder why it didn't go in.

I didn't see any response to Zach's request for code that actually
tests out the shared ring buffer.  Do you have such code?

-Jeff