[Bug 275348] Lots of aio issues

From: <bugzilla-noreply_at_freebsd.org>
Date: Sun, 26 Nov 2023 08:57:31 UTC
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=275348

            Bug ID: 275348
           Summary: Lots of aio issues
           Product: Base System
           Version: 13.2-RELEASE
          Hardware: Any
                OS: Any
            Status: New
          Severity: Affects Only Me
          Priority: ---
         Component: kern
          Assignee: bugs@FreeBSD.org
          Reporter: pjfloyd@wanadoo.fr
 Attachment #246576 text/plain
         mime type:

Created attachment 246576
  --> https://bugs.freebsd.org/bugzilla/attachment.cgi?id=246576&action=edit
macOS aio testcase

FreeBSD 13.2-RELEASE-p5 amd64

I've been trying to improved the handling of aio_read(v) and aio_write(v) in
Valgrind.

One of the tests that I've been using is an existing aio test for Darwin. There
seem to be lots of issues when I run it on FreeBSD.

1. aio_read with an unaddressable buffer fails on macOS but not of FreeBSD. OK
that could be an implementation detail, I don't think that POSIX requires it to
fail.
2. aio_return doesn't seem to clear the status after such an aio_read, et seems
to remain stuck on EFAULT. To make that clear the sequence is

aio_read with NULL buf
aio_return
aio_read with valid buf

3. doing 2 consecutive aio_reads doesn't get a fail from the second - I would
expect the second to fail unless there has been a successful aio_return

4. doing a duplicate aio_return also does not fail

5. all the above points also apply to aio_write

I haven't checked the v versions yet

Added the Darwin testcase as an attachment.

-- 
You are receiving this mail because:
You are the assignee for the bug.