[Bug tapsets/15554] New: nfs.proc.commit_done probe returning invalid data

[Bug tapsets/15554] New: nfs.proc.commit_done probe returning invalid data

[dsmith at redhat dot com]

http://sourceware.org/bugzilla/show_bug.cgi?id=15554

            Bug ID: 15554
           Summary: nfs.proc.commit_done probe returning invalid data
           Product: systemtap
           Version: unspecified
            Status: NEW
          Severity: normal
          Priority: P2
         Component: tapsets
          Assignee: systemtap at sourceware dot org
          Reporter: dsmith at redhat dot com

The following kernel commit changed the nfs[34]_commit_done functions to take
the new 'nfs_commit_data' structures instead of 'nfs_write_data' structures as
they had in the past. But, both parameters were called 'data'.

====
commit 0b7c01533aa9f4a228d07d2768d084acb3a387bc
Author: Fred Isaman <iisaman@netapp.com>
Date:   Fri Apr 20 14:47:39 2012 -0400

    NFS: add a struct nfs_commit_data to replace nfs_write_data in commits

    Commits don't need the vectors of pages, etc. that writes do. Split out
    a separate structure for the commit operation.

    Signed-off-by: Fred Isaman <iisaman@netapp.com>
    Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
====

In the nfs[34]_commit_done probes, the $data variable in passed to the
__nfs_write_data_info() function, which casts the pointer to a 'nfs_write_data'
pointer. Since the 'nfs_write_data' and 'nfs_commit_data' structures are laid
out differently, the output of __nfs_write_data_info() function is wrong.

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

[Bug tapsets/15554] nfs.proc.commit_done probe returning invalid data

[dsmith at redhat dot com]

http://sourceware.org/bugzilla/show_bug.cgi?id=15554

David Smith <dsmith at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|---                         |FIXED

--- Comment #1 from David Smith <dsmith at redhat dot com> ---
(In some respects this bug is a symptom of bug #10300.)

This problem was fixed in commit fb68439. The underlying problem here is that
we had a typed pointer, passed it to a function (which loses type information),
then cast that address to the wrong pointer type. The fix changes the
nfs_proc.stp tapset to use macros, which preserves type information. If the
arguments to any of the nfs functions change again, we should get compile
errors and realize what has happened.

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