[Bug tapsets/17688] New: probe nfs.fop.aio_read no longer valid

["dsmith at redhat dot com" <sourceware-bugzilla at sourceware dot org>]

https://sourceware.org/bugzilla/show_bug.cgi?id=17688

            Bug ID: 17688
           Summary: probe nfs.fop.aio_read no longer valid
           Product: systemtap
           Version: unspecified
            Status: NEW
          Severity: normal
          Priority: P2
         Component: tapsets
          Assignee: systemtap at sourceware dot org
          Reporter: dsmith at redhat dot com

In tapset/linux/nfs.stp, we've got:

====
probe nfs.fop.aio_read = kernel.function ("nfs_file_read") !,
                         module("nfs").function("nfs_file_read")
====

The following kernel commit (first present in 3.16) makes the above probe
invalid:

<https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/fs/nfs/file.c?id=3aa2d199f8eb8149a88005e88736d583cbc39d31>

Here's the first part of that patch:

====
>From 3aa2d199f8eb8149a88005e88736d583cbc39d31 Mon Sep 17 00:00:00 2001
From: Al Viro <viro@zeniv.linux.org.uk>
Date: Wed, 2 Apr 2014 20:14:12 -0400
Subject: nfs: switch to ->read_iter()

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

diff --git a/fs/nfs/file.c b/fs/nfs/file.c
index ead8f44..200bdb0 100644
--- a/fs/nfs/file.c
+++ b/fs/nfs/file.c
@@ -165,26 +165,21 @@ nfs_file_flush(struct file *file, fl_owner_t id)
 EXPORT_SYMBOL_GPL(nfs_file_flush);

 ssize_t
-nfs_file_read(struct kiocb *iocb, const struct iovec *iov,
-        unsigned long nr_segs, loff_t pos)
+nfs_file_read(struct kiocb *iocb, struct iov_iter *to)
 {
     struct inode *inode = file_inode(iocb->ki_filp);
-    size_t count = iov_length(iov, nr_segs);
     ssize_t result;
-    struct iov_iter to;
-
-    iov_iter_init(&to, READ, iov, nr_segs, count);

     if (iocb->ki_filp->f_flags & O_DIRECT)
-        return nfs_file_direct_read(iocb, &to, pos, true);
+        return nfs_file_direct_read(iocb, to, iocb->ki_pos, true);

     dprintk("NFS: read(%pD2, %zu@%lu)\n",
         iocb->ki_filp,
-        count, (unsigned long) pos);
+        iov_iter_count(to), (unsigned long) iocb->ki_pos);

     result = nfs_revalidate_mapping(inode, iocb->ki_filp->f_mapping);
     if (!result) {
-        result = generic_file_read_iter(iocb, &to);
+        result = generic_file_read_iter(iocb, to);
         if (result > 0)
             nfs_add_stats(inode, NFSIOS_NORMALREADBYTES, result);
     }
@@ -945,9 +940,9 @@ EXPORT_SYMBOL_GPL(nfs_setlease);

 const struct file_operations nfs_file_operations = {
     .llseek        = nfs_file_llseek,
-    .read        = do_sync_read,
+    .read        = new_sync_read,
     .write        = do_sync_write,
-    .aio_read    = nfs_file_read,
+    .read_iter    = nfs_file_read,
     .aio_write    = nfs_file_write,
     .mmap        = nfs_file_mmap,
     .open        = nfs_file_open,
====

So not only does this patch change the function signature of nfs_file_read(),
it really isn't the 'aio_read' member of the nfs file operations structure
anymore. Instead it is the 'read_iter' member.

Some rethinking will be needed here.

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