Re: 2.6 kernel failing lofs mount from lsb-runtime-test

On Wed, 4 Feb 2004, Wichmann, Mats D wrote:

> Posted here in case anyone knows about it
> and can save us some time chasing I down.
> 
> It appears the the lsb-runtime-test shows
> a number of additional failures when running
> on a 2.6 kernel.
> 
> The new errors appear to be in mounting the
> loopback filesystem that is constructed by
> the test suite to test handling of out-of-
> space conditions by various interfaces.
> 
> The construction and mounting of the disk
> image of a small ext2 filesystem works just
> fine if done by hand.
> 
> Any ideas?
> 

Just some more data points since I'm currently beating my head against 
this wall.

First failure: (all very repeatable)

      /tset/ANSI.os/streamio/Mgetc/T.getc 11  Unresolved
        
                Test Description:
        If the implementation supports a read-only file system:
           When the file pointer references a file on a read-only file 
system,
           then getc() does not update the time related element st_atime 
of
           the file.
        Posix Ref: Component GETC Assertion 8.2.3-X1(A)
        
                Test Information:
        fopen("getc-d.11/getc-t.11", "r") failed - errno 2 (ENOENT)

After that, a similar error in:

/tset/ANSI.os/streamio/Mgetc/T.getchar 11

and then:

       /tset/ANSI.os/streamio/Mputc/T.fputc 11 Unresolved
        
                Test Description:
        ENOSPC in errno and a return value of EOF when fputc() attempts to
        write to a device on which there is no more space available.
        Posix Ref: Component FPUTC Assertion 8.2.3.11-09(B)
        
                Test Information:
        creat failed in setupnospc

for several of the streamio tests

Once it's all done, /proc/mounts shows a number of /dev/loop0 mount points 
for these tests, although "mount" doesn't show anything mounted.

Using an abrreviated scen.exec for just the relevant tests, leaves 2 mount 
points behind. stat -f shows:

stat -f d.putc/putc-d.11 (or d.fputc/putc-d.11)
ID: 0 Namelen: 255 Type: ext2/ext3
Blocks: Total: 1963	Free: 16 Available: 0 Size: 1024
Inodes: Total: 256	Free: 3

I've found that I need to uninstall/reinstall the test suite to get back 
to clean runs. Rebooting back to a 2.4 kernel isn't sufficient.

This behavior is the same for a generic kernel.org 2.6.1, or our cooker 
2.6.x kernels. Base filesystem mounted either ext2 or ext3.

-- 
Stew Benedict
--
MandrakeSoft