fs core fixes (was Re: TFTP driver changes)
Eugeny S. Mints
jack at oktet.ru
Mon Oct 15 09:58:10 UTC 2001
Please, have a look at my letter to Till with CC to you.
There is the patch to unmount.c: in addition to your
comments for unmount.c below there are: order of unmount_h
and fs_unmounmt_h calls changed and call
rtems_filesystem_freenode for (&fs_root_loc) added in a
proper(as it seems to me) place in unmoun function.
Call to free fs_root_loc.mt_entry->mt_fs_root in unmount
On Fri, 12 Oct 2001, jennifer averett wrote:
> I am currently attempting to go through the patch.
> Please correct me if I am wrong about what the patch
> does to each of the files, these will be the comments
> put into the log files.
> utime.c - Added verification that utime is supported by the filesystem
> Added missing call to rtems_filesystem_freenode
> close.c - Modified such that node information remained valid while
> the node was open.
Are you going to apply patch for these files? It seems to me
arose disscuusion. For example, I think changes to open and
close are wrong because of ideology. ( see appropriate
letters, please )
> link.c - Remove calls to freenode when the node was not successfully
> unmount.c - In the method file_systems_below_this_mountpoint added calls to
> correctly free fs_root_loc when a failure occurs.
> Note: The call to free fs_root_loc.mt_entry->mt_fs_root in unmount was removed
> from the patch
> I was thinking of adding a macro that compares two loc nodes
> by using the mt_entry and a call to stat. This would require
> all filesystem implementations to support fstat_h in order for
> mount, unmount, and rtems_libio_is_file_open to work. It
> would however remove the problem where node_access
> is being used for node identification. Any suggestions/ comments?
It seems, this is the right way.
Eugeny S. Mints
1 Ulianovskaya st., Petergof, St.Petersburg, 198904 Russia
Phone: +7(812)428-4384 Fax: +7(812)327-2246
mailto:Eugeny.Mints at oktet.ru
More information about the users