This may look strange, since this buffer is originally allocated using
the TAG_IO_NAME tag. However, it happens that file-system drivers are
allowed to re-allocate this buffer: this is what the MS' open-sourced
CDFS driver does, see e.g. CdCommonCreate() and CdNormalizeFileNames()
in cdfs/create.c .
This fixes a pool tag mismatch 'mNoI' != 'nFdC' BSOD when resources
are freed when closing a file that has been opened with a relative name
on a CDFS-mounted volume.
IopCloseFile can be called by IopDeleteFile. In that situation, it
doesn't set any process as first parameter. Furthermore, we are in a
situation where it's not required to lock the file object (see the
assert before the call).
So, bring back the infamous IopParseDevice() hack. Dismounting is to be fixed in FastFAT.
Even though it might not be the last remaining bug...
CORE-14124
CORE-14126
CORE-14133
For the record, the only place it was remaining was 1st stage, notably because
FSD (FastAT <3) was missing a few features. As this features weren't triggered
in 3rd stage (unless forced), it was not needed there any longer.
Now that they are implemented, and seem working out of the box, this hack might
not be longer anymore.
This is my ... ?! pfff attempt at killing it. It might not be the last, so just
disabling the hack for now. If there are no regressions (who can really believe that?)
we'll be clear for dropping it once for all.
