nfsd patches

[David Walter]

I tried testing the patches that Marcus submitted yesterday.

The short version is that the nfsd daemon seems  to have problems, but
they don't all appear to be new, the translator works mostly correctly
for me.

sequence:

set up a local mount on a hurd box in /etc/exports: /src
  /path/to/test/build/hurd/nfsd/nfsd
  settrans -fgc /tmp/nfs /path/to/test/build/hurd/nfs/nfs localhost:/src
  cp /etc/profile /src
  ls /tmp/nfs
reports the size correctly from
   cat /tmp/nfs/profile
cat reads the text of the file 

This also works for a larger file.

now the following
  cp /src/xemacs21-bin.....deb /tmp/nfs
  ls /tmp/nfs
reports the size as UINT_MAX
  less of /src/xemacs21-bin.....deb

starts with null characters, then about two screens down the beginning
of the .deb archive header. 

Okay that is with the current cvs version.

Now with a version prior to the changes.

First I did cvs  co -D"Sun Sept 29 2002 00:00". /hurd/nfsd

Then build and run this nfsd
Now I get the following:

  cp /src/xemacs21-bin.....deb /tmp/nfs
  ls /tmp/nfs
reports the size as UINT_MAX
  less of /src/xemacs21-bin.....deb

starts with null characters, 10 screens and still the same.
#copy via nfsd
  cp /etc/exports /tmp/nfs
  cat /tmp/nfs
no data read, but the size is reported correctly.
# copy w/o the translator or daemon
  cp /etc/exports /src
  cat /tmp/nfs
no data read, but the size is reported correctly.

rm seems to work correctly in both cases.

Now using the older version of the nfs translator.

I get the same results, as with the new version.

Okay, finally using the nfsd from a linux mount and the nfs
translators. starting with the old one (just making sure :-)

File creation, read, write are okay. with the old version of nfs
translator.

The nfs translator fails to start up from a passively set translator.

settrans -fga /tmp/nfs
ls /tmp/nfs

nfs: translator died.

It  appears that the hurd's nfs  translator fails to startup correctly
at  times,  but   I haven't  been  able  to   identify   this as being
consistent, or what combination of  servers or translator versions are
affected.


I will continue to try to find these problems a bit later this week.

-- 
/^\
\ /     ASCII RIBBON CAMPAIGN
 X        AGAINST HTML MAIL
/ \