RTEMS snapshot ss-20010126 available
osv at javad.ru
Mon Jan 29 11:21:06 UTC 2001
I'll take a look at these problems.
Meanwhile, here are some comments:
Jake Janovetz <janovetz at uiuc.edu> writes:
> I've noticed a couple problems with the revised ftpd. I'm
> trying to track them down... I figure since Sergei has made
> the most recent modifications, he may know better what changes
> broke them (or if I'm setting things up incorrectly)...
> 1. LIST is sometimes sent from a client without a path. Currently,
> FTPD expects LIST to have some path. This should be a matter
> of checking the number of arguments filled by sscanf. Using the
> command "dir ." with ftp clients will fix this problem, but it
> is inconvenient.
It works for me with my sources (will check it with snapshot).
> 2. Multiple connections sometimes cause a crash. I ftp with one
> client and do: "dir ." and get a directory. A second ftp client
> (while the first one is open or even after the first one is closed)
> causes a crash at "dir .". I don't know the source of this one,
Never encountered this. Need to check with snapshot.
> 3. Hooks are broken (the filename comparison doesn't appear to
> work anymore -- I put to a hook and it just creates a file with
> that name. [this one particularly sucked because I could not
> download an old code image... I had to dig out the debug cable
> and do the gdb song and dance]
> -- This one occurs because you take the filename and the CWD and
> build a new filename. This new filename may/may not match the
> hook. Originally, I planned that hooks would not depend on the
> current working directory. I don't know what you have in mind
> for the revised setup.
Is easy to fix, I believe. Will do.
> I'm using the following configuration:
> struct rtems_ftpd_configuration rtems_ftpd_configuration =
> FTPD_TASK_PRIORITY, /* FTPD task priority */
> 512*1024, /* Maximum hook 'file' size */
> 0, /* Default port */
> ftpd_hooks, /* Local ftpd hooks */
> 0, /* Use / for home directory */
> 5, /* Up to 5 connections */
> 30, /* Timeout */
> 0 /* Read-write access */
> Sergei, you mention that the hooks should be dropped in favor
> of RTEMS pseudo-devices. What are these and how easily can they
> be built and installed? I'm familiar with the device setup with
> RTEMS, but it is substantially more complex than just associating
> some action with a buffer. If the pseudo-devices are easy, I'd
> be more than happy to switch to them, but right now, creating
> devices for the hooks I have would be a pain.
As for pseudo-devices, I'm not familiar with them, but I think I understand an
idea. Yes, they are not so simple to create as FTP hooks, but they *are*
general way to do such things and I believe that if general method exists,
there should be very strong reasons to use specific extensions
instead. However, as I already said before, I don't insist on removing FTPD
hooks. I just wanted to drop a notice, and decided to put it into comments in
the FTPD source code.
> On Fri, Jan 26, 2001 at 03:03:20PM -0600, Joel Sherrill wrote:
> > As far as I know, this includes every patch/submission sent
> > in during the past few weeks. It fixes a bug or two in recent
> > snapshots (unix port works again for example). It also includes
> > the "multiuser" patches. From my perspective, the core of these
> > patches is a neater organization of the POSIX process related
> > parameters.
> > ftp://ftp.oarcorp.com/pub/rtems/snapshots/rtems/current
> > Enjoy.
> > --
> > Joel Sherrill, Ph.D. Director of Research & Development
> > joel at OARcorp.com On-Line Applications Research
> > Ask me about RTEMS: a free RTOS Huntsville AL 35805
> > Support Available (256) 722-9985
More information about the users