Re: [AVFS] Segfault using #ulha#utar

> I'm currently hunting down a bug in the archive handler. I noticed a
> segfault when accessing a file named "t.tar.lha#" which is resolved to
> "t.tar.lha#ulha#utar".
> 
> I backtraced the whole virt_open call and came to the following problem:
> Looking up the "utar" segment leads to the function "parse_tarfile"
> which opens the base file, in this case the ulha-handled file. Nothing
> wrong so far.
> Opening the ulha file leads to arch_do_open. Around line 359 in
> archive.c in this function a new archfile struct is created. The nod for
> this struct is the default node for the ulha entry. Especially its data
> entry is NULL (the struct extfsnode entry).
> 
> Finally ap->open is called (it's actually extfs_open) and the
> get_extfs_file function accesses the extfsnode which is NULL.
> 
> I looked over this problem to find a good solution. "t.tar.lha#ulha" is
> a directory and open() on directories is allowed. However, since we
> can't return any handle for an extfs entry without a path I think the
> only solution is to return an error in get_extfs_file. Returning -EISDIR
> is perhaps a good choice.

Yes, that seems to be the right solution, since #utar needs a regular
file to operate on, so when it tries to read a directory it should get
back an error.

> What do you think about this? I hope my description of this problem is
> enough to understand it but if you need more information I have plenty
> of it :-)
> 
> Besides this, I'm looking into some other minor problems but I will
> write about it later :-)

Thanks!

Miklos


-------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc. Do you grep through log files
for problems?  Stop!  Download the new AJAX search engine that makes
searching your log files as easy as surfing the  web.  DOWNLOAD SPLUNK!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=103432&bid=230486&dat=121642