[prev in list] [next in list] [prev in thread] [next in thread] 

List:       busybox
Subject:    Re: [BusyBox] tar and gunzip changes
From:       Glenn McGrath <bug1 () optushome ! com ! au>
Date:       2001-03-28 23:38:16
[Download RAW message or body]

Robert Kaiser wrote:
> 
> Hi Glenn,
> 
> On Mit, 28 Mär 2001 you wrote:
> > tar -z should be working again now (it does for me).
> 
> The tar -z race condition that I reported yesterday is definitely gone now :-)
> 
> >
> > If you find something is broken, do let me know, but i hope everything
> > is alright
> 
> In tar.c, if -z is not given (i.e. unzipFlag = FALSE), you end up calling
> gz_close() with an uninitialized pid parameter. I would suggest to apply
> the attached "patch-tar".
> 
> > I commented out the signal handling, im not too sure about that...
> 
> me too :-)
> 
> The signal handling in routine unzip() looks OK to me, no reason to comment it
> out. Also, the signal handler abort_gzip() should do an exit() as originally
> intended.
> 
> But gz_open() should _not_ setup any signal handling. gz_open() gets called
> by the _parent_ process, but abort_gzip() is a signal handler intended to
> handle signals sent to unzip, not to unzip's parent. Setting up abort_gzip() as
> SIGCHLD handler for the parent as the commented-out code suggests IMHO
> makes no sense at all.
> Attached is another patch "patch-gunzip" that changes gunzip.c accordingly.
> 
> Cheers
> 
> Rob
> 


Thanks Rob, will probabably apply them shortly


Glenn



[prev in list] [next in list] [prev in thread] [next in thread]