[Bug translator/4281] two problems with "stap -m"

[hunt at redhat dot com]

------- Additional Comments From hunt at redhat dot com  2007-03-26 18:09 -------
(In reply to comment #1)
> Problem 1.  There is at least one reason for using '-m NAME' without '-k' - to
> disable caching.  Admittedly there is another way to disable caching
> ("SYSTEMTAP_DIR=/dev/null stap close.stp"), but it is uglier.

I don't see how its a good idea to combine to options like this. Instead of
relying on a side-effect of the way "-m" is currently implemented to disable
cache, why not give it its own option?

Regardless, you can still disable cache and compile a module, then place a copy
of it in the current directory. That is what I think it should do.

> Problem 2.  Does this happen with any extension or just '.ko'?  Is any extension
> valid?  For instance, are any of the following valid?
> 
> stap -m close.c close.stp
> stap -m close.o close.stp
> stap -m my.close.module close.stp
> 
> We should probably worry about other special characters in the module name, such
> as '/'.

Consulting the source code in rmmod, "." and "-" (dash) are not used in module
names. Anything after the dot is ignored.  There might be other characters that,
if quoted just right, would make it through the systemtap command line and
compilation process but even if they end up in the module name probably won't be
a problem.

I recommend detecting names ending in ".ko" and silently doing the right thing.
Otherwise if the module name contains "." or "-", print an error and exit.

-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=4281

------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.