On Mon, Feb 10, 2014 at 2:30 AM, Dylan Baker <[email protected]> wrote: > On Friday, February 07, 2014 10:22:48 PM Ilia Mirkin wrote: >> On Fri, Feb 7, 2014 at 10:08 PM, Dylan Baker <[email protected]> > wrote: >> > On Friday, February 07, 2014 09:42:05 PM Ilia Mirkin wrote: >> >> This makes it possible to run the summary on e.g. compressed files or >> >> otherwise piped in with the <( ... ) shell construct. >> >> >> >> There should be no difference between open() on a path before and after >> >> the realpath call. >> >> >> >> Signed-off-by: Ilia Mirkin <[email protected]> >> >> --- >> >> >> >> framework/core.py | 2 -- >> >> 1 file changed, 2 deletions(-) >> >> >> >> diff --git a/framework/core.py b/framework/core.py >> >> index 45eea12..6a122f5 100644 >> >> --- a/framework/core.py >> >> +++ b/framework/core.py >> >> >> >> @@ -647,8 +647,6 @@ def load_results(filename): >> >> "main" >> >> >> >> """ >> >> >> >> - filename = os.path.realpath(filename) >> >> - >> >> >> >> try: >> >> with open(filename, 'r') as resultsfile: >> >> testrun = TestrunResult(resultsfile) >> > >> > I know that some people install piglit and add it's programs to their >> > $PATH, is this going to break any of those use cases? It didn't seem to >> > when I tested it, but some of those people might want to weigh in. >> >> I can't see how it would matter one way or another. Stuff like >> realpath is to deal with people feeding symlinks/etc that end up >> pointing outside of a tree (so you want to deny that for security >> reasons). But perhaps there's something I'm missing... >> >> -ilia > > That was the reason this was originally added IIRC. However, no one seems to
The reason was to try to deny people the ability to use symlinks that pointed outside of a fixed tree? Seem unlikely... The problem here though is that <(xzcat file.xz) gives it like /dev/fd/63 which is a "fake" symlink, i.e. it reads as a symlink but it points to a non-sensical location (a "pipe" object in /proc). And open() doesn't work on that. > be complaining, your code seems fine, it doesn't break anything that I can > see. > > Reviewed-by: Dylan Baker <[email protected]> Thanks! I still don't have piglit commit access (perhaps it's time I asked for it), could you check this in? -ilia _______________________________________________ Piglit mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/piglit
