Bugfix for CompPhdr::operator() that could cause sort to fail. #139
Conversation
weak ordering as required by std::sort.
|
Do you have a test case? |
src/patchelf.cc
Outdated
| if (x.p_type == PT_PHDR) { | ||
| if (y.p_type == PT_PHDR) return false; | ||
| return true; | ||
| } |
There was a problem hiding this comment.
This could be simplified a bit (mostly removing the duplicated check of y.p_type == PT_PHDR:
// Nothing is less than a PHDR:
if (y.p_type == PT_PHDR) return false;
// PHDR's are less than everything (that's not a PHDR)
if (x.p_type == PT_PHDR) return true;
// Sort non-PHDR's by address:
...
Comments are just suggestions.
There was a problem hiding this comment.
While it does seem worth fixing this, according to documents such as [1][2] it seems that only a single PT_PHDR entry may legally exist in a file.
Thinking about what PT_PHDR actually means, this makes sense :). Also my comments are indeed a bit silly, I didn't have my "ELF-knowledge hat" on.
So this is probably harmless as-is: for any legal set of segments this operator will define a valid (strict weak) ordering.
Might be worth ensuring we reject files that attempt to have multiple PT_PHDR segments?
[1] http://www.sco.com/developers/gabi/latest/ch5.pheader.html
[2] http://www.skyfree.org/linux/references/ELF_Format.pdf
There was a problem hiding this comment.
I've simplified the code now. Thanks.
Yes, this was found using a debug version of std::sort, not by obviously incorrect results. Since there's only a single PT_PHDR entry, It probably doesn't any difference on any sane implementation of std::sort, but it is undefined behavior, so there's no guarantees about that.
|
@domenkozar see my comment on the commit -- I'm not sure a test case (as an ELF input file) could be produced with normal tools. Even if such a file were constructed, Although if @kmillar has such a file that certainly would be interesting! :) |
Modified CompPhdr::operator() so that it provides a strict weak ordering as required by std::sort.