This is the mail archive of the gdb-patches@sourceware.org mailing list for the GDB project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

[review v3] Replace some more qsort calls with std::sort


Simon Marchi has posted comments on this change.

Change URL: https://gnutoolchain-gerrit.osci.io/r/c/binutils-gdb/+/131
......................................................................


Patch Set 3:

It looks like this patch causes some regressions when building with -D_GLIBCXX_DEBUG=1.  One such failure is in gdb.dwarf2/ada-linkage-name.exp (make check TESTS="gdb.dwarf2/ada-linkage-name.exp").

The problem is that the comparison functions don't have the required properties for std::sort:

 49 /usr/include/c++/9.2.0/bits/stl_algo.h:4858:^M
 50 In function:^M
 51     void std::sort(_RAIter, _RAIter, _Compare) [with _RAIter = ^M
 52     obj_section**; _Compare = bool (*)(const obj_section*, const ^M
 53     obj_section*)]^M
 54 ^M
 55 Error: comparison doesn't meet irreflexive requirements, assert(!(a < a)).^M
 56 ^M
 57 Objects involved in the operation:^M
 58     instance "functor" @ 0x0x7ffda9ccc9a0 {^M
 59       type = bool (*)(obj_section const*, obj_section const*);^M
 60     }^M
 61     iterator::value_type "ordered type" {^M
 62       type = obj_section*;^M
 63     }^M

It is probably a pre-existing bad behavior with the comparison functions, but using std::sort points those out.

I would encourage you to use -D_GLIBCXX_DEBUG=1 all the time for your development builds (except for when you want to do benchmarks), it catches a bunch of things.



Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]