Re: Implement removal of malloc/free pairs with NULL check

[Sam James]

Jan Hubicka <hubi...@ucw.cz> writes:

> [...]
> The attached patch adds code to track size of allocated block and
> disable the transformation when the block is not known to be smaller
> then half of the address space by ranger.  We can do the runtime check
> discussed on the top of that.
>
> I have bootstrap&regresst running (it will still need testusite
> compensation).  However I am not convinced this is a good direction.
> this patch will still break the example where one allocates 5x 1/4th of
> address space, leaves it unused, and verify that at least one of
> allocation fails.
>

With this patch, I get a bootstrap failure:

during GIMPLE pass: cddce
/var/tmp/portage/sys-devel/gcc-15.0.9999/work/gcc-15.0.9999/libstdc++-v3/src/c++98/locale.cc:
 In constructor ‘std::locale::_Impl::_Impl(const std::locale::_Impl&, 
std::size_t)’:
/var/tmp/portage/sys-devel/gcc-15.0.9999/work/gcc-15.0.9999/libstdc++-v3/src/c++98/locale.cc:549:1:
 internal compiler error: Segmentation fault
  549 | } // namespace
      | ^
0x5574ada2b4d3 diagnostic_context::diagnostic_impl(rich_location*, 
diagnostic_metadata const*, diagnostic_option_id, char const*, __va_list_tag 
(*) [1], diagnostic_t)
        ???:0
0x5574ada426b5 internal_error(char const*, ...)
        ???:0
0x5574ac1f70b0 crash_signal(int)
        ???:0
0x5574ac34de9d is_removable_allocation_p(gcall*, tree_node**, tree_node**)
        ???:0
0x5574ac34e14e mark_stmt_if_obviously_necessary(gimple*, bool)
        ???:0
0x5574ac34fb4e perform_tree_ssa_dce(bool)
        ???:0
0x5574ac351c79 (anonymous namespace)::pass_cd_dce::execute(function*)
        ???:0
Please submit a full bug report, with preprocessed source (by using 
-freport-bug).
Please include the complete backtrace with any bug report.

Attached preprocessed source.

thanks,
sam

locale.ii.xz
Description: application/xz