Re: [PATCH] Fortran: fix issues with class(*) assignment [PR114827]

[Paul Richard Thomas <paul.richard.thomas@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 1084 bytes --]

Hi Harald,

This patch is verging on 'obvious', ..... once one sees it :-)

Yes, it's good for mainline and all active branches, when available.

Thanks

Paul

PS The fall-out pr114874 is so peculiar that I am dropping everything to
find the source.


On Mon, 29 Apr 2024 at 19:39, Harald Anlauf <anlauf@gmx.de> wrote:

> Dear all,
>
> the attached patch fixes issues with assignments of unlimited polymorphic
> entities that were found with the help of valgrind or asan, see PR.
> Looking
> further into it, it turns out that allocation sizes as well as array spans
> could be set incorrectly, leading to wrong results or heap corruption.
>
> The fix is rather straightforward: take into the _len of unlimited
> polymorphic entities when it is non-zero to get the correct allocation
> sizes and array spans.
>
> The patch has been tested by the reporter, see PR.
>
> Regtested on x86_64-pc-linux-gnu.  OK for 15-mainline?
>
> I would like to backport this to active branches where appropriate,
> starting with 14 after it reopens after release.  Is this OK?
>
> Thanks,
> Harald
>
>