Re: [PATCH] fortran: Assume there is no cyclic reference with submodule symbols [PR99798]

2024-05-12 Thread Paul Richard Thomas
Hi Mikael, That is an ingenious solution. Given the complexity, I think that the comments are well warranted. OK for master and, I would suggest, 14-branch after a few weeks. Thanks! Paul On Sun, 12 May 2024 at 14:16, Mikael Morin wrote: > Hello, > > Here is my final patch to fix the ICE of

[PATCH] fortran: Assume there is no cyclic reference with submodule symbols [PR99798]

2024-05-12 Thread Mikael Morin
Hello, Here is my final patch to fix the ICE of PR99798. It's maybe overly verbose with comments, but the memory management is hopefully clarified. I tested this with a full fortran regression test on x86_64-linux and a manual check with valgrind on the testcase. OK for master? -- 8< -- This