[Bug tree-optimization/124151] [13/14/15/16 Regression] internal compiler error: in analyze_access_subtree, at tree-sra.cc:2935

[tobiausgais at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=124151

--- Comment #10 from Totto <tobiausgais at gmail dot com> ---
Yes, It seems really like that, I also tried the original source file with a
gcc with PR 117217 in it, than it worked, but enabling `-fchecking` makes it
another ICE.

Thanks for creating a reduced test case, as I don't know the SRA that well, I
wasn't sure, what struct / union / bitfield created that ICE. 

The original reason I didn't consider PR 117217 to be the fix for my ICE was,
that it happened in verify_sra_access_forest and not in analyze_access_subtree
like mine, but I might have interpreted that wrong. I also didn't remember
using bool: 1 bitfields, so I thought it had something to do with nested union
and smaller than int enums. But that were just guesses from my side :(

Anyways, PR 117217 only fixes it without `-fchecking`, so I guess this is still
a bug. 

Thank you for your work and the patience, as I am not that familiar with
compilers, so I really thought this was something new :( 

I am sorry if I reported a duplicate issue, as in time of AI it is already hard
enough for maintainers, so I'm "glad" that it is at least still a bug with
`-fchecking` ':)