Re: Fix ICEs related to VM types in C [PR106465, PR107557, PR108424, PR109450]

2023-04-11 Thread Martin Uecker via Gcc-patches
hes/2023-February/611562.html While revising it I realized that recursing for parameter when gimplifying is incorrect for the same reason as recursing in other cases (see the comment in gimplify_type_sizes). This lead to PR109450. > > > Fix ICEs related to VM types in C [PR106465, P

Re: Fix ICEs related to VM types in C [PR106465, PR107557, PR108424, PR109450]

2023-04-11 Thread Joseph Myers
ease give a reference to whatever previous patch posting / discussion there may have been? > Fix ICEs related to VM types in C [PR106465, PR107557, PR108424, > PR109450] 108424 seems to be the wrong PR number. -- Joseph S. Myers jos...@codesourcery.com

Re: Fix ICEs related to VM types in C [PR106465, PR107557, PR108424, PR109450]

2023-04-11 Thread Andrew Pinski via Gcc-patches
nd before), though I wonder how much value considering something which didn't work in GCC 4.0+ these days as a regression. Also splitting them up if possible is always welcomed to be able to review each thing separately. Thanks, Andrew > > Martin > > > > Fix ICEs related

Fix ICEs related to VM types in C [PR106465, PR107557, PR108424, PR109450]

2023-04-11 Thread Martin Uecker via Gcc-patches
Ok, here is another attempt on fixing issues with size expression. Not all are regressions, but it does not make sense to try to split it up. Martin Fix ICEs related to VM types in C [PR106465, PR107557, PR108424, PR109450] Size expressions were sometimes lost