Re: [committed] avoid assuming cfun is nonnull [PR102243]

2021-09-20 Thread Richard Biener via Gcc-patches
On Mon, Sep 20, 2021 at 1:33 AM Martin Sebor via Gcc-patches wrote: > > After the front end passes control to the middle end cfun is never > null (I'm pretty sure) cfun is NULL during IPA pass execution, it's set to non-NULL by the pass manager when we execute a non-IPA pass on a specific

[committed] avoid assuming cfun is nonnull [PR102243]

2021-09-19 Thread Martin Sebor via Gcc-patches
After the front end passes control to the middle end cfun is never null (I'm pretty sure) but when a middle end helper is called from the C++ front end cfun can be null for a namespace scope initializer expression. A recent change of mine to the helper introduced an assumption to the contrary,