| Subject: | Re: [PATCH] Add new built-in: __builtin_unreachable() |
|---|---|
| From: | Paolo Bonzini |
| Date: | Mon, 15 Jun 2009 10:15:45 +0200 |
Could you explain why it is allowed to remove the trapping instruction?If this were java (which admittedly it is not), that would not be allowed. Because of -fdelete-null-pointer-checks. Without non-call exceptions, *f does not throw an exception but still the compiler is allowed to assume its address is not null in anything that postdominates it. Admittedly this case is a bit borderline. Paolo |
| Previous by Date: | Re: [PATCH: PR/40314] extract common address for memory access to fields of large struct, Carrot Wei |
|---|---|
| Next by Date: | Re: [PATCH] Fix recent -gdwarf-2 failures, Dominique Dhumieres |
| Previous by Thread: | Re: [PATCH] Add new built-in: __builtin_unreachable(), David Daney |
| Next by Thread: | Re: [PATCH] Add new built-in: __builtin_unreachable(), Paul Brook |
| Indexes: | [Date] [Thread] [Top] [All Lists] |