[llvm-dev] Difference in EHType between ARM and AArch64 (original) (raw)
Grang, Mandeep Singh via llvm-dev llvm-dev at lists.llvm.org
Thu Apr 6 14:39:59 PDT 2017
- Previous message: [llvm-dev] Difference in EHType between ARM and AArch64
- Next message: [llvm-dev] LLVM buildmaster will be updated and restarted tonight
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Thanks for your explanation Renato. So I guess with Joerg's patch which changes emitting .cfi_sections, my AArch64 test is now "supposed" to fail.
--Mandeep
On 4/6/2017 2:22 AM, Renato Golin wrote:
On 5 April 2017 at 21:35, Grang, Mandeep Singh via llvm-dev <llvm-dev at lists.llvm.org> wrote:
Could you please clarify, why the behavior for handling .cfisections needs to be different between ARM and AArch64? Hi Mandeep, Many years ago, ARM devised a specially crafted exception handling system tuned for code size and embedded targets, which was the whole point of their hardware designs. This was called the EHABI. AArch64 was never meant to be an embedded architecture (ARMv8M is still AArch32), and the complexity added by having a completely different exception handling just wasn't worth the trouble. For that reason, AArch64 follows the Dwarf EH while AArch32 follows the EHABI. cheers, --renato
- Previous message: [llvm-dev] Difference in EHType between ARM and AArch64
- Next message: [llvm-dev] LLVM buildmaster will be updated and restarted tonight
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]