AaronBallman wrote:
> @AaronBallman I know of no other CFI attribute requests.
Excellent, glad to hear there's not some massive set of changes waiting in the
wings.
> (But keep in mind that I'm not the CFI expert I play on TV.)
Sorry, but you touched it last, so you own it. I don't make the
bwendling wrote:
@AaronBallman I know of no other CFI attribute requests. (But keep in mind that
I'm not the CFI expert I play on TV.)
https://github.com/llvm/llvm-project/pull/141846
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lis
AaronBallman wrote:
> Also, this is ANOTHER bit on function types that we're using for CFI. I'm
> wondering if this is actually a generally useful enough feature to be using
> this much compiler resources.
>
> @AaronBallman ?
I think CFI is pretty important for security posture (it basically
bwendling wrote:
@erichkeane If the Arm attribute code were merged into `ExtraAttributeInfo`, it
would save us a bit in `FunctionTypeExtraBitfields`.
I can add more testing I suppose, but the tests I have here are a copy of the
current CFI tests, which test all of the ways CFI is used. So I'm
https://github.com/erichkeane commented:
Testing seems insufficient, particularly around conversion rules/etc of these
types. I would like to see quite a bit more for Sema/inside of templates/how
this applies to instantiations of all these things, etc.
Also, this is ANOTHER bit on function ty
https://github.com/bwendling updated
https://github.com/llvm/llvm-project/pull/141846
>From bfdf16cee252dc6491f15c0715c7a39538fbf120 Mon Sep 17 00:00:00 2001
From: Bill Wendling
Date: Wed, 23 Apr 2025 13:14:24 -0700
Subject: [PATCH 1/2] [Clang][attr] Add cfi_salt attribute
The new 'cfi_salt' a
bwendling wrote:
I moved the `cfi_salt` into the `FunctionProtoType` fields. I left it open for
further additions. One bit of optimization I could do is merge the Arm stuff
into this new struct, but wanted to do something a bit less cluttered first.
https://github.com/llvm/llvm-project/pull/14
https://github.com/bwendling updated
https://github.com/llvm/llvm-project/pull/141846
>From bfdf16cee252dc6491f15c0715c7a39538fbf120 Mon Sep 17 00:00:00 2001
From: Bill Wendling
Date: Wed, 23 Apr 2025 13:14:24 -0700
Subject: [PATCH 1/2] [Clang][attr] Add cfi_salt attribute
The new 'cfi_salt' a
efriedma-quic wrote:
Something along the lines of FunctionTypeArmAttributes seems fine.
https://github.com/llvm/llvm-project/pull/141846
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
bwendling wrote:
@efriedma-quic Okay, I'm not entirely sure I understand @erichkeane's concerns,
but that's fine. Are you suggesting adding the salt information to the
`FunctionType` instead?
https://github.com/llvm/llvm-project/pull/141846
___
cfe-c
efriedma-quic wrote:
See #135836 for discussion of why you can't use AttributedType like this. CC
@PiJoules
https://github.com/llvm/llvm-project/pull/141846
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailma
llvmbot wrote:
@llvm/pr-subscribers-clang
Author: Bill Wendling (bwendling)
Changes
The new 'cfi_salt' attribute is used to differentiate between two functions
with the same type signature. It's applied to the function declaration,
function definition, and function typedefs of the functi
https://github.com/bwendling created
https://github.com/llvm/llvm-project/pull/141846
The new 'cfi_salt' attribute is used to differentiate between two functions
with the same type signature. It's applied to the function declaration,
function definition, and function typedefs of the function t
13 matches
Mail list logo
