Due to bug 1522678, we will not be able to enable -fstack-clash-protection on armhfp.
This is unfortunate, but I don't think the lack of upstream support on armhfp should block this feature for all other architectures (including aarch64).
Thanks, Florian
On Wed, Jan 3, 2018 at 7:46 PM, Florian Weimer fweimer@redhat.com wrote:
Due to bug 1522678, we will not be able to enable -fstack-clash-protection on armhfp.
Is there an upstream bug report/thread for this? It would be useful to at least make the appropriate ARM maintainers aware of the problem so if they're interested enough or see value in it they can deal with it and then it can be enabled later.
Can you please also make sure any arm specific bugs are linked against ARMTracker, I can't escalate unknown issues to contacts I have with in relevant organisations if I don't know they exist, and it's impossible to track all bugs.
This is unfortunate, but I don't think the lack of upstream support on armhfp should block this feature for all other architectures (including aarch64).
I think that's a reasonable compromise, if/when it gets fixed upstream we can always enable it then and it'll get applied with the next mass rebuild.
Peter
On 01/04/2018 04:11 AM, Peter Robinson wrote:
On Wed, Jan 3, 2018 at 7:46 PM, Florian Weimer fweimer@redhat.com wrote:
Due to bug 1522678, we will not be able to enable -fstack-clash-protection on armhfp.
Is there an upstream bug report/thread for this? It would be useful to at least make the appropriate ARM maintainers aware of the problem so if they're interested enough or see value in it they can deal with it and then it can be enabled later.
Nobody has written a patch, so 32-bit ARM uses the generic bits, which are suboptimal (in terms of ABI compliance and protection level) and, it appears, not really usable.
Even the final aarch64 bits are still pending review by the ARM maintainers.
Can you please also make sure any arm specific bugs are linked against ARMTracker, I can't escalate unknown issues to contacts I have with in relevant organisations if I don't know they exist, and it's impossible to track all bugs.
Okay, I will do that in the future.
Thanks, Florian
On 01/04/2018 04:48 PM, Florian Weimer wrote:
On 01/04/2018 04:11 AM, Peter Robinson wrote:
On Wed, Jan 3, 2018 at 7:46 PM, Florian Weimer fweimer@redhat.com wrote:
Due to bug 1522678, we will not be able to enable -fstack-clash-protection on armhfp.
Is there an upstream bug report/thread for this? It would be useful to at least make the appropriate ARM maintainers aware of the problem so if they're interested enough or see value in it they can deal with it and then it can be enabled later.
Nobody has written a patch, so 32-bit ARM uses the generic bits, which are suboptimal (in terms of ABI compliance and protection level) and, it appears, not really usable.
Even the final aarch64 bits are still pending review by the ARM maintainers.
The aarch64 bits still have not been committed. It might help if you pinged your Arm contacts again.
Thanks, Florian
-
Florian Weimer -
Peter Robinson