13 Dec
2023
13 Dec
'23
8:02 p.m.
On Wed, Dec 13, 2023 at 11:59:45AM -0800, Deepak Gupta wrote:
On Wed, Nov 22, 2023 at 1:45 AM Mark Brown broonie@kernel.org wrote:
if (task->thread.gcs_el0_mode & PR_SHADOW_STACK_ENABLE)gcscre0_el1 |= GCSCRE0_EL1_RVCHKEN | GCSCRE0_EL1_PCRSEL;
If the intent is to disable, is the GCS stack freed or kept around? I expect if libc is taking the decision to disable, kernel should free it up. Is it freed in some other flow?
Kept around and freed on thread exit. There is a potential race between for example disabling in a signal handler and something trying to walk the stack so we err on the side of caution.