Re: [PATCH] arm64/gcs: Don't call gcs_free() during flush_gcs()

[Catalin Marinas <catalin.marinas@arm.com>]

On Wed, Jun 11, 2025 at 05:28:13PM +0100, Mark Brown wrote:
> Currently we call gcs_free() during flush_gcs() to reset the thread state
> for GCS. This includes unmapping any kernel allocated GCS, but this is
> redundant when doing a flush_thread() since we are reinitialisng the thread
> memory too. Inline the reinitialisaton of the thread struct.
> 
> Signed-off-by: Mark Brown <broonie@kernel.org>
> ---
>  arch/arm64/kernel/process.c | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
> 
> diff --git a/arch/arm64/kernel/process.c b/arch/arm64/kernel/process.c
> index a5ca15daeb8a..5954cec19660 100644
> --- a/arch/arm64/kernel/process.c
> +++ b/arch/arm64/kernel/process.c
> @@ -288,7 +288,9 @@ static void flush_gcs(void)
>  	if (!system_supports_gcs())
>  		return;
>  
> -	gcs_free(current);
> +	current->thread.gcspr_el0 = 0;
> +	current->thread.gcs_base = 0;
> +	current->thread.gcs_size = 0;
>  	current->thread.gcs_el0_mode = 0;
>  	write_sysreg_s(GCSCRE0_EL1_nTR, SYS_GCSCRE0_EL1);
>  	write_sysreg_s(0, SYS_GCSPR_EL0);

I think this makes sense.

However, I thought there was another slightly misplaced call to
gcs_free() via arch_release_task_struct(). I wouldn't touch the user
memory with vm_munmap() when releasing a task structure. Is this needed
because the shadow stack is allocated automatically on thread creation,
so we need something to free it when the thread died?

Another caller of gcs_free() is deactivate_mm(). It's not clear to me
when we need to free the shadow stack on this path. On the exit_mm()
path for example we have mmput() -> exit_mmap() that takes care of
unmapping everything. Similarly on the exec_mmap() path.

-- 
Catalin