On Fri, Jul 18, 2025 at 11:37:36PM -0500, Jeremy Linton wrote:
quoted hunk ↗ jump to hunk
+/*
+ * Unlike put_user_gcs() above, the use of copy_from_user() may provide
+ * an opening for non GCS pages to be used to source data. Therefore this
+ * should only be used in contexts where that is acceptable.
+ */
Even in user space, the GCS pages can be read with normal loads, so
already usable as a data source if one wants to (not that it's of much
use). So not sure the comment here is needed.
+static inline u64 load_user_gcs(unsigned long __user *addr, int *err)
Nitpick: name this get_user_gcs() for symmetry with put_user_gcs().
quoted hunk ↗ jump to hunk
+{
+ unsigned long ret;
+ u64 load = 0;
+
+ gcsb_dsync();
Might be worth a comment here, see the one in gcs_restore_signal().
quoted hunk ↗ jump to hunk
+ ret = copy_from_user(&load, addr, sizeof(load));
+ if (ret != 0)
+ *err = ret;
+ return load;
+}
Otherwise the patch looks fine:
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>