svn commit: r359672 - head/sys/riscv/riscv

[Jessica Clarke]

Author: jrtc27
Date: Mon Apr  6 22:31:30 2020
New Revision: 359672
URL: https://svnweb.freebsd.org/changeset/base/359672

Log:
  riscv: Make sure local hart's icache is synced in pmap_sync_icache
  
  The only way to flush the local hart's icache is with a FENCE.I (or an
  equivalent SBI call); a normal FENCE is insufficient and, for the
  single-hart case, unnecessary.
  
  Reviewed by:	jhb (mentor), markj
  Approved by:	jhb (mentor), markj
  Differential Revision:	https://reviews.freebsd.org/D24317

Modified:
  head/sys/riscv/riscv/pmap.c

Modified: head/sys/riscv/riscv/pmap.c
==============================================================================
--- head/sys/riscv/riscv/pmap.c	Mon Apr  6 22:29:15 2020	(r359671)
+++ head/sys/riscv/riscv/pmap.c	Mon Apr  6 22:31:30 2020	(r359672)
@@ -4335,13 +4335,20 @@ pmap_sync_icache(pmap_t pmap, vm_offset_t va, vm_size_
 	 * RISC-V harts, the writing hart has to execute a data FENCE
 	 * before requesting that all remote RISC-V harts execute a
 	 * FENCE.I."
+	 *
+	 * However, this is slightly misleading; we still need to
+	 * perform a FENCE.I for the local hart, as FENCE does nothing
+	 * for its icache. FENCE.I alone is also sufficient for the
+	 * local hart.
 	 */
 	sched_pin();
 	mask = all_harts;
 	CPU_CLR(PCPU_GET(hart), &mask);
-	fence();
-	if (!CPU_EMPTY(&mask) && smp_started)
+	fence_i()
+	if (!CPU_EMPTY(&mask) && smp_started) {
+		fence();
 		sbi_remote_fence_i(mask.__bits);
+	}
 	sched_unpin();
 }