[PATCH V6 02/49] x86/entry: Use the correct fence macro after swapgs in kernel CR3

[Lai Jiangshan]

From: Lai Jiangshan <laijs@xxxxxxxxxxxxxxxxx>

The commit c75890700455 ("x86/entry/64: Remove unneeded kernel CR3
switching") removes a CR3 write in the faulting path of load_gs_index.

But the path's FENCE_SWAPGS_USER_ENTRY has no fence operation if PTI
is enabled.  Rahter, it depends on the CR3 write of SWITCH_TO_KERNEL_CR3.
So the path should use FENCE_SWAPGS_KERNEL_ENTRY if SWITCH_TO_KERNEL_CR3
is removed.

Fixes: c75890700455 ("x86/entry/64: Remove unneeded kernel CR3 switching")
Signed-off-by: Lai Jiangshan <laijs@xxxxxxxxxxxxxxxxx>
---
 arch/x86/entry/entry_64.S | 16 +++++++++-------
 1 file changed, 9 insertions(+), 7 deletions(-)

diff --git a/arch/x86/entry/entry_64.S b/arch/x86/entry/entry_64.S
index 8582709576bf..4967edded48d 100644
--- a/arch/x86/entry/entry_64.S
+++ b/arch/x86/entry/entry_64.S
@@ -991,11 +991,6 @@ SYM_CODE_START_LOCAL(error_entry)
 	pushq	%r12
 	ret
 
-.Lerror_entry_done_lfence:
-	FENCE_SWAPGS_KERNEL_ENTRY
-.Lerror_entry_done:
-	ret
-
 	/*
 	 * There are two places in the kernel that can potentially fault with
 	 * usergs. Handle them here.  B stepping K8s sometimes report a
@@ -1018,8 +1013,15 @@ SYM_CODE_START_LOCAL(error_entry)
 	 * .Lgs_change's error handler with kernel gsbase.
 	 */
 	SWAPGS
-	FENCE_SWAPGS_USER_ENTRY
-	jmp .Lerror_entry_done
+
+	/*
+	 * The above code has no serializing instruction.  So do an lfence
+	 * to prevent GS speculation, regardless of whether it is kernel
+	 * gsbase or user gsbase.
+	 */
+.Lerror_entry_done_lfence:
+	FENCE_SWAPGS_KERNEL_ENTRY
+	ret
 
 .Lbstep_iret:
 	/* Fix truncated RIP */
-- 
2.19.1.6.gb485710b