Skip to contentSkip to conversation

x86_32: fix extable entry for iret

PATCH In review Lai Jiangshan 2 participants 2 messages Nov 7, 2019

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 x86_32: fix extable entry for iret

3c88c692c287(x86/stackframe/32: Provide consistent pt_regs) added code after label .Lirq_return and before 'iret', an instruction which should be expected to be found in the extable when there is an exception on it. But the extable entry stores the address of .Lirq_return not the new address of 'iret', which disables the corresponding fixup. This patch fixes the extable entry by using a new label.

---
Purely accidently found, untested.

 arch/x86/entry/entry_32.S | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
arch/x86/entry/entry_32.S +2 -1
index f83ca5aa8b77..f62aa6655cfb 100644
@@ -1081,6 +1081,7 @@ restore_all:
 	 * when returning from IPI handler and when returning from
 	 * scheduler to user-space.
 	 */
+.Lirq_return_ex:
 	INTERRUPT_RETURN
 
 restore_all_kernel:
@@ -1118,7 +1119,7 @@ ENTRY(iret_exc	)
 
 	jmp	common_exception
 .previous
-	_ASM_EXTABLE(.Lirq_return, iret_exc)
+	_ASM_EXTABLE(.Lirq_return_ex, iret_exc)
 ENDPROC(entry_INT80_32)
 
 .macro FIXUP_ESPFIX_STACK
Signature · 1 line
2.20.1
maintainer
#1 lore raw
Quoted 8 lines · Lai Jiangshan

3c88c692c287(x86/stackframe/32: Provide consistent pt_regs) added code after label .Lirq_return and before 'iret', an instruction which should be expected to be found in the extable when there is an exception on it. But the extable entry stores the address of .Lirq_return not the new address of 'iret', which disables the corresponding fixup. This patch fixes the extable entry by using a new label.

Egads!

What happens if you run tools/testing/selftests/x86/sigreturn_32 with and without this patch?

--Andy