ARM: 9297/1: vfp: avoid unbalanced stack on 'success' return path

Commit c76c6c4 ("ARM: 9294/2: vfp: Fix broken softirq handling
with instrumentation enabled") updated the VFP exception entry logic to
go via a C function, so that we get the compiler's version of
local_bh_disable(), which may be instrumented, and isn't generally
callable from assembler.

However, this assumes that passing an alternative 'success' return
address works in C as it does in asm, and this is only the case if the C
calls in question are tail calls, as otherwise, the stack will need some
unwinding as well.

I have already sent patches to the list that replace most of the asm
logic with C code, and so it is preferable to have a minimal fix that
addresses the issue and can be backported along with the commit that it

fixes to v6.3 from v6.4. Hopefully, we can land the C conversion for v6.5.

So instead of passing the 'success' return address as a function
argument, pass the stack address from where to pop it so that both LR
and SP have the expected value.

Fixes: c76c6c4 ("ARM: 9294/2: vfp: Fix broken softirq handling with ...")
Reported-by: syzbot+d4b00edc2d0c910d4bf4@syzkaller.appspotmail.com
Tested-by: syzbot+d4b00edc2d0c910d4bf4@syzkaller.appspotmail.com
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Tested-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Tested-by: Andre Przywara <andre.przywara@arm.com>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>

Author: ardbiesheuvel

arch/arm/vfp/entry.S

@@ -23,6 +23,9 @@
 @
 ENTRY(do_vfp)
 	mov	r1, r10
-	mov	r3, r9
-	b	vfp_entry
+	str	lr, [sp, #-8]!
+	add	r3, sp, #4
+	str	r9, [r3]
+	bl	vfp_entry
+	ldr	pc, [sp], #8
 ENDPROC(do_vfp)

arch/arm/vfp/vfphw.S

@@ -172,13 +172,14 @@ vfp_hw_state_valid:
 					@ out before setting an FPEXC that
 					@ stops us reading stuff
 	VFPFMXR	FPEXC, r1		@ Restore FPEXC last
+	mov	sp, r3			@ we think we have handled things
+	pop	{lr}
 	sub	r2, r2, #4		@ Retry current instruction - if Thumb
 	str	r2, [sp, #S_PC]		@ mode it's two 16-bit instructions,
 					@ else it's one 32-bit instruction, so
 					@ always subtract 4 from the following
 					@ instruction address.
 
-	mov	lr, r3			@ we think we have handled things
 local_bh_enable_and_ret:
 	adr	r0, .
 	mov	r1, #SOFTIRQ_DISABLE_OFFSET
@@ -209,8 +210,9 @@ skip:
 
 process_exception:
 	DBGSTR	"bounce"
+	mov	sp, r3			@ setup for a return to the user code.
+	pop	{lr}
 	mov	r2, sp			@ nothing stacked - regdump is at TOS
-	mov	lr, r3			@ setup for a return to the user code.
 
 	@ Now call the C code to package up the bounce to the support code
 	@   r0 holds the trigger instruction