[PATCH 4.18] MIPS: memset: Fix CPU_DADDI_WORKAROUNDS `small_fixup' regression

[PATCH 4.18] MIPS: memset: Fix CPU_DADDI_WORKAROUNDS `small_fixup' regression

[Maciej W. Rozycki]

Fix a commit 8a8158c85e1e ("MIPS: memset.S: EVA & fault support for 
small_memset") regression and remove assembly warnings:

arch/mips/lib/memset.S: Assembler messages:
arch/mips/lib/memset.S:243: Warning: Macro instruction expanded into multiple instructions in a branch delay slot

triggering with the CPU_DADDI_WORKAROUNDS option set and this code:

	PTR_SUBU	a2, t1, a0
	jr		ra
	 PTR_ADDIU	a2, 1

This is because with that option in place the DADDIU instruction, which 
the PTR_ADDIU CPP macro expands to, becomes a GAS macro, which in turn 
expands to an LI/DADDU (or actually ADDIU/DADDU) sequence:

 13c:	01a4302f 	dsubu	a2,t1,a0
 140:	03e00008 	jr	ra
 144:	24010001 	li	at,1
 148:	00c1302d 	daddu	a2,a2,at
	...

Correct this by switching off the `noreorder' assembly mode and letting 
GAS schedule this jump's delay slot, as there is nothing special about 
it that would require manual scheduling.  With this change in place 
correct code is produced:

 13c:	01a4302f 	dsubu	a2,t1,a0
 140:	24010001 	li	at,1
 144:	03e00008 	jr	ra
 148:	00c1302d 	daddu	a2,a2,at
	...

Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org>
Fixes: 8a8158c85e1e ("MIPS: memset.S: EVA & fault support for small_memset")
Cc: stable@vger.kernel.org # 4.17+
---
Hi,

 This is a version (one I originally made actually) of commit 148b9aba99e0 
for 4.18-stable and before.  No functional changes.  Please apply.

  Maciej
---
 arch/mips/lib/memset.S |    4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

linux-mips-memset-jr-ra-nodaddi-fix.patch
Index: linux-20180812-4maxp64-defconfig/arch/mips/lib/memset.S
===================================================================
--- linux-20180812-4maxp64-defconfig.orig/arch/mips/lib/memset.S
+++ linux-20180812-4maxp64-defconfig/arch/mips/lib/memset.S
@@ -261,9 +261,11 @@
 	 nop
 
 .Lsmall_fixup\@:
+	.set		reorder
 	PTR_SUBU	a2, t1, a0
+	PTR_ADDIU	a2, 1
 	jr		ra
-	 PTR_ADDIU	a2, 1
+	.set		noreorder
 
 	.endm
 

Re: [PATCH 4.18] MIPS: memset: Fix CPU_DADDI_WORKAROUNDS `small_fixup' regression

[Greg KH]

On Tue, Oct 16, 2018 at 10:51:10PM +0100, Maciej W. Rozycki wrote:
> Fix a commit 8a8158c85e1e ("MIPS: memset.S: EVA & fault support for 
> small_memset") regression and remove assembly warnings:
> 
> arch/mips/lib/memset.S: Assembler messages:
> arch/mips/lib/memset.S:243: Warning: Macro instruction expanded into multiple instructions in a branch delay slot
> 
> triggering with the CPU_DADDI_WORKAROUNDS option set and this code:
> 
> 	PTR_SUBU	a2, t1, a0
> 	jr		ra
> 	 PTR_ADDIU	a2, 1
> 
> This is because with that option in place the DADDIU instruction, which 
> the PTR_ADDIU CPP macro expands to, becomes a GAS macro, which in turn 
> expands to an LI/DADDU (or actually ADDIU/DADDU) sequence:
> 
>  13c:	01a4302f 	dsubu	a2,t1,a0
>  140:	03e00008 	jr	ra
>  144:	24010001 	li	at,1
>  148:	00c1302d 	daddu	a2,a2,at
> 	...
> 
> Correct this by switching off the `noreorder' assembly mode and letting 
> GAS schedule this jump's delay slot, as there is nothing special about 
> it that would require manual scheduling.  With this change in place 
> correct code is produced:
> 
>  13c:	01a4302f 	dsubu	a2,t1,a0
>  140:	24010001 	li	at,1
>  144:	03e00008 	jr	ra
>  148:	00c1302d 	daddu	a2,a2,at
> 	...
> 
> Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org>
> Fixes: 8a8158c85e1e ("MIPS: memset.S: EVA & fault support for small_memset")
> Cc: stable@vger.kernel.org # 4.17+
> ---
> Hi,
> 
>  This is a version (one I originally made actually) of commit 148b9aba99e0 
> for 4.18-stable and before.  No functional changes.  Please apply.

Now queued up, thanks.

greg k-h