Skip to contentSkip to conversation

arm64: Define name for the original RES1 bit but now functinal bit

PATCH Superseded Shaoqin Huang 3 participants 8 messages Jul 19, 2023 – Jul 20, 2023

Subsystems (2)

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

cover arm64: Define name for the original RES1 bit but now functinal bit

According the talk[1], because the architecture get updated, what used to be a RES1 bit becomes a functinal bit. So we can define the name for these bits, this also increase the readability.

[1] lore.kernel.org/[email protected]

Shaoqin Huang (2):
arm64: Replace the SCTLR_EL1 filed definition by _BITUL()
arm64: Define name for the bits used in SCTLR_EL1_RES1

 lib/arm64/asm/sysreg.h | 31 ++++++++++++++++++++-----------
 1 file changed, 20 insertions(+), 11 deletions(-)
Signature · 1 line
2.39.1
#1 lore raw

PATCH 1/2 arm64: Replace the SCTLR_EL1 filed definition by _BITUL()

Currently the SCTLR_EL1_* is defined by (1 << x), all of them can be replaced by the _BITUL() macro to make the format consistent with the SCTLR_EL1_RES1 definition.

---
 lib/arm64/asm/sysreg.h | 16 ++++++++--------
 1 file changed, 8 insertions(+), 8 deletions(-)
lib/arm64/asm/sysreg.h +8 -8
index 18c4ed3..c7f529d 100644
@@ -80,14 +80,14 @@ asm(
 #define ICC_GRPEN1_EL1			sys_reg(3, 0, 12, 12, 7)
 
 /* System Control Register (SCTLR_EL1) bits */
-#define SCTLR_EL1_EE	(1 << 25)
-#define SCTLR_EL1_WXN	(1 << 19)
-#define SCTLR_EL1_I	(1 << 12)
-#define SCTLR_EL1_SA0	(1 << 4)
-#define SCTLR_EL1_SA	(1 << 3)
-#define SCTLR_EL1_C	(1 << 2)
-#define SCTLR_EL1_A	(1 << 1)
-#define SCTLR_EL1_M	(1 << 0)
+#define SCTLR_EL1_EE		_BITUL(25)
+#define SCTLR_EL1_WXN		_BITUL(19)
+#define SCTLR_EL1_I		_BITUL(12)
+#define SCTLR_EL1_SA0		_BITUL(4)
+#define SCTLR_EL1_SA		_BITUL(3)
+#define SCTLR_EL1_C		_BITUL(2)
+#define SCTLR_EL1_A		_BITUL(1)
+#define SCTLR_EL1_M		_BITUL(0)
 
 #define SCTLR_EL1_RES1	(_BITUL(7) | _BITUL(8) | _BITUL(11) | _BITUL(20) | \
 			 _BITUL(22) | _BITUL(23) | _BITUL(28) | _BITUL(29))
Signature · 1 line
2.39.1
#2 lore raw

Re: [kvm-unit-tests PATCH v1 1/2] arm64: Replace the SCTLR_EL1 filed definition by _BITUL()

Hi Shaoqin,

Quoted 3 lines · Shaoqin Huang

Currently the SCTLR_EL1_* is defined by (1 << x), all of them can be replaced by the _BITUL() macro to make the format consistent with the SCTLR_EL1_RES1 definition.

I would rephrase the commit title into arm64: Use _BITUL() to define SCTLR_EL1 bit fields

Besides, since SCTLR_EL1 is 64b shouldn't we have _BITULL() everywhere instead?

Eric

Quoted 33 lines
---
 lib/arm64/asm/sysreg.h | 16 ++++++++--------
 1 file changed, 8 insertions(+), 8 deletions(-)
lib/arm64/asm/sysreg.h +8 -8
index 18c4ed3..c7f529d 100644
@@ -80,14 +80,14 @@ asm(
 #define ICC_GRPEN1_EL1			sys_reg(3, 0, 12, 12, 7)
 
 /* System Control Register (SCTLR_EL1) bits */
-#define SCTLR_EL1_EE	(1 << 25)
-#define SCTLR_EL1_WXN	(1 << 19)
-#define SCTLR_EL1_I	(1 << 12)
-#define SCTLR_EL1_SA0	(1 << 4)
-#define SCTLR_EL1_SA	(1 << 3)
-#define SCTLR_EL1_C	(1 << 2)
-#define SCTLR_EL1_A	(1 << 1)
-#define SCTLR_EL1_M	(1 << 0)
+#define SCTLR_EL1_EE		_BITUL(25)
+#define SCTLR_EL1_WXN		_BITUL(19)
+#define SCTLR_EL1_I		_BITUL(12)
+#define SCTLR_EL1_SA0		_BITUL(4)
+#define SCTLR_EL1_SA		_BITUL(3)
+#define SCTLR_EL1_C		_BITUL(2)
+#define SCTLR_EL1_A		_BITUL(1)
+#define SCTLR_EL1_M		_BITUL(0)
 
 #define SCTLR_EL1_RES1	(_BITUL(7) | _BITUL(8) | _BITUL(11) | _BITUL(20) | \
 			 _BITUL(22) | _BITUL(23) | _BITUL(28) | _BITUL(29))
#3 lore raw

Re: [kvm-unit-tests PATCH v1 1/2] arm64: Replace the SCTLR_EL1 filed definition by _BITUL()

Hi Eric,

Quoted 10 lines · Eric Auger

Hi Shaoqin,

Quoted 3 lines · Shaoqin Huang

Currently the SCTLR_EL1_* is defined by (1 << x), all of them can be replaced by the _BITUL() macro to make the format consistent with the SCTLR_EL1_RES1 definition.

I would rephrase the commit title into arm64: Use _BITUL() to define SCTLR_EL1 bit fields

This would be good. I will change the title to it.

Quoted 2 lines

Besides, since SCTLR_EL1 is 64b shouldn't we have _BITULL() everywhere instead?

Yeah. Although the _BITUL() can cover the 64 bit value, it would be clear to use the _BITULL(), let me change it in next version.

Thanks,
Shaoqin

Quoted 36 lines

Eric

Quoted 33 lines
---
  lib/arm64/asm/sysreg.h | 16 ++++++++--------
  1 file changed, 8 insertions(+), 8 deletions(-)
lib/arm64/asm/sysreg.h +8 -8
index 18c4ed3..c7f529d 100644
@@ -80,14 +80,14 @@ asm(
  #define ICC_GRPEN1_EL1			sys_reg(3, 0, 12, 12, 7)
 
  /* System Control Register (SCTLR_EL1) bits */
-#define SCTLR_EL1_EE	(1 << 25)
-#define SCTLR_EL1_WXN	(1 << 19)
-#define SCTLR_EL1_I	(1 << 12)
-#define SCTLR_EL1_SA0	(1 << 4)
-#define SCTLR_EL1_SA	(1 << 3)
-#define SCTLR_EL1_C	(1 << 2)
-#define SCTLR_EL1_A	(1 << 1)
-#define SCTLR_EL1_M	(1 << 0)
+#define SCTLR_EL1_EE		_BITUL(25)
+#define SCTLR_EL1_WXN		_BITUL(19)
+#define SCTLR_EL1_I		_BITUL(12)
+#define SCTLR_EL1_SA0		_BITUL(4)
+#define SCTLR_EL1_SA		_BITUL(3)
+#define SCTLR_EL1_C		_BITUL(2)
+#define SCTLR_EL1_A		_BITUL(1)
+#define SCTLR_EL1_M		_BITUL(0)
 
  #define SCTLR_EL1_RES1	(_BITUL(7) | _BITUL(8) | _BITUL(11) | _BITUL(20) | \
  			 _BITUL(22) | _BITUL(23) | _BITUL(28) | _BITUL(29))
Signature · 1 line
Shaoqin
#4 lore raw

PATCH 2/2 arm64: Define name for the bits used in SCTLR_EL1_RES1

In reply to Shaoqin Huang

Currently some fields in SCTLR_EL1 don't define a name and directly used in the SCTLR_EL1_RES1, that's not good now since these fields have been functional and have a name.

According to the ARM DDI 0487J.a, define the name related to these fields.

---
 lib/arm64/asm/sysreg.h | 13 +++++++++++--
 1 file changed, 11 insertions(+), 2 deletions(-)
lib/arm64/asm/sysreg.h +11 -2
index c7f529d..9c68698 100644
@@ -80,17 +80,26 @@ asm(
 #define ICC_GRPEN1_EL1			sys_reg(3, 0, 12, 12, 7)
 
 /* System Control Register (SCTLR_EL1) bits */
+#define SCTLR_EL1_LSMAOE	_BITUL(29)
+#define SCTLR_EL1_NTLSMD	_BITUL(28)
 #define SCTLR_EL1_EE		_BITUL(25)
+#define SCTLR_EL1_SPAN		_BITUL(23)
+#define SCTLR_EL1_EIS		_BITUL(22)
+#define SCTLR_EL1_TSCXT		_BITUL(20)
 #define SCTLR_EL1_WXN		_BITUL(19)
 #define SCTLR_EL1_I		_BITUL(12)
+#define SCTLR_EL1_EOS		_BITUL(11)
+#define SCTLR_EL1_SED		_BITUL(8)
+#define SCTLR_EL1_ITD		_BITUL(7)
 #define SCTLR_EL1_SA0		_BITUL(4)
 #define SCTLR_EL1_SA		_BITUL(3)
 #define SCTLR_EL1_C		_BITUL(2)
 #define SCTLR_EL1_A		_BITUL(1)
 #define SCTLR_EL1_M		_BITUL(0)
 
-#define SCTLR_EL1_RES1	(_BITUL(7) | _BITUL(8) | _BITUL(11) | _BITUL(20) | \
-			 _BITUL(22) | _BITUL(23) | _BITUL(28) | _BITUL(29))
+#define SCTLR_EL1_RES1	(SCTLR_EL1_ITD | SCTLR_EL1_SED | SCTLR_EL1_EOS | \
+			 SCTLR_EL1_TSCXT | SCTLR_EL1_EIS | SCTLR_EL1_SPAN | \
+			 SCTLR_EL1_NTLSMD | SCTLR_EL1_LSMAOE)
 #define INIT_SCTLR_EL1_MMU_OFF	\
 			SCTLR_EL1_RES1
 
Signature · 1 line
2.39.1
#5 lore raw

Re: [kvm-unit-tests PATCH v1 2/2] arm64: Define name for the bits used in SCTLR_EL1_RES1

Quoted 8 lines · Shaoqin Huang

Currently some fields in SCTLR_EL1 don't define a name and directly used in the SCTLR_EL1_RES1, that's not good now since these fields have been functional and have a name.

Quoted 5 lines

According to the ARM DDI 0487J.a, define the name related to these fields.

Thanks for doing that!

While I debugged my SCTLR issue on EFI, I did try to grep “SPAN", and this patch would’ve certainly helped.

#6 lore raw

Re: [kvm-unit-tests PATCH v1 2/2] arm64: Define name for the bits used in SCTLR_EL1_RES1

In reply to Shaoqin Huang

Hi Shaoqin,

Quoted 36 lines · Shaoqin Huang

Currently some fields in SCTLR_EL1 don't define a name and directly used in the SCTLR_EL1_RES1, that's not good now since these fields have been functional and have a name.

According to the ARM DDI 0487J.a, define the name related to these fields.

---
 lib/arm64/asm/sysreg.h | 13 +++++++++++--
 1 file changed, 11 insertions(+), 2 deletions(-)
lib/arm64/asm/sysreg.h +8 -0
index c7f529d..9c68698 100644
@@ -80,17 +80,26 @@ asm(
 #define ICC_GRPEN1_EL1			sys_reg(3, 0, 12, 12, 7)
 
 /* System Control Register (SCTLR_EL1) bits */
+#define SCTLR_EL1_LSMAOE	_BITUL(29)
+#define SCTLR_EL1_NTLSMD	_BITUL(28)
 #define SCTLR_EL1_EE		_BITUL(25)
+#define SCTLR_EL1_SPAN		_BITUL(23)
+#define SCTLR_EL1_EIS		_BITUL(22)
+#define SCTLR_EL1_TSCXT		_BITUL(20)
 #define SCTLR_EL1_WXN		_BITUL(19)
 #define SCTLR_EL1_I		_BITUL(12)
+#define SCTLR_EL1_EOS		_BITUL(11)
+#define SCTLR_EL1_SED		_BITUL(8)
+#define SCTLR_EL1_ITD		_BITUL(7)
 #define SCTLR_EL1_SA0		_BITUL(4)
 #define SCTLR_EL1_SA		_BITUL(3)
 #define SCTLR_EL1_C		_BITUL(2)
 #define SCTLR_EL1_A		_BITUL(1)

lib/arm64/asm/sysreg.h asm( · line 105

Quoted 10 lines
lib/arm64/asm/sysreg.h +3 -2
 #define SCTLR_EL1_M		_BITUL(0)
 
-#define SCTLR_EL1_RES1	(_BITUL(7) | _BITUL(8) | _BITUL(11) | _BITUL(20) | \
-			 _BITUL(22) | _BITUL(23) | _BITUL(28) | _BITUL(29))
+#define SCTLR_EL1_RES1	(SCTLR_EL1_ITD | SCTLR_EL1_SED | SCTLR_EL1_EOS | \
+			 SCTLR_EL1_TSCXT | SCTLR_EL1_EIS | SCTLR_EL1_SPAN | \
+			 SCTLR_EL1_NTLSMD | SCTLR_EL1_LSMAOE)
 #define INIT_SCTLR_EL1_MMU_OFF	\
 			SCTLR_EL1_RES1
 

The change looks good to me (although _BITULL remark still holds).

Independently on this patch the _RES1 terminology looks odd to me. For example ESO bit is RES1 only if FEAT_ExS is not implemented. Maybe I misunderstand why it was named that way but to me RES1 means another thing. If confirmed we could simply drop SCTLR_EL1_RES1 which is not used elsewhere and directly define INIT_SCTLR_EL1_MMU_OF.

Thanks

Eric

#7 lore raw

Re: [kvm-unit-tests PATCH v1 2/2] arm64: Define name for the bits used in SCTLR_EL1_RES1

Hi Eric,

Quoted 51 lines · Eric Auger

Hi Shaoqin,

Quoted 36 lines · Shaoqin Huang

Currently some fields in SCTLR_EL1 don't define a name and directly used in the SCTLR_EL1_RES1, that's not good now since these fields have been functional and have a name.

According to the ARM DDI 0487J.a, define the name related to these fields.

---
  lib/arm64/asm/sysreg.h | 13 +++++++++++--
  1 file changed, 11 insertions(+), 2 deletions(-)
lib/arm64/asm/sysreg.h +8 -0
index c7f529d..9c68698 100644
@@ -80,17 +80,26 @@ asm(
  #define ICC_GRPEN1_EL1			sys_reg(3, 0, 12, 12, 7)
 
  /* System Control Register (SCTLR_EL1) bits */
+#define SCTLR_EL1_LSMAOE	_BITUL(29)
+#define SCTLR_EL1_NTLSMD	_BITUL(28)
  #define SCTLR_EL1_EE		_BITUL(25)
+#define SCTLR_EL1_SPAN		_BITUL(23)
+#define SCTLR_EL1_EIS		_BITUL(22)
+#define SCTLR_EL1_TSCXT		_BITUL(20)
  #define SCTLR_EL1_WXN		_BITUL(19)
  #define SCTLR_EL1_I		_BITUL(12)
+#define SCTLR_EL1_EOS		_BITUL(11)
+#define SCTLR_EL1_SED		_BITUL(8)
+#define SCTLR_EL1_ITD		_BITUL(7)
  #define SCTLR_EL1_SA0		_BITUL(4)
  #define SCTLR_EL1_SA		_BITUL(3)
  #define SCTLR_EL1_C		_BITUL(2)
  #define SCTLR_EL1_A		_BITUL(1)

lib/arm64/asm/sysreg.h asm( · line 105

Quoted 10 lines
lib/arm64/asm/sysreg.h +3 -2
  #define SCTLR_EL1_M		_BITUL(0)
 
-#define SCTLR_EL1_RES1	(_BITUL(7) | _BITUL(8) | _BITUL(11) | _BITUL(20) | \
-			 _BITUL(22) | _BITUL(23) | _BITUL(28) | _BITUL(29))
+#define SCTLR_EL1_RES1	(SCTLR_EL1_ITD | SCTLR_EL1_SED | SCTLR_EL1_EOS | \
+			 SCTLR_EL1_TSCXT | SCTLR_EL1_EIS | SCTLR_EL1_SPAN | \
+			 SCTLR_EL1_NTLSMD | SCTLR_EL1_LSMAOE)
  #define INIT_SCTLR_EL1_MMU_OFF	\
  			SCTLR_EL1_RES1
 

The change looks good to me (although _BITULL remark still holds).

Quoted 5 lines

Independently on this patch the _RES1 terminology looks odd to me. For example ESO bit is RES1 only if FEAT_ExS is not implemented. Maybe I misunderstand why it was named that way but to me RES1 means another thing. If confirmed we could simply drop SCTLR_EL1_RES1 which is not used elsewhere and directly define INIT_SCTLR_EL1_MMU_OF.

The SCTLR_EL1_RES1 was originally defined by commit 10b65ce ("arm64: Configure SCTLR_EL1 at boot"). It seems at that time the BIT(11), yes the EOS bit is undefined, and must be reserved for 1, but now it has a definition which is EOS. It should have no effect if set EOS bit to 1 since its originally res to 1.

SCTLR_EL1_RES1 can simply be dropped since it has no other user.

Thanks,
Shaoqin

Quoted 5 lines

Thanks

Eric

Signature · 1 line
Shaoqin