Skip to contentSkip to conversation

Cleanups and improvements

PATCH In review Mark Rutland 3 participants 34 messages Jan 25, 2022 – Jan 28, 2022 Kbuild

lore
Share & tools
lore
#0 lore raw

v3 cover Cleanups and improvements

This series reworks the aarch64 boot-wrapper, moving a fair amount of initialization code to C. This has a few benefits:

1) This makes the code more legible and easier to maintain.

Current feature detection and system register configuration is written in assembly, requiring runs of *almost* identical blocks of assembly to read ID registers and conditionally initialize register values. This requires a bunch of labels (which are all named numerically), and all the magic numbers are hard coded, so this gets pretty painful to read:

| mrs x1, id_aa64isar0_el1
| ubfx x1, x1, #24, #4
| cbz x1, 1f
|
| orr x0, x0, #(1 << 34) // TME enable
|
| 1:

In C, it's much easier to add helpers which use mnemonics, which makes the code much easier to read, and avoids the need to manually allocate registers, etc:

| if (mrs_field(ID_AA64ISAR0_EL1, TME))
| scr |= SCR_EL3_TME;

This should make it easier to handle new architectural extensions (and/or architecture variants such as ARMv8-R) in future.

2) This allows for better diagnostics.

Currently a mis-configured boot-wrapper is rather unforgiving, and provides no indication as to what might have gone wrong. By moving initialization to C, we can make use to the UART code to log diagnostic information, and we can more easily add additional error handling and conditional logic.

This series adds diagnostic information and error handling that can be used to identify problems such as the boot-wrapper being launched at the wrong exception level:

| Boot-wrapper v0.2
| Entered at EL2
| Memory layout:
| [0000000080000000..0000000080001f90] => boot-wrapper | [000000008000fff8..0000000080010000] => mbox | [0000000080200000..00000000822af200] => kernel | [0000000088000000..0000000088002857] => dtb |
| WARNING: PSCI could not be initialized. Boot may fail

Originally I had planned for this to be a more expansive set of changes, unifying some early control-flow, fixing some latent bugs, and making the boot-wrapper dynamically handle being entered at any of EL{3,2,1} with automated selection of a suitable DT. As the series has already become pretty long, I'd like to get this preparatory cleanup out of the way first, and handle those cases with subsequent patches.

If there are no objections I'll apply this by the end of this week.

I've pushed the series to the cleanup branch in the boot-wrapper repo:

https://git.kernel.org/pub/scm/linux/kernel/git/mark/boot-wrapper-aarch64.git/
git://git.kernel.org/pub/scm/linux/kernel/git/mark/boot-wrapper-aarch64.git

... and it should apply cleanup atop the master branch.

Since v1 [1]:
* Add comments for bit-field macros
* Use BIT() for *_RES1 definitions
* Complete start_el3/start_no_el3 cleanup
* Remove extraneous macro definition

Since v2 [2]:
* Accumulate tags
* Rebase atop other recent patches
* Correct ZCR_EL3.LEN initialization
* Drop BITS_INSERT()
* Split SCTLR_EL1 values used for AArch32 kernels
* Add missing SCTLR_EL1 RES1 bits
* Refactor print_ulong_hex()
* Refactor PSCI arch initialization

[1] https://lore.kernel.org/r/[email protected]/
[2] https://lore.kernel.org/r/[email protected]/
Thanks,
Mark.

Mark Rutland (15):
aarch64: correct ZCR_EL3.LEN initialization
aarch64: correct SCTLR_EL1_KERNEL for AA32 kernels Document entry requirements
Add bit-field macros
aarch64: add system register accessors
aarch32: add coprocessor accessors
aarch64: add mov_64 macro
aarch64: initialize SCTLR_ELx for the boot-wrapper Rework common init C code
Announce boot-wrapper mode / exception level
aarch64: move the bulk of EL3 initialization to C
aarch32: move the bulk of Secure PL1 initialization to C Announce locations of memory objects
Rework bootmethod initialization
Unify start_el3 & start_no_el3

 Makefile.am                       |   7 +-
 arch/aarch32/boot.S               |  33 ++++--
 arch/aarch32/include/asm/cpu.h    |  62 ++++++++---
 arch/aarch32/include/asm/gic-v3.h |   6 +-
 arch/aarch32/init.c               |  57 ++++++++++
 arch/aarch32/psci.S               |  16 +--
 arch/aarch32/utils.S              |   9 --
 arch/aarch64/boot.S               | 177 ++++++++++++------------------
 arch/aarch64/common.S             |  10 +-
 arch/aarch64/include/asm/cpu.h    | 117 +++++++++++++++++---
 arch/aarch64/include/asm/gic-v3.h |  10 +-
 arch/aarch64/init.c               | 106 ++++++++++++++++++
 arch/aarch64/psci.S               |  22 +---
 arch/aarch64/spin.S               |   6 +-
 arch/aarch64/utils.S              |   9 --
 common/boot.c                     |   4 -
 common/init.c                     |  60 ++++++++++
 common/platform.c                 |  45 +++++---
 common/psci.c                     |  20 +++-
 include/bits.h                    |  66 +++++++++++
 include/boot.h                    |   7 ++
 include/platform.h                |  20 ++++
 model.lds.S                       |  20 +++-
 23 files changed, 648 insertions(+), 241 deletions(-)
 create mode 100644 arch/aarch32/init.c
 create mode 100644 arch/aarch64/init.c
 create mode 100644 common/init.c
 create mode 100644 include/bits.h
 create mode 100644 include/platform.h
Signature · 7 lines
2.30.2


_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
#1 lore raw

PATCH v3 1/15 aarch64: correct ZCR_EL3.LEN initialization

The ZCR_ELx.LEN field covers bits[3:0] of ZCR_ELx, and bits[8:4] are RAZ/WI, likely so that in future these can be used to extend LEN without complicating the probing of the maximum available vector length.

Currently ZCR_EL3_LEN_MASK is defined as 0x1ff, covering both the LEN field and the RAZ/WI bits.

To match the architecture as documented, reduce this down to 0xf, only covering the bits currently allocated to the LEN field. We can extend this in future if the LEN field is widened.

As the same time, rename ZCR_EL3_LEN_MASK to ZCR_EL3_LEN_MAX, as it is used as a value rather than as a bit mask.

Cc Andre Przywara <[email protected]>
Cc Mark Brown <[email protected]>
---
 arch/aarch64/boot.S            | 2 +-
 arch/aarch64/include/asm/cpu.h | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)
arch/aarch64/boot.S +1 -1
index 72c6b79..ee9ddaa 100644
@@ -123,7 +123,7 @@ ASM_FUNC(_start)
 	msr	cptr_el3, x0
 	isb
 
-	mov	x0, #ZCR_EL3_LEN_MASK		// SVE: Enable full vector len
+	mov	x0, #ZCR_EL3_LEN_MAX		// SVE: Enable full vector len
 	msr	ZCR_EL3, x0			// for EL2.
 
 1:
arch/aarch64/include/asm/cpu.h +1 -1
index 63eb1c3..e3269b8 100644
@@ -30,7 +30,7 @@
 #define CPTR_EL3_EZ		(1 << 8)
 
 #define ZCR_EL3			s3_6_c1_c2_0
-#define ZCR_EL3_LEN_MASK	0x1ff
+#define ZCR_EL3_LEN_MAX		0xf
 
 #define SCTLR_EL1_CP15BEN	(1 << 5)
 #define SCTLR_EL1_RES1		(3 << 28 | 3 << 22 | 1 << 11)
Signature · 7 lines
2.30.2


_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
#2 lore raw

Re: [bootwrapper PATCH v3 01/15] aarch64: correct ZCR_EL3.LEN initialization

Quoted 2 lines · Mark Rutland

As the same time, rename ZCR_EL3_LEN_MASK to ZCR_EL3_LEN_MAX, as it is used as a value rather than as a bit mask.

Well, it's both and we do call it a mask in both the kernel and TF-A too...

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

#3 lore raw

Re: [bootwrapper PATCH v3 01/15] aarch64: correct ZCR_EL3.LEN initialization

Quoted 7 lines · Mark Brown
Quoted 2 lines · Mark Rutland

As the same time, rename ZCR_EL3_LEN_MASK to ZCR_EL3_LEN_MAX, as it is used as a value rather than as a bit mask.

Well, it's both and we do call it a mask in both the kernel and TF-A too...

Here specifically in the boot-wrapper I'm just using it as a value, so I don't see that those are relevant?

The idea was to be consistent with the <register>_<field>_<valuename> naming, and avoiding the "why are you programming a mask as the value" questions from last time around, where the unambiguous options are:

a) Use <register>_<field>_<valuename> definition, as here.

b) Use <register>_<field>_MASK, and add comments at each usage as to why we use a mask as a value, to explain why that isn't a bug.

c) Have both <register>_<field>_MASK and some value definition, and use some insertion helper to insert the value.

... and I went with (a) because it was the simplest.

Is there a problem with this?

Thanks,
Mark.

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

#4 lore raw

Re: [bootwrapper PATCH v3 01/15] aarch64: correct ZCR_EL3.LEN initialization

Quoted 11 lines · Mark Rutland
Quoted 7 lines · Mark Brown
Quoted 2 lines · Mark Rutland

As the same time, rename ZCR_EL3_LEN_MASK to ZCR_EL3_LEN_MAX, as it is used as a value rather than as a bit mask.

Well, it's both and we do call it a mask in both the kernel and TF-A too...

Here specifically in the boot-wrapper I'm just using it as a value, so I don't see that those are relevant?

That's also a bunch of the other usage - the kernel uses it to set the maximum available value by virtue of setting every possible bit in the bitfield, though it does also use it as a mask in places.

Quoted 3 lines

The idea was to be consistent with the <register>_<field>_<valuename> naming, and avoiding the "why are you programming a mask as the value" questions from last time around, where the unambiguous options are:

I've not seen that - I did see that there had been some discussion as to why the code had been including the reserved bits but hadn't seen any discussion of confusion between mask and max.

Quoted 11 lines

a) Use <register>_<field>_<valuename> definition, as here.

b) Use <register>_<field>_MASK, and add comments at each usage as to why we use a mask as a value, to explain why that isn't a bug.

c) Have both <register>_<field>_MASK and some value definition, and use some insertion helper to insert the value.

... and I went with (a) because it was the simplest.

Is there a problem with this?

My concern is continuity of the enumeration algorithm between the various implementations we have, it's something we're not currently great with and this creates a separation between the kernel and the boot wrapper implementations. There's no change in what's actually being done but it creates some additional effort to figure out why we're setting a maximum here and not trying to set all the bits as we do in the kernel. TF-A is doing something else again (actually two different things if you include SME which uses the same algorithm for the streaming vector length), for SVE it uses a maximum length too but it uses a lower length than the maximum representable one that's used here.

It may be that the best thing to do is to make this change but also update the kernel to have separate but identical mask and maximum definitions, though TBH I'd rather not actually have that update happen while the SME series is pending.

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

#5 lore raw

Re: [bootwrapper PATCH v3 01/15] aarch64: correct ZCR_EL3.LEN initialization

Quoted 17 lines · Mark Brown
Quoted 11 lines · Mark Rutland
Quoted 7 lines · Mark Brown
Quoted 2 lines · Mark Rutland

As the same time, rename ZCR_EL3_LEN_MASK to ZCR_EL3_LEN_MAX, as it is used as a value rather than as a bit mask.

Well, it's both and we do call it a mask in both the kernel and TF-A too...

Here specifically in the boot-wrapper I'm just using it as a value, so I don't see that those are relevant?

That's also a bunch of the other usage - the kernel uses it to set the maximum available value by virtue of setting every possible bit in the bitfield, though it does also use it as a mask in places.

Quoted 7 lines
Quoted 3 lines

The idea was to be consistent with the <register>_<field>_<valuename> naming, and avoiding the "why are you programming a mask as the value" questions from last time around, where the unambiguous options are:

I've not seen that - I did see that there had been some discussion as to why the code had been including the reserved bits but hadn't seen any discussion of confusion between mask and max.

In:

https://lore.kernel.org/all/[email protected]/#t

w.r.t. the use of ZCR_EL3_LEN, Andre said:

| So when comparing this to the other uses of XXX_EL3_YYY, they typically | describe a mask, but here we seems to abuse this as a value?

... which begat the rename. In this cleanup I've tried to use <register>_<field> for masks, and <register>_<field>_<valuename> for specific values to program in, and hence I added ZCR_EL3_LEN_MAX.

I'm going to stick with ZCR_EL3_LEN_MAX.

In that same paragraph he said:

| the existing code writes 0x1ff into that register, presumable to cover future | vector length extensions

... which begat the conversation of which length to use.

I later noted that TF-A currently uses 0xf, and so I aligned with that and plan to stick with it in the absence of any published architectural guidance that we *should* set the RAZ/WI bits. We can always add that in future when necessary if/when the field grows, and until then it's less surprising to only be setting the bits that are documented as needing to be set in this manner.

Quoted 12 lines
Quoted 11 lines

a) Use <register>_<field>_<valuename> definition, as here.

b) Use <register>_<field>_MASK, and add comments at each usage as to why we use a mask as a value, to explain why that isn't a bug.

c) Have both <register>_<field>_MASK and some value definition, and use some insertion helper to insert the value.

... and I went with (a) because it was the simplest.

Is there a problem with this?

Quoted 7 lines

My concern is continuity of the enumeration algorithm between the various implementations we have, it's something we're not currently great with and this creates a separation between the kernel and the boot wrapper implementations. There's no change in what's actually being done but it creates some additional effort to figure out why we're setting a maximum here and not trying to set all the bits as we do in the kernel.

TBH, I'd argue if we're setting those bits in the kernel it's probably a bug, because we don't know *exactly* what effect they'll have when allocated in the future.

Thanks,
Mark.

Quoted 9 lines

TF-A is doing something else again (actually two different things if you include SME which uses the same algorithm for the streaming vector length), for SVE it uses a maximum length too but it uses a lower length than the maximum representable one that's used here.

It may be that the best thing to do is to make this change but also update the kernel to have separate but identical mask and maximum definitions, though TBH I'd rather not actually have that update happen while the SME series is pending.

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

#6 lore raw

Re: [bootwrapper PATCH v3 01/15] aarch64: correct ZCR_EL3.LEN initialization

Quoted 24 lines · Mark Rutland
Quoted 15 lines · Mark Brown
Quoted 7 lines · Mark Rutland

On Tue, Jan 25, 2022 at 03:59:38PM +0000, Mark Brown wrote:

a) Use <register>_<field>_<valuename> definition, as here.

b) Use <register>_<field>_MASK, and add comments at each usage as to why we use a mask as a value, to explain why that isn't a bug.

Quoted 6 lines

c) Have both <register>_<field>_MASK and some value definition, and use some insertion helper to insert the value.

... and I went with (a) because it was the simplest.

Is there a problem with this?

Quoted 7 lines

My concern is continuity of the enumeration algorithm between the various implementations we have, it's something we're not currently great with and this creates a separation between the kernel and the boot wrapper implementations. There's no change in what's actually being done but it creates some additional effort to figure out why we're setting a maximum here and not trying to set all the bits as we do in the kernel.

Quoted 3 lines

TBH, I'd argue if we're setting those bits in the kernel it's probably a bug, because we don't know *exactly* what effect they'll have when allocated in the future.

My comments there are related purely to the renaming of the constant, not to the narrowing of the bitmask which is less of an issue. Basically avoiding the question "why is that a maximum and not a mask" and if there's limits on the valid values that are more restrictive than the representable values.

The ideal thing for the boot wrapper would be option (c) with either a configuration option to set the maximum to something lower than all bits, unless someone was unduly enthusiastic and wanted to have an all CPUs enumeration algorithm, but both are probably more trouble than it's worth.

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

#7 lore raw

Re: [bootwrapper PATCH v3 01/15] aarch64: correct ZCR_EL3.LEN initialization

Quoted 25 lines · Mark Brown
Quoted 23 lines · Mark Rutland
Quoted 22 lines · Mark Brown
Quoted 7 lines · Mark Rutland

On Tue, Jan 25, 2022 at 03:59:38PM +0000, Mark Brown wrote:

a) Use <register>_<field>_<valuename> definition, as here.

b) Use <register>_<field>_MASK, and add comments at each usage as to why we use a mask as a value, to explain why that isn't a bug.

Quoted 6 lines

c) Have both <register>_<field>_MASK and some value definition, and use some insertion helper to insert the value.

... and I went with (a) because it was the simplest.

Is there a problem with this?

My concern is continuity of the enumeration algorithm between the various implementations we have, it's something we're not currently great with and this creates a separation between the kernel and the boot wrapper implementations. There's no change in what's actually being done but it creates some additional effort to figure out why we're setting a maximum here and not trying to set all the bits as we do in the kernel.

Quoted 6 lines
Quoted 3 lines

TBH, I'd argue if we're setting those bits in the kernel it's probably a bug, because we don't know *exactly* what effect they'll have when allocated in the future.

My comments there are related purely to the renaming of the constant, not to the narrowing of the bitmask which is less of an issue.

Ok.

Quoted 9 lines

Basically avoiding the question "why is that a maximum and not a mask" and if there's limits on the valid values that are more restrictive than the representable values.

The ideal thing for the boot wrapper would be option (c) with either a configuration option to set the maximum to something lower than all bits, unless someone was unduly enthusiastic and wanted to have an all CPUs enumeration algorithm, but both are probably more trouble than it's worth.

FWIW, if it's useful for testing to be able to configure the ZCR_EL3.LEN to some smaller value, I'd be happy to move to (c) and get that plumbed into the build system.

For now, I plan to take the series with this as-is, since having ZCR_EL3_LEN_MAX is consistent with other field values we only insert, e.g. MDCR_EL3_NSPB_NS_NOTRAP.

I'm happy to rejig that in future.

Thanks,
Mark

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

#8 lore raw

Re: [bootwrapper PATCH v3 01/15] aarch64: correct ZCR_EL3.LEN initialization

Quoted 11 lines · Mark Rutland
Quoted 5 lines · Mark Brown

The ideal thing for the boot wrapper would be option (c) with either a configuration option to set the maximum to something lower than all bits, unless someone was unduly enthusiastic and wanted to have an all CPUs enumeration algorithm, but both are probably more trouble than it's worth.

FWIW, if it's useful for testing to be able to configure the ZCR_EL3.LEN to some smaller value, I'd be happy to move to (c) and get that plumbed into the build system.

For Linux it'd mainly be useful in the case where the boot wrapper got used on systems which have asymmetric maximum vector lengths.

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

Reviewed-by on v3 1/15
#9 lore raw

Re: [bootwrapper PATCH v3 01/15] aarch64: correct ZCR_EL3.LEN initialization

In reply to Mark Rutland
Quoted 11 lines · On Tue, 25 Jan 2022 15:00:43 +0000 Mark Rutland <[email protected]> wrote:

The ZCR_ELx.LEN field covers bits[3:0] of ZCR_ELx, and bits[8:4] are RAZ/WI, likely so that in future these can be used to extend LEN without complicating the probing of the maximum available vector length.

Currently ZCR_EL3_LEN_MASK is defined as 0x1ff, covering both the LEN field and the RAZ/WI bits.

To match the architecture as documented, reduce this down to 0xf, only covering the bits currently allocated to the LEN field. We can extend this in future if the LEN field is widened.

Quoted 6 lines

As the same time, rename ZCR_EL3_LEN_MASK to ZCR_EL3_LEN_MAX, as it is used as a value rather than as a bit mask.

Cc Andre Przywara <[email protected]>
Cc Mark Brown <[email protected]>

Looks good to me:

Cheers,
Andre

Quoted 23 lines
---
 arch/aarch64/boot.S            | 2 +-
 arch/aarch64/include/asm/cpu.h | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)
arch/aarch64/boot.S +1 -1
index 72c6b79..ee9ddaa 100644
@@ -123,7 +123,7 @@ ASM_FUNC(_start)
 	msr	cptr_el3, x0
 	isb
 
-	mov	x0, #ZCR_EL3_LEN_MASK		// SVE: Enable full vector len
+	mov	x0, #ZCR_EL3_LEN_MAX		// SVE: Enable full vector len
 	msr	ZCR_EL3, x0			// for EL2.
 
 1:
arch/aarch64/include/asm/cpu.h +0 -0
index 63eb1c3..e3269b8 100644

arch/aarch64/include/asm/cpu.h line 36

Quoted 8 lines
arch/aarch64/include/asm/cpu.h +1 -1
@@ -30,7 +30,7 @@
 #define CPTR_EL3_EZ		(1 << 8)
 
 #define ZCR_EL3			s3_6_c1_c2_0
-#define ZCR_EL3_LEN_MASK	0x1ff
+#define ZCR_EL3_LEN_MAX		0xf
 
 #define SCTLR_EL1_CP15BEN	(1 << 5)
 #define SCTLR_EL1_RES1		(3 << 28 | 3 << 22 | 1 << 11)

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

#10 lore raw

PATCH v3 2/15 aarch64: correct SCTLR_EL1_KERNEL for AA32 kernels

In reply to Mark Rutland

Bits [31:0] of the AArch64 SCTLR_EL1 register are architecturally mapped to bits [31:0] of the AArch32 SCTLR register. This means that any individual bit always has the same value across the two registers.

Across the two registers, the same bit may have distinct meanings, distinct RESx behaviour, and distinct reset behaviour. For example, SCTLR_EL1[28] is nTLSMD, which we wich to initialize to 0b1, whereas SCTLR[28] is TRE, which we wish to initialize to 0b0.

To avoid setting bits which we did not intend to, and in preparation for configuring more AArch64 RES1 bits, this patch decouples the SCTLR_EL1_KERNEL value used for AArch32 kernels from the SCTLR_EL1_RES1 definitions such that is does not inherit AArch64 SCTLR_EL1 bits which would be problematic in the AArch32 SCTLR. For now we open-code a copy of the bits defined in arch/aarch64/include/asm/cpu.h.

Cc Andre Przywara <[email protected]>
---
 arch/aarch64/include/asm/cpu.h | 11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)
arch/aarch64/include/asm/cpu.h +9 -2
index e3269b8..49d3f86 100644
@@ -36,8 +36,15 @@
 #define SCTLR_EL1_RES1		(3 << 28 | 3 << 22 | 1 << 11)
 
 #ifdef KERNEL_32
-/* 32-bit kernel decompressor uses CP15 barriers */
-#define SCTLR_EL1_KERNEL	(SCTLR_EL1_RES1 | SCTLR_EL1_CP15BEN)
+/*
+ * When booting a 32-bit kernel, EL1 uses AArch32 and registers which are
+ * architecturally mapped must be configured with the AArch32 layout.
+ *
+ * We copy the AArch32 definition of SCTLR_KERNEL here.
+ *
+ * TODO: restructure the headers to share a single definition.
+ */
+#define SCTLR_EL1_KERNEL	(3 << 22 | 1 << 11 | 1 << 5 | 3 << 4)
 #define SPSR_KERNEL		(SPSR_A | SPSR_I | SPSR_F | SPSR_HYP)
 #else
 #define SCTLR_EL1_KERNEL	SCTLR_EL1_RES1
Signature · 7 lines
2.30.2


_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
Reviewed-by on v3 2/15
#11 lore raw

Re: [bootwrapper PATCH v3 02/15] aarch64: correct SCTLR_EL1_KERNEL for AA32 kernels

On Tue, 25 Jan 2022 15:00:44 +0000
Mark Rutland <[email protected]> wrote:

Hi,

Quoted 31 lines

Bits [31:0] of the AArch64 SCTLR_EL1 register are architecturally mapped to bits [31:0] of the AArch32 SCTLR register. This means that any individual bit always has the same value across the two registers.

Across the two registers, the same bit may have distinct meanings, distinct RESx behaviour, and distinct reset behaviour. For example, SCTLR_EL1[28] is nTLSMD, which we wich to initialize to 0b1, whereas SCTLR[28] is TRE, which we wish to initialize to 0b0.

To avoid setting bits which we did not intend to, and in preparation for configuring more AArch64 RES1 bits, this patch decouples the SCTLR_EL1_KERNEL value used for AArch32 kernels from the SCTLR_EL1_RES1 definitions such that is does not inherit AArch64 SCTLR_EL1 bits which would be problematic in the AArch32 SCTLR. For now we open-code a copy of the bits defined in arch/aarch64/include/asm/cpu.h.

Cc Andre Przywara <[email protected]>
---
 arch/aarch64/include/asm/cpu.h | 11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)
arch/aarch64/include/asm/cpu.h +0 -1
index e3269b8..49d3f86 100644
@@ -36,8 +36,15 @@
 #define SCTLR_EL1_RES1		(3 << 28 | 3 << 22 | 1 << 11)
 
 #ifdef KERNEL_32
-/* 32-bit kernel decompressor uses CP15 barriers */

arch/aarch64/include/asm/cpu.h line 47

Quoted 10 lines
arch/aarch64/include/asm/cpu.h +9 -1
-#define SCTLR_EL1_KERNEL	(SCTLR_EL1_RES1 | SCTLR_EL1_CP15BEN)
+/*
+ * When booting a 32-bit kernel, EL1 uses AArch32 and registers which are
+ * architecturally mapped must be configured with the AArch32 layout.
+ *
+ * We copy the AArch32 definition of SCTLR_KERNEL here.
+ *
+ * TODO: restructure the headers to share a single definition.
+ */
+#define SCTLR_EL1_KERNEL	(3 << 22 | 1 << 11 | 1 << 5 | 3 << 4)

Is the verbatim copying of the value, including the redundancy, intended? Maybe that meant 3 << 3, as bits[4:3] are documented as "Reserved, RAO/SBOP" in the ARMv7 ARM?
It's fine with me either way, since it copies the aarch32 value, just wanted to check:

Cheers,
Andre

arch/aarch64/include/asm/cpu.h line 50

Quoted 3 lines
arch/aarch64/include/asm/cpu.h +0 -0
 #define SPSR_KERNEL		(SPSR_A | SPSR_I | SPSR_F | SPSR_HYP)
 #else
 #define SCTLR_EL1_KERNEL	SCTLR_EL1_RES1

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

#12 lore raw

Re: [bootwrapper PATCH v3 02/15] aarch64: correct SCTLR_EL1_KERNEL for AA32 kernels

Quoted 34 lines · Andre Przywara

On Tue, 25 Jan 2022 15:00:44 +0000
Mark Rutland <[email protected]> wrote:

Hi,

Quoted 29 lines

Bits [31:0] of the AArch64 SCTLR_EL1 register are architecturally mapped to bits [31:0] of the AArch32 SCTLR register. This means that any individual bit always has the same value across the two registers.

Across the two registers, the same bit may have distinct meanings, distinct RESx behaviour, and distinct reset behaviour. For example, SCTLR_EL1[28] is nTLSMD, which we wich to initialize to 0b1, whereas SCTLR[28] is TRE, which we wish to initialize to 0b0.

To avoid setting bits which we did not intend to, and in preparation for configuring more AArch64 RES1 bits, this patch decouples the SCTLR_EL1_KERNEL value used for AArch32 kernels from the SCTLR_EL1_RES1 definitions such that is does not inherit AArch64 SCTLR_EL1 bits which would be problematic in the AArch32 SCTLR. For now we open-code a copy of the bits defined in arch/aarch64/include/asm/cpu.h.

Cc Andre Przywara <[email protected]>
---
 arch/aarch64/include/asm/cpu.h | 11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)
arch/aarch64/include/asm/cpu.h +0 -0
index e3269b8..49d3f86 100644
@@ -36,8 +36,15 @@
 #define SCTLR_EL1_RES1		(3 << 28 | 3 << 22 | 1 << 11)
 
Quoted 14 lines

arch/aarch64/include/asm/cpu.h line 47

Quoted 12 lines
arch/aarch64/include/asm/cpu.h +9 -2
 #ifdef KERNEL_32
-/* 32-bit kernel decompressor uses CP15 barriers */
-#define SCTLR_EL1_KERNEL	(SCTLR_EL1_RES1 | SCTLR_EL1_CP15BEN)
+/*
+ * When booting a 32-bit kernel, EL1 uses AArch32 and registers which are
+ * architecturally mapped must be configured with the AArch32 layout.
+ *
+ * We copy the AArch32 definition of SCTLR_KERNEL here.
+ *
+ * TODO: restructure the headers to share a single definition.
+ */
+#define SCTLR_EL1_KERNEL	(3 << 22 | 1 << 11 | 1 << 5 | 3 << 4)

Is the verbatim copying of the value, including the redundancy, intended?

Yup, the intent is to have the exact same definition, overlapping bits and all.

I would like to clean that up in future, but for now I want this to clearly be the same.

Quoted 6 lines

Maybe that meant 3 << 3, as bits[4:3] are documented as "Reserved, RAO/SBOP" in the ARMv7 ARM?
It's fine with me either way, since it copies the aarch32 value, just wanted to check:

Thanks!

Mark.

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

#13 lore raw

PATCH v3 3/15 Document entry requirements

In reply to Mark Rutland

Currently the boot-wrapper only supports some combinations of exception levels, with other combinations not being supported.

While we generally expect the boot-wrapper to be entered at the highest implemented exception level, the AArch32 boot-wrapper has a comment implying it supports being entered with something else owning EL3. As this would require such EL3 firmware to always be in sync with the boot-wrapper's requirements, which change over time, we don't actually support such a configuration.

Some CPU state (such as CNTFRQ/CNTFRQ_EL0) needs to be initialized at the highest implemented exception level, but today the boot-wrapper only does so when entered at EL3 / Secure-PL1. Thus, today the only completely supported configurations are EL3 / Secure-PL1, and entering in other configurations is not entirely supported.

The aarch64 jump_kernel function always writes to SCTLR_EL2, which is UNDEFINED at EL1. Hence, the aarch64 boot-wrapper does not support being entered at EL1.

The aarch32 code assumes that any non-hyp mode is Secure PL1, and attempt to switch to monitor mode. If entered on a system without the security extensions, where the highest privileged mode is Non-secure PL1, this will not work. Hence the aarch32 boot-wrapper does not support being entered at Non-secure PL1.

Actually supporting all of these configurations requires restructuring much of the boot-wrapper. For now, document the supported configurations in each architecture's boot.S, and remove the misleading comment from arch/aarch32/boot.S. Subsequent patches will improve the support and add support for additional configurations.

There should be no functional change as a result of this patch.

---
 arch/aarch32/boot.S | 14 ++++++++++++++
 arch/aarch32/psci.S |  5 -----
 arch/aarch64/boot.S | 13 +++++++++++++
 3 files changed, 27 insertions(+), 5 deletions(-)
arch/aarch32/boot.S +14 -0
index 4add338..00c432d 100644
@@ -16,6 +16,20 @@
 	.arch_extension virt
 
 	.section .init
+
+	/*
+	 * The boot-wrapper must be entered from the reset vector at the
+	 * highest implemented exception level. The boot-wrapper only supports
+	 * being entered in the following modes:
+	 *
+	 * - PL1 / EL3 (Secure) Supervisor mode
+	 *   Entering in this mode is strongly recommended.
+	 *   PL2 must be implemented.
+	 *
+	 * - PL2 / EL2 (Non-secure) Hypervisor mode
+	 *   Entering in this mode is partially supported.
+	 *   PSCI is not supported when entered in this mode.
+	 */
 ASM_FUNC(_start)
 	/* Stack initialisation */
 	cpuid	r0, r1
arch/aarch32/psci.S +0 -5
index dc7aeb7..e0d2972 100644
@@ -44,11 +44,6 @@ ASM_FUNC(start_el3)
 	/* pass through */
 
 ASM_FUNC(start_no_el3)
-	/*
-	 * For no-el3, we assume that firmware launched the boot-wrapper in
-	 * non-secure EL2 or EL1. We assume it has its own PSCI implementation
-	 * sitting at EL3, and that this path is only taken by primary CPU.
-	 */
 	cpuid	r0, r1
 	blx	find_logical_id
 	b	psci_first_spin
arch/aarch64/boot.S +13 -0
index ee9ddaa..d682ba5 100644
@@ -12,6 +12,19 @@
 
 	.section .init
 
+	/*
+	 * The boot-wrapper must be entered from the reset vector at the
+	 * highest implemented exception level. The boot-wrapper only supports
+	 * being entered at the following exception levels:
+	 *
+	 * - EL3 (Secure)
+	 *   Entering at EL3 is strongly recommended.
+	 *   EL2 must be implemented.
+	 *
+	 * - EL2 (Non-secure)
+	 *   Entering at EL2 is partially supported.
+	 *   PSCI is not supported when entered in this exception level.
+	 */
 ASM_FUNC(_start)
 	cpuid	x0, x1
 	bl	find_logical_id
Signature · 7 lines
2.30.2


_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
Reviewed-by on v3 3/15
#14 lore raw

Re: [bootwrapper PATCH v3 03/15] Document entry requirements

On Tue, 25 Jan 2022 15:00:45 +0000
Mark Rutland <[email protected]> wrote:

Hi,

Quoted 32 lines

Currently the boot-wrapper only supports some combinations of exception levels, with other combinations not being supported.

While we generally expect the boot-wrapper to be entered at the highest implemented exception level, the AArch32 boot-wrapper has a comment implying it supports being entered with something else owning EL3. As this would require such EL3 firmware to always be in sync with the boot-wrapper's requirements, which change over time, we don't actually support such a configuration.

Some CPU state (such as CNTFRQ/CNTFRQ_EL0) needs to be initialized at the highest implemented exception level, but today the boot-wrapper only does so when entered at EL3 / Secure-PL1. Thus, today the only completely supported configurations are EL3 / Secure-PL1, and entering in other configurations is not entirely supported.

The aarch64 jump_kernel function always writes to SCTLR_EL2, which is UNDEFINED at EL1. Hence, the aarch64 boot-wrapper does not support being entered at EL1.

The aarch32 code assumes that any non-hyp mode is Secure PL1, and attempt to switch to monitor mode. If entered on a system without the security extensions, where the highest privileged mode is Non-secure PL1, this will not work. Hence the aarch32 boot-wrapper does not support being entered at Non-secure PL1.

Actually supporting all of these configurations requires restructuring much of the boot-wrapper. For now, document the supported configurations in each architecture's boot.S, and remove the misleading comment from arch/aarch32/boot.S. Subsequent patches will improve the support and add support for additional configurations.

Quoted 3 lines

There should be no functional change as a result of this patch.

I like that idea of documenting what we can sanely support, instead of trying to support everyone and their dog, when this won't work in practice anyway or cannot be well tested.

Thanks!
Andre

Quoted 61 lines
---
 arch/aarch32/boot.S | 14 ++++++++++++++
 arch/aarch32/psci.S |  5 -----
 arch/aarch64/boot.S | 13 +++++++++++++
 3 files changed, 27 insertions(+), 5 deletions(-)
arch/aarch32/boot.S +14 -0
index 4add338..00c432d 100644
@@ -16,6 +16,20 @@
 	.arch_extension virt
 
 	.section .init
+
+	/*
+	 * The boot-wrapper must be entered from the reset vector at the
+	 * highest implemented exception level. The boot-wrapper only supports
+	 * being entered in the following modes:
+	 *
+	 * - PL1 / EL3 (Secure) Supervisor mode
+	 *   Entering in this mode is strongly recommended.
+	 *   PL2 must be implemented.
+	 *
+	 * - PL2 / EL2 (Non-secure) Hypervisor mode
+	 *   Entering in this mode is partially supported.
+	 *   PSCI is not supported when entered in this mode.
+	 */
 ASM_FUNC(_start)
 	/* Stack initialisation */
 	cpuid	r0, r1
arch/aarch32/psci.S +0 -5
index dc7aeb7..e0d2972 100644
@@ -44,11 +44,6 @@ ASM_FUNC(start_el3)
 	/* pass through */
 
 ASM_FUNC(start_no_el3)
-	/*
-	 * For no-el3, we assume that firmware launched the boot-wrapper in
-	 * non-secure EL2 or EL1. We assume it has its own PSCI implementation
-	 * sitting at EL3, and that this path is only taken by primary CPU.
-	 */
 	cpuid	r0, r1
 	blx	find_logical_id
 	b	psci_first_spin
arch/aarch64/boot.S +6 -0
index ee9ddaa..d682ba5 100644
@@ -12,6 +12,19 @@
 
 	.section .init
 
+	/*
+	 * The boot-wrapper must be entered from the reset vector at the
+	 * highest implemented exception level. The boot-wrapper only supports
+	 * being entered at the following exception levels:
+	 *
+	 * - EL3 (Secure)

arch/aarch64/boot.S line 30

Quoted 10 lines
arch/aarch64/boot.S +7 -0
+	 *   Entering at EL3 is strongly recommended.
+	 *   EL2 must be implemented.
+	 *
+	 * - EL2 (Non-secure)
+	 *   Entering at EL2 is partially supported.
+	 *   PSCI is not supported when entered in this exception level.
+	 */
 ASM_FUNC(_start)
 	cpuid	x0, x1
 	bl	find_logical_id

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

#15 lore raw

PATCH v3 4/15 Add bit-field macros

In reply to Mark Rutland

Arm architectural documentation typically defines bit-fields as [msb,lsb] and single-bit fields as [bit]. For clarity it would be helpful if we could define fields in the same way.

Add helpers so that we can do so, along with helper to extract/insert bit-field values.

There should be no functional change as a result of this patch.

---
 include/bits.h | 66 ++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 66 insertions(+)
 create mode 100644 include/bits.h
include/bits.h new file +66 -0
new file mode 100644
index 0000000..5f0ad8f
@@ -0,0 +1,66 @@
+/*
+ * include/bits.h - helpers for bit-field definitions.
+ *
+ * Copyright (C) 2021 ARM Limited. All rights reserved.
+ *
+ * Use of this source code is governed by a BSD-style license that can be
+ * found in the LICENSE.txt file.
+ */
+#ifndef __BITS_H
+#define __BITS_H
+
+#ifdef __ASSEMBLY__
+#define UL(x)	x
+#define ULL(x)	x
+#else
+#define UL(x)	x##UL
+#define ULL(x)	x##ULL
+#endif
+
+/*
+ * Define a contiguous mask of bits with `msb` as the most significant bit and
+ * `lsb` as the least significant bit. The `msb` value must be greater than or
+ * equal to `lsb`.
+ *
+ * For example:
+ * - BITS(63, 63) is 0x8000000000000000
+ * - BITS(63, 0)  is 0xFFFFFFFFFFFFFFFF
+ * - BITS(0, 0)   is 0x0000000000000001
+ * - BITS(49, 17) is 0x0001FFFFFFFE0000
+ */
+#define BITS(msb, lsb) \
+	((~ULL(0) >> (63 - msb)) & (~ULL(0) << lsb))
+
+/*
+ * Define a mask of a single set bit `b`.
+ *
+ * For example:
+ * - BIT(63) is 0x8000000000000000
+ * - BIT(0)  is 0x0000000000000001
+ * - BIT(32) is 0x0000000100000000
+ */
+#define BIT(b)	BITS(b, b)
+
+/*
+ * Find the least significant set bit in the contiguous set of bits in `mask`.
+ *
+ * For example:
+ * - BITS_LSB(0x0000000000000001) is 0
+ * - BITS_LSB(0x000000000000ff00) is 8
+ * - BITS_LSB(0x8000000000000000) is 63
+ */
+#define BITS_LSB(mask)	(__builtin_ffsll(mask) - 1)
+
+/*
+ * Extract a bit-field out of `val` described by the contiguous set of bits in
+ * `mask`.
+ *
+ * For example:
+ * - BITS_EXTRACT(0xABCD, BITS(15, 12)) is 0xA
+ * - BITS_EXTRACT(0xABCD, BITS(11, 8))  is 0xB
+ * - BITS_EXTRACT(0xABCD, BIT(7))       is 0x1
+ */
+#define BITS_EXTRACT(val, mask) \
+	(((val) & (mask)) >> BITS_LSB(mask))
+
+#endif
Signature · 7 lines
2.30.2


_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
Reviewed-by on v3 4/15
#16 lore raw

Re: [bootwrapper PATCH v3 04/15] Add bit-field macros

Quoted 10 lines · On Tue, 25 Jan 2022 15:00:46 +0000 Mark Rutland <[email protected]> wrote:

Arm architectural documentation typically defines bit-fields as [msb,lsb] and single-bit fields as [bit]. For clarity it would be helpful if we could define fields in the same way.

Add helpers so that we can do so, along with helper to extract/insert bit-field values.

There should be no functional change as a result of this patch.

I extracted the code and checked the examples, see below.

Quoted 30 lines
---
 include/bits.h | 66 ++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 66 insertions(+)
 create mode 100644 include/bits.h
include/bits.h new file +19 -0
new file mode 100644
index 0000000..5f0ad8f
@@ -0,0 +1,66 @@
+/*
+ * include/bits.h - helpers for bit-field definitions.
+ *
+ * Copyright (C) 2021 ARM Limited. All rights reserved.
+ *
+ * Use of this source code is governed by a BSD-style license that can be
+ * found in the LICENSE.txt file.
+ */
+#ifndef __BITS_H
+#define __BITS_H
+
+#ifdef __ASSEMBLY__
+#define UL(x)	x
+#define ULL(x)	x
+#else
+#define UL(x)	x##UL
+#define ULL(x)	x##ULL
+#endif
+

include/bits.h line 29

Quoted 10 lines
include/bits.h new file +10 -0
+/*
+ * Define a contiguous mask of bits with `msb` as the most significant bit and
+ * `lsb` as the least significant bit. The `msb` value must be greater than or
+ * equal to `lsb`.
+ *
+ * For example:
+ * - BITS(63, 63) is 0x8000000000000000
+ * - BITS(63, 0)  is 0xFFFFFFFFFFFFFFFF
+ * - BITS(0, 0)   is 0x0000000000000001
+ * - BITS(49, 17) is 0x0001FFFFFFFE0000

This should be 0x0003fffffffe0000, in real life and with the below implementation. So just a comment fix.

The rest of the examples checked out, so with that fixed:

Cheers,
Andre

Quoted 27 lines
include/bits.h new file +27 -0
+ */
+#define BITS(msb, lsb) \
+	((~ULL(0) >> (63 - msb)) & (~ULL(0) << lsb))
+
+/*
+ * Define a mask of a single set bit `b`.
+ *
+ * For example:
+ * - BIT(63) is 0x8000000000000000
+ * - BIT(0)  is 0x0000000000000001
+ * - BIT(32) is 0x0000000100000000
+ */
+#define BIT(b)	BITS(b, b)
+
+/*
+ * Find the least significant set bit in the contiguous set of bits in `mask`.
+ *
+ * For example:
+ * - BITS_LSB(0x0000000000000001) is 0
+ * - BITS_LSB(0x000000000000ff00) is 8
+ * - BITS_LSB(0x8000000000000000) is 63
+ */
+#define BITS_LSB(mask)	(__builtin_ffsll(mask) - 1)
+
+/*
+ * Extract a bit-field out of `val` described by the contiguous set of bits in
+ * `mask`.

include/bits.h line 66

Quoted 10 lines
include/bits.h new file +10 -0
+ *
+ * For example:
+ * - BITS_EXTRACT(0xABCD, BITS(15, 12)) is 0xA
+ * - BITS_EXTRACT(0xABCD, BITS(11, 8))  is 0xB
+ * - BITS_EXTRACT(0xABCD, BIT(7))       is 0x1
+ */
+#define BITS_EXTRACT(val, mask) \
+	(((val) & (mask)) >> BITS_LSB(mask))
+
+#endif

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

#17 lore raw

Re: [bootwrapper PATCH v3 04/15] Add bit-field macros

Quoted 27 lines · Andre Przywara
Quoted 10 lines · On Tue, 25 Jan 2022 15:00:46 +0000 Mark Rutland <[email protected]> wrote:

Arm architectural documentation typically defines bit-fields as [msb,lsb] and single-bit fields as [bit]. For clarity it would be helpful if we could define fields in the same way.

Add helpers so that we can do so, along with helper to extract/insert bit-field values.

There should be no functional change as a result of this patch.

I extracted the code and checked the examples, see below.

Quoted 10 lines

+/*
+ * Define a contiguous mask of bits with msb as the most significant bit and
+ * lsb as the least significant bit. The msb value must be greater than or
+ * equal to lsb.
+ *
+ * For example:
+ * - BITS(63, 63) is 0x8000000000000000
+ * - BITS(63, 0) is 0xFFFFFFFFFFFFFFFF
+ * - BITS(0, 0) is 0x0000000000000001
+ * - BITS(49, 17) is 0x0001FFFFFFFE0000

Quoted 2 lines

This should be 0x0003fffffffe0000, in real life and with the below implementation. So just a comment fix.

Good spot; I'll fix that up.

Quoted 3 lines

The rest of the examples checked out, so with that fixed:

Thanks!

Mark.

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

#18 lore raw

PATCH v3 5/15 aarch64: add system register accessors

In reply to Mark Rutland

We open code the use of mrs/msr for specific registers, which is somewhat tedious. Add macros to do this generically, along with a helper to extract a specific register field. Existing C usage is converted to the new helpers, and register definitions moved to a common location.

There should be no functional change as a result of this patch.

---
 arch/aarch64/include/asm/cpu.h    | 41 ++++++++++++++++++++++---------
 arch/aarch64/include/asm/gic-v3.h | 10 +++-----
 2 files changed, 32 insertions(+), 19 deletions(-)
arch/aarch64/include/asm/cpu.h +29 -12
index 49d3f86..341a545 100644
@@ -9,10 +9,14 @@
 #ifndef __ASM_AARCH64_CPU_H
 #define __ASM_AARCH64_CPU_H
 
+#include <bits.h>
+
 #define MPIDR_ID_BITS		0xff00ffffff
 
 #define CURRENTEL_EL3		(3 << 2)
 
+#define ID_AA64PFR0_EL1_GIC	BITS(27, 24)
+
 /*
  * RES1 bits,  little-endian, caches and MMU off, no alignment checking,
  * no WXN.
@@ -29,6 +33,12 @@
 
 #define CPTR_EL3_EZ		(1 << 8)
 
+#define ICC_SRE_EL2		S3_4_C12_C9_5
+#define ICC_SRE_EL3		S3_6_C12_C12_5
+#define ICC_CTLR_EL1		S3_0_C12_C12_4
+#define ICC_CTLR_EL3		S3_6_C12_C12_4
+#define ICC_PMR_EL1		S3_0_C4_C6_0
+
 #define ZCR_EL3			s3_6_c1_c2_0
 #define ZCR_EL3_LEN_MAX		0xf
 
@@ -57,20 +67,27 @@
 
 #define sevl()		asm volatile ("sevl\n" : : : "memory")
 
-static inline unsigned long read_mpidr(void)
-{
-	unsigned long mpidr;
+#define __str(def)	#def
 
-	asm volatile ("mrs	%0, mpidr_el1\n" : "=r" (mpidr));
-	return mpidr & MPIDR_ID_BITS;
-}
+#define mrs(reg)							\
+({									\
+	unsigned long __mrs_val;					\
+	asm volatile("mrs %0, " __str(reg) : "=r" (__mrs_val));		\
+	__mrs_val;							\
+})
 
-static inline uint64_t read_id_aa64pfr0(void)
-{
-	uint64_t val;
+#define msr(reg, val)							\
+do {									\
+	unsigned long __msr_val = val;					\
+	asm volatile("msr " __str(reg) ", %0" : : "r" (__msr_val));	\
+} while (0)
+
+#define mrs_field(reg, field) \
+	BITS_EXTRACT(mrs(reg), (reg##_##field))
 
-	asm volatile ("mrs	%0, id_aa64pfr0_el1\n" : "=r" (val));
-	return val;
+static inline unsigned long read_mpidr(void)
+{
+	return mrs(mpidr_el1) & MPIDR_ID_BITS;
 }
 
 static inline void iciallu(void)
@@ -80,7 +97,7 @@ static inline void iciallu(void)
 
 static inline int has_gicv3_sysreg(void)
 {
-	return !!((read_id_aa64pfr0() >> 24) & 0xf);
+	return !!mrs_field(ID_AA64PFR0_EL1, GIC);
 }
 
 #endif /* !__ASSEMBLY__ */
arch/aarch64/include/asm/gic-v3.h +3 -7
index 5b32380..2447480 100644
@@ -9,20 +9,16 @@
 #ifndef __ASM_AARCH64_GICV3_H
 #define __ASM_AARCH64_GICV3_H
 
-#define ICC_SRE_EL2	"S3_4_C12_C9_5"
-#define ICC_SRE_EL3	"S3_6_C12_C12_5"
-#define ICC_CTLR_EL1	"S3_0_C12_C12_4"
-#define ICC_CTLR_EL3	"S3_6_C12_C12_4"
-#define ICC_PMR_EL1	"S3_0_C4_C6_0"
+#include <asm/cpu.h>
 
 static inline void gic_write_icc_sre(uint32_t val)
 {
-	asm volatile ("msr " ICC_SRE_EL3 ", %0" : : "r" (val));
+	msr(ICC_SRE_EL3, val);
 }
 
 static inline void gic_write_icc_ctlr(uint32_t val)
 {
-	asm volatile ("msr " ICC_CTLR_EL3 ", %0" : : "r" (val));
+	msr(ICC_CTLR_EL3, val);
 }
 
 #endif
Signature · 7 lines
2.30.2


_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
#19 lore raw

PATCH v3 6/15 aarch32: add coprocessor accessors

In reply to Mark Rutland

We open code the use of mrc/mcr for specific registers, which is somewhat tedious. Add macros to do this generically, along with a helper to extract a specific register field. Existing C usage is converted to the new helpers, and register definitions moved to a common location.

There should be no functional change as a result of this patch.

---
 arch/aarch32/include/asm/cpu.h    | 45 ++++++++++++++++++++-----------
 arch/aarch32/include/asm/gic-v3.h |  6 +++--
 2 files changed, 34 insertions(+), 17 deletions(-)
arch/aarch32/include/asm/cpu.h +30 -15
index 105cae5..d691c7b 100644
@@ -9,9 +9,13 @@
 #ifndef __ASM_AARCH32_CPU_H
 #define __ASM_AARCH32_CPU_H
 
+#include <bits.h>
+
 #define MPIDR_ID_BITS		0x00ffffff
 #define MPIDR_INVALID		(-1)
 
+#define ID_PFR1_GIC		BITS(31, 28)
+
 /* Only RES1 bits and CP15 barriers for the kernel */
 #define HSCTLR_KERNEL		(3 << 28 | 3 << 22 | 1 << 18 | 1 << 16 | 1 << 11 | 3 << 4)
 #define SCTLR_KERNEL		(3 << 22 | 1 << 11 | 1 << 5 | 3 << 4)
@@ -40,32 +44,43 @@
 #define sevl()		asm volatile ("sev" : : : "memory")
 #endif
 
-static inline unsigned long read_mpidr(void)
-{
-	unsigned long mpidr;
+#define MPIDR		"p15, 0, %0, c0, c0, 5"
+#define ID_PFR1		"p15, 0, %0, c0, c1, 1"
+#define ICIALLU		"p15, 0, %0, c7, c5, 0"
 
-	asm volatile ("mrc	p15, 0, %0, c0, c0, 5\n" : "=r" (mpidr));
-	return mpidr & MPIDR_ID_BITS;
-}
+#define ICC_SRE		"p15, 6, %0, c12, c12, 5"
+#define ICC_CTLR	"p15, 6, %0, c12, c12, 4"
 
-static inline uint32_t read_id_pfr1(void)
-{
-	uint32_t val;
+#define mrc(reg)						\
+({								\
+	unsigned long __mrc_val;				\
+	asm volatile("mrc " reg : "=r" (__mrc_val));		\
+	__mrc_val;						\
+})
 
-	asm volatile ("mrc	p15, 0, %0, c0, c1, 1\n" : "=r" (val));
-	return val;
+#define mcr(reg, val)						\
+do {								\
+	unsigned long __mcr_val = val;				\
+	asm volatile("mcr " reg : : "r" (__mcr_val));		\
+} while (0)
+
+
+#define mrc_field(reg, field) \
+	BITS_EXTRACT(mrc(reg), (reg##_##field))
+
+static inline unsigned long read_mpidr(void)
+{
+	return mrc(MPIDR) & MPIDR_ID_BITS;
 }
 
 static inline void iciallu(void)
 {
-	uint32_t val = 0;
-
-	asm volatile ("mcr	p15, 0, %0, c7, c5, 0" : : "r" (val));
+	mcr(ICIALLU, 0);
 }
 
 static inline int has_gicv3_sysreg(void)
 {
-	return !!((read_id_pfr1() >> 28) & 0xf);
+	return !!mrc_field(ID_PFR1, GIC);
 }
 
 #endif /* __ASSEMBLY__ */
arch/aarch32/include/asm/gic-v3.h +4 -2
index 65f38de..b28136a 100644
@@ -9,14 +9,16 @@
 #ifndef __ASM_AARCH32_GICV3_H
 #define __ASM_AARCH32_GICV3_H
 
+#include <asm/cpu.h>
+
 static inline void gic_write_icc_sre(uint32_t val)
 {
-	asm volatile ("mcr p15, 6, %0, c12, c12, 5" : : "r" (val));
+	mcr(ICC_SRE, val);
 }
 
 static inline void gic_write_icc_ctlr(uint32_t val)
 {
-	asm volatile ("mcr p15, 6, %0, c12, c12, 4" : : "r" (val));
+	mcr(ICC_CTLR, val);
 }
 
 #endif
Signature · 7 lines
2.30.2


_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
Reviewed-by on v3 6/15
#20 lore raw

Re: [bootwrapper PATCH v3 06/15] aarch32: add coprocessor accessors

On Tue, 25 Jan 2022 15:00:48 +0000
Mark Rutland <[email protected]> wrote:

Hi,

Quoted 8 lines

We open code the use of mrc/mcr for specific registers, which is somewhat tedious. Add macros to do this generically, along with a helper to extract a specific register field. Existing C usage is converted to the new helpers, and register definitions moved to a common location.

There should be no functional change as a result of this patch.

The bits look alright:

Cheers,
Andre

Quoted 95 lines
---
 arch/aarch32/include/asm/cpu.h    | 45 ++++++++++++++++++++-----------
 arch/aarch32/include/asm/gic-v3.h |  6 +++--
 2 files changed, 34 insertions(+), 17 deletions(-)
arch/aarch32/include/asm/cpu.h +30 -15
index 105cae5..d691c7b 100644
@@ -9,9 +9,13 @@
 #ifndef __ASM_AARCH32_CPU_H
 #define __ASM_AARCH32_CPU_H
 
+#include <bits.h>
+
 #define MPIDR_ID_BITS		0x00ffffff
 #define MPIDR_INVALID		(-1)
 
+#define ID_PFR1_GIC		BITS(31, 28)
+
 /* Only RES1 bits and CP15 barriers for the kernel */
 #define HSCTLR_KERNEL		(3 << 28 | 3 << 22 | 1 << 18 | 1 << 16 | 1 << 11 | 3 << 4)
 #define SCTLR_KERNEL		(3 << 22 | 1 << 11 | 1 << 5 | 3 << 4)
@@ -40,32 +44,43 @@
 #define sevl()		asm volatile ("sev" : : : "memory")
 #endif
 
-static inline unsigned long read_mpidr(void)
-{
-	unsigned long mpidr;
+#define MPIDR		"p15, 0, %0, c0, c0, 5"
+#define ID_PFR1		"p15, 0, %0, c0, c1, 1"
+#define ICIALLU		"p15, 0, %0, c7, c5, 0"
 
-	asm volatile ("mrc	p15, 0, %0, c0, c0, 5\n" : "=r" (mpidr));
-	return mpidr & MPIDR_ID_BITS;
-}
+#define ICC_SRE		"p15, 6, %0, c12, c12, 5"
+#define ICC_CTLR	"p15, 6, %0, c12, c12, 4"
 
-static inline uint32_t read_id_pfr1(void)
-{
-	uint32_t val;
+#define mrc(reg)						\
+({								\
+	unsigned long __mrc_val;				\
+	asm volatile("mrc " reg : "=r" (__mrc_val));		\
+	__mrc_val;						\
+})
 
-	asm volatile ("mrc	p15, 0, %0, c0, c1, 1\n" : "=r" (val));
-	return val;
+#define mcr(reg, val)						\
+do {								\
+	unsigned long __mcr_val = val;				\
+	asm volatile("mcr " reg : : "r" (__mcr_val));		\
+} while (0)
+
+
+#define mrc_field(reg, field) \
+	BITS_EXTRACT(mrc(reg), (reg##_##field))
+
+static inline unsigned long read_mpidr(void)
+{
+	return mrc(MPIDR) & MPIDR_ID_BITS;
 }
 
 static inline void iciallu(void)
 {
-	uint32_t val = 0;
-
-	asm volatile ("mcr	p15, 0, %0, c7, c5, 0" : : "r" (val));
+	mcr(ICIALLU, 0);
 }
 
 static inline int has_gicv3_sysreg(void)
 {
-	return !!((read_id_pfr1() >> 28) & 0xf);
+	return !!mrc_field(ID_PFR1, GIC);
 }
 
 #endif /* __ASSEMBLY__ */
arch/aarch32/include/asm/gic-v3.h +2 -1
index 65f38de..b28136a 100644
@@ -9,14 +9,16 @@
 #ifndef __ASM_AARCH32_GICV3_H
 #define __ASM_AARCH32_GICV3_H
 
+#include <asm/cpu.h>
+
 static inline void gic_write_icc_sre(uint32_t val)
 {
-	asm volatile ("mcr p15, 6, %0, c12, c12, 5" : : "r" (val));

arch/aarch32/include/asm/gic-v3.h line 24

Quoted 10 lines
arch/aarch32/include/asm/gic-v3.h +2 -1
+	mcr(ICC_SRE, val);
 }
 
 static inline void gic_write_icc_ctlr(uint32_t val)
 {
-	asm volatile ("mcr p15, 6, %0, c12, c12, 4" : : "r" (val));
+	mcr(ICC_CTLR, val);
 }
 
 #endif

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

#21 lore raw

PATCH v3 7/15 aarch64: add mov_64 macro

In reply to Mark Rutland

In subsequent patches we'll need to load 64-bit values into GPRs before the CPU is in a known endianness, where we cannot use literal pools.

In preparation for that, this patch adds a new mov_64 macro to load a 64-bit value into a GPR using a sequence of MOV and MOVKs, which will function the same regardless of the CPU's endianness.

At the same time, move the cpuid macro to use mov_64 internally.

---
 arch/aarch64/common.S | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)
arch/aarch64/common.S +9 -1
index c7171a9..3279fa9 100644
@@ -9,9 +9,17 @@
 
 #include <cpu.h>
 
+	/* Load a 64-bit value using immediates */
+	.macro	mov_64 dest, val
+	mov	\dest, #(((\val) >>  0) & 0xffff)
+	movk	\dest, #(((\val) >> 16) & 0xffff), lsl #16
+	movk	\dest, #(((\val) >> 32) & 0xffff), lsl #32
+	movk	\dest, #(((\val) >> 48) & 0xffff), lsl #48
+	.endm
+
 	/* Put MPIDR into \dest, clobber \tmp and flags */
 	.macro cpuid dest, tmp
 	mrs	\dest, mpidr_el1
-	ldr	\tmp, =MPIDR_ID_BITS
+	mov_64	\tmp, MPIDR_ID_BITS
 	ands	\dest, \dest, \tmp
 	.endm
Signature · 7 lines
2.30.2


_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
#22 lore raw

PATCH v3 8/15 aarch64: initialize SCTLR_ELx for the boot-wrapper

In reply to Mark Rutland

The SCTLR_ELx registers contain fields which are UNKNOWN or IMPLEMENTATION DEFINED out of reset. This includes SCTLR_ELx.EE, which defines the endianness of memory accesses (e.g. reads from literal pools). Due to this, portions of boot-wrapper code are not guaranteed to work correctly.

Rework the startup code to explicitly initialize SCTLR_ELx for the exception level the boot-wrapper was entered at. When entered at EL2 it's necessary to first initialise HCR_EL2.E2H as this affects the RESx behaviour of bits in SCTLR_EL2, and also aliases SCTLR_EL1 to SCTLR_EL2, which would break the initialization performed in jump_kernel.

As we plan to eventually support the highest implemented EL being any of EL3/EL2/EL1, code is added to handle all of these exception levels, even though we do not currently support starting at EL1.

We'll initialize other registers in subsequent patches.

---
 arch/aarch64/boot.S            | 74 +++++++++++++++++++++++++++-------
 arch/aarch64/include/asm/cpu.h | 30 +++++++++++++-
 2 files changed, 88 insertions(+), 16 deletions(-)
arch/aarch64/boot.S +59 -15
index d682ba5..204c199 100644
@@ -26,26 +26,26 @@
 	 *   PSCI is not supported when entered in this exception level.
 	 */
 ASM_FUNC(_start)
-	cpuid	x0, x1
-	bl	find_logical_id
-	cmp	x0, #MPIDR_INVALID
-	beq	err_invalid_id
-	bl	setup_stack
-
-	/*
-	 * EL3 initialisation
-	 */
 	mrs	x0, CurrentEL
 	cmp	x0, #CURRENTEL_EL3
-	b.eq	1f
+	b.eq	reset_at_el3
+	cmp	x0, #CURRENTEL_EL2
+	b.eq	reset_at_el2
+	cmp	x0, #CURRENTEL_EL1
+	b.eq	reset_at_el1
 
-	mov	w0, #1
-	ldr	x1, =flag_no_el3
-	str	w0, [x1]
+	/* Booting at EL0 is not supported */
+	b	.
 
-	b	start_no_el3
+	/*
+	 * EL3 initialisation
+	 */
+reset_at_el3:
+	mov_64	x0, SCTLR_EL3_RESET
+	msr	sctlr_el3, x0
+	isb
 
-1:	mov	x0, #0x30			// RES1
+	mov	x0, #0x30			// RES1
 	orr	x0, x0, #(1 << 0)		// Non-secure EL1
 	orr	x0, x0, #(1 << 8)		// HVC enable
 
@@ -143,10 +143,54 @@ ASM_FUNC(_start)
 	ldr	x0, =COUNTER_FREQ
 	msr	cntfrq_el0, x0
 
+	cpuid	x0, x1
+	bl	find_logical_id
+	cmp	x0, #MPIDR_INVALID
+	b.eq	err_invalid_id
+	bl	setup_stack
+
 	bl	gic_secure_init
 
 	b	start_el3
 
+	/*
+	 * EL2 initialization
+	 */
+reset_at_el2:
+	// Ensure E2H is not in use
+	mov_64	x0, HCR_EL2_RESET
+	msr	hcr_el2, x0
+	isb
+
+	mov_64	x0, SCTLR_EL2_RESET
+	msr	sctlr_el2, x0
+	isb
+
+	b	reset_no_el3
+
+	/*
+	 * EL1 initialization
+	 */
+reset_at_el1:
+	mov_64	x0, SCTLR_EL1_RESET
+	msr	sctlr_el1, x0
+	isb
+
+	b	reset_no_el3
+
+reset_no_el3:
+	cpuid	x0, x1
+	bl	find_logical_id
+	cmp	x0, #MPIDR_INVALID
+	b.eq	err_invalid_id
+	bl	setup_stack
+
+	mov	w0, #1
+	ldr	x1, =flag_no_el3
+	str	w0, [x1]
+
+	b	start_no_el3
+
 err_invalid_id:
 	b	.
 
arch/aarch64/include/asm/cpu.h +29 -1
index 341a545..0a4085b 100644
@@ -14,6 +14,35 @@
 #define MPIDR_ID_BITS		0xff00ffffff
 
 #define CURRENTEL_EL3		(3 << 2)
+#define CURRENTEL_EL2		(2 << 2)
+#define CURRENTEL_EL1		(1 << 2)
+
+/*
+ * RES1 bit definitions definitions as of ARM DDI 0487G.b
+ *
+ * These includes bits which are RES1 in some configurations.
+ */
+#define SCTLR_EL3_RES1							\
+	(BIT(29) | BIT(28) | BIT(23) | BIT(22) | BIT(18) | BIT(16) |	\
+	 BIT(11) | BIT(5) | BIT(4))
+
+#define SCTLR_EL2_RES1							\
+	(BIT(29) | BIT(28) | BIT(23) | BIT(22) | BIT(20) | BIT(18) |	\
+	 BIT(16) | BIT(11) | BIT(5) | BIT(4))
+
+#define SCTLR_EL1_RES1							\
+	(BIT(29) | BIT(28) | BIT(23) | BIT(22) | BIT(20) | BIT(11) |	\
+	 BIT(8) | BIT(7) | BIT(4))
+
+#define HCR_EL2_RES1		(BIT(1))
+
+/*
+ * Initial register values required for the boot-wrapper to run out-of-reset.
+ */
+#define SCTLR_EL3_RESET		SCTLR_EL3_RES1
+#define SCTLR_EL2_RESET		SCTLR_EL2_RES1
+#define SCTLR_EL1_RESET		SCTLR_EL1_RES1
+#define HCR_EL2_RESET		HCR_EL2_RES1
 
 #define ID_AA64PFR0_EL1_GIC	BITS(27, 24)
 
@@ -43,7 +72,6 @@
 #define ZCR_EL3_LEN_MAX		0xf
 
 #define SCTLR_EL1_CP15BEN	(1 << 5)
-#define SCTLR_EL1_RES1		(3 << 28 | 3 << 22 | 1 << 11)
 
 #ifdef KERNEL_32
 /*
Signature · 7 lines
2.30.2


_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
Reviewed-by on v3 8/15
#23 lore raw

Re: [bootwrapper PATCH v3 08/15] aarch64: initialize SCTLR_ELx for the boot-wrapper

On Tue, 25 Jan 2022 15:00:50 +0000
Mark Rutland <[email protected]> wrote:

Hi,

Quoted 12 lines

The SCTLR_ELx registers contain fields which are UNKNOWN or IMPLEMENTATION DEFINED out of reset. This includes SCTLR_ELx.EE, which defines the endianness of memory accesses (e.g. reads from literal pools). Due to this, portions of boot-wrapper code are not guaranteed to work correctly.

Rework the startup code to explicitly initialize SCTLR_ELx for the exception level the boot-wrapper was entered at. When entered at EL2 it's necessary to first initialise HCR_EL2.E2H as this affects the RESx behaviour of bits in SCTLR_EL2, and also aliases SCTLR_EL1 to SCTLR_EL2, which would break the initialization performed in jump_kernel.

Quoted 5 lines

As we plan to eventually support the highest implemented EL being any of EL3/EL2/EL1, code is added to handle all of these exception levels, even though we do not currently support starting at EL1.

We'll initialize other registers in subsequent patches.

While I still would like to have seen bit 7 cleared and bit 5 set in SCTLR_EL1, that is admittedly debatable, and any EL1 user should set that for their EL0 themselves. The rest is fine, so:

Quoted 1 line

Cheers,
Andre

Quoted 147 lines
---
 arch/aarch64/boot.S            | 74 +++++++++++++++++++++++++++-------
 arch/aarch64/include/asm/cpu.h | 30 +++++++++++++-
 2 files changed, 88 insertions(+), 16 deletions(-)
arch/aarch64/boot.S +59 -15
index d682ba5..204c199 100644
@@ -26,26 +26,26 @@
 	 *   PSCI is not supported when entered in this exception level.
 	 */
 ASM_FUNC(_start)
-	cpuid	x0, x1
-	bl	find_logical_id
-	cmp	x0, #MPIDR_INVALID
-	beq	err_invalid_id
-	bl	setup_stack
-
-	/*
-	 * EL3 initialisation
-	 */
 	mrs	x0, CurrentEL
 	cmp	x0, #CURRENTEL_EL3
-	b.eq	1f
+	b.eq	reset_at_el3
+	cmp	x0, #CURRENTEL_EL2
+	b.eq	reset_at_el2
+	cmp	x0, #CURRENTEL_EL1
+	b.eq	reset_at_el1
 
-	mov	w0, #1
-	ldr	x1, =flag_no_el3
-	str	w0, [x1]
+	/* Booting at EL0 is not supported */
+	b	.
 
-	b	start_no_el3
+	/*
+	 * EL3 initialisation
+	 */
+reset_at_el3:
+	mov_64	x0, SCTLR_EL3_RESET
+	msr	sctlr_el3, x0
+	isb
 
-1:	mov	x0, #0x30			// RES1
+	mov	x0, #0x30			// RES1
 	orr	x0, x0, #(1 << 0)		// Non-secure EL1
 	orr	x0, x0, #(1 << 8)		// HVC enable
 
@@ -143,10 +143,54 @@ ASM_FUNC(_start)
 	ldr	x0, =COUNTER_FREQ
 	msr	cntfrq_el0, x0
 
+	cpuid	x0, x1
+	bl	find_logical_id
+	cmp	x0, #MPIDR_INVALID
+	b.eq	err_invalid_id
+	bl	setup_stack
+
 	bl	gic_secure_init
 
 	b	start_el3
 
+	/*
+	 * EL2 initialization
+	 */
+reset_at_el2:
+	// Ensure E2H is not in use
+	mov_64	x0, HCR_EL2_RESET
+	msr	hcr_el2, x0
+	isb
+
+	mov_64	x0, SCTLR_EL2_RESET
+	msr	sctlr_el2, x0
+	isb
+
+	b	reset_no_el3
+
+	/*
+	 * EL1 initialization
+	 */
+reset_at_el1:
+	mov_64	x0, SCTLR_EL1_RESET
+	msr	sctlr_el1, x0
+	isb
+
+	b	reset_no_el3
+
+reset_no_el3:
+	cpuid	x0, x1
+	bl	find_logical_id
+	cmp	x0, #MPIDR_INVALID
+	b.eq	err_invalid_id
+	bl	setup_stack
+
+	mov	w0, #1
+	ldr	x1, =flag_no_el3
+	str	w0, [x1]
+
+	b	start_no_el3
+
 err_invalid_id:
 	b	.
 
arch/aarch64/include/asm/cpu.h +29 -0
index 341a545..0a4085b 100644
@@ -14,6 +14,35 @@
 #define MPIDR_ID_BITS		0xff00ffffff
 
 #define CURRENTEL_EL3		(3 << 2)
+#define CURRENTEL_EL2		(2 << 2)
+#define CURRENTEL_EL1		(1 << 2)
+
+/*
+ * RES1 bit definitions definitions as of ARM DDI 0487G.b
+ *
+ * These includes bits which are RES1 in some configurations.
+ */
+#define SCTLR_EL3_RES1							\
+	(BIT(29) | BIT(28) | BIT(23) | BIT(22) | BIT(18) | BIT(16) |	\
+	 BIT(11) | BIT(5) | BIT(4))
+
+#define SCTLR_EL2_RES1							\
+	(BIT(29) | BIT(28) | BIT(23) | BIT(22) | BIT(20) | BIT(18) |	\
+	 BIT(16) | BIT(11) | BIT(5) | BIT(4))
+
+#define SCTLR_EL1_RES1							\
+	(BIT(29) | BIT(28) | BIT(23) | BIT(22) | BIT(20) | BIT(11) |	\
+	 BIT(8) | BIT(7) | BIT(4))
+
+#define HCR_EL2_RES1		(BIT(1))
+
+/*
+ * Initial register values required for the boot-wrapper to run out-of-reset.
+ */
+#define SCTLR_EL3_RESET		SCTLR_EL3_RES1
+#define SCTLR_EL2_RESET		SCTLR_EL2_RES1
+#define SCTLR_EL1_RESET		SCTLR_EL1_RES1
+#define HCR_EL2_RESET		HCR_EL2_RES1
 
 #define ID_AA64PFR0_EL1_GIC	BITS(27, 24)
 

arch/aarch64/include/asm/cpu.h line 77

Quoted 7 lines
arch/aarch64/include/asm/cpu.h +0 -1
@@ -43,7 +72,6 @@
 #define ZCR_EL3_LEN_MAX		0xf
 
 #define SCTLR_EL1_CP15BEN	(1 << 5)
-#define SCTLR_EL1_RES1		(3 << 28 | 3 << 22 | 1 << 11)
 
 #ifdef KERNEL_32
 /*

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

#24 lore raw

PATCH v3 9/15 Rework common init C code

In reply to Mark Rutland

In init_platform() we initialize a UART and announce the presence of the bootwrapper to the world. We do this relatively late in the boot-flow, and prior to this will silently ignore errors (e.g. in gic_secure_init).

To make it possible to provide improved diagnostics, and to allow us to move more initialization into C, this patch reworks the init code to call a C function earlier, where we can announce the presence of the boot-wrapper and perform other initialization.

In subsequent patches this will be expanded with more CPU initialization.

---
 Makefile.am         |  2 +-
 arch/aarch32/boot.S |  5 +++++
 arch/aarch64/boot.S |  4 ++++
 common/boot.c       |  4 ----
 common/init.c       | 24 ++++++++++++++++++++++++
 common/platform.c   | 12 +++++++-----
 include/platform.h  | 17 +++++++++++++++++
 7 files changed, 58 insertions(+), 10 deletions(-)
 create mode 100644 common/init.c
 create mode 100644 include/platform.h
Makefile.am +1 -1
index 5731a19..91a6a02 100644
@@ -34,7 +34,7 @@ endif
 PSCI_CPU_OFF	:= 0x84000002
 
 COMMON_SRC	:= common/
-COMMON_OBJ	:= boot.o bakery_lock.o platform.o lib.o
+COMMON_OBJ	:= boot.o bakery_lock.o platform.o lib.o init.o
 
 ARCH_OBJ	:= boot.o stack.o utils.o
 
arch/aarch32/boot.S +5 -0
index 00c432d..ee073ea 100644
@@ -48,6 +48,9 @@ ASM_FUNC(_start)
 	mov	r0, #1
 	ldr	r1, =flag_no_el3
 	str	r0, [r1]
+
+	bl	cpu_init_bootwrapper
+
 	b	start_no_el3
 
 _switch_monitor:
@@ -71,6 +74,8 @@ _monitor:
 	ldr	r0, =COUNTER_FREQ
 	mcr	p15, 0, r0, c14, c0, 0		@ CNTFRQ
 
+	bl	cpu_init_bootwrapper
+
 	bl	gic_secure_init
 
 	/* Initialise boot method */
arch/aarch64/boot.S +4 -0
index 204c199..7872daf 100644
@@ -149,6 +149,8 @@ reset_at_el3:
 	b.eq	err_invalid_id
 	bl	setup_stack
 
+	bl	cpu_init_bootwrapper
+
 	bl	gic_secure_init
 
 	b	start_el3
@@ -189,6 +191,8 @@ reset_no_el3:
 	ldr	x1, =flag_no_el3
 	str	w0, [x1]
 
+	bl	cpu_init_bootwrapper
+
 	b	start_no_el3
 
 err_invalid_id:
common/boot.c +0 -4
index c74d34c..29d53a4 100644
@@ -12,8 +12,6 @@
 extern unsigned long entrypoint;
 extern unsigned long dtb;
 
-void init_platform(void);
-
 void __noreturn jump_kernel(unsigned long address,
 			    unsigned long a0,
 			    unsigned long a1,
@@ -62,8 +60,6 @@ void __noreturn first_spin(unsigned int cpu, unsigned long *mbox,
 			   unsigned long invalid)
 {
 	if (cpu == 0) {
-		init_platform();
-
 		*mbox = (unsigned long)&entrypoint;
 		sevl();
 		spin(mbox, invalid, 1);
common/init.c new file +24 -0
new file mode 100644
index 0000000..9c471c9
@@ -0,0 +1,24 @@
+/*
+ * init.c - common boot-wrapper initialization
+ *
+ * Copyright (C) 2021 ARM Limited. All rights reserved.
+ *
+ * Use of this source code is governed by a BSD-style license that can be
+ * found in the LICENSE.txt file.
+ */
+#include <cpu.h>
+#include <platform.h>
+
+static void announce_bootwrapper(void)
+{
+	print_string("Boot-wrapper v0.2\r\n\r\n");
+}
+
+void cpu_init_bootwrapper(void)
+{
+	if (this_cpu_logical_id() == 0) {
+		init_uart();
+		announce_bootwrapper();
+		init_platform();
+	}
+}
common/platform.c +7 -5
index d11f568..47bf547 100644
@@ -1,5 +1,5 @@
 /*
- * platform.c - code to initialise everything required when first booting.
+ * platform.c - Platform initialization and I/O.
  *
  * Copyright (C) 2015 ARM Limited. All rights reserved.
  *
@@ -7,6 +7,7 @@
  * found in the LICENSE.txt file.
  */
 
+#include <cpu.h>
 #include <stdint.h>
 
 #include <asm/io.h>
@@ -30,7 +31,7 @@
 #define V2M_SYS(reg)	((void *)SYSREGS_BASE + V2M_SYS_##reg)
 #endif
 
-static void print_string(const char *str)
+void print_string(const char *str)
 {
 	uint32_t flags;
 
@@ -47,7 +48,7 @@ static void print_string(const char *str)
 	}
 }
 
-void init_platform(void)
+void init_uart(void)
 {
 	/*
 	 * UART initialisation (38400 8N1)
@@ -58,9 +59,10 @@ void init_platform(void)
 	raw_writel(0x70,	PL011(UART_LCR_H));
 	/* Enable the UART, TXen and RXen */
 	raw_writel(0x301,	PL011(UARTCR));
+}
 
-	print_string("Boot-wrapper v0.2\r\n\r\n");
-
+void init_platform(void)
+{
 #ifdef SYSREGS_BASE
 	/*
 	 * CLCD output site MB
include/platform.h new file +17 -0
new file mode 100644
index 0000000..e5248e1
@@ -0,0 +1,17 @@
+/*
+ * include/platform.h - Platform initialization and I/O.
+ *
+ * Copyright (C) 2021 ARM Limited. All rights reserved.
+ *
+ * Use of this source code is governed by a BSD-style license that can be
+ * found in the LICENSE.txt file.
+ */
+#ifndef __PLATFORM_H
+#define __PLATFORM_H
+
+void print_string(const char *str);
+void init_uart(void);
+
+void init_platform(void);
+
+#endif /* __PLATFORM_H */
Signature · 7 lines
2.30.2


_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
#25 lore raw

PATCH v3 10/15 Announce boot-wrapper mode / exception level

In reply to Mark Rutland

When something goes wrong within the boot-wrapper, it can be very helpful to know where we started from. Add an arch_announce() function to log this early in the boot process. More information can be added here in future.

This is logged ot the serial console as:

| Boot-wrapper v0.2
| Entered at EL3

---
 Makefile.am                    |  2 +-
 arch/aarch32/include/asm/cpu.h |  9 +++++++++
 arch/aarch32/init.c            | 29 +++++++++++++++++++++++++++++
 arch/aarch64/init.c            | 19 +++++++++++++++++++
 common/init.c                  |  6 +++++-
 common/platform.c              | 24 ++++++++++++++----------
 include/platform.h             |  1 +
 7 files changed, 78 insertions(+), 12 deletions(-)
 create mode 100644 arch/aarch32/init.c
 create mode 100644 arch/aarch64/init.c
Makefile.am +1 -1
index 91a6a02..5a0ebd2 100644
@@ -36,7 +36,7 @@ PSCI_CPU_OFF	:= 0x84000002
 COMMON_SRC	:= common/
 COMMON_OBJ	:= boot.o bakery_lock.o platform.o lib.o init.o
 
-ARCH_OBJ	:= boot.o stack.o utils.o
+ARCH_OBJ	:= boot.o stack.o utils.o init.o
 
 if BOOTWRAPPER_32
 CPPFLAGS	+= -DBOOTWRAPPER_32
arch/aarch32/include/asm/cpu.h +9 -0
index d691c7b..aa72204 100644
@@ -44,6 +44,15 @@
 #define sevl()		asm volatile ("sev" : : : "memory")
 #endif
 
+static inline unsigned long read_cpsr(void)
+{
+	unsigned long cpsr;
+	asm volatile ("mrs      %0, cpsr\n" : "=r" (cpsr));
+	return cpsr;
+}
+
+#define read_cpsr_mode()       (read_cpsr() & PSR_MODE_MASK)
+
 #define MPIDR		"p15, 0, %0, c0, c0, 5"
 #define ID_PFR1		"p15, 0, %0, c0, c1, 1"
 #define ICIALLU		"p15, 0, %0, c7, c5, 0"
arch/aarch32/init.c new file +29 -0
new file mode 100644
index 0000000..785a428
@@ -0,0 +1,29 @@
+/*
+ * init.c - common boot-wrapper initialization
+ *
+ * Copyright (C) 2021 ARM Limited. All rights reserved.
+ *
+ * Use of this source code is governed by a BSD-style license that can be
+ * found in the LICENSE.txt file.
+ */
+#include <cpu.h>
+#include <platform.h>
+
+static const char *mode_string(void)
+{
+	switch (read_cpsr_mode()) {
+	case PSR_MON:
+		return "PL1";
+	case PSR_HYP:
+		return "PL2 (Non-secure)";
+	default:
+		return "<UNKNOWN MODE>";
+	}
+}
+
+void announce_arch(void)
+{
+	print_string("Entered at ");
+	print_string(mode_string());
+	print_string("\r\n");
+}
arch/aarch64/init.c new file +19 -0
new file mode 100644
index 0000000..82816e7
@@ -0,0 +1,19 @@
+/*
+ * init.c - common boot-wrapper initialization
+ *
+ * Copyright (C) 2021 ARM Limited. All rights reserved.
+ *
+ * Use of this source code is governed by a BSD-style license that can be
+ * found in the LICENSE.txt file.
+ */
+#include <cpu.h>
+#include <platform.h>
+
+void announce_arch(void)
+{
+	unsigned char el = mrs(CurrentEl) >> 2;
+
+	print_string("Entered at EL");
+	print_char('0' + el);
+	print_string("\r\n");
+}
common/init.c +5 -1
index 9c471c9..2600f73 100644
@@ -11,14 +11,18 @@
 
 static void announce_bootwrapper(void)
 {
-	print_string("Boot-wrapper v0.2\r\n\r\n");
+	print_string("Boot-wrapper v0.2\r\n");
 }
 
+void announce_arch(void);
+
 void cpu_init_bootwrapper(void)
 {
 	if (this_cpu_logical_id() == 0) {
 		init_uart();
 		announce_bootwrapper();
+		announce_arch();
+		print_string("\r\n");
 		init_platform();
 	}
 }
common/platform.c +14 -10
index 47bf547..80d0562 100644
@@ -31,21 +31,25 @@
 #define V2M_SYS(reg)	((void *)SYSREGS_BASE + V2M_SYS_##reg)
 #endif
 
-void print_string(const char *str)
+void print_char(char c)
 {
 	uint32_t flags;
 
-	while (*str) {
-		do
-			flags = raw_readl(PL011(UARTFR));
-		while (flags & PL011_UARTFR_FIFO_FULL);
+	do {
+		flags = raw_readl(PL011(UARTFR));
+	} while (flags & PL011_UARTFR_FIFO_FULL);
+
+	raw_writel(c, PL011(UARTDR));
 
-		raw_writel(*str++, PL011(UARTDR));
+	do {
+		flags = raw_readl(PL011(UARTFR));
+	} while (flags & PL011_UARTFR_BUSY);
+}
 
-		do
-			flags = raw_readl(PL011(UARTFR));
-		while (flags & PL011_UARTFR_BUSY);
-	}
+void print_string(const char *str)
+{
+	while (*str)
+		print_char(*str++);
 }
 
 void init_uart(void)
include/platform.h +1 -0
index e5248e1..237b481 100644
@@ -9,6 +9,7 @@
 #ifndef __PLATFORM_H
 #define __PLATFORM_H
 
+void print_char(char c);
 void print_string(const char *str);
 void init_uart(void);
 
Signature · 7 lines
2.30.2


_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
#26 lore raw

PATCH v3 11/15 aarch64: move the bulk of EL3 initialization to C

In reply to Mark Rutland

The majority of state that we initialize at EL3 is necessary for code at lower ELs to function, but isnt' necessary for the boot-wrapper itself. Given that, it would be better to write this in C where it can be written mode clearly, and where it will be possible to add logging/debug logic.

This patch migrates the AArch64 EL3 initialization to C.

There should be no functional change as a result of this patch.

---
 Makefile.am                    |   2 +-
 arch/aarch64/boot.S            | 100 +--------------------------------
 arch/aarch64/include/asm/cpu.h |  35 +++++++++++-
 arch/aarch64/init.c            |  72 ++++++++++++++++++++++++
 4 files changed, 109 insertions(+), 100 deletions(-)
Makefile.am +1 -1
index 5a0ebd2..a5e8e8b 100644
@@ -155,7 +155,7 @@ $(COMMON_SRC):
 %.o: %.S Makefile | $(ARCH_SRC)
 	$(CC) $(CPPFLAGS) -D__ASSEMBLY__ $(CFLAGS) $(DEFINES) -c -o $@ $<
 
-%.o: %.c Makefile | $(COMMON_SRC)
+%.o: %.c Makefile | $(ARCH_SRC) $(COMMON_SRC)
 	$(CC) $(CPPFLAGS) $(CFLAGS) $(DEFINES) -c -o $@ $<
 
 model.lds: $(LD_SCRIPT) Makefile
arch/aarch64/boot.S +2 -98

112 lines not on this page · raw message

arch/aarch64/include/asm/cpu.h +34 -1
index 0a4085b..1be2d54 100644
@@ -34,7 +34,40 @@
 	(BIT(29) | BIT(28) | BIT(23) | BIT(22) | BIT(20) | BIT(11) |	\
 	 BIT(8) | BIT(7) | BIT(4))
 
-#define HCR_EL2_RES1		(BIT(1))
+#define MDCR_EL3_NSPB_NS_NOTRAP			(UL(3) << 12)
+#define MDCR_EL3_NSTB_NS_NOTRAP			(UL(3) << 24)
+#define MDCR_EL3_SBRBE_NOTRAP_NOPROHIBIT	(UL(3) << 32)
+#define MDCR_EL3_ENPMSN				BIT(36)
+
+#define SCR_EL3_RES1			BITS(5, 4)
+#define SCR_EL3_NS			BIT(0)
+#define SCR_EL3_HCE			BIT(8)
+#define SCR_EL3_RW			BIT(10)
+#define SCR_EL3_APK			BIT(16)
+#define SCR_EL3_API			BIT(17)
+#define SCR_EL3_ATA			BIT(26)
+#define SCR_EL3_FGTEN			BIT(27)
+#define SCR_EL3_ECVEN			BIT(28)
+#define SCR_EL3_TME			BIT(34)
+
+#define HCR_EL2_RES1			BIT(1)
+
+#define ID_AA64DFR0_EL1_PMSVER		BITS(35, 32)
+#define ID_AA64DFR0_EL1_TRACEBUFFER	BITS(47, 44)
+#define ID_AA64DFR0_EL1_BRBE		BITS(55, 52)
+
+#define ID_AA64ISAR0_EL1_TME		BITS(27, 24)
+
+#define ID_AA64ISAR1_EL1_APA		BITS(7, 4)
+#define ID_AA64ISAR1_EL1_API		BITS(11, 8)
+#define ID_AA64ISAR1_EL1_GPA		BITS(27, 24)
+#define ID_AA64ISAR1_EL1_GPI		BITS(31, 28)
+
+#define ID_AA64MMFR0_EL1_FGT		BITS(59, 56)
+#define ID_AA64MMFR0_EL1_ECV		BITS(63, 60)
+
+#define ID_AA64PFR1_EL1_MTE		BITS(11, 8)
+#define ID_AA64PFR0_EL1_SVE		BITS(35, 32)
 
 /*
  * Initial register values required for the boot-wrapper to run out-of-reset.
arch/aarch64/init.c +72 -0

81 lines not on this page · raw message

Signature · 7 lines
2.30.2


_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
Reviewed-by on v3 11/15
#27 lore raw

Re: [bootwrapper PATCH v3 11/15] aarch64: move the bulk of EL3 initialization to C

Quoted 11 lines · On Tue, 25 Jan 2022 15:00:53 +0000 Mark Rutland <[email protected]> wrote:

The majority of state that we initialize at EL3 is necessary for code at lower ELs to function, but isnt' necessary for the boot-wrapper itself. Given that, it would be better to write this in C where it can be written mode clearly, and where it will be possible to add logging/debug logic.

This patch migrates the AArch64 EL3 initialization to C.

There should be no functional change as a result of this patch.

With the changes it looks good now, also checked for the BRBE addition.

Cheers,
Andre

Quoted 261 lines
---
 Makefile.am                    |   2 +-
 arch/aarch64/boot.S            | 100 +--------------------------------
 arch/aarch64/include/asm/cpu.h |  35 +++++++++++-
 arch/aarch64/init.c            |  72 ++++++++++++++++++++++++
 4 files changed, 109 insertions(+), 100 deletions(-)
Makefile.am +1 -1
index 5a0ebd2..a5e8e8b 100644
@@ -155,7 +155,7 @@ $(COMMON_SRC):
 %.o: %.S Makefile | $(ARCH_SRC)
 	$(CC) $(CPPFLAGS) -D__ASSEMBLY__ $(CFLAGS) $(DEFINES) -c -o $@ $<
 
-%.o: %.c Makefile | $(COMMON_SRC)
+%.o: %.c Makefile | $(ARCH_SRC) $(COMMON_SRC)
 	$(CC) $(CPPFLAGS) $(CFLAGS) $(DEFINES) -c -o $@ $<
 
 model.lds: $(LD_SCRIPT) Makefile
arch/aarch64/boot.S +2 -98
index 7872daf..c0ec518 100644
@@ -45,104 +45,6 @@ reset_at_el3:
 	msr	sctlr_el3, x0
 	isb
 
-	mov	x0, #0x30			// RES1
-	orr	x0, x0, #(1 << 0)		// Non-secure EL1
-	orr	x0, x0, #(1 << 8)		// HVC enable
-
-	/* Enable pointer authentication if present */
-	mrs	x1, id_aa64isar1_el1
-	/* We check for APA+API and GPA+GPI */
-	ldr	x2, =((0xff << 24) | (0xff << 4))
-	and	x1, x1, x2
-	cbz	x1, 1f
-
-	orr	x0, x0, #(1 << 16)		// AP key enable
-	orr	x0, x0, #(1 << 17)		// AP insn enable
-1:
-	/* Enable TME if present */
-	mrs	x1, id_aa64isar0_el1
-	ubfx	x1, x1, #24, #4
-	cbz	x1, 1f
-
-	orr	x0, x0, #(1 << 34)		// TME enable
-1:
-	/* Enable FGT if present */
-	mrs	x1, id_aa64mmfr0_el1
-	ubfx	x1, x1, #56, #4
-	cbz	x1, 1f
-
-	orr	x0, x0, #(1 << 27)		// FGT enable
-1:
-	/* Enable ECV2 if present (allows CNTPOFF_EL2) */
-	mrs	x1, id_aa64mmfr0_el1
-	ubfx	x1, x1, #60, #4
-	cmp	x1, #2
-	b.lt	1f
-
-	orr	x0, x0, #(1 << 28)		// ECV enable
-1:
-	/* Enable MTE if present */
-	mrs	x10, id_aa64pfr1_el1
-	ubfx	x10, x10, #8, #4
-	cmp	x10, #2
-	b.lt	1f
-
-	orr	x0, x0, #(1 << 26)		// ATA enable
-1:
-#ifndef KERNEL_32
-	orr	x0, x0, #(1 << 10)		// 64-bit EL2
-#endif
-	msr	scr_el3, x0
-
-	msr	cptr_el3, xzr			// Disable copro. traps to EL3
-
-	mov	x0, xzr
-	mrs	x1, id_aa64dfr0_el1
-	ubfx	x1, x1, #32, #4
-	cbz	x1, 1f
-
-	// Enable SPE for the non-secure world.
-	orr	x0, x0, #(0x3 << 12)
-
-	// Do not trap PMSNEVFR_EL1 if present
-	cmp	x1, #3
-	b.lt	1f
-	orr	x0, x0, #(1 << 36)
-
-1:	mrs	x1, id_aa64dfr0_el1
-	ubfx	x1, x1, #44, #4
-	cbz	x1, 1f
-
-	// Enable TRBE for the non-secure world.
-	ldr	x1, =(0x3 << 24)
-	orr	x0, x0, x1
-
-1:	mrs     x1, id_aa64dfr0_el1
-	ubfx    x1, x1, #52, #4
-	cbz     x1, 1f
-
-	// Enable BRBE for the non-secure world.
-	ldr     x1, =(0x3 << 32)
-	orr     x0, x0, x1
-
-1:	msr	mdcr_el3, x0			// Disable traps to EL3
-
-	mrs	x0, id_aa64pfr0_el1
-	ubfx	x0, x0, #32, #4			// SVE present?
-	cbz	x0, 1f				// Skip SVE init if not
-
-	mrs	x0, cptr_el3
-	orr	x0, x0, #CPTR_EL3_EZ		// enable SVE
-	msr	cptr_el3, x0
-	isb
-
-	mov	x0, #ZCR_EL3_LEN_MAX		// SVE: Enable full vector len
-	msr	ZCR_EL3, x0			// for EL2.
-
-1:
-	ldr	x0, =COUNTER_FREQ
-	msr	cntfrq_el0, x0
-
 	cpuid	x0, x1
 	bl	find_logical_id
 	cmp	x0, #MPIDR_INVALID
@@ -151,6 +53,8 @@ reset_at_el3:
 
 	bl	cpu_init_bootwrapper
 
+	bl	cpu_init_el3
+
 	bl	gic_secure_init
 
 	b	start_el3
arch/aarch64/include/asm/cpu.h +34 -1
index 0a4085b..1be2d54 100644
@@ -34,7 +34,40 @@
 	(BIT(29) | BIT(28) | BIT(23) | BIT(22) | BIT(20) | BIT(11) |	\
 	 BIT(8) | BIT(7) | BIT(4))
 
-#define HCR_EL2_RES1		(BIT(1))
+#define MDCR_EL3_NSPB_NS_NOTRAP			(UL(3) << 12)
+#define MDCR_EL3_NSTB_NS_NOTRAP			(UL(3) << 24)
+#define MDCR_EL3_SBRBE_NOTRAP_NOPROHIBIT	(UL(3) << 32)
+#define MDCR_EL3_ENPMSN				BIT(36)
+
+#define SCR_EL3_RES1			BITS(5, 4)
+#define SCR_EL3_NS			BIT(0)
+#define SCR_EL3_HCE			BIT(8)
+#define SCR_EL3_RW			BIT(10)
+#define SCR_EL3_APK			BIT(16)
+#define SCR_EL3_API			BIT(17)
+#define SCR_EL3_ATA			BIT(26)
+#define SCR_EL3_FGTEN			BIT(27)
+#define SCR_EL3_ECVEN			BIT(28)
+#define SCR_EL3_TME			BIT(34)
+
+#define HCR_EL2_RES1			BIT(1)
+
+#define ID_AA64DFR0_EL1_PMSVER		BITS(35, 32)
+#define ID_AA64DFR0_EL1_TRACEBUFFER	BITS(47, 44)
+#define ID_AA64DFR0_EL1_BRBE		BITS(55, 52)
+
+#define ID_AA64ISAR0_EL1_TME		BITS(27, 24)
+
+#define ID_AA64ISAR1_EL1_APA		BITS(7, 4)
+#define ID_AA64ISAR1_EL1_API		BITS(11, 8)
+#define ID_AA64ISAR1_EL1_GPA		BITS(27, 24)
+#define ID_AA64ISAR1_EL1_GPI		BITS(31, 28)
+
+#define ID_AA64MMFR0_EL1_FGT		BITS(59, 56)
+#define ID_AA64MMFR0_EL1_ECV		BITS(63, 60)
+
+#define ID_AA64PFR1_EL1_MTE		BITS(11, 8)
+#define ID_AA64PFR0_EL1_SVE		BITS(35, 32)
 
 /*
  * Initial register values required for the boot-wrapper to run out-of-reset.
arch/aarch64/init.c +62 -0
index 82816e7..74190be 100644
@@ -8,6 +8,7 @@
  */
 #include <cpu.h>
 #include <platform.h>
+#include <stdbool.h>
 
 void announce_arch(void)
 {
@@ -17,3 +18,74 @@ void announce_arch(void)
 	print_char('0' + el);
 	print_string("\r\n");
 }
+
+static inline bool kernel_is_32bit(void)
+{
+#ifdef KERNEL_32
+	return true;
+#else
+	return false;
+#endif
+}
+
+static inline bool cpu_has_pauth(void)
+{
+	const unsigned long id_pauth = ID_AA64ISAR1_EL1_APA |
+				       ID_AA64ISAR1_EL1_API |
+				       ID_AA64ISAR1_EL1_GPA |
+				       ID_AA64ISAR1_EL1_GPI;
+
+	return mrs(ID_AA64ISAR1_EL1) & id_pauth;
+}
+
+void cpu_init_el3(void)
+{
+	unsigned long scr = SCR_EL3_RES1 | SCR_EL3_NS | SCR_EL3_HCE;
+	unsigned long mdcr = 0;
+	unsigned long cptr = 0;
+
+	if (cpu_has_pauth())
+		scr |= SCR_EL3_APK | SCR_EL3_API;
+
+	if (mrs_field(ID_AA64ISAR0_EL1, TME))
+		scr |= SCR_EL3_TME;
+
+	if (mrs_field(ID_AA64MMFR0_EL1, FGT))
+		scr |= SCR_EL3_FGTEN;
+
+	if (mrs_field(ID_AA64MMFR0_EL1, ECV) >= 2)
+		scr |= SCR_EL3_ECVEN;
+
+	if (mrs_field(ID_AA64PFR1_EL1, MTE) >= 2)
+		scr |= SCR_EL3_ATA;
+
+	if (!kernel_is_32bit())
+		scr |= SCR_EL3_RW;
+
+	msr(SCR_EL3, scr);
+
+	msr(CPTR_EL3, cptr);
+
+	if (mrs_field(ID_AA64DFR0_EL1, PMSVER))
+		mdcr |= MDCR_EL3_NSPB_NS_NOTRAP;
+
+	if (mrs_field(ID_AA64DFR0_EL1, PMSVER) >= 3)
+		mdcr |= MDCR_EL3_ENPMSN;
+
+	if (mrs_field(ID_AA64DFR0_EL1, TRACEBUFFER))
+		mdcr |= MDCR_EL3_NSTB_NS_NOTRAP;
+
+	if (mrs_field(ID_AA64DFR0_EL1, BRBE))
+		mdcr |= MDCR_EL3_SBRBE_NOTRAP_NOPROHIBIT;
+
+	msr(MDCR_EL3, mdcr);

arch/aarch64/init.c void announce_arch(void) · line 91

Quoted 10 lines
arch/aarch64/init.c +10 -0
+
+	if (mrs_field(ID_AA64PFR0_EL1, SVE)) {
+		cptr |= CPTR_EL3_EZ;
+		msr(CPTR_EL3, cptr);
+		isb();
+		msr(ZCR_EL3, ZCR_EL3_LEN_MAX);
+	}
+
+	msr(CNTFRQ_EL0, COUNTER_FREQ);
+}

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

#28 lore raw

PATCH v3 12/15 aarch32: move the bulk of Secure PL1 initialization to C

In reply to Mark Rutland

The majority of state that we initialize at Secure PL1 is necessary for code at lower PLs to function, but isnt' necessary for the boot-wrapper itself. Given that, it would be better to write this in C where it can be written mode clearly, and where it will be possible to add logging/debug logic.

This patch migrates the AArch32 Secure PL1 initialization to C.

There should be no functional change as a result of this patch.

---
 arch/aarch32/boot.S            | 11 +----------
 arch/aarch32/include/asm/cpu.h |  9 +++++++++
 arch/aarch32/init.c            | 12 ++++++++++++
 3 files changed, 22 insertions(+), 10 deletions(-)
arch/aarch32/boot.S +1 -10
index ee073ea..820957b 100644
@@ -63,16 +63,7 @@ _monitor:
 	/* Move the stack to Monitor mode*/
 	mrs	sp, sp_svc
 
-	/* Setup secure registers and devices */
-	mov	r0, #1				@ Non-secure lower level
-	orr	r0, #(1 << 8)			@ HVC enable
-	mcr	p15, 0, r0, c1, c1, 0		@ SCR
-
-	mov	r0, #(1 << 10 | 1 << 11)	@ Enable NS access to CPACR
-	mcr	p15, 0, r0, c1, c1, 2		@ NSACR
-
-	ldr	r0, =COUNTER_FREQ
-	mcr	p15, 0, r0, c14, c0, 0		@ CNTFRQ
+	bl	cpu_init_secure_pl1
 
 	bl	cpu_init_bootwrapper
 
arch/aarch32/include/asm/cpu.h +9 -0
index aa72204..c1bce9a 100644
@@ -30,6 +30,11 @@
 #define PSR_I			(1 << 7)
 #define PSR_A			(1 << 8)
 
+#define SCR_NS			BIT(0)
+#define SCR_HCE			BIT(8)
+
+#define NSACR_CP10		BIT(10)
+#define NSACR_CP11		BIT(11)
 
 #define SPSR_KERNEL		(PSR_A | PSR_I | PSR_F | PSR_HYP)
 
@@ -55,11 +60,15 @@ static inline unsigned long read_cpsr(void)
 
 #define MPIDR		"p15, 0, %0, c0, c0, 5"
 #define ID_PFR1		"p15, 0, %0, c0, c1, 1"
+#define SCR		"p15, 0, %0, c1, c1, 0"
+#define NSACR		"p15, 0, %0, c1, c1, 2"
 #define ICIALLU		"p15, 0, %0, c7, c5, 0"
 
 #define ICC_SRE		"p15, 6, %0, c12, c12, 5"
 #define ICC_CTLR	"p15, 6, %0, c12, c12, 4"
 
+#define CNTFRQ		"p15, 0, %0, c14, c0, 0"
+
 #define mrc(reg)						\
 ({								\
 	unsigned long __mrc_val;				\
arch/aarch32/init.c +12 -0
index 785a428..7143c66 100644
@@ -27,3 +27,15 @@ void announce_arch(void)
 	print_string(mode_string());
 	print_string("\r\n");
 }
+
+void cpu_init_secure_pl1(void)
+{
+	unsigned long scr = SCR_NS | SCR_HCE;
+	unsigned long nsacr = NSACR_CP10 | NSACR_CP11;
+
+	mcr(SCR, scr);
+
+	mcr(NSACR, nsacr);
+
+	mcr(CNTFRQ, COUNTER_FREQ);
+}
Signature · 7 lines
2.30.2


_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
#29 lore raw

PATCH v3 13/15 Announce locations of memory objects

In reply to Mark Rutland

To make it easier to debug boot failures, log the location of memory objects at boot time.

This is logged to the serial console as:

| Boot-wrapper v0.2
| Entered at EL3
| Memory layout:
| [0000000080000000..0000000080001f90] => boot-wrapper
| [000000008000fff8..0000000080010000] => mbox
| [0000000080200000..00000000822af200] => kernel
| [0000000088000000..0000000088002857] => dtb
---
 common/init.c      | 27 +++++++++++++++++++++++++++
 common/platform.c  | 13 +++++++++++++
 include/platform.h |  2 ++
 model.lds.S        | 20 ++++++++++++++------
 4 files changed, 56 insertions(+), 6 deletions(-)
common/init.c +27 -0
index 2600f73..fc74b9e 100644
@@ -14,6 +14,32 @@ static void announce_bootwrapper(void)
 	print_string("Boot-wrapper v0.2\r\n");
 }
 
+#define announce_object(object, desc)				\
+do {								\
+	extern char object##__start[];				\
+	extern char object##__end[];				\
+	print_string("[");					\
+	print_ulong_hex((unsigned long)object##__start);	\
+	print_string("..");					\
+	print_ulong_hex((unsigned long)object##__end);		\
+	print_string("] => " desc "\r\n");			\
+} while (0)
+
+static void announce_objects(void)
+{
+	print_string("Memory layout:\r\n");
+	announce_object(text, "boot-wrapper");
+	announce_object(mbox, "mbox");
+	announce_object(kernel, "kernel");
+#ifdef XEN
+	announce_object(xen, "xen");
+#endif
+	announce_object(dtb, "dtb");
+#ifdef USE_INITRD
+	announce_object(filesystem, "initrd");
+#endif
+}
+
 void announce_arch(void);
 
 void cpu_init_bootwrapper(void)
@@ -22,6 +48,7 @@ void cpu_init_bootwrapper(void)
 		init_uart();
 		announce_bootwrapper();
 		announce_arch();
+		announce_objects();
 		print_string("\r\n");
 		init_platform();
 	}
common/platform.c +13 -0
index 80d0562..1607ee6 100644
@@ -52,6 +52,19 @@ void print_string(const char *str)
 		print_char(*str++);
 }
 
+#define HEX_CHARS_PER_LONG	(2 * sizeof(long))
+#define HEX_CHARS		"0123456789abcdef"
+
+void print_ulong_hex(unsigned long val)
+{
+	int i;
+
+	for (i = HEX_CHARS_PER_LONG - 1; i >= 0; i--) {
+		int v = (val >> (4 * i)) & 0xf;
+		print_char(HEX_CHARS[v]);
+	}
+}
+
 void init_uart(void)
 {
 	/*
include/platform.h +2 -0
index 237b481..c88e124 100644
@@ -11,6 +11,8 @@
 
 void print_char(char c);
 void print_string(const char *str);
+void print_ulong_hex(unsigned long val);
+
 void init_uart(void);
 
 void init_platform(void);
model.lds.S +14 -6
index d4e7e13..dacaa25 100644
@@ -35,46 +35,54 @@ SECTIONS
 	 * the boot section's *(.data)
 	 */
 	.kernel (PHYS_OFFSET + KERNEL_OFFSET): {
-		kernel = .;
+		kernel__start = .;
 		KERNEL
+		kernel__end = .;
 	}
 
 #ifdef XEN
 	.xen (PHYS_OFFSET + XEN_OFFSET): {
-		xen = .;
+		xen__start = .;
 		XEN
+		xen__end = .;
 	}
 
-	entrypoint = xen;
+	entrypoint = xen__start;
 #else
-	entrypoint = kernel;
+	entrypoint = kernel__start;
 #endif
 
 	.dtb (PHYS_OFFSET + FDT_OFFSET): {
+		dtb__start = .;
 		dtb = .;
 		./fdt.dtb
+		dtb__end = .;
 	}
 
 #ifdef USE_INITRD
 	.filesystem (PHYS_OFFSET + FS_OFFSET): {
-		filesystem = .;
+		filesystem__start = .;
 		FILESYSTEM
-		fs_size = . - filesystem;
+		filesystem__end = .;
 	}
 #endif
 
 	.boot PHYS_OFFSET: {
+		text__start = .;
 		*(.init)
 		*(.text*)
 		*(.data* .rodata* .bss* COMMON)
 		*(.vectors)
 		*(.stack)
 		PROVIDE(etext = .);
+		text__end = .;
 	}
 
 	.mbox (PHYS_OFFSET + MBOX_OFFSET): {
+		mbox__start = .;
 		mbox = .;
 		QUAD(0x0)
+		mbox__end = .;
 	}
 
 	ASSERT(etext <= (PHYS_OFFSET + TEXT_LIMIT), ".text overflow!")
Signature · 7 lines
2.30.2


_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
#30 lore raw

PATCH v3 14/15 Rework bootmethod initialization

In reply to Mark Rutland

We currently initialize the bootmethod late, in assembly code. This requires us to maintain the el3/no_el3 distintion late into the boot process, and means we cannot produce any helpful diagnostic when booted at an unexpected exception level.

Rework things so that we initialize the bootmethod early, with a warning when things are wrong. The el3/no_el3 distinction is now irrelevant to the bootmethod code, and can be removed in subsequent patches.

When a boot-wrapper configured for PSCI is entered at EL2, a warning is looged to the serial console as:

| Boot-wrapper v0.2
| Entered at EL2
| Memory layout:
| [0000000080000000..0000000080001f90] => boot-wrapper
| [000000008000fff8..0000000080010000] => mbox
| [0000000080200000..00000000822af200] => kernel
| [0000000088000000..0000000088002857] => dtb
|
| WARNING: PSCI could not be initialized. Boot may fail
---
 Makefile.am                    |  1 +
 arch/aarch32/include/asm/cpu.h |  1 +
 arch/aarch32/init.c            | 16 ++++++++++++++++
 arch/aarch32/psci.S            |  8 +-------
 arch/aarch32/utils.S           |  9 ---------
 arch/aarch64/init.c            | 15 +++++++++++++++
 arch/aarch64/psci.S            | 21 ++-------------------
 arch/aarch64/spin.S            |  3 +++
 arch/aarch64/utils.S           |  9 ---------
 common/init.c                  |  7 ++++++-
 common/psci.c                  | 20 +++++++++++++++++---
 include/boot.h                 |  7 +++++++
 12 files changed, 69 insertions(+), 48 deletions(-)
Makefile.am +1 -0
index a5e8e8b..40bc5d6 100644
@@ -48,6 +48,7 @@ ARCH_SRC	:= arch/aarch64/
 endif
 
 if PSCI
+DEFINES		+= -DPSCI
 ARCH_OBJ	+= psci.o
 COMMON_OBJ	+= psci.o
 PSCI_NODE	:= psci {				\
arch/aarch32/include/asm/cpu.h +1 -0
index c1bce9a..3426075 100644
@@ -63,6 +63,7 @@ static inline unsigned long read_cpsr(void)
 #define SCR		"p15, 0, %0, c1, c1, 0"
 #define NSACR		"p15, 0, %0, c1, c1, 2"
 #define ICIALLU		"p15, 0, %0, c7, c5, 0"
+#define MVBAR		"p15, 0, %0, c12, c0, 1"
 
 #define ICC_SRE		"p15, 6, %0, c12, c12, 5"
 #define ICC_CTLR	"p15, 6, %0, c12, c12, 4"
arch/aarch32/init.c +16 -0
index 7143c66..e25f0c7 100644
@@ -8,6 +8,7 @@
  */
 #include <cpu.h>
 #include <platform.h>
+#include <stdbool.h>
 
 static const char *mode_string(void)
 {
@@ -39,3 +40,18 @@ void cpu_init_secure_pl1(void)
 
 	mcr(CNTFRQ, COUNTER_FREQ);
 }
+
+#ifdef PSCI
+extern char psci_vectors[];
+
+bool cpu_init_psci_arch(void)
+{
+	if (read_cpsr_mode() != PSR_MON)
+		return false;
+
+	mcr(MVBAR, (unsigned long)psci_vectors);
+	isb();
+
+	return true;
+}
+#endif
arch/aarch32/psci.S +1 -7
index e0d2972..cdc36b0 100644
@@ -15,7 +15,7 @@
 
 	.section .vectors
 	.align 6
-smc_vectors:
+ASM_DATA(psci_vectors)
 	b	err_exception			@ Reset
 	b	err_exception			@ Undef
 	b	handle_smc			@ SMC
@@ -39,11 +39,5 @@ handle_smc:
 	movs	pc, lr
 
 ASM_FUNC(start_el3)
-	ldr	r0, =smc_vectors
-	blx	setup_vector
-	/* pass through */
-
 ASM_FUNC(start_no_el3)
-	cpuid	r0, r1
-	blx	find_logical_id
 	b	psci_first_spin
arch/aarch32/utils.S +0 -9
index 5809f48..58279aa 100644
@@ -35,12 +35,3 @@ ASM_FUNC(find_logical_id)
 	bx	lr
 3:	mov	r0, #MPIDR_INVALID
 	bx	lr
-
-/*
- * Setup EL3 vectors.
- * r0: vector address
- */
-ASM_FUNC(setup_vector)
-	mcr	p15, 0, r0, c12, c0, 1	@ MVBAR
-	isb
-	bx	lr
arch/aarch64/init.c +15 -0
index 74190be..6677f2b 100644
@@ -89,3 +89,18 @@ void cpu_init_el3(void)
 
 	msr(CNTFRQ_EL0, COUNTER_FREQ);
 }
+
+#ifdef PSCI
+extern char psci_vectors[];
+
+bool cpu_init_psci_arch(void)
+{
+	if (mrs(CurrentEL) != CURRENTEL_EL3)
+		return false;
+
+	msr(VBAR_EL3, (unsigned long)psci_vectors);
+	isb();
+
+	return true;
+}
+#endif
arch/aarch64/psci.S +2 -19
index 8bd224b..d6ca2eb 100644
@@ -19,7 +19,7 @@
 	.section .vectors, "w"
 
 	.align 11
-vector:
+ASM_DATA(psci_vectors)
 	// current EL, SP_EL0
 	ventry	err_exception	// synchronous
 	ventry	err_exception	// IRQ
@@ -80,22 +80,5 @@ smc_exit:
 	eret
 
 ASM_FUNC(start_el3)
-	ldr	x0, =vector
-	bl	setup_vector
-
-	/* only boot the primary cpu (entry 0 in the table) */
-	cpuid	x0, x1
-	bl	find_logical_id
-	b	psci_first_spin
-
-/*
- * This PSCI implementation requires EL3. Without EL3 we'll only boot the
- * primary cpu, all others will be trapped in an infinite loop.
- */
 ASM_FUNC(start_no_el3)
-	cpuid	x0, x1
-	bl	find_logical_id
-	cbz	x0, psci_first_spin
-spin_dead:
-	wfe
-	b	spin_dead
+	b	psci_first_spin
arch/aarch64/spin.S +3 -0
index 1ea1c0b..764c532 100644
@@ -12,6 +12,9 @@
 
 	.text
 
+ASM_FUNC(cpu_init_bootmethod)
+	ret
+
 ASM_FUNC(start_el3)
 ASM_FUNC(start_no_el3)
 	cpuid	x0, x1
arch/aarch64/utils.S +0 -9
index 85c7f8a..32393cc 100644
@@ -32,12 +32,3 @@ ASM_FUNC(find_logical_id)
 	ret
 3:	mov	x0, #MPIDR_INVALID
 	ret
-
-/*
- * Setup EL3 vectors
- * x0: vector address
- */
-ASM_FUNC(setup_vector)
-	msr	VBAR_EL3, x0
-	isb
-	ret
common/init.c +6 -1
index fc74b9e..3c05ac3 100644
@@ -6,6 +6,7 @@
  * Use of this source code is governed by a BSD-style license that can be
  * found in the LICENSE.txt file.
  */
+#include <boot.h>
 #include <cpu.h>
 #include <platform.h>
 
@@ -44,7 +45,9 @@ void announce_arch(void);
 
 void cpu_init_bootwrapper(void)
 {
-	if (this_cpu_logical_id() == 0) {
+	unsigned int cpu = this_cpu_logical_id();
+
+	if (cpu == 0) {
 		init_uart();
 		announce_bootwrapper();
 		announce_arch();
@@ -52,4 +55,6 @@ void cpu_init_bootwrapper(void)
 		print_string("\r\n");
 		init_platform();
 	}
+
+	cpu_init_bootmethod(cpu);
 }
common/psci.c +17 -3
index a0e8700..5ae4255 100644
@@ -12,6 +12,7 @@
 #include <bakery_lock.h>
 #include <boot.h>
 #include <cpu.h>
+#include <platform.h>
 #include <psci.h>
 
 #ifndef CPU_IDS
@@ -78,12 +79,25 @@ long psci_call(unsigned long fid, unsigned long arg1, unsigned long arg2)
 	}
 }
 
-void __noreturn psci_first_spin(unsigned int cpu)
+void __noreturn psci_first_spin(void)
 {
-	if (cpu == MPIDR_INVALID)
-		while (1);
+	unsigned int cpu = this_cpu_logical_id();
 
 	first_spin(cpu, branch_table + cpu, PSCI_ADDR_INVALID);
 
 	unreachable();
 }
+
+void cpu_init_bootmethod(unsigned int cpu)
+{
+	if (cpu_init_psci_arch())
+		return;
+
+	if (cpu == 0) {
+		print_string("WARNING: PSCI could not be initialized. Boot may fail\r\n\r\n");
+		return;
+	}
+
+	while (1)
+		wfe();
+}
include/boot.h +7 -0
index d75e013..459d1d5 100644
@@ -10,10 +10,17 @@
 #define __BOOT_H
 
 #include <compiler.h>
+#include <stdbool.h>
 
 void __noreturn spin(unsigned long *mbox, unsigned long invalid, int is_entry);
 
 void __noreturn first_spin(unsigned int cpu, unsigned long *mbox,
 			   unsigned long invalid_addr);
 
+void cpu_init_bootmethod(unsigned int cpu);
+
+#ifdef PSCI
+bool cpu_init_psci_arch(void);
+#endif
+
 #endif
Signature · 7 lines
2.30.2


_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
Reviewed-by on v3 14/15
#31 lore raw

Re: [bootwrapper PATCH v3 14/15] Rework bootmethod initialization

On Tue, 25 Jan 2022 15:00:56 +0000
Mark Rutland <[email protected]> wrote:

Hi Mark,

Quoted 22 lines

We currently initialize the bootmethod late, in assembly code. This requires us to maintain the el3/no_el3 distintion late into the boot process, and means we cannot produce any helpful diagnostic when booted at an unexpected exception level.

Rework things so that we initialize the bootmethod early, with a warning when things are wrong. The el3/no_el3 distinction is now irrelevant to the bootmethod code, and can be removed in subsequent patches.

When a boot-wrapper configured for PSCI is entered at EL2, a warning is looged to the serial console as:

| Boot-wrapper v0.2
| Entered at EL2
| Memory layout:
| [0000000080000000..0000000080001f90] => boot-wrapper
| [000000008000fff8..0000000080010000] => mbox
| [0000000080200000..00000000822af200] => kernel
| [0000000088000000..0000000088002857] => dtb
|
| WARNING: PSCI could not be initialized. Boot may fail
Quoted 1 line

Thanks for the changes, looks good now:

Cheers,
Andre

Quoted 286 lines
---
 Makefile.am                    |  1 +
 arch/aarch32/include/asm/cpu.h |  1 +
 arch/aarch32/init.c            | 16 ++++++++++++++++
 arch/aarch32/psci.S            |  8 +-------
 arch/aarch32/utils.S           |  9 ---------
 arch/aarch64/init.c            | 15 +++++++++++++++
 arch/aarch64/psci.S            | 21 ++-------------------
 arch/aarch64/spin.S            |  3 +++
 arch/aarch64/utils.S           |  9 ---------
 common/init.c                  |  7 ++++++-
 common/psci.c                  | 20 +++++++++++++++++---
 include/boot.h                 |  7 +++++++
 12 files changed, 69 insertions(+), 48 deletions(-)
Makefile.am +1 -0
index a5e8e8b..40bc5d6 100644
@@ -48,6 +48,7 @@ ARCH_SRC	:= arch/aarch64/
 endif
 
 if PSCI
+DEFINES		+= -DPSCI
 ARCH_OBJ	+= psci.o
 COMMON_OBJ	+= psci.o
 PSCI_NODE	:= psci {				\
arch/aarch32/include/asm/cpu.h +1 -0
index c1bce9a..3426075 100644
@@ -63,6 +63,7 @@ static inline unsigned long read_cpsr(void)
 #define SCR		"p15, 0, %0, c1, c1, 0"
 #define NSACR		"p15, 0, %0, c1, c1, 2"
 #define ICIALLU		"p15, 0, %0, c7, c5, 0"
+#define MVBAR		"p15, 0, %0, c12, c0, 1"
 
 #define ICC_SRE		"p15, 6, %0, c12, c12, 5"
 #define ICC_CTLR	"p15, 6, %0, c12, c12, 4"
arch/aarch32/init.c +16 -0
index 7143c66..e25f0c7 100644
@@ -8,6 +8,7 @@
  */
 #include <cpu.h>
 #include <platform.h>
+#include <stdbool.h>
 
 static const char *mode_string(void)
 {
@@ -39,3 +40,18 @@ void cpu_init_secure_pl1(void)
 
 	mcr(CNTFRQ, COUNTER_FREQ);
 }
+
+#ifdef PSCI
+extern char psci_vectors[];
+
+bool cpu_init_psci_arch(void)
+{
+	if (read_cpsr_mode() != PSR_MON)
+		return false;
+
+	mcr(MVBAR, (unsigned long)psci_vectors);
+	isb();
+
+	return true;
+}
+#endif
arch/aarch32/psci.S +1 -7
index e0d2972..cdc36b0 100644
@@ -15,7 +15,7 @@
 
 	.section .vectors
 	.align 6
-smc_vectors:
+ASM_DATA(psci_vectors)
 	b	err_exception			@ Reset
 	b	err_exception			@ Undef
 	b	handle_smc			@ SMC
@@ -39,11 +39,5 @@ handle_smc:
 	movs	pc, lr
 
 ASM_FUNC(start_el3)
-	ldr	r0, =smc_vectors
-	blx	setup_vector
-	/* pass through */
-
 ASM_FUNC(start_no_el3)
-	cpuid	r0, r1
-	blx	find_logical_id
 	b	psci_first_spin
arch/aarch32/utils.S +0 -9
index 5809f48..58279aa 100644
@@ -35,12 +35,3 @@ ASM_FUNC(find_logical_id)
 	bx	lr
 3:	mov	r0, #MPIDR_INVALID
 	bx	lr
-
-/*
- * Setup EL3 vectors.
- * r0: vector address
- */
-ASM_FUNC(setup_vector)
-	mcr	p15, 0, r0, c12, c0, 1	@ MVBAR
-	isb
-	bx	lr
arch/aarch64/init.c +15 -0
index 74190be..6677f2b 100644
@@ -89,3 +89,18 @@ void cpu_init_el3(void)
 
 	msr(CNTFRQ_EL0, COUNTER_FREQ);
 }
+
+#ifdef PSCI
+extern char psci_vectors[];
+
+bool cpu_init_psci_arch(void)
+{
+	if (mrs(CurrentEL) != CURRENTEL_EL3)
+		return false;
+
+	msr(VBAR_EL3, (unsigned long)psci_vectors);
+	isb();
+
+	return true;
+}
+#endif
arch/aarch64/psci.S +2 -19
index 8bd224b..d6ca2eb 100644
@@ -19,7 +19,7 @@
 	.section .vectors, "w"
 
 	.align 11
-vector:
+ASM_DATA(psci_vectors)
 	// current EL, SP_EL0
 	ventry	err_exception	// synchronous
 	ventry	err_exception	// IRQ
@@ -80,22 +80,5 @@ smc_exit:
 	eret
 
 ASM_FUNC(start_el3)
-	ldr	x0, =vector
-	bl	setup_vector
-
-	/* only boot the primary cpu (entry 0 in the table) */
-	cpuid	x0, x1
-	bl	find_logical_id
-	b	psci_first_spin
-
-/*
- * This PSCI implementation requires EL3. Without EL3 we'll only boot the
- * primary cpu, all others will be trapped in an infinite loop.
- */
 ASM_FUNC(start_no_el3)
-	cpuid	x0, x1
-	bl	find_logical_id
-	cbz	x0, psci_first_spin
-spin_dead:
-	wfe
-	b	spin_dead
+	b	psci_first_spin
arch/aarch64/spin.S +3 -0
index 1ea1c0b..764c532 100644
@@ -12,6 +12,9 @@
 
 	.text
 
+ASM_FUNC(cpu_init_bootmethod)
+	ret
+
 ASM_FUNC(start_el3)
 ASM_FUNC(start_no_el3)
 	cpuid	x0, x1
arch/aarch64/utils.S +0 -9
index 85c7f8a..32393cc 100644
@@ -32,12 +32,3 @@ ASM_FUNC(find_logical_id)
 	ret
 3:	mov	x0, #MPIDR_INVALID
 	ret
-
-/*
- * Setup EL3 vectors
- * x0: vector address
- */
-ASM_FUNC(setup_vector)
-	msr	VBAR_EL3, x0
-	isb
-	ret
common/init.c +6 -1
index fc74b9e..3c05ac3 100644
@@ -6,6 +6,7 @@
  * Use of this source code is governed by a BSD-style license that can be
  * found in the LICENSE.txt file.
  */
+#include <boot.h>
 #include <cpu.h>
 #include <platform.h>
 
@@ -44,7 +45,9 @@ void announce_arch(void);
 
 void cpu_init_bootwrapper(void)
 {
-	if (this_cpu_logical_id() == 0) {
+	unsigned int cpu = this_cpu_logical_id();
+
+	if (cpu == 0) {
 		init_uart();
 		announce_bootwrapper();
 		announce_arch();
@@ -52,4 +55,6 @@ void cpu_init_bootwrapper(void)
 		print_string("\r\n");
 		init_platform();
 	}
+
+	cpu_init_bootmethod(cpu);
 }
common/psci.c +17 -3
index a0e8700..5ae4255 100644
@@ -12,6 +12,7 @@
 #include <bakery_lock.h>
 #include <boot.h>
 #include <cpu.h>
+#include <platform.h>
 #include <psci.h>
 
 #ifndef CPU_IDS
@@ -78,12 +79,25 @@ long psci_call(unsigned long fid, unsigned long arg1, unsigned long arg2)
 	}
 }
 
-void __noreturn psci_first_spin(unsigned int cpu)
+void __noreturn psci_first_spin(void)
 {
-	if (cpu == MPIDR_INVALID)
-		while (1);
+	unsigned int cpu = this_cpu_logical_id();
 
 	first_spin(cpu, branch_table + cpu, PSCI_ADDR_INVALID);
 
 	unreachable();
 }
+
+void cpu_init_bootmethod(unsigned int cpu)
+{
+	if (cpu_init_psci_arch())
+		return;
+
+	if (cpu == 0) {
+		print_string("WARNING: PSCI could not be initialized. Boot may fail\r\n\r\n");
+		return;
+	}
+
+	while (1)
+		wfe();
+}
include/boot.h +1 -0
index d75e013..459d1d5 100644
@@ -10,10 +10,17 @@
 #define __BOOT_H
 
 #include <compiler.h>
+#include <stdbool.h>
 
 void __noreturn spin(unsigned long *mbox, unsigned long invalid, int is_entry);
 

include/boot.h line 26

Quoted 10 lines
include/boot.h +6 -0
 void __noreturn first_spin(unsigned int cpu, unsigned long *mbox,
 			   unsigned long invalid_addr);
 
+void cpu_init_bootmethod(unsigned int cpu);
+
+#ifdef PSCI
+bool cpu_init_psci_arch(void);
+#endif
+
 #endif

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

#32 lore raw

PATCH v3 15/15 Unify start_el3 & start_no_el3

In reply to Mark Rutland

Now that the start_el3 and start_no_el3 labels point at the same place, unify them into a start_bootmethod label and update callers.

There should be no functional change as a result of this patch.

---
 arch/aarch32/boot.S | 5 ++---
 arch/aarch32/psci.S | 3 +--
 arch/aarch64/boot.S | 4 ++--
 arch/aarch64/psci.S | 3 +--
 arch/aarch64/spin.S | 3 +--
 5 files changed, 7 insertions(+), 11 deletions(-)
arch/aarch32/boot.S +2 -3
index 820957b..4d16c9c 100644
@@ -51,7 +51,7 @@ ASM_FUNC(_start)
 
 	bl	cpu_init_bootwrapper
 
-	b	start_no_el3
+	b	start_bootmethod
 
 _switch_monitor:
 	adr	lr, _monitor
@@ -69,8 +69,7 @@ _monitor:
 
 	bl	gic_secure_init
 
-	/* Initialise boot method */
-	b	start_el3
+	b	start_bootmethod
 
 err_invalid_id:
 	b	.
arch/aarch32/psci.S +1 -2
index cdc36b0..6613b6f 100644
@@ -38,6 +38,5 @@ handle_smc:
 	pop	{r4 - r12, lr}
 	movs	pc, lr
 
-ASM_FUNC(start_el3)
-ASM_FUNC(start_no_el3)
+ASM_FUNC(start_bootmethod)
 	b	psci_first_spin
arch/aarch64/boot.S +2 -2
index c0ec518..da5fa65 100644
@@ -57,7 +57,7 @@ reset_at_el3:
 
 	bl	gic_secure_init
 
-	b	start_el3
+	b	start_bootmethod
 
 	/*
 	 * EL2 initialization
@@ -97,7 +97,7 @@ reset_no_el3:
 
 	bl	cpu_init_bootwrapper
 
-	b	start_no_el3
+	b	start_bootmethod
 
 err_invalid_id:
 	b	.
arch/aarch64/psci.S +1 -2
index d6ca2eb..9709dbb 100644
@@ -79,6 +79,5 @@ smc_exit:
 	ldp	x18, x19, [sp], #16
 	eret
 
-ASM_FUNC(start_el3)
-ASM_FUNC(start_no_el3)
+ASM_FUNC(start_bootmethod)
 	b	psci_first_spin
arch/aarch64/spin.S +1 -2
index 764c532..375f732 100644
@@ -15,8 +15,7 @@
 ASM_FUNC(cpu_init_bootmethod)
 	ret
 
-ASM_FUNC(start_el3)
-ASM_FUNC(start_no_el3)
+ASM_FUNC(start_bootmethod)
 	cpuid	x0, x1
 	bl	find_logical_id
 
Signature · 7 lines
2.30.2


_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
#33 lore raw
In reply to Mark Rutland
Quoted 1 line · Mark Rutland

If there are no objections I'll apply this by the end of this week.

I've pushed this out to master branch now (with the comment fix Andre suggested). As before, the repo is:

Quoted 2 lines

https://git.kernel.org/pub/scm/linux/kernel/git/mark/boot-wrapper-aarch64.git/
git://git.kernel.org/pub/scm/linux/kernel/git/mark/boot-wrapper-aarch64.git

Thanks,
Mark.

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel