Skip to contentSkip to conversation

mips: Prepare MIPS-arch code for Baikal-T1 SoC support

PATCH Superseded [email protected] 11 participants 89 messages Mar 6, 2020 – May 20, 2020

Subsystems (8)

v1 of 5 Superseded Latest is v5 →

lore
Share & tools
lore
#0 lore raw

cover mips: Prepare MIPS-arch code for Baikal-T1 SoC support

From: Serge Semin <[email protected]>

This is a first patchset of a series of about 25 ones, which are intended to add the full Baikal-T1 SoC [1] support to the Linux kernel. Since they will concern various kernel subsystems, I decided to split the whole work up into the patchesets in accordance with the subsystems/devices their changes are introduced for. Nearly 2/3 of the work is already done and will be sent out very soon. While the rest of the changes specifically related to the fast-speed interfaces (DW 12G PHY, PCIe, SATA, xGBE, GMAC, USB, DDRC, IC) are still in refactoring and preparation for integration into the mainline kernel. Hopefully I'll finish them up in the next two-three months, and submit them straight away.

Getting back to this patchset. As the subject states this is intended to prepare the MIPS-arch and generic kernel code for further Baikal-T1 SoC platform support integration (note the Baikal-T1 SoC platform code will be submitted last after the whole series of patchsets as a closure of the submission process). First of all the patchset starts with a set of changes to the dt-bindings kernel concerning MIPS CPC and CDMM nodes as being described by the trivial device bindings. In addition we updated the vendors prefix schema with Baikal Electronics JSC prefix so being further committed device drivers would be correctly accepted by the checkpatch-script.

Then I found out that dtc-compiler integrated into the kernel doesn't support all the possible values of the 'reg'-property used to define the i2c-slave devices address. It may have 10-bits and own-slave flags, so the compiler has to deal with them. Even though the patch might be better to be integrated into the mainline dtc repo I decided to send it to the kernel first.

While I was working with the MIPS architecture code, I discovered, that there is a bug in the Coherency Manager v2 error causes declaration and the errors handler lacked of CM2 L2 ECC/parity errors support. So the fixes are here in the patchset.

Baikal-T1 SoC is based on the MIPS P5600 Warrior IP-core, which itself has MIPS32 Release 5 architecture. Even though on ISA level it doesn't differ much from the MIPS32 Release 2 release, there are still some peculiarities, which make it's justified to add the direct MIPS32r5 support into the kernel (see the specific patch for details). In addition seeing there is more than one real chip based on the MIPS P5600 core on the market, it would be good to have the direct P5600 CPU config in the MIPS-arch.

There were some issues we discovered while were working with MIPS-arch code. So the cleanups and fixes are introduced in this patchset. First of all the Write-Merge CPU feature hasn't been handled in a generic way. Even if a platform defined the writecombine flag as _CACHE_UNCACHED_ACCELERATED, the feature might have been disabled in the CP0 register. We either enable it or leave it as is in accordance with the knowledge of whether the corresponding platform really supports it. Secondly Memory Accessibility Attribute Registers (MAAR) haven't been properly initialized when Extended Physical Address (XPA) mode was enabled. Thirdly since some of the platforms may have a very strict limitations on the IO-memory access instructions. For instance Baikal-T1 SoC IO-memory can be accessed by the lw/sw instructions only. In this case for early-printk and CPS-debug code we suggest to use the instructions in accordance with the UART-registers offset (lb/sb if offset = 0, lh/sh if offset = 1 and so on). Fourthly in case if CPUFREQ feature is enabled and frequency of the CPU is changed by the reference clock alteration, we must make sure that MIPS r4k timer related services are properly updated when CPU-frequency changes. It concerns udelay lpj adjustment, MIPS timer clockevent frequency update. In addition when CPU reference frequency changes it isn't recommended to use the timer as clocksource at all, since currently the subsystem isn't tolerant to the unstable clock sources. So in this case we suggest to use the r4k timer for clocksourcing only as a last resort. Fifthly we discovered a bug in a method of CPUFREQ boost feature enable procedure and fixed it in one of the patches within this patchset. And finally there are a few fixups/cleanups we suggest to integrate into the MIPS FDC and CDMM related code (see the patches for details).

This patchset is rebased and tested on the mainline Linux kernel 5.6-rc4:
commit 98d54f81e36b ("Linux 5.6-rc4").

[1] http://www.baikalelectronics.com/products/168/

P.S. Sorry for the previous emails burst. I forgot to start the series with this cover-letter and the corporate smtp broke the transmission in the middle anyway. Please don't pay attention to them. Here is the proper emails resubmission.

Cc Maxim Kaurkin <[email protected]>
Cc Pavel Parkhomenko <[email protected]>
Cc Ramil Zaripov <[email protected]>
Cc Ekaterina Skachko <[email protected]>
Cc Vadim Vlasov <[email protected]>
Cc Thomas Bogendoerfer <[email protected]>
Cc Paul Burton <[email protected]>
Cc Ralf Baechle <[email protected]>
Cc Rob Herring <[email protected]>
Cc Frank Rowand <[email protected]>
cc Tony Lindgren <[email protected]>
Cc Arnd Bergmann <[email protected]>
Cc Greg Kroah-Hartman <[email protected]>
Cc Jiri Slaby <[email protected]>

Serge Semin (22):
dt-bindings: Permit platform devices in the trivial-devices bindings
dt-bindings: Add MIPS CPC controller as a trivial devices
dt-bindings: Add MIPS CDMM controller as a trivial device
dt-bindings: Add vendor prefix for Baikal Electronics, JSC
mips: cm: Fix an invalid error code of INTVN_*_ERR
mips: cm: Add L2 ECC/parity errors reporting
mips: Add MIPS32 Release 5 support
mips: Add MIPS Warrior P5600 support
mips: Fix cpu_has_mips64r1/2 activation for MIPS32 CPUs
mips: Add CP0 Write Merge config support
mips: Add CONFIG/CONFIG6 reg fields macro
mips: MAAR: Use more precise address mask
mips: MAAR: Add XPA mode support
mips: early_printk_8250: Use offset-sized IO-mem accessors
mips: Use offset-sized IO-mem accessors in CPS debug printout
mips: cdmm: Add mti,mips-cdmm dtb node support
bus: cdmm: Add MIPS R5 arch support
tty: mips_ejtag_fdc: Mark expected switch fall-through
mips: Add udelay lpj numbers adjustment
mips: csrc-r4k: Decrease r4k-clocksource rating if CPU_FREQ enabled
mips: cevt-r4k: Update the r4k-clockevent frequency in sync with CPU
cpufreq: Return zero on success in boost sw setting

 .../bindings/power/mti,mips-cpc.txt           |  8 ---
 .../devicetree/bindings/trivial-devices.yaml  | 12 ++--
 .../devicetree/bindings/vendor-prefixes.yaml  |  2 +
 arch/mips/Kconfig                             | 62 ++++++++++++++--
 arch/mips/Makefile                            |  2 +
 arch/mips/include/asm/asmmacro.h              | 18 ++---
 arch/mips/include/asm/compiler.h              |  5 ++
 arch/mips/include/asm/cpu-features.h          | 34 ++++++---
 arch/mips/include/asm/cpu-info.h              |  2 +-
 arch/mips/include/asm/cpu-type.h              |  6 +-
 arch/mips/include/asm/cpu.h                   | 11 +--
 arch/mips/include/asm/fpu.h                   |  4 +-
 arch/mips/include/asm/hazards.h               |  8 ++-
 arch/mips/include/asm/maar.h                  | 17 ++++-
 arch/mips/include/asm/mipsregs.h              | 33 ++++++++-
 arch/mips/include/asm/module.h                |  4 ++
 arch/mips/include/asm/stackframe.h            |  2 +-
 arch/mips/include/asm/switch_to.h             |  8 +--
 arch/mips/kernel/cevt-r4k.c                   | 44 ++++++++++++
 arch/mips/kernel/cps-vec-ns16550.S            | 18 ++++-
 arch/mips/kernel/cpu-probe.c                  | 60 ++++++++++++++++
 arch/mips/kernel/csrc-r4k.c                   |  4 ++
 arch/mips/kernel/early_printk_8250.c          | 34 ++++++++-
 arch/mips/kernel/entry.S                      |  6 +-
 arch/mips/kernel/mips-cm.c                    | 66 +++++++++++++++--
 arch/mips/kernel/proc.c                       |  2 +
 arch/mips/kernel/r4k_fpu.S                    | 14 ++--
 arch/mips/kernel/spram.c                      |  4 +-
 arch/mips/kernel/time.c                       | 70 +++++++++++++++++++
 arch/mips/kvm/vz.c                            |  6 +-
 arch/mips/lib/csum_partial.S                  |  6 +-
 arch/mips/mm/c-r4k.c                          |  7 +-
 arch/mips/mm/init.c                           |  8 ++-
 arch/mips/mm/sc-mips.c                        |  7 +-
 drivers/bus/Kconfig                           |  2 +-
 drivers/bus/mips_cdmm.c                       | 15 ++++
 drivers/cpufreq/cpufreq.c                     |  2 +-
 drivers/tty/mips_ejtag_fdc.c                  |  1 +
 38 files changed, 529 insertions(+), 85 deletions(-)
 delete mode 100644 Documentation/devicetree/bindings/power/mti,mips-cpc.txt
Signature · 1 line
2.25.1
#1 lore raw

v2 cover mips: Prepare MIPS-arch code for Baikal-T1 SoC support

From: Serge Semin <[email protected]>

This is a first patchset of a series of about 25 ones, which are intended to add the full Baikal-T1 SoC [1] support to the Linux kernel. Since they will concern various kernel subsystems, I decided to split the whole work up into the patchesets in accordance with the subsystems/devices their changes are introduced for. Nearly 2/3 of the work is already done and will be sent out very soon. While the rest of the changes specifically related to the fast-speed interfaces (DW 12G PHY, PCIe, SATA, xGBE, GMAC, USB, DDRC, IC) are still in refactoring and preparation for integration into the mainline kernel. Hopefully I'll finish them up in the next two-three months, and submit them straight away.

Getting back to this patchset. As the subject states this is intended to prepare the MIPS-arch and generic kernel code for further Baikal-T1 SoC platform support integration (note the Baikal-T1 SoC platform code will be submitted last after the whole series of patchsets as a closure of the submission process). First of all the patchset starts with a set of changes to the dt-bindings kernel concerning MIPS CPC and CDMM nodes to make them being represented by dt schemas. In addition we updated the vendors prefix schema with Baikal Electronics JSC prefix so being further committed vendor-specific device drivers would be correctly accepted by the checkpatch script.

While I was working with the MIPS architecture code, I discovered, that there is a bug in the Coherency Manager v2 error causes declaration and the errors handler lacked of CM2 L2 ECC/parity errors support. So the fixes are here in the patchset.

Baikal-T1 SoC is based on the MIPS P5600 Warrior IP-core, which itself has MIPS32 Release 5 architecture. Even though on ISA level it doesn't differ much from the MIPS32 Release 2 release, there are still some peculiarities, which make it's justified to add the direct MIPS32r5 support into the kernel (see the specific patch for details). In addition seeing there is more than one real chip based on the MIPS P5600 core on the market, it would be good to have the direct P5600 CPU config in the MIPS-arch.

There were some issues we discovered while were working with MIPS-arch code. So the cleanups and fixes are introduced in this patchset. First of all the Write-Merge CPU feature hasn't been handled in a generic way. Even if a platform defined the writecombine flag as _CACHE_UNCACHED_ACCELERATED, the feature might have been disabled in the CP0 register. We either enable it or leave it as is in accordance with the knowledge of whether the corresponding platform really supports it. Secondly Memory Accessibility Attribute Registers (MAAR) haven't been properly initialized when Extended Physical Address (XPA) mode was enabled. Thirdly since some of the platforms may have a very strict limitations on the IO-memory access instructions. For instance Baikal-T1 SoC IO-memory can be accessed by the lw/sw instructions only. In this case for early-printk and CPS-debug code we suggest to use the instructions in accordance with the UART-registers offset (lb/sb if offset = 0, lh/sh if offset = 1 and so on). Fourthly in case if CPUFREQ feature is enabled and frequency of the CPU is changed by the reference clock alteration, we must make sure that MIPS r4k timer related services are properly updated when CPU-frequency changes. It concerns udelay lpj adjustment, MIPS timer clockevent frequency update. In addition when CPU reference frequency changes it isn't recommended to use the timer as clocksource at all, since currently the subsystem isn't tolerant to the unstable clock sources. So in this case we suggest to use the r4k timer for clocksourcing only as a last resort. Fifthly we discovered a bug in a method of CPUFREQ boost feature enable procedure and fixed it in one of the patches within this patchset. And finally there are a few fixups/cleanups we suggest to integrate into the MIPS FDC and CDMM related code (see the patches for details).

This patchset is rebased and tested on the mainline Linux kernel 5.7-rc4:
0e698dfa2822 ("Linux 5.7-rc4")
tag: v5.7-rc4

[1] http://www.baikalelectronics.com/products/168/

Changelog v2:
- Replace "be" vendor prefix with "baikal" one.
- Remove patches adding the platform devices to the trivial-devices.yaml bindings file (Rob nacked it).
- Add yaml-based bindings file for MIPS CDMM dt-node.
- Convert mti,mips-cpc to DT schema.
- Use a shorter summary describing the bindings modification patches.
- Rearrange the SoBs with adding Alexey' co-development tag.
- Lowercase the hex numbers in the dt-bindings.
- Fix author and SoB emails mismatch in the patch of adding the Baikal Electronis JSC prefix to the dt schema of the prefixes.
- Remove patch "tty: mips_ejtag_fdc: Mark expected switch fall-through" from the patchset as being already applied to the Greg' tty-next branch.
- Our corporate email server doesn't change Message-Id anymore, so the patchset is resubmitted being in the cover-letter-threaded format.

Cc Maxim Kaurkin <[email protected]>
Cc Pavel Parkhomenko <[email protected]>
Cc Ramil Zaripov <[email protected]>
Cc Ekaterina Skachko <[email protected]>
Cc Vadim Vlasov <[email protected]>
Cc Alexey Kolotnikov <[email protected]>
Cc Paul Burton <[email protected]>
Cc Ralf Baechle <[email protected]>
Cc Rob Herring <[email protected]>
Cc Frank Rowand <[email protected]>
Cc Tony Lindgren <[email protected]>
Cc Arnd Bergmann <[email protected]>
Cc Olof Johansson <[email protected]>
Cc Greg Kroah-Hartman <[email protected]>
Cc Jiri Slaby <[email protected]>
Cc "Rafael J. Wysocki" <[email protected]>
Cc Viresh Kumar <[email protected]>

Serge Semin (20):
dt-bindings: power: Convert mti,mips-cpc to DT schema
dt-bindings: bus: Add MIPS CDMM controller
dt-bindings: Add vendor prefix for Baikal Electronics, JSC
mips: cm: Fix an invalid error code of INTVN_*_ERR
mips: cm: Add L2 ECC/parity errors reporting
mips: Add MIPS32 Release 5 support
mips: Add MIPS Warrior P5600 support
mips: Fix cpu_has_mips64r1/2 activation for MIPS32 CPUs
mips: Add CP0 Write Merge config support
mips: Add CONFIG/CONFIG6/Cause reg fields macro
mips: MAAR: Use more precise address mask
mips: MAAR: Add XPA mode support
mips: early_printk_8250: Use offset-sized IO-mem accessors
mips: Use offset-sized IO-mem accessors in CPS debug printout
mips: cdmm: Add mti,mips-cdmm dtb node support
bus: cdmm: Add MIPS R5 arch support
mips: Add udelay lpj numbers adjustment
mips: csrc-r4k: Decrease r4k-clocksource rating if CPU_FREQ enabled
mips: cevt-r4k: Update the r4k-clockevent frequency in sync with CPU
cpufreq: Return zero on success in boost sw setting

 .../bindings/bus/mti,mips-cdmm.yaml           | 35 ++++++++++
 .../bindings/power/mti,mips-cpc.txt           |  8 ---
 .../bindings/power/mti,mips-cpc.yaml          | 35 ++++++++++
 .../devicetree/bindings/vendor-prefixes.yaml  |  2 +
 arch/mips/Kconfig                             | 62 ++++++++++++++--
 arch/mips/Makefile                            |  2 +
 arch/mips/include/asm/asmmacro.h              | 18 ++---
 arch/mips/include/asm/compiler.h              |  5 ++
 arch/mips/include/asm/cpu-features.h          | 34 ++++++---
 arch/mips/include/asm/cpu-info.h              |  2 +-
 arch/mips/include/asm/cpu-type.h              |  6 +-
 arch/mips/include/asm/cpu.h                   | 11 +--
 arch/mips/include/asm/fpu.h                   |  4 +-
 arch/mips/include/asm/hazards.h               |  8 ++-
 arch/mips/include/asm/maar.h                  | 17 ++++-
 arch/mips/include/asm/mipsregs.h              | 34 ++++++++-
 arch/mips/include/asm/stackframe.h            |  2 +-
 arch/mips/include/asm/switch_to.h             |  8 +--
 arch/mips/include/asm/vermagic.h              |  4 ++
 arch/mips/kernel/cevt-r4k.c                   | 44 ++++++++++++
 arch/mips/kernel/cps-vec-ns16550.S            | 18 ++++-
 arch/mips/kernel/cpu-probe.c                  | 60 ++++++++++++++++
 arch/mips/kernel/csrc-r4k.c                   |  4 ++
 arch/mips/kernel/early_printk_8250.c          | 34 ++++++++-
 arch/mips/kernel/entry.S                      |  6 +-
 arch/mips/kernel/mips-cm.c                    | 66 +++++++++++++++--
 arch/mips/kernel/proc.c                       |  2 +
 arch/mips/kernel/r4k_fpu.S                    | 14 ++--
 arch/mips/kernel/spram.c                      |  4 +-
 arch/mips/kernel/time.c                       | 70 +++++++++++++++++++
 arch/mips/kvm/vz.c                            |  6 +-
 arch/mips/lib/csum_partial.S                  |  6 +-
 arch/mips/mm/c-r4k.c                          |  7 +-
 arch/mips/mm/init.c                           |  8 ++-
 arch/mips/mm/sc-mips.c                        |  7 +-
 drivers/bus/Kconfig                           |  2 +-
 drivers/bus/mips_cdmm.c                       | 15 ++++
 drivers/cpufreq/cpufreq.c                     |  2 +-
 38 files changed, 591 insertions(+), 81 deletions(-)
 create mode 100644 Documentation/devicetree/bindings/bus/mti,mips-cdmm.yaml
 delete mode 100644 Documentation/devicetree/bindings/power/mti,mips-cpc.txt
 create mode 100644 Documentation/devicetree/bindings/power/mti,mips-cpc.yaml
Signature · 1 line
2.25.1
#2 lore raw

PATCH v2 1/20 dt-bindings: power: Convert mti,mips-cpc to DT schema

From: Serge Semin <[email protected]>

It's a Cluster Power Controller embedded into the MIPS IP cores. Currently the corresponding dts node is supposed to have compatible and reg properties.

Cc Alexey Malahov <[email protected]>
Cc Thomas Bogendoerfer <[email protected]>
Cc Paul Burton <[email protected]>
Cc Ralf Baechle <[email protected]>
Cc Arnd Bergmann <[email protected]>

---

Changelog v2:
- Reword the changelog summary - use shorter version.
- Lowercase the example hex'es.

---
 .../bindings/power/mti,mips-cpc.txt           |  8 -----
 .../bindings/power/mti,mips-cpc.yaml          | 35 +++++++++++++++++++
 2 files changed, 35 insertions(+), 8 deletions(-)
 delete mode 100644 Documentation/devicetree/bindings/power/mti,mips-cpc.txt
 create mode 100644 Documentation/devicetree/bindings/power/mti,mips-cpc.yaml
Documentation/devicetree/bindings/power/mti,mips-cpc.txt deleted +0 -8
deleted file mode 100644
index c6b82511ae8a..000000000000
@@ -1,8 +0,0 @@
-Binding for MIPS Cluster Power Controller (CPC).
-
-This binding allows a system to specify where the CPC registers are
-located.
-
-Required properties:
-compatible : Should be "mti,mips-cpc".
-regs: Should describe the address & size of the CPC register region.
Documentation/devicetree/bindings/power/mti,mips-cpc.yaml new file +35 -0
new file mode 100644
index 000000000000..9cd92a57130c
@@ -0,0 +1,35 @@
+# SPDX-License-Identifier: (GPL-2.0 OR BSD-2-Clause)
+%YAML 1.2
+---
+$id: http://devicetree.org/schemas/power/mti,mips-cpc.yaml#
+$schema: http://devicetree.org/meta-schemas/core.yaml#
+
+title: MIPS Cluster Power Controller
+
+description: |
+  Defines a location of the MIPS Cluster Power Controller registers.
+
+maintainers:
+  - Paul Burton <[email protected]>
+
+properties:
+  compatible:
+    const: mti,mips-cpc
+
+  reg:
+    description: |
+      Base address and size of an unoccupied memory region, which will be
+      used to map the MIPS CPC registers block.
+    maxItems: 1
+
+required:
+  - compatible
+  - reg
+
+examples:
+  - |
+    cpc@1bde0000 {
+      compatible = "mti,mips-cpc";
+      reg = <0 0x1bde0000 0 0x8000>;
+    };
+...
Signature · 1 line
2.25.1
maintainer Reviewed-by on v2 1/20
#3 lore raw

Re: [PATCH v2 01/20] dt-bindings: power: Convert mti,mips-cpc to DT schema

Quoted 19 lines · On Wed, 6 May 2020 20:42:19 +0300, wrote:

From: Serge Semin <[email protected]>

It's a Cluster Power Controller embedded into the MIPS IP cores. Currently the corresponding dts node is supposed to have compatible and reg properties.

Cc Alexey Malahov <[email protected]>
Cc Thomas Bogendoerfer <[email protected]>
Cc Paul Burton <[email protected]>
Cc Ralf Baechle <[email protected]>
Cc Arnd Bergmann <[email protected]>

---

Changelog v2:
- Reword the changelog summary - use shorter version.
- Lowercase the example hex'es.

Quoted 7 lines
---
 .../bindings/power/mti,mips-cpc.txt           |  8 -----
 .../bindings/power/mti,mips-cpc.yaml          | 35 +++++++++++++++++++
 2 files changed, 35 insertions(+), 8 deletions(-)
 delete mode 100644 Documentation/devicetree/bindings/power/mti,mips-cpc.txt
 create mode 100644 Documentation/devicetree/bindings/power/mti,mips-cpc.yaml
#4 lore raw

Re: [PATCH v2 01/20] dt-bindings: power: Convert mti,mips-cpc to DT schema

Quoted 28 lines · Rob Herring
Quoted 19 lines · On Wed, 6 May 2020 20:42:19 +0300, wrote:

From: Serge Semin <[email protected]>

It's a Cluster Power Controller embedded into the MIPS IP cores. Currently the corresponding dts node is supposed to have compatible and reg properties.

Cc Alexey Malahov <[email protected]>
Cc Thomas Bogendoerfer <[email protected]>
Cc Paul Burton <[email protected]>
Cc Ralf Baechle <[email protected]>
Cc Arnd Bergmann <[email protected]>

---

Changelog v2:
- Reword the changelog summary - use shorter version.
- Lowercase the example hex'es.

Quoted 7 lines
---
 .../bindings/power/mti,mips-cpc.txt           |  8 -----
 .../bindings/power/mti,mips-cpc.yaml          | 35 +++++++++++++++++++
 2 files changed, 35 insertions(+), 8 deletions(-)
 delete mode 100644 Documentation/devicetree/bindings/power/mti,mips-cpc.txt
 create mode 100644 Documentation/devicetree/bindings/power/mti,mips-cpc.yaml
Quoted 1 line

Great! Thanks.

-Sergey

#5 lore raw

PATCH v2 2/20 dt-bindings: bus: Add MIPS CDMM controller

In reply to [email protected]

From: Serge Semin <[email protected]>

It's a Common Device Memory Map controller embedded into the MIPS IP cores, which dts node is supposed to have compatible and reg properties.

Cc Alexey Malahov <[email protected]>
Cc Thomas Bogendoerfer <[email protected]>
Cc Paul Burton <[email protected]>
Cc Ralf Baechle <[email protected]>
Cc Arnd Bergmann <[email protected]>

---

Changelog v2:
- Lowercase the example hex'es.

---
 .../bindings/bus/mti,mips-cdmm.yaml           | 35 +++++++++++++++++++
 1 file changed, 35 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/bus/mti,mips-cdmm.yaml
Documentation/devicetree/bindings/bus/mti,mips-cdmm.yaml new file +35 -0
new file mode 100644
index 000000000000..d28d65ae57b2
@@ -0,0 +1,35 @@
+# SPDX-License-Identifier: (GPL-2.0 OR BSD-2-Clause)
+%YAML 1.2
+---
+$id: http://devicetree.org/schemas/bus/mti,mips-cdmm.yaml#
+$schema: http://devicetree.org/meta-schemas/core.yaml#
+
+title: MIPS Common Device Memory Map
+
+description: |
+  Defines a location of the MIPS Common Device Memory Map registers.
+
+maintainers:
+  - James Hogan <[email protected]>
+
+properties:
+  compatible:
+    const: mti,mips-cdmm
+
+  reg:
+    description: |
+      Base address and size of an unoccupied memory region, which will be
+      used to map the MIPS CDMM registers block.
+    maxItems: 1
+
+required:
+  - compatible
+  - reg
+
+examples:
+  - |
+    cdmm@1bde8000 {
+      compatible = "mti,mips-cdmm";
+      reg = <0 0x1bde8000 0 0x8000>;
+    };
+...
Signature · 1 line
2.25.1
maintainer Reviewed-by on v2 2/20
#6 lore raw

Re: [PATCH v2 02/20] dt-bindings: bus: Add MIPS CDMM controller

Quoted 14 lines · On Wed, 6 May 2020 20:42:20 +0300, wrote:

From: Serge Semin <[email protected]>

It's a Common Device Memory Map controller embedded into the MIPS IP cores, which dts node is supposed to have compatible and reg properties.

Cc Alexey Malahov <[email protected]>
Cc Thomas Bogendoerfer <[email protected]>
Cc Paul Burton <[email protected]>
Cc Ralf Baechle <[email protected]>
Cc Arnd Bergmann <[email protected]>
Quoted 9 lines

---

Changelog v2:
- Lowercase the example hex'es.

---
 .../bindings/bus/mti,mips-cdmm.yaml           | 35 +++++++++++++++++++
 1 file changed, 35 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/bus/mti,mips-cdmm.yaml
#7 lore raw

Re: [PATCH v2 02/20] dt-bindings: bus: Add MIPS CDMM controller

Quoted 25 lines · Rob Herring
Quoted 14 lines · On Wed, 6 May 2020 20:42:20 +0300, wrote:

From: Serge Semin <[email protected]>

It's a Common Device Memory Map controller embedded into the MIPS IP cores, which dts node is supposed to have compatible and reg properties.

Cc Alexey Malahov <[email protected]>
Cc Thomas Bogendoerfer <[email protected]>
Cc Paul Burton <[email protected]>
Cc Ralf Baechle <[email protected]>
Cc Arnd Bergmann <[email protected]>
Quoted 9 lines

---

Changelog v2:
- Lowercase the example hex'es.

---
 .../bindings/bus/mti,mips-cdmm.yaml           | 35 +++++++++++++++++++
 1 file changed, 35 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/bus/mti,mips-cdmm.yaml
Quoted 1 line

Great! Thanks.

-Sergey

#8 lore raw

PATCH v2 3/20 dt-bindings: Add vendor prefix for Baikal Electronics, JSC

In reply to [email protected]

From: Serge Semin <[email protected]>

Add "BAIKAL ELECTRONICS, JSC" to the list of devicetree vendor prefixes as "baikal".

Website: http://www.baikalelectronics.com

Cc Alexey Malahov <[email protected]>
Cc Thomas Bogendoerfer <[email protected]>
Cc Paul Burton <[email protected]>
Cc Ralf Baechle <[email protected]>
Cc Arnd Bergmann <[email protected]>

---

Changelog v2:
- Fix author and SoB emails mismatch.
- Add 'baikal' vendor prefix instead of ambiguous 'be'.

---
 Documentation/devicetree/bindings/vendor-prefixes.yaml | 2 ++
 1 file changed, 2 insertions(+)
Documentation/devicetree/bindings/vendor-prefixes.yaml +2 -0
index d3891386d671..674c0d07c0ad 100644
@@ -139,6 +139,8 @@ patternProperties:
     description: Azoteq (Pty) Ltd
   "^azw,.*":
     description: Shenzhen AZW Technology Co., Ltd.
+  "^baikal,.*":
+    description: BAIKAL ELECTRONICS, JSC
   "^bananapi,.*":
     description: BIPAI KEJI LIMITED
   "^beacon,.*":
Signature · 1 line
2.25.1
#9 lore raw

Re: [PATCH v2 03/20] dt-bindings: Add vendor prefix for Baikal Electronics, JSC

Hi Sergey.

Quoted 16 lines · On Wed, May 06, 2020 at 08:42:21PM +0300, [email protected]…

From: Serge Semin <[email protected]>

Add "BAIKAL ELECTRONICS, JSC" to the list of devicetree vendor prefixes as "baikal".

Website: http://www.baikalelectronics.com

Cc Alexey Malahov <[email protected]>
Cc Thomas Bogendoerfer <[email protected]>
Cc Paul Burton <[email protected]>
Cc Ralf Baechle <[email protected]>
Cc Arnd Bergmann <[email protected]>
Quoted 6 lines

---

Changelog v2:
- Fix author and SoB emails mismatch.

- Add 'baikal' vendor prefix instead of ambiguous 'be'.

Agree, much better.

Quoted 9 lines
---
 Documentation/devicetree/bindings/vendor-prefixes.yaml | 2 ++
 1 file changed, 2 insertions(+)
Documentation/devicetree/bindings/vendor-prefixes.yaml +0 -0
index d3891386d671..674c0d07c0ad 100644

Documentation/devicetree/bindings/vendor-prefixes.yaml patternProperties: · line 143

Quoted 5 lines
Documentation/devicetree/bindings/vendor-prefixes.yaml +2 -0
@@ -139,6 +139,8 @@ patternProperties:
     description: Azoteq (Pty) Ltd
   "^azw,.*":
     description: Shenzhen AZW Technology Co., Ltd.
+  "^baikal,.*":
+    description: BAIKAL ELECTRONICS, JSC

Baikal do not use ALL UPPSECASE on their website for their name. So please use same case use as they do themself.

Sam

Quoted 5 lines
Documentation/devicetree/bindings/vendor-prefixes.yaml +0 -0
   "^bananapi,.*":
     description: BIPAI KEJI LIMITED
   "^beacon,.*":
Signature · 1 line
2.25.1
#10 lore raw

Re: [PATCH v2 03/20] dt-bindings: Add vendor prefix for Baikal Electronics, JSC

Hi Sam,

Quoted 31 lines · Sam Ravnborg

Hi Sergey.

Quoted 16 lines · On Wed, May 06, 2020 at 08:42:21PM +0300, [email protected]…

From: Serge Semin <[email protected]>

Add "BAIKAL ELECTRONICS, JSC" to the list of devicetree vendor prefixes as "baikal".

Website: http://www.baikalelectronics.com

Cc Alexey Malahov <[email protected]>
Cc Thomas Bogendoerfer <[email protected]>
Cc Paul Burton <[email protected]>
Cc Ralf Baechle <[email protected]>
Cc Arnd Bergmann <[email protected]>
Quoted 6 lines

---

Changelog v2:
- Fix author and SoB emails mismatch.

- Add 'baikal' vendor prefix instead of ambiguous 'be'.

Agree, much better.

Quoted 4 lines
---
 Documentation/devicetree/bindings/vendor-prefixes.yaml | 2 ++
 1 file changed, 2 insertions(+)
Quoted 13 lines

Documentation/devicetree/bindings/vendor-prefixes.yaml patternProperties: · line 143

Quoted 10 lines
Documentation/devicetree/bindings/vendor-prefixes.yaml +2 -0
index d3891386d671..674c0d07c0ad 100644
@@ -139,6 +139,8 @@ patternProperties:
     description: Azoteq (Pty) Ltd
   "^azw,.*":
     description: Shenzhen AZW Technology Co., Ltd.
+  "^baikal,.*":
+    description: BAIKAL ELECTRONICS, JSC

Baikal do not use ALL UPPSECASE on their website for their name. So please use same case use as they do themself.

It's not like me can't be considered as part of them.) I discussed the upper-case and normal version with our managers half a year ago and we came up to use the upper-case version. From Russian legal point of view it's also the upper-cased version what counts. I don't really know why the site use different naming, but in the internal documents it's always as submitted. Anyway I asked this question one more time to our managers. If they say to use as you suggest, then I'll resend an update in v3 patchset, if v3 doesn't get to be necessary I'll send a followup patch with fix.

-Sergey

Quoted 8 lines

Sam

Quoted 5 lines
Documentation/devicetree/bindings/vendor-prefixes.yaml +0 -0
   "^bananapi,.*":
     description: BIPAI KEJI LIMITED
   "^beacon,.*":
Signature · 1 line
2.25.1
Acked-by
#11 lore raw

Re: [PATCH v2 03/20] dt-bindings: Add vendor prefix for Baikal Electronics, JSC

Hi Sergey

Quoted 11 lines
Quoted 10 lines

Documentation/devicetree/bindings/vendor-prefixes.yaml patternProperties: · line 143

Quoted 7 lines

+++ b/Documentation/devicetree/bindings/vendor-prefixes.yaml

Documentation/devicetree/bindings/vendor-prefixes.yaml +2 -0
@@ -139,6 +139,8 @@ patternProperties:
     description: Azoteq (Pty) Ltd
   "^azw,.*":
     description: Shenzhen AZW Technology Co., Ltd.
+  "^baikal,.*":
+    description: BAIKAL ELECTRONICS, JSC

Baikal do not use ALL UPPSECASE on their website for their name. So please use same case use as they do themself.

Quoted 9 lines

It's not like me can't be considered as part of them.) I discussed the upper-case and normal version with our managers half a year ago and we came up to use the upper-case version. From Russian legal point of view it's also the upper-cased version what counts. I don't really know why the site use different naming, but in the internal documents it's always as submitted. Anyway I asked this question one more time to our managers. If they say to use as you suggest, then I'll resend an update in v3 patchset, if v3 doesn't get to be necessary I'll send a followup patch with fix.

I had expected it was upper case because others used upper case. But there is a good explanation and then all is fine wiht me.

And it is an alphabetic order - so

Quoted 11 lines

-Sergey

Quoted 8 lines

Sam

Quoted 5 lines
Documentation/devicetree/bindings/vendor-prefixes.yaml +0 -0
   "^bananapi,.*":
     description: BIPAI KEJI LIMITED
   "^beacon,.*":
Signature · 1 line
2.25.1
#12 lore raw

Re: [PATCH v2 03/20] dt-bindings: Add vendor prefix for Baikal Electronics, JSC

Quoted 22 lines · Sam Ravnborg

Hi Sergey

Quoted 11 lines
Quoted 10 lines

Documentation/devicetree/bindings/vendor-prefixes.yaml patternProperties: · line 143

Quoted 7 lines

+++ b/Documentation/devicetree/bindings/vendor-prefixes.yaml

Documentation/devicetree/bindings/vendor-prefixes.yaml +2 -0
@@ -139,6 +139,8 @@ patternProperties:
     description: Azoteq (Pty) Ltd
   "^azw,.*":
     description: Shenzhen AZW Technology Co., Ltd.
+  "^baikal,.*":
+    description: BAIKAL ELECTRONICS, JSC

Baikal do not use ALL UPPSECASE on their website for their name. So please use same case use as they do themself.

Quoted 9 lines

It's not like me can't be considered as part of them.) I discussed the upper-case and normal version with our managers half a year ago and we came up to use the upper-case version. From Russian legal point of view it's also the upper-cased version what counts. I don't really know why the site use different naming, but in the internal documents it's always as submitted. Anyway I asked this question one more time to our managers. If they say to use as you suggest, then I'll resend an update in v3 patchset, if v3 doesn't get to be necessary I'll send a followup patch with fix.

Quoted 5 lines

I had expected it was upper case because others used upper case. But there is a good explanation and then all is fine wiht me.

And it is an alphabetic order - so

Great. Thanks. I've just got an answer from our legal department:
"The Company has a Charter in which the name in English is:
“BAIKAL ELECTRONICS, JSC.” Charter is the only thing you need to focus on when specifying the name."

I also pointed out that it would be good to have the site updated if the upper-cased name is required by the Charter. Hope it will be done soon.)

-Sergey

Quoted 12 lines
Quoted 12 lines

-Sergey

Quoted 8 lines

Sam

Quoted 5 lines
Documentation/devicetree/bindings/vendor-prefixes.yaml +0 -0
   "^bananapi,.*":
     description: BIPAI KEJI LIMITED
   "^beacon,.*":
Signature · 1 line
2.25.1
#13 lore raw

Re: [PATCH v2 03/20] dt-bindings: Add vendor prefix for Baikal Electronics, JSC

In reply to [email protected]

Rob,

Could you also take a look at this patch? There are several patchsets I've sent which depend on the vendor-prefix it defines. So when you get to check those patchsets DT files, the dt_binding_check will fail without it. Is it possible somehow to pick this patch up from here and apply it before checking those Baikal-T1-specific binding files?

-Sergey

Quoted 41 lines · On Wed, May 06, 2020 at 08:42:21PM +0300, [email protected]…

From: Serge Semin <[email protected]>

Add "BAIKAL ELECTRONICS, JSC" to the list of devicetree vendor prefixes as "baikal".

Website: http://www.baikalelectronics.com

Cc Alexey Malahov <[email protected]>
Cc Thomas Bogendoerfer <[email protected]>
Cc Paul Burton <[email protected]>
Cc Ralf Baechle <[email protected]>
Cc Arnd Bergmann <[email protected]>

---

Changelog v2:
- Fix author and SoB emails mismatch.
- Add 'baikal' vendor prefix instead of ambiguous 'be'.

---
 Documentation/devicetree/bindings/vendor-prefixes.yaml | 2 ++
 1 file changed, 2 insertions(+)
Documentation/devicetree/bindings/vendor-prefixes.yaml +2 -0
index d3891386d671..674c0d07c0ad 100644
@@ -139,6 +139,8 @@ patternProperties:
     description: Azoteq (Pty) Ltd
   "^azw,.*":
     description: Shenzhen AZW Technology Co., Ltd.
+  "^baikal,.*":
+    description: BAIKAL ELECTRONICS, JSC
   "^bananapi,.*":
     description: BIPAI KEJI LIMITED
   "^beacon,.*":
Signature · 1 line
2.25.1
maintainer Applied
#14 lore raw

Re: [PATCH v2 03/20] dt-bindings: Add vendor prefix for Baikal Electronics, JSC

In reply to [email protected]
Quoted 16 lines · On Wed, 6 May 2020 20:42:21 +0300, wrote:

From: Serge Semin <[email protected]>

Add "BAIKAL ELECTRONICS, JSC" to the list of devicetree vendor prefixes as "baikal".

Website: http://www.baikalelectronics.com

Cc Alexey Malahov <[email protected]>
Cc Thomas Bogendoerfer <[email protected]>
Cc Paul Burton <[email protected]>
Cc Ralf Baechle <[email protected]>
Cc Arnd Bergmann <[email protected]>
Quoted 9 lines

---

Changelog v2:
- Fix author and SoB emails mismatch.
- Add 'baikal' vendor prefix instead of ambiguous 'be'.

---
 Documentation/devicetree/bindings/vendor-prefixes.yaml | 2 ++
 1 file changed, 2 insertions(+)

Applied, thanks!

#15 lore raw

PATCH v2 4/20 mips: cm: Fix an invalid error code of INTVN_*_ERR

In reply to [email protected]

From: Serge Semin <[email protected]>

Commit 3885c2b463f6 ("MIPS: CM: Add support for reporting CM cache errors") adds cm2_causes[] array with map of error type ID and pointers to the short description string. There is a mistake in the table, since according to MIPS32 manual CM2_ERROR_TYPE = {17,18} correspond to INTVN_WR_ERR and INTVN_RD_ERR, while the table claims they have {0x17,0x18} codes. This is obviously hex-dec copy-paste bug. Moreover codes {0x18 - 0x1a} indicate L2 ECC errors.

Fixes 3885c2b463f6 ("MIPS: CM: Add support for reporting CM cache errors")
Cc Alexey Malahov <[email protected]>
Cc Thomas Bogendoerfer <[email protected]>
Cc Paul Burton <[email protected]>
Cc Ralf Baechle <[email protected]>
Cc Arnd Bergmann <[email protected]>
Cc Rob Herring <[email protected]>
---
 arch/mips/kernel/mips-cm.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
arch/mips/kernel/mips-cm.c +3 -3
index cdb93ed91cde..361bfc91a0e6 100644
@@ -119,9 +119,9 @@ static char *cm2_causes[32] = {
 	"COH_RD_ERR", "MMIO_WR_ERR", "MMIO_RD_ERR", "0x07",
 	"0x08", "0x09", "0x0a", "0x0b",
 	"0x0c", "0x0d", "0x0e", "0x0f",
-	"0x10", "0x11", "0x12", "0x13",
-	"0x14", "0x15", "0x16", "INTVN_WR_ERR",
-	"INTVN_RD_ERR", "0x19", "0x1a", "0x1b",
+	"0x10", "INTVN_WR_ERR", "INTVN_RD_ERR", "0x13",
+	"0x14", "0x15", "0x16", "0x17",
+	"0x18", "0x19", "0x1a", "0x1b",
 	"0x1c", "0x1d", "0x1e", "0x1f"
 };
 
Signature · 1 line
2.25.1
maintainer Applied
#16 lore raw

Re: [PATCH v2 04/20] mips: cm: Fix an invalid error code of INTVN_*_ERR

Quoted 20 lines · On Wed, May 06, 2020 at 08:42:22PM +0300, [email protected]…

From: Serge Semin <[email protected]>

Commit 3885c2b463f6 ("MIPS: CM: Add support for reporting CM cache errors") adds cm2_causes[] array with map of error type ID and pointers to the short description string. There is a mistake in the table, since according to MIPS32 manual CM2_ERROR_TYPE = {17,18} correspond to INTVN_WR_ERR and INTVN_RD_ERR, while the table claims they have {0x17,0x18} codes. This is obviously hex-dec copy-paste bug. Moreover codes {0x18 - 0x1a} indicate L2 ECC errors.

Fixes 3885c2b463f6 ("MIPS: CM: Add support for reporting CM cache errors")
Cc Alexey Malahov <[email protected]>
Cc Thomas Bogendoerfer <[email protected]>
Cc Paul Burton <[email protected]>
Cc Ralf Baechle <[email protected]>
Cc Arnd Bergmann <[email protected]>
Cc Rob Herring <[email protected]>
Quoted 3 lines
---
 arch/mips/kernel/mips-cm.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

applied to mips-next.

Thomas.

Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]
#17 lore raw

Re: [PATCH v2 04/20] mips: cm: Fix an invalid error code of INTVN_*_ERR

Quoted 25 lines · Thomas Bogendoerfer
Quoted 20 lines · On Wed, May 06, 2020 at 08:42:22PM +0300, [email protected]…

From: Serge Semin <[email protected]>

Commit 3885c2b463f6 ("MIPS: CM: Add support for reporting CM cache errors") adds cm2_causes[] array with map of error type ID and pointers to the short description string. There is a mistake in the table, since according to MIPS32 manual CM2_ERROR_TYPE = {17,18} correspond to INTVN_WR_ERR and INTVN_RD_ERR, while the table claims they have {0x17,0x18} codes. This is obviously hex-dec copy-paste bug. Moreover codes {0x18 - 0x1a} indicate L2 ECC errors.

Fixes 3885c2b463f6 ("MIPS: CM: Add support for reporting CM cache errors")
Cc Alexey Malahov <[email protected]>
Cc Thomas Bogendoerfer <[email protected]>
Cc Paul Burton <[email protected]>
Cc Ralf Baechle <[email protected]>
Cc Arnd Bergmann <[email protected]>
Cc Rob Herring <[email protected]>
Quoted 3 lines
---
 arch/mips/kernel/mips-cm.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
Quoted 1 line

applied to mips-next.

Great! Thanks.

-Sergey

Quoted 6 lines

Thomas.

Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]
#18 lore raw

PATCH v2 5/20 mips: cm: Add L2 ECC/parity errors reporting

In reply to [email protected]

From: Serge Semin <[email protected]>

According to the MIPS32 InterAptiv software manual error codes 24 - 26 of CM2 indicate L2 ECC/parity error with switching to a corresponding errors info fields. This patch provides these errors parsing code, which handles the read/write uncorrectable and correctable ECC/parity errors, and prints instruction causing the fault, RAM array type, cache way/dword and syndrome associated with the faulty data.

---
 arch/mips/kernel/mips-cm.c | 62 ++++++++++++++++++++++++++++++++++++--
 1 file changed, 60 insertions(+), 2 deletions(-)
arch/mips/kernel/mips-cm.c +60 -2
index 361bfc91a0e6..f60af512c877 100644
@@ -114,6 +114,48 @@ static char *cm2_core[8] = {
 	"Exclusive/OK", "Exclusive/Data"
 };
 
+static char *cm2_l2_type[4] = {
+	[0x0] = "None",
+	[0x1] = "Tag RAM single/double ECC error",
+	[0x2] = "Data RAM single/double ECC error",
+	[0x3] = "WS RAM uncorrectable dirty parity"
+};
+
+static char *cm2_l2_instr[32] = {
+	[0x00] = "L2_NOP",
+	[0x01] = "L2_ERR_CORR",
+	[0x02] = "L2_TAG_INV",
+	[0x03] = "L2_WS_CLEAN",
+	[0x04] = "L2_RD_MDYFY_WR",
+	[0x05] = "L2_WS_MRU",
+	[0x06] = "L2_EVICT_LN2",
+	[0x07] = "0x07",
+	[0x08] = "L2_EVICT",
+	[0x09] = "L2_REFL",
+	[0x0a] = "L2_RD",
+	[0x0b] = "L2_WR",
+	[0x0c] = "L2_EVICT_MRU",
+	[0x0d] = "L2_SYNC",
+	[0x0e] = "L2_REFL_ERR",
+	[0x0f] = "0x0f",
+	[0x10] = "L2_INDX_WB_INV",
+	[0x11] = "L2_INDX_LD_TAG",
+	[0x12] = "L2_INDX_ST_TAG",
+	[0x13] = "L2_INDX_ST_DATA",
+	[0x14] = "L2_INDX_ST_ECC",
+	[0x15] = "0x15",
+	[0x16] = "0x16",
+	[0x17] = "0x17",
+	[0x18] = "L2_FTCH_AND_LCK",
+	[0x19] = "L2_HIT_INV",
+	[0x1a] = "L2_HIT_WB_INV",
+	[0x1b] = "L2_HIT_WB",
+	[0x1c] = "0x1c",
+	[0x1d] = "0x1d",
+	[0x1e] = "0x1e",
+	[0x1f] = "0x1f"
+};
+
 static char *cm2_causes[32] = {
 	"None", "GC_WR_ERR", "GC_RD_ERR", "COH_WR_ERR",
 	"COH_RD_ERR", "MMIO_WR_ERR", "MMIO_RD_ERR", "0x07",
@@ -121,7 +163,7 @@ static char *cm2_causes[32] = {
 	"0x0c", "0x0d", "0x0e", "0x0f",
 	"0x10", "INTVN_WR_ERR", "INTVN_RD_ERR", "0x13",
 	"0x14", "0x15", "0x16", "0x17",
-	"0x18", "0x19", "0x1a", "0x1b",
+	"L2_RD_UNCORR", "L2_WR_UNCORR", "L2_CORR", "0x1b",
 	"0x1c", "0x1d", "0x1e", "0x1f"
 };
 
@@ -360,7 +402,7 @@ void mips_cm_error_report(void)
 				 "CCA=%lu TR=%s MCmd=%s STag=%lu "
 				 "SPort=%lu\n", cca_bits, cm2_tr[tr_bits],
 				 cm2_cmd[cmd_bits], stag_bits, sport_bits);
-		} else {
+		} else if (cause < 24) {
 			/* glob state & sresp together */
 			unsigned long c3_bits = (cm_error >> 18) & 7;
 			unsigned long c2_bits = (cm_error >> 15) & 7;
@@ -377,6 +419,22 @@ void mips_cm_error_report(void)
 				 cm2_core[c1_bits], cm2_core[c0_bits],
 				 sc_bit ? "True" : "False",
 				 cm2_cmd[cmd_bits], sport_bits);
+		} else {
+			unsigned long muc_bit = (cm_error >> 23) & 1;
+			unsigned long ins_bits = (cm_error >> 18) & 0x1f;
+			unsigned long arr_bits = (cm_error >> 16) & 3;
+			unsigned long dw_bits = (cm_error >> 12) & 15;
+			unsigned long way_bits = (cm_error >> 9) & 7;
+			unsigned long mway_bit = (cm_error >> 8) & 1;
+			unsigned long syn_bits = (cm_error >> 0) & 0xFF;
+
+			snprintf(buf, sizeof(buf),
+				 "Type=%s%s Instr=%s DW=%lu Way=%lu "
+				 "MWay=%s Syndrome=0x%02lx",
+				 muc_bit ? "Multi-UC " : "",
+				 cm2_l2_type[arr_bits],
+				 cm2_l2_instr[ins_bits], dw_bits, way_bits,
+				 mway_bit ? "True" : "False", syn_bits);
 		}
 		pr_err("CM_ERROR=%08llx %s <%s>\n", cm_error,
 		       cm2_causes[cause], buf);
Signature · 1 line
2.25.1
maintainer Applied
#19 lore raw

Re: [PATCH v2 05/20] mips: cm: Add L2 ECC/parity errors reporting

Quoted 19 lines · On Wed, May 06, 2020 at 08:42:23PM +0300, [email protected]…

From: Serge Semin <[email protected]>

According to the MIPS32 InterAptiv software manual error codes 24 - 26 of CM2 indicate L2 ECC/parity error with switching to a corresponding errors info fields. This patch provides these errors parsing code, which handles the read/write uncorrectable and correctable ECC/parity errors, and prints instruction causing the fault, RAM array type, cache way/dword and syndrome associated with the faulty data.

Quoted 3 lines
---
 arch/mips/kernel/mips-cm.c | 62 ++++++++++++++++++++++++++++++++++++--
 1 file changed, 60 insertions(+), 2 deletions(-)

applied to mips-next.

Thomas.

Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]
#20 lore raw

Re: [PATCH v2 05/20] mips: cm: Add L2 ECC/parity errors reporting

Quoted 24 lines · Thomas Bogendoerfer
Quoted 19 lines · On Wed, May 06, 2020 at 08:42:23PM +0300, [email protected]…

From: Serge Semin <[email protected]>

According to the MIPS32 InterAptiv software manual error codes 24 - 26 of CM2 indicate L2 ECC/parity error with switching to a corresponding errors info fields. This patch provides these errors parsing code, which handles the read/write uncorrectable and correctable ECC/parity errors, and prints instruction causing the fault, RAM array type, cache way/dword and syndrome associated with the faulty data.

Quoted 3 lines
---
 arch/mips/kernel/mips-cm.c | 62 ++++++++++++++++++++++++++++++++++++--
 1 file changed, 60 insertions(+), 2 deletions(-)
Quoted 1 line

applied to mips-next.

Great! Thanks.

-Sergey

Quoted 6 lines

Thomas.

Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]
#21 lore raw

PATCH v2 6/20 mips: Add MIPS32 Release 5 support

In reply to [email protected]

From: Serge Semin <[email protected]>

There are five MIPS32/64 architecture releases currently available:
from 1 to 6 except fourth one, which was intentionally skipped. Three of them can be called as major: 1st, 2nd and 6th, that not only have some system level alterations, but also introduced significant core/ISA level updates. The rest of the MIPS architecture releases are minor.

Even though they don't have as much ISA/system/core level changes as the major ones with respect to the previous releases, they still provide a set of updates (I'd say they were intended to be the intermediate releases before a major one) that might be useful for the kernel and user-level code, when activated by the kernel or compiler. In particular the following features were introduced or ended up being available at/after MIPS32 Release 5 architecture:
+ the last release of the misaligned memory access instructions,
+ virtualisation - VZ ASE - is optional component of the arch,
+ SIMD - MSA ASE - is optional component of the arch,
+ DSP ASE is optional component of the arch,
+ CP0.Status.FR=1 for CP1.FIR.F64=1 (pure 64-bit FPU general registers) must be available if FPU is implemented,
+ CP1.FIR.Has2008 support is required so CP1.FCSR.{ABS2008,NAN2008} bits are available.
+ UFR/UNFR aliases to access CP0.Status.FR from user-space by means of ctc1/cfc1 instructions (enabled by CP0.Config5.UFR),
+ CP0.COnfig5.LLB=1 and eretnc instruction are implemented to without accidentally clearing LL-bit when returning from an interrupt, exception, or error trap,
+ XPA feature together with extended versions of CPx registers is introduced, which needs to have mfhc0/mthc0 instructions available.

So due to these changes GNU GCC provides an extended instructions set support for MIPS32 Release 5 by default like eretnc/mfhc0/mthc0. Even though the architecture alteration isn't that big, it still worth to be taken into account by the kernel software. Finally we can't deny that some optimization/limitations might be found in future and implemented on some level in kernel or compiler. In this case having even intermediate MIPS architecture releases support would be more than useful.

So the most of the changes provided by this commit can be split into either compile- or runtime configs related. The compile-time related changes are caused by adding the new CONFIG_CPU_MIPS32_R5/CONFIG_CPU_MIPSR5 configs and concern the code activating MIPSR2 or MIPSR6 already implemented features (like eretnc/LLbit, mthc0/mfhc0). In addition CPU_HAS_MSA can be now freely enabled for MIPS32 release 5 based platforms as this is done for CPU_MIPS32_R6 CPUs. The runtime changes concerns the features which are handled with respect to the MIPS ISA revision detected at run-time by means of CP0.Config.{AT,AR} bits. Alas these fields can be used to detect either r1 or r2 or r6 releases. But since we know which CPUs in fact support the R5 arch, we can manually set MIPS_CPU_ISA_M32R5 bit of c->isa_level and then use cpu_has_mips32_r5 where it's appropriate.

Since XPA/EVA provide too complex alterationss and to have them used with MIPS32 Release 2 charged kernels (for compatibility with current platform configs) they are left to be setup as a separate kernel configs.

---
Even without this patch the code contains too many patterns like:
+#if defined(CONFIG_MIPS32R2) || defined(CONFIG_MIPS32R6) What about switching it to a simpler
+#if CONFIG_TARGET_ISA_REV >= 2 ?
Though I'd prefer this config to be named like CONFIG_CPU_MIPS_REV. What do you think?

Similarly the pattern like:

(MIPS_CPU_ISA_M32R1 | MIPS_CPU_ISA_M64R1 |
 MIPS_CPU_ISA_M32R2 | MIPS_CPU_ISA_M64R2 |
 MIPS_CPU_ISA_M32R5 |
 MIPS_CPU_ISA_M32R6 | MIPS_CPU_ISA_M64R6)

could be replaced with simpler one:
(MIPS_CPU_ISA_M32 | MIPS_CPU_ISA_M64)
if corresponding macro were available. What do think about adding such?

We also could add CPU_MIPS64_R5 config support here, but I don't think it's necessary at the moment seeing there is no any real chip ever produced with that arch. Right?

---
 arch/mips/Kconfig                    | 34 ++++++++++++++++++++++++----
 arch/mips/Makefile                   |  1 +
 arch/mips/include/asm/asmmacro.h     | 18 ++++++++-------
 arch/mips/include/asm/compiler.h     |  5 ++++
 arch/mips/include/asm/cpu-features.h | 20 +++++++++++-----
 arch/mips/include/asm/cpu-info.h     |  2 +-
 arch/mips/include/asm/cpu-type.h     |  6 ++++-
 arch/mips/include/asm/cpu.h          |  7 +++---
 arch/mips/include/asm/fpu.h          |  4 ++--
 arch/mips/include/asm/hazards.h      |  8 ++++---
 arch/mips/include/asm/stackframe.h   |  2 +-
 arch/mips/include/asm/switch_to.h    |  8 +++----
 arch/mips/include/asm/vermagic.h     |  2 ++
 arch/mips/kernel/cpu-probe.c         | 12 ++++++++++
 arch/mips/kernel/entry.S             |  6 ++---
 arch/mips/kernel/proc.c              |  2 ++
 arch/mips/kernel/r4k_fpu.S           | 14 ++++++------
 arch/mips/kvm/vz.c                   |  6 ++---
 arch/mips/lib/csum_partial.S         |  6 +++--
 arch/mips/mm/c-r4k.c                 |  7 +++---
 arch/mips/mm/sc-mips.c               |  7 +++---
 21 files changed, 123 insertions(+), 54 deletions(-)
arch/mips/Kconfig +30 -4
index 690718b3701a..55c3dbfea336 100644
@@ -1580,6 +1580,21 @@ config CPU_MIPS32_R2
 	  specific type of processor in your system, choose those that one
 	  otherwise CPU_MIPS32_R1 is a safe bet for any MIPS32 system.
 
+config CPU_MIPS32_R5
+	bool "MIPS32 Release 5"
+	depends on SYS_HAS_CPU_MIPS32_R5
+	select CPU_HAS_PREFETCH
+	select CPU_SUPPORTS_32BIT_KERNEL
+	select CPU_SUPPORTS_HIGHMEM
+	select CPU_SUPPORTS_MSA
+	select HAVE_KVM
+	select MIPS_O32_FP64_SUPPORT
+	help
+	  Choose this option to build a kernel for release 5 or later of the
+	  MIPS32 architecture.  New MIPS processors, starting with the Warrior
+	  family, are based on a MIPS32r5 processor. If you own an older
+	  processor, you probably need to select MIPS32r1 or MIPS32r2 instead.
+
 config CPU_MIPS32_R6
 	bool "MIPS32 Release 6"
 	depends on SYS_HAS_CPU_MIPS32_R6
@@ -1826,7 +1841,7 @@ endchoice
 config CPU_MIPS32_3_5_FEATURES
 	bool "MIPS32 Release 3.5 Features"
 	depends on SYS_HAS_CPU_MIPS32_R3_5
-	depends on CPU_MIPS32_R2 || CPU_MIPS32_R6
+	depends on CPU_MIPS32_R2 || CPU_MIPS32_R5 || CPU_MIPS32_R6
 	help
 	  Choose this option to build a kernel for release 2 or later of the
 	  MIPS32 architecture including features from the 3.5 release such as
@@ -1846,7 +1861,7 @@ config CPU_MIPS32_3_5_EVA
 config CPU_MIPS32_R5_FEATURES
 	bool "MIPS32 Release 5 Features"
 	depends on SYS_HAS_CPU_MIPS32_R5
-	depends on CPU_MIPS32_R2
+	depends on CPU_MIPS32_R2 || CPU_MIPS32_R5
 	help
 	  Choose this option to build a kernel for release 2 or later of the
 	  MIPS32 architecture including features from release 5 such as
@@ -2084,7 +2099,8 @@ endmenu
 #
 config CPU_MIPS32
 	bool
-	default y if CPU_MIPS32_R1 || CPU_MIPS32_R2 || CPU_MIPS32_R6
+	default y if CPU_MIPS32_R1 || CPU_MIPS32_R2 || CPU_MIPS32_R5 || \
+		     CPU_MIPS32_R6
 
 config CPU_MIPS64
 	bool
@@ -2104,6 +2120,13 @@ config CPU_MIPSR2
 	select CPU_HAS_DIEI if !CPU_DIEI_BROKEN
 	select MIPS_SPRAM
 
+config CPU_MIPSR5
+	bool
+	default y if CPU_MIPS32_R5
+	select CPU_HAS_RIXI
+	select CPU_HAS_DIEI if !CPU_DIEI_BROKEN
+	select MIPS_SPRAM
+
 config CPU_MIPSR6
 	bool
 	default y if CPU_MIPS32_R6 || CPU_MIPS64_R6
@@ -2118,6 +2141,7 @@ config TARGET_ISA_REV
 	int
 	default 1 if CPU_MIPSR1
 	default 2 if CPU_MIPSR2
+	default 5 if CPU_MIPSR5
 	default 6 if CPU_MIPSR6
 	default 0
 	help
@@ -2707,7 +2731,9 @@ config NEED_PER_CPU_EMBED_FIRST_CHUNK
 
 config RELOCATABLE
 	bool "Relocatable kernel"
-	depends on SYS_SUPPORTS_RELOCATABLE && (CPU_MIPS32_R2 || CPU_MIPS64_R2 || CPU_MIPS32_R6 || CPU_MIPS64_R6 || CAVIUM_OCTEON_SOC)
+	depends on SYS_SUPPORTS_RELOCATABLE
+	depends on CPU_MIPS32_R2 || CPU_MIPS64_R2 || CPU_MIPS32_R5 || \
+		   CPU_MIPS32_R6 || CPU_MIPS64_R6 || CAVIUM_OCTEON_SOC
 	help
 	  This builds a kernel image that retains relocation information
 	  so it can be loaded someplace besides the default 1MB.
arch/mips/Makefile +1 -0
index e1c44aed8156..9172fb0f630b 100644
@@ -171,6 +171,7 @@ cflags-$(CONFIG_CPU_R4X00)	+= -march=r4600 -Wa,--trap
 cflags-$(CONFIG_CPU_TX49XX)	+= -march=r4600 -Wa,--trap
 cflags-$(CONFIG_CPU_MIPS32_R1)	+= -march=mips32 -Wa,--trap
 cflags-$(CONFIG_CPU_MIPS32_R2)	+= -march=mips32r2 -Wa,--trap
+cflags-$(CONFIG_CPU_MIPS32_R5)	+= -march=mips32r5 -Wa,--trap -modd-spreg
 cflags-$(CONFIG_CPU_MIPS32_R6)	+= -march=mips32r6 -Wa,--trap -modd-spreg
 cflags-$(CONFIG_CPU_MIPS64_R1)	+= -march=mips64 -Wa,--trap
 cflags-$(CONFIG_CPU_MIPS64_R2)	+= -march=mips64r2 -Wa,--trap
arch/mips/include/asm/asmmacro.h +10 -8
index 655f40ddb6d1..86f2323ebe6b 100644
@@ -44,7 +44,8 @@
 	.endm
 #endif
 
-#if defined(CONFIG_CPU_MIPSR2) || defined(CONFIG_CPU_MIPSR6)
+#if defined(CONFIG_CPU_MIPSR2) || defined(CONFIG_CPU_MIPSR5) || \
+    defined(CONFIG_CPU_MIPSR6)
 	.macro	local_irq_enable reg=t0
 	ei
 	irq_enable_hazard
@@ -54,7 +55,7 @@
 	di
 	irq_disable_hazard
 	.endm
-#else
+#else /* !CONFIG_CPU_MIPSR2 && !CONFIG_CPU_MIPSR5 && !CONFIG_CPU_MIPSR6 */
 	.macro	local_irq_enable reg=t0
 	mfc0	\reg, CP0_STATUS
 	ori	\reg, \reg, 1
@@ -79,7 +80,7 @@
 	sw      \reg, TI_PRE_COUNT($28)
 #endif
 	.endm
-#endif /* CONFIG_CPU_MIPSR2 */
+#endif  /* !CONFIG_CPU_MIPSR2 && !CONFIG_CPU_MIPSR5 && !CONFIG_CPU_MIPSR6 */
 
 	.macro	fpu_save_16even thread tmp=t0
 	.set	push
@@ -131,7 +132,7 @@
 
 	.macro	fpu_save_double thread status tmp
 #if defined(CONFIG_64BIT) || defined(CONFIG_CPU_MIPSR2) || \
-		defined(CONFIG_CPU_MIPSR6)
+    defined(CONFIG_CPU_MIPSR5) || defined(CONFIG_CPU_MIPSR6)
 	sll	\tmp, \status, 5
 	bgez	\tmp, 10f
 	fpu_save_16odd \thread
@@ -190,7 +191,7 @@
 
 	.macro	fpu_restore_double thread status tmp
 #if defined(CONFIG_64BIT) || defined(CONFIG_CPU_MIPSR2) || \
-		defined(CONFIG_CPU_MIPSR6)
+    defined(CONFIG_CPU_MIPSR5) || defined(CONFIG_CPU_MIPSR6)
 	sll	\tmp, \status, 5
 	bgez	\tmp, 10f				# 16 register mode?
 
@@ -200,16 +201,17 @@
 	fpu_restore_16even \thread \tmp
 	.endm
 
-#if defined(CONFIG_CPU_MIPSR2) || defined(CONFIG_CPU_MIPSR6)
+#if defined(CONFIG_CPU_MIPSR2) || defined(CONFIG_CPU_MIPSR5) || \
+    defined(CONFIG_CPU_MIPSR6)
 	.macro	_EXT	rd, rs, p, s
 	ext	\rd, \rs, \p, \s
 	.endm
-#else /* !CONFIG_CPU_MIPSR2 || !CONFIG_CPU_MIPSR6 */
+#else /* !CONFIG_CPU_MIPSR2 && !CONFIG_CPU_MIPSR5 && !CONFIG_CPU_MIPSR6 */
 	.macro	_EXT	rd, rs, p, s
 	srl	\rd, \rs, \p
 	andi	\rd, \rd, (1 << \s) - 1
 	.endm
-#endif /* !CONFIG_CPU_MIPSR2 || !CONFIG_CPU_MIPSR6 */
+#endif /* !CONFIG_CPU_MIPSR2 && !CONFIG_CPU_MIPSR5 && !CONFIG_CPU_MIPSR6 */
 
 /*
  * Temporary until all gas have MT ASE support
arch/mips/include/asm/compiler.h +5 -0
index f77e99f1722e..a2cb2d2b1c07 100644
@@ -57,6 +57,11 @@
 #define MIPS_ISA_ARCH_LEVEL MIPS_ISA_LEVEL
 #define MIPS_ISA_LEVEL_RAW mips64r6
 #define MIPS_ISA_ARCH_LEVEL_RAW MIPS_ISA_LEVEL_RAW
+#elif defined(CONFIG_CPU_MIPSR5)
+#define MIPS_ISA_LEVEL "mips64r5"
+#define MIPS_ISA_ARCH_LEVEL MIPS_ISA_LEVEL
+#define MIPS_ISA_LEVEL_RAW mips64r5
+#define MIPS_ISA_ARCH_LEVEL_RAW MIPS_ISA_LEVEL_RAW
 #else
 /* MIPS64 is a superset of MIPS32 */
 #define MIPS_ISA_LEVEL "mips64r2"
arch/mips/include/asm/cpu-features.h +14 -6
index de44c92b1c1f..e2f31bd6363b 100644
@@ -284,6 +284,9 @@
 #ifndef cpu_has_mips32r2
 # define cpu_has_mips32r2	__isa_range_or_flag(2, 6, MIPS_CPU_ISA_M32R2)
 #endif
+#ifndef cpu_has_mips32r5
+# define cpu_has_mips32r5	__isa_range_or_flag(5, 6, MIPS_CPU_ISA_M32R5)
+#endif
 #ifndef cpu_has_mips32r6
 # define cpu_has_mips32r6	__isa_ge_or_flag(6, MIPS_CPU_ISA_M32R6)
 #endif
@@ -313,19 +316,24 @@
 				(cpu_has_mips_3 | cpu_has_mips_4_5_64_r2_r6)
 #define cpu_has_mips_4_5_64_r2_r6					\
 				(cpu_has_mips_4_5 | cpu_has_mips64r1 |	\
-				 cpu_has_mips_r2 | cpu_has_mips_r6)
+				 cpu_has_mips_r2 | cpu_has_mips_r5 | \
+				 cpu_has_mips_r6)
 
-#define cpu_has_mips32	(cpu_has_mips32r1 | cpu_has_mips32r2 | cpu_has_mips32r6)
+#define cpu_has_mips32	(cpu_has_mips32r1 | cpu_has_mips32r2 | \
+			 cpu_has_mips32r5 | cpu_has_mips32r6)
 #define cpu_has_mips64	(cpu_has_mips64r1 | cpu_has_mips64r2 | cpu_has_mips64r6)
 #define cpu_has_mips_r1 (cpu_has_mips32r1 | cpu_has_mips64r1)
 #define cpu_has_mips_r2 (cpu_has_mips32r2 | cpu_has_mips64r2)
+#define cpu_has_mips_r5	(cpu_has_mips32r5)
 #define cpu_has_mips_r6	(cpu_has_mips32r6 | cpu_has_mips64r6)
 #define cpu_has_mips_r	(cpu_has_mips32r1 | cpu_has_mips32r2 | \
-			 cpu_has_mips32r6 | cpu_has_mips64r1 | \
-			 cpu_has_mips64r2 | cpu_has_mips64r6)
+			 cpu_has_mips32r5 | cpu_has_mips32r6 | \
+			 cpu_has_mips64r1 | cpu_has_mips64r2 | \
+			 cpu_has_mips64r6)
 
-/* MIPSR2 and MIPSR6 have a lot of similarities */
-#define cpu_has_mips_r2_r6	(cpu_has_mips_r2 | cpu_has_mips_r6)
+/* MIPSR2 - MIPSR6 have a lot of similarities */
+#define cpu_has_mips_r2_r6	(cpu_has_mips_r2 | cpu_has_mips_r5 | \
+				 cpu_has_mips_r6)
 
 /*
  * cpu_has_mips_r2_exec_hazard - return if IHB is required on current processor
arch/mips/include/asm/cpu-info.h +1 -1
index ed7ffe4e63a3..bce3ea7fff7c 100644
@@ -142,7 +142,7 @@ struct proc_cpuinfo_notifier_args {
 static inline unsigned int cpu_cluster(struct cpuinfo_mips *cpuinfo)
 {
 	/* Optimisation for systems where multiple clusters aren't used */
-	if (!IS_ENABLED(CONFIG_CPU_MIPSR6))
+	if (!IS_ENABLED(CONFIG_CPU_MIPSR5) && !IS_ENABLED(CONFIG_CPU_MIPSR6))
 		return 0;
 
 	return (cpuinfo->globalnumber & MIPS_GLOBALNUMBER_CLUSTER) >>
arch/mips/include/asm/cpu-type.h +5 -1
index 49f0061a6051..11090bc20464 100644
@@ -51,11 +51,15 @@ static inline int __pure __get_cpu_type(const int cpu_type)
 	case CPU_M14KEC:
 	case CPU_INTERAPTIV:
 	case CPU_PROAPTIV:
-	case CPU_P5600:
+#endif
+
+#ifdef CONFIG_SYS_HAS_CPU_MIPS32_R5
 	case CPU_M5150:
+	case CPU_P5600:
 #endif
 
 #if defined(CONFIG_SYS_HAS_CPU_MIPS32_R2) || \
+    defined(CONFIG_SYS_HAS_CPU_MIPS32_R5) || \
     defined(CONFIG_SYS_HAS_CPU_MIPS32_R6) || \
     defined(CONFIG_SYS_HAS_CPU_MIPS64_R2) || \
     defined(CONFIG_SYS_HAS_CPU_MIPS64_R6)
arch/mips/include/asm/cpu.h +4 -3
index 216a22916740..9bae99b568c9 100644
@@ -343,11 +343,12 @@ enum cpu_type_enum {
 #define MIPS_CPU_ISA_M32R2	0x00000020
 #define MIPS_CPU_ISA_M64R1	0x00000040
 #define MIPS_CPU_ISA_M64R2	0x00000080
-#define MIPS_CPU_ISA_M32R6	0x00000100
-#define MIPS_CPU_ISA_M64R6	0x00000200
+#define MIPS_CPU_ISA_M32R5	0x00000100
+#define MIPS_CPU_ISA_M32R6	0x00000400
+#define MIPS_CPU_ISA_M64R6	0x00000800
 
 #define MIPS_CPU_ISA_32BIT (MIPS_CPU_ISA_II | MIPS_CPU_ISA_M32R1 | \
-	MIPS_CPU_ISA_M32R2 | MIPS_CPU_ISA_M32R6)
+	MIPS_CPU_ISA_M32R2 | MIPS_CPU_ISA_M32R5 | MIPS_CPU_ISA_M32R6)
 #define MIPS_CPU_ISA_64BIT (MIPS_CPU_ISA_III | MIPS_CPU_ISA_IV | \
 	MIPS_CPU_ISA_V | MIPS_CPU_ISA_M64R1 | MIPS_CPU_ISA_M64R2 | \
 	MIPS_CPU_ISA_M64R6)
arch/mips/include/asm/fpu.h +2 -2
index 9476e0498d59..f0b37663fade 100644
@@ -71,8 +71,8 @@ static inline int __enable_fpu(enum fpu_mode mode)
 		goto fr_common;
 
 	case FPU_64BIT:
-#if !(defined(CONFIG_CPU_MIPSR2) || defined(CONFIG_CPU_MIPSR6) \
-      || defined(CONFIG_64BIT))
+#if !(defined(CONFIG_CPU_MIPSR2) || defined(CONFIG_CPU_MIPSR5) || \
+      defined(CONFIG_CPU_MIPSR6) || defined(CONFIG_64BIT))
 		/* we only have a 32-bit FPU */
 		return SIGFPE;
 #endif
arch/mips/include/asm/hazards.h +5 -3
index a0b92205f933..f855478d12fa 100644
@@ -22,8 +22,9 @@
 /*
  * TLB hazards
  */
-#if (defined(CONFIG_CPU_MIPSR2) || defined(CONFIG_CPU_MIPSR6)) && \
-	!defined(CONFIG_CPU_CAVIUM_OCTEON) && !defined(CONFIG_CPU_LOONGSON64)
+#if (defined(CONFIG_CPU_MIPSR2) || defined(CONFIG_CPU_MIPSR5) || \
+     defined(CONFIG_CPU_MIPSR6)) && \
+    !defined(CONFIG_CPU_CAVIUM_OCTEON) && !defined(CONFIG_CPU_LOONGSON64)
 
 /*
  * MIPSR2 defines ehb for hazard avoidance
@@ -278,7 +279,8 @@ do {									\
 
 #define __disable_fpu_hazard
 
-#elif defined(CONFIG_CPU_MIPSR2) || defined(CONFIG_CPU_MIPSR6)
+#elif defined(CONFIG_CPU_MIPSR2) || defined(CONFIG_CPU_MIPSR5) || \
+      defined(CONFIG_CPU_MIPSR6)
 
 #define __enable_fpu_hazard						\
 	___ehb
arch/mips/include/asm/stackframe.h +1 -1
index 4d6ad907ae54..3e8d2aaf96af 100644
@@ -424,7 +424,7 @@
 
 		.macro	RESTORE_SP_AND_RET docfi=0
 		RESTORE_SP \docfi
-#ifdef CONFIG_CPU_MIPSR6
+#if defined(CONFIG_CPU_MIPSR5) || defined(CONFIG_CPU_MIPSR6)
 		eretnc
 #else
 		.set	push
arch/mips/include/asm/switch_to.h +4 -4
index 09cbe9042828..0b0a93bf83cd 100644
@@ -67,11 +67,11 @@ do {									\
 #endif
 
 /*
- * Clear LLBit during context switches on MIPSr6 such that eretnc can be used
+ * Clear LLBit during context switches on MIPSr5+ such that eretnc can be used
  * unconditionally when returning to userland in entry.S.
  */
-#define __clear_r6_hw_ll_bit() do {					\
-	if (cpu_has_mips_r6)						\
+#define __clear_r5_hw_ll_bit() do {					\
+	if (cpu_has_mips_r5 || cpu_has_mips_r6)				\
 		write_c0_lladdr(0);					\
 } while (0)
 
@@ -129,7 +129,7 @@ do {									\
 		}							\
 		clear_c0_status(ST0_CU2);				\
 	}								\
-	__clear_r6_hw_ll_bit();						\
+	__clear_r5_hw_ll_bit();						\
 	__clear_software_ll_bit();					\
 	if (cpu_has_userlocal)						\
 		write_c0_userlocal(task_thread_info(next)->tp_value);	\
arch/mips/include/asm/vermagic.h +2 -0
index 24dc3d35161c..5a0e739f597a 100644
@@ -8,6 +8,8 @@
 #define MODULE_PROC_FAMILY "MIPS32_R1 "
 #elif defined CONFIG_CPU_MIPS32_R2
 #define MODULE_PROC_FAMILY "MIPS32_R2 "
+#elif defined CONFIG_CPU_MIPS32_R5
+#define MODULE_PROC_FAMILY "MIPS32_R5 "
 #elif defined CONFIG_CPU_MIPS32_R6
 #define MODULE_PROC_FAMILY "MIPS32_R6 "
 #elif defined CONFIG_CPU_MIPS64_R1
arch/mips/kernel/cpu-probe.c +12 -0
index f21a2304401f..a2dafef2df45 100644
@@ -92,6 +92,7 @@ static void cpu_set_fpu_2008(struct cpuinfo_mips *c)
 {
 	if (c->isa_level & (MIPS_CPU_ISA_M32R1 | MIPS_CPU_ISA_M64R1 |
 			    MIPS_CPU_ISA_M32R2 | MIPS_CPU_ISA_M64R2 |
+			    MIPS_CPU_ISA_M32R5 |
 			    MIPS_CPU_ISA_M32R6 | MIPS_CPU_ISA_M64R6)) {
 		unsigned long sr, fir, fcsr, fcsr0, fcsr1;
 
@@ -172,6 +173,7 @@ static void cpu_set_nofpu_2008(struct cpuinfo_mips *c)
 	case STRICT:
 		if (c->isa_level & (MIPS_CPU_ISA_M32R1 | MIPS_CPU_ISA_M64R1 |
 				    MIPS_CPU_ISA_M32R2 | MIPS_CPU_ISA_M64R2 |
+				    MIPS_CPU_ISA_M32R5 |
 				    MIPS_CPU_ISA_M32R6 | MIPS_CPU_ISA_M64R6)) {
 			c->options |= MIPS_CPU_NAN_2008 | MIPS_CPU_NAN_LEGACY;
 		} else {
@@ -263,9 +265,11 @@ static void cpu_set_nofpu_id(struct cpuinfo_mips *c)
 	value = 0;
 	if (c->isa_level & (MIPS_CPU_ISA_M32R1 | MIPS_CPU_ISA_M64R1 |
 			    MIPS_CPU_ISA_M32R2 | MIPS_CPU_ISA_M64R2 |
+			    MIPS_CPU_ISA_M32R5 |
 			    MIPS_CPU_ISA_M32R6 | MIPS_CPU_ISA_M64R6))
 		value |= MIPS_FPIR_D | MIPS_FPIR_S;
 	if (c->isa_level & (MIPS_CPU_ISA_M32R2 | MIPS_CPU_ISA_M64R2 |
+			    MIPS_CPU_ISA_M32R5 |
 			    MIPS_CPU_ISA_M32R6 | MIPS_CPU_ISA_M64R6))
 		value |= MIPS_FPIR_F64 | MIPS_FPIR_L | MIPS_FPIR_W;
 	if (c->options & MIPS_CPU_NAN_2008)
@@ -286,6 +290,7 @@ static void cpu_set_fpu_opts(struct cpuinfo_mips *c)
 
 	if (c->isa_level & (MIPS_CPU_ISA_M32R1 | MIPS_CPU_ISA_M64R1 |
 			    MIPS_CPU_ISA_M32R2 | MIPS_CPU_ISA_M64R2 |
+			    MIPS_CPU_ISA_M32R5 |
 			    MIPS_CPU_ISA_M32R6 | MIPS_CPU_ISA_M64R6)) {
 		if (c->fpu_id & MIPS_FPIR_3D)
 			c->ases |= MIPS_ASE_MIPS3D;
@@ -563,6 +568,9 @@ static void set_isa(struct cpuinfo_mips *c, unsigned int isa)
 		set_elf_base_platform("mips32r6");
 		/* Break here so we don't add incompatible ISAs */
 		break;
+	case MIPS_CPU_ISA_M32R5:
+		c->isa_level |= MIPS_CPU_ISA_M32R5;
+		/* fall through */
 	case MIPS_CPU_ISA_M32R2:
 		c->isa_level |= MIPS_CPU_ISA_M32R2;
 		set_elf_base_platform("mips32r2");
@@ -1751,6 +1759,10 @@ static inline void cpu_probe_mips(struct cpuinfo_mips *c, unsigned int cpu)
 	spram_config();
 
 	switch (__get_cpu_type(c->cputype)) {
+	case CPU_M5150:
+	case CPU_P5600:
+		set_isa(c, MIPS_CPU_ISA_M32R5);
+		break;
 	case CPU_I6500:
 		c->options |= MIPS_CPU_SHARED_FTLB_ENTRIES;
 		/* fall-through */
arch/mips/kernel/entry.S +3 -3
index 4849a48afc0f..4b896f5023ff 100644
@@ -169,8 +169,8 @@ syscall_exit_work:
 	jal	syscall_trace_leave
 	b	resume_userspace
 
-#if defined(CONFIG_CPU_MIPSR2) || defined(CONFIG_CPU_MIPSR6) || \
-    defined(CONFIG_MIPS_MT)
+#if defined(CONFIG_CPU_MIPSR2) || defined(CONFIG_CPU_MIPSR5) || \
+    defined(CONFIG_CPU_MIPSR6) || defined(CONFIG_MIPS_MT)
 
 /*
  * MIPS32R2 Instruction Hazard Barrier - must be called
@@ -183,4 +183,4 @@ LEAF(mips_ihb)
 	nop
 	END(mips_ihb)
 
-#endif /* CONFIG_CPU_MIPSR2 or CONFIG_CPU_MIPSR6 or CONFIG_MIPS_MT */
+#endif /* CONFIG_CPU_MIPSR2 - CONFIG_CPU_MIPSR6 or CONFIG_MIPS_MT */
arch/mips/kernel/proc.c +2 -0
index f8d36710cd58..b0ffb968f07d 100644
@@ -98,6 +98,8 @@ static int show_cpuinfo(struct seq_file *m, void *v)
 		seq_printf(m, "%s", " mips32r1");
 	if (cpu_has_mips32r2)
 		seq_printf(m, "%s", " mips32r2");
+	if (cpu_has_mips32r5)
+		seq_printf(m, "%s", " mips32r5");
 	if (cpu_has_mips32r6)
 		seq_printf(m, "%s", " mips32r6");
 	if (cpu_has_mips64r1)
arch/mips/kernel/r4k_fpu.S +7 -7
index 59be5c812aa2..b91e91106475 100644
@@ -41,7 +41,7 @@
 LEAF(_save_fp)
 EXPORT_SYMBOL(_save_fp)
 #if defined(CONFIG_64BIT) || defined(CONFIG_CPU_MIPSR2) || \
-		defined(CONFIG_CPU_MIPSR6)
+    defined(CONFIG_CPU_MIPSR5) || defined(CONFIG_CPU_MIPSR6)
 	mfc0	t0, CP0_STATUS
 #endif
 	fpu_save_double a0 t0 t1		# clobbers t1
@@ -53,7 +53,7 @@ EXPORT_SYMBOL(_save_fp)
  */
 LEAF(_restore_fp)
 #if defined(CONFIG_64BIT) || defined(CONFIG_CPU_MIPSR2) || \
-		defined(CONFIG_CPU_MIPSR6)
+    defined(CONFIG_CPU_MIPSR5) || defined(CONFIG_CPU_MIPSR6)
 	mfc0	t0, CP0_STATUS
 #endif
 	fpu_restore_double a0 t0 t1		# clobbers t1
@@ -103,10 +103,10 @@ LEAF(_save_fp_context)
 	.set	pop
 
 #if defined(CONFIG_64BIT) || defined(CONFIG_CPU_MIPSR2) || \
-		defined(CONFIG_CPU_MIPSR6)
+    defined(CONFIG_CPU_MIPSR5) || defined(CONFIG_CPU_MIPSR6)
 	.set	push
 	SET_HARDFLOAT
-#ifdef CONFIG_CPU_MIPSR2
+#if defined(CONFIG_CPU_MIPSR2) || defined(CONFIG_CPU_MIPSR5)
 	.set	mips32r2
 	.set	fp=64
 	mfc0	t0, CP0_STATUS
@@ -170,11 +170,11 @@ LEAF(_save_fp_context)
 LEAF(_restore_fp_context)
 	EX	lw t1, 0(a1)
 
-#if defined(CONFIG_64BIT) || defined(CONFIG_CPU_MIPSR2)  || \
-		defined(CONFIG_CPU_MIPSR6)
+#if defined(CONFIG_64BIT) || defined(CONFIG_CPU_MIPSR2) || \
+    defined(CONFIG_CPU_MIPSR5) || defined(CONFIG_CPU_MIPSR6)
 	.set	push
 	SET_HARDFLOAT
-#ifdef CONFIG_CPU_MIPSR2
+#if defined(CONFIG_CPU_MIPSR2) || defined(CONFIG_CPU_MIPSR5)
 	.set	mips32r2
 	.set	fp=64
 	mfc0	t0, CP0_STATUS
arch/mips/kvm/vz.c +3 -3
index dde20887a70d..66432b5ab229 100644
@@ -2980,7 +2980,7 @@ static int kvm_vz_vcpu_setup(struct kvm_vcpu *vcpu)
 	 */
 
 	/* PageGrain */
-	if (cpu_has_mips_r6)
+	if (cpu_has_mips_r5 || cpu_has_mips_r6)
 		kvm_write_sw_gc0_pagegrain(cop0, PG_RIE | PG_XIE | PG_IEC);
 	/* Wired */
 	if (cpu_has_mips_r6)
@@ -2988,7 +2988,7 @@ static int kvm_vz_vcpu_setup(struct kvm_vcpu *vcpu)
 				       read_gc0_wired() & MIPSR6_WIRED_LIMIT);
 	/* Status */
 	kvm_write_sw_gc0_status(cop0, ST0_BEV | ST0_ERL);
-	if (cpu_has_mips_r6)
+	if (cpu_has_mips_r5 || cpu_has_mips_r6)
 		kvm_change_sw_gc0_status(cop0, ST0_FR, read_gc0_status());
 	/* IntCtl */
 	kvm_write_sw_gc0_intctl(cop0, read_gc0_intctl() &
@@ -3086,7 +3086,7 @@ static int kvm_vz_vcpu_setup(struct kvm_vcpu *vcpu)
 	}
 
 	/* reset HTW registers */
-	if (cpu_guest_has_htw && cpu_has_mips_r6) {
+	if (cpu_guest_has_htw && (cpu_has_mips_r5 || cpu_has_mips_r6)) {
 		/* PWField */
 		kvm_write_sw_gc0_pwfield(cop0, 0x0c30c302);
 		/* PWSize */
arch/mips/lib/csum_partial.S +4 -2
index fda7b57b826e..87fda0713b84 100644
@@ -279,7 +279,8 @@ EXPORT_SYMBOL(csum_partial)
 #endif
 
 	/* odd buffer alignment? */
-#if defined(CONFIG_CPU_MIPSR2) || defined(CONFIG_CPU_LOONGSON64)
+#if defined(CONFIG_CPU_MIPSR2) || defined(CONFIG_CPU_MIPSR5) || \
+    defined(CONFIG_CPU_LOONGSON64)
 	.set	push
 	.set	arch=mips32r2
 	wsbh	v1, sum
@@ -732,7 +733,8 @@ EXPORT_SYMBOL(csum_partial)
 	addu	sum, v1
 #endif
 
-#if defined(CONFIG_CPU_MIPSR2) || defined(CONFIG_CPU_LOONGSON64)
+#if defined(CONFIG_CPU_MIPSR2) || defined(CONFIG_CPU_MIPSR5) || \
+    defined(CONFIG_CPU_LOONGSON64)
 	.set	push
 	.set	arch=mips32r2
 	wsbh	v1, sum
arch/mips/mm/c-r4k.c +4 -3
index 36a311348739..99146abfbbca 100644
@@ -1703,9 +1703,10 @@ static void setup_scache(void)
 		return;
 
 	default:
-		if (c->isa_level & (MIPS_CPU_ISA_M32R1 | MIPS_CPU_ISA_M32R2 |
-				    MIPS_CPU_ISA_M32R6 | MIPS_CPU_ISA_M64R1 |
-				    MIPS_CPU_ISA_M64R2 | MIPS_CPU_ISA_M64R6)) {
+		if (c->isa_level & (MIPS_CPU_ISA_M32R1 | MIPS_CPU_ISA_M64R1 |
+				    MIPS_CPU_ISA_M32R2 | MIPS_CPU_ISA_M64R2 |
+				    MIPS_CPU_ISA_M32R5 |
+				    MIPS_CPU_ISA_M32R6 | MIPS_CPU_ISA_M64R6)) {
 #ifdef CONFIG_MIPS_CPU_SCACHE
 			if (mips_sc_init ()) {
 				scache_size = c->scache.ways * c->scache.sets * c->scache.linesz;
arch/mips/mm/sc-mips.c +4 -3
index dbdbfe5d8408..6fae3014ad3e 100644
@@ -194,9 +194,10 @@ static inline int __init mips_sc_probe(void)
 		return mips_sc_probe_cm3();
 
 	/* Ignore anything but MIPSxx processors */
-	if (!(c->isa_level & (MIPS_CPU_ISA_M32R1 | MIPS_CPU_ISA_M32R2 |
-			      MIPS_CPU_ISA_M32R6 | MIPS_CPU_ISA_M64R1 |
-			      MIPS_CPU_ISA_M64R2 | MIPS_CPU_ISA_M64R6)))
+	if (!(c->isa_level & (MIPS_CPU_ISA_M32R1 | MIPS_CPU_ISA_M64R1 |
+			      MIPS_CPU_ISA_M32R2 | MIPS_CPU_ISA_M64R2 |
+			      MIPS_CPU_ISA_M32R5 |
+			      MIPS_CPU_ISA_M32R6 | MIPS_CPU_ISA_M64R6)))
 		return 0;
 
 	/* Does this MIPS32/MIPS64 CPU have a config2 register? */
Signature · 1 line
2.25.1
#22 lore raw

Re: [PATCH v2 06/20] mips: Add MIPS32 Release 5 support

Quoted 3 lines · On Wed, May 06, 2020 at 08:42:24PM +0300, [email protected]…

We also could add CPU_MIPS64_R5 config support here, but I don't think it's necessary at the moment seeing there is no any real chip ever produced with that arch. Right?

how much is missing ? Looks like not too much, so it might be worth to add it at least for symmetry to the other ISAs...

Thomas.

Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]
#23 lore raw

Re: [PATCH v2 06/20] mips: Add MIPS32 Release 5 support

Quoted 7 lines · Thomas Bogendoerfer
Quoted 3 lines · On Wed, May 06, 2020 at 08:42:24PM +0300, [email protected]…

We also could add CPU_MIPS64_R5 config support here, but I don't think it's necessary at the moment seeing there is no any real chip ever produced with that arch. Right?

how much is missing ? Looks like not too much, so it might be worth to add it at least for symmetry to the other ISAs...

Yeah, just a few more alteration to add together with new CPU_MIPS64_R5 config. I'll do this in v3.

-Sergey

Quoted 6 lines

Thomas.

Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]
#24 lore raw

PATCH v2 7/20 mips: Add MIPS Warrior P5600 support

In reply to [email protected]

From: Serge Semin <[email protected]>

This is a MIPS32 Release 5 based IP core with XPA, EVA, dual/quad issue exec pipes, MMU with two-levels TLB, UCA, MSA, MDU core level features and system level features like up to six P5600 calculation cores, CM2 with L2 cache, IOCU/IOMMU (though might be unused depending on the system-specific IP core configuration), GIC, CPC, virtualisation module, eJTAG and PDtrace.

As being MIPS32 Release 5 based core it provides all the features available by the CPU_MIPS32_R5 config, while adding a few more like UCA attribute support, availability of CPU-freq (by means of L2/CM clock ratio setting), EI/VI GIC modes detection at runtime.

In addition to this if P5600 architecture is enabled modern GNU GCC provides a specific tuning for P5600 processors with respect to the classic MIPS32 Release 5. First of all branch-likely avoidance is activated only when the code is compiled with the speed optimization (avoidance is always enabled for the pure MIPS32 Release 5 architecture). Secondly the madd/msub avoidance is enabled since madd/msub utilization isn't profitable due to overhead of getting the result out of the HI/LO registers. Multiply-accumulate instructions are activated and utilized together with the necessary code reorder when multiply-add/multiply-subtract statements are met. Finally load/store bonding is activated by default. All of these optimizations may make the code relatively faster than if just MIP32 release 5 architecture was requested.

---
 arch/mips/Kconfig                | 38 +++++++++++++++++++++++++++-----
 arch/mips/Makefile               |  1 +
 arch/mips/include/asm/vermagic.h |  2 ++
 3 files changed, 36 insertions(+), 5 deletions(-)
arch/mips/Kconfig +33 -5
index 55c3dbfea336..e3b780a389a9 100644
@@ -1665,6 +1665,28 @@ config CPU_MIPS64_R6
 	  family, are based on a MIPS64r6 processor. If you own an older
 	  processor, you probably need to select MIPS64r1 or MIPS64r2 instead.
 
+config CPU_P5600
+	bool "MIPS Warrior P5600"
+	depends on SYS_HAS_CPU_P5600
+	select CPU_HAS_PREFETCH
+	select CPU_SUPPORTS_32BIT_KERNEL
+	select CPU_SUPPORTS_HIGHMEM
+	select CPU_SUPPORTS_MSA
+	select CPU_SUPPORTS_UNCACHED_ACCELERATED
+	select CPU_SUPPORTS_CPUFREQ
+	select CPU_MIPSR2_IRQ_VI
+	select CPU_MIPSR2_IRQ_EI
+	select HAVE_KVM
+	select MIPS_O32_FP64_SUPPORT
+	help
+	  Choose this option to build a kernel for MIPS Warrior P5600 CPU.
+	  It's based on MIPS32r5 ISA with XPA, EVA, dual/quad issue exec pipes,
+	  MMU with two-levels TLB, UCA, MSA, MDU core level features and system
+	  level features like up to six P5600 calculation cores, CM2 with L2
+	  cache, IOCU/IOMMU (though might be unused depending on the system-
+	  specific IP core configuration), GIC, CPC, virtualisation module,
+	  eJTAG and PDtrace.
+
 config CPU_R3000
 	bool "R3000"
 	depends on SYS_HAS_CPU_R3000
@@ -1841,7 +1863,8 @@ endchoice
 config CPU_MIPS32_3_5_FEATURES
 	bool "MIPS32 Release 3.5 Features"
 	depends on SYS_HAS_CPU_MIPS32_R3_5
-	depends on CPU_MIPS32_R2 || CPU_MIPS32_R5 || CPU_MIPS32_R6
+	depends on CPU_MIPS32_R2 || CPU_MIPS32_R5 || CPU_MIPS32_R6 || \
+		   CPU_P5600
 	help
 	  Choose this option to build a kernel for release 2 or later of the
 	  MIPS32 architecture including features from the 3.5 release such as
@@ -1861,7 +1884,7 @@ config CPU_MIPS32_3_5_EVA
 config CPU_MIPS32_R5_FEATURES
 	bool "MIPS32 Release 5 Features"
 	depends on SYS_HAS_CPU_MIPS32_R5
-	depends on CPU_MIPS32_R2 || CPU_MIPS32_R5
+	depends on CPU_MIPS32_R2 || CPU_MIPS32_R5 || CPU_P5600
 	help
 	  Choose this option to build a kernel for release 2 or later of the
 	  MIPS32 architecture including features from release 5 such as
@@ -2016,6 +2039,10 @@ config SYS_HAS_CPU_MIPS64_R6
 	bool
 	select ARCH_HAS_SYNC_DMA_FOR_CPU if DMA_NONCOHERENT
 
+config SYS_HAS_CPU_P5600
+	bool
+	select ARCH_HAS_SYNC_DMA_FOR_CPU if DMA_NONCOHERENT
+
 config SYS_HAS_CPU_R3000
 	bool
 
@@ -2100,7 +2127,7 @@ endmenu
 config CPU_MIPS32
 	bool
 	default y if CPU_MIPS32_R1 || CPU_MIPS32_R2 || CPU_MIPS32_R5 || \
-		     CPU_MIPS32_R6
+		     CPU_MIPS32_R6 || CPU_P5600
 
 config CPU_MIPS64
 	bool
@@ -2122,7 +2149,7 @@ config CPU_MIPSR2
 
 config CPU_MIPSR5
 	bool
-	default y if CPU_MIPS32_R5
+	default y if CPU_MIPS32_R5 || CPU_P5600
 	select CPU_HAS_RIXI
 	select CPU_HAS_DIEI if !CPU_DIEI_BROKEN
 	select MIPS_SPRAM
@@ -2733,7 +2760,8 @@ config RELOCATABLE
 	bool "Relocatable kernel"
 	depends on SYS_SUPPORTS_RELOCATABLE
 	depends on CPU_MIPS32_R2 || CPU_MIPS64_R2 || CPU_MIPS32_R5 || \
-		   CPU_MIPS32_R6 || CPU_MIPS64_R6 || CAVIUM_OCTEON_SOC
+		   CPU_MIPS32_R6 || CPU_MIPS64_R6 || CPU_P5600 || \
+		   CAVIUM_OCTEON_SOC
 	help
 	  This builds a kernel image that retains relocation information
 	  so it can be loaded someplace besides the default 1MB.
arch/mips/Makefile +1 -0
index 9172fb0f630b..264dead560f4 100644
@@ -176,6 +176,7 @@ cflags-$(CONFIG_CPU_MIPS32_R6)	+= -march=mips32r6 -Wa,--trap -modd-spreg
 cflags-$(CONFIG_CPU_MIPS64_R1)	+= -march=mips64 -Wa,--trap
 cflags-$(CONFIG_CPU_MIPS64_R2)	+= -march=mips64r2 -Wa,--trap
 cflags-$(CONFIG_CPU_MIPS64_R6)	+= -march=mips64r6 -Wa,--trap
+cflags-$(CONFIG_CPU_P5600)	+= -march=p5600 -Wa,--trap -modd-spreg
 cflags-$(CONFIG_CPU_R5000)	+= -march=r5000 -Wa,--trap
 cflags-$(CONFIG_CPU_R5500)	+= $(call cc-option,-march=r5500,-march=r5000) \
 			-Wa,--trap
arch/mips/include/asm/vermagic.h +2 -0
index 5a0e739f597a..d03f97350f91 100644
@@ -48,6 +48,8 @@
 #define MODULE_PROC_FAMILY "LOONGSON64 "
 #elif defined CONFIG_CPU_CAVIUM_OCTEON
 #define MODULE_PROC_FAMILY "OCTEON "
+#elif defined CONFIG_CPU_P5600
+#define MODULE_PROC_FAMILY "P5600 "
 #elif defined CONFIG_CPU_XLR
 #define MODULE_PROC_FAMILY "XLR "
 #elif defined CONFIG_CPU_XLP
Signature · 1 line
2.25.1
#25 lore raw

Re: [PATCH v2 07/20] mips: Add MIPS Warrior P5600 support

Quoted 46 lines · On Wed, May 06, 2020 at 08:42:25PM +0300, [email protected]…
arch/mips/Kconfig +25 -2
+config CPU_P5600
+	bool "MIPS Warrior P5600"
+	depends on SYS_HAS_CPU_P5600
+	select CPU_HAS_PREFETCH
+	select CPU_SUPPORTS_32BIT_KERNEL
+	select CPU_SUPPORTS_HIGHMEM
+	select CPU_SUPPORTS_MSA
+	select CPU_SUPPORTS_UNCACHED_ACCELERATED
+	select CPU_SUPPORTS_CPUFREQ
+	select CPU_MIPSR2_IRQ_VI
+	select CPU_MIPSR2_IRQ_EI
+	select HAVE_KVM
+	select MIPS_O32_FP64_SUPPORT
+	help
+	  Choose this option to build a kernel for MIPS Warrior P5600 CPU.
+	  It's based on MIPS32r5 ISA with XPA, EVA, dual/quad issue exec pipes,
+	  MMU with two-levels TLB, UCA, MSA, MDU core level features and system
+	  level features like up to six P5600 calculation cores, CM2 with L2
+	  cache, IOCU/IOMMU (though might be unused depending on the system-
+	  specific IP core configuration), GIC, CPC, virtualisation module,
+	  eJTAG and PDtrace.
+
 config CPU_R3000
 	bool "R3000"
 	depends on SYS_HAS_CPU_R3000
@@ -1841,7 +1863,8 @@ endchoice
 config CPU_MIPS32_3_5_FEATURES
 	bool "MIPS32 Release 3.5 Features"
 	depends on SYS_HAS_CPU_MIPS32_R3_5
-	depends on CPU_MIPS32_R2 || CPU_MIPS32_R5 || CPU_MIPS32_R6
+	depends on CPU_MIPS32_R2 || CPU_MIPS32_R5 || CPU_MIPS32_R6 || \
+		   CPU_P5600
 	help
 	  Choose this option to build a kernel for release 2 or later of the
 	  MIPS32 architecture including features from the 3.5 release such as
@@ -1861,7 +1884,7 @@ config CPU_MIPS32_3_5_EVA
 config CPU_MIPS32_R5_FEATURES
 	bool "MIPS32 Release 5 Features"
 	depends on SYS_HAS_CPU_MIPS32_R5
-	depends on CPU_MIPS32_R2 || CPU_MIPS32_R5
+	depends on CPU_MIPS32_R2 || CPU_MIPS32_R5 || CPU_P5600
 	help
 	  Choose this option to build a kernel for release 2 or later of the
 	  MIPS32 architecture including features from release 5 such as

arch/mips/Kconfig config SYS_HAS_CPU_MIPS64_R6 · line 2045

Quoted 7 lines
arch/mips/Kconfig +4 -0
@@ -2016,6 +2039,10 @@ config SYS_HAS_CPU_MIPS64_R6
 	bool
 	select ARCH_HAS_SYNC_DMA_FOR_CPU if DMA_NONCOHERENT
 
+config SYS_HAS_CPU_P5600
+	bool
+	select ARCH_HAS_SYNC_DMA_FOR_CPU if DMA_NONCOHERENT
+

P5600 is CPU_MIPS_R5 so can't you select it here and drop all the || CPU_5600 above/below ?

Thomas.

Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]
#26 lore raw

Re: [PATCH v2 07/20] mips: Add MIPS Warrior P5600 support

Quoted 50 lines · Thomas Bogendoerfer
Quoted 49 lines · On Wed, May 06, 2020 at 08:42:25PM +0300, [email protected]…
arch/mips/Kconfig +25 -2
+config CPU_P5600
+	bool "MIPS Warrior P5600"
+	depends on SYS_HAS_CPU_P5600
+	select CPU_HAS_PREFETCH
+	select CPU_SUPPORTS_32BIT_KERNEL
+	select CPU_SUPPORTS_HIGHMEM
+	select CPU_SUPPORTS_MSA
+	select CPU_SUPPORTS_UNCACHED_ACCELERATED
+	select CPU_SUPPORTS_CPUFREQ
+	select CPU_MIPSR2_IRQ_VI
+	select CPU_MIPSR2_IRQ_EI
+	select HAVE_KVM
+	select MIPS_O32_FP64_SUPPORT
+	help
+	  Choose this option to build a kernel for MIPS Warrior P5600 CPU.
+	  It's based on MIPS32r5 ISA with XPA, EVA, dual/quad issue exec pipes,
+	  MMU with two-levels TLB, UCA, MSA, MDU core level features and system
+	  level features like up to six P5600 calculation cores, CM2 with L2
+	  cache, IOCU/IOMMU (though might be unused depending on the system-
+	  specific IP core configuration), GIC, CPC, virtualisation module,
+	  eJTAG and PDtrace.
+
 config CPU_R3000
 	bool "R3000"
 	depends on SYS_HAS_CPU_R3000
@@ -1841,7 +1863,8 @@ endchoice
 config CPU_MIPS32_3_5_FEATURES
 	bool "MIPS32 Release 3.5 Features"
 	depends on SYS_HAS_CPU_MIPS32_R3_5
-	depends on CPU_MIPS32_R2 || CPU_MIPS32_R5 || CPU_MIPS32_R6
+	depends on CPU_MIPS32_R2 || CPU_MIPS32_R5 || CPU_MIPS32_R6 || \
+		   CPU_P5600
 	help
 	  Choose this option to build a kernel for release 2 or later of the
 	  MIPS32 architecture including features from the 3.5 release such as
@@ -1861,7 +1884,7 @@ config CPU_MIPS32_3_5_EVA
 config CPU_MIPS32_R5_FEATURES
 	bool "MIPS32 Release 5 Features"
 	depends on SYS_HAS_CPU_MIPS32_R5
-	depends on CPU_MIPS32_R2 || CPU_MIPS32_R5
+	depends on CPU_MIPS32_R2 || CPU_MIPS32_R5 || CPU_P5600
 	help
 	  Choose this option to build a kernel for release 2 or later of the
 	  MIPS32 architecture including features from release 5 such as
@@ -2016,6 +2039,10 @@ config SYS_HAS_CPU_MIPS64_R6
 	bool
 	select ARCH_HAS_SYNC_DMA_FOR_CPU if DMA_NONCOHERENT
 
Quoted 7 lines

arch/mips/Kconfig config SYS_HAS_CPU_MIPS64_R6 · line 2045

Quoted 4 lines
arch/mips/Kconfig +4 -0
+config SYS_HAS_CPU_P5600
+	bool
+	select ARCH_HAS_SYNC_DMA_FOR_CPU if DMA_NONCOHERENT
+

P5600 is CPU_MIPS_R5 so can't you select it here and drop all the || CPU_5600 above/below ?

Alas, We can't do this so easy. CONFIG_CPU_MIPS32_{R2,R5,R6} and any other CONFIG_CPU_* configs is something that kernel config-file is supposed to select. Their availability is enabled by the CONFIG_SYS_HAS_CPU_* configs. CONFIG_CPU_* is supposed to activate CPU-specific features and there is only one CONFIG_CPU_x can be enabled at a time seeing it's a part of the "CPU type" choice kconfig menu. In addition the CPU config also tunes a compiler to activate the arch-specific ISA and optimizations in the arch/mips/Makefile by setting -march=cpu-name (where cpu-name can be p5600, mips32r5, etc).

Yes, P5600 is based on the MIPS32r5, but it also has got some specific features (see config CPU_P5600 and config MIPS32_R5), which makes it to be different from the ancestor. So In addition to the difficulties described above IMHO converting CPU_P5600 to a set of features activated on top of the CPU_MIPS32_R5 config would contradict the design of the CPU-support configs implemented in the MIPS arch subsystem.

-Sergey

Quoted 6 lines

Thomas.

Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]
#27 lore raw

Re: [PATCH v2 07/20] mips: Add MIPS Warrior P5600 support

Quoted 13 lines · Serge Semin
Quoted 2 lines · Thomas Bogendoerfer

P5600 is CPU_MIPS_R5 so can't you select it here and drop all the || CPU_5600 above/below ?

Alas, We can't do this so easy. CONFIG_CPU_MIPS32_{R2,R5,R6} and any other CONFIG_CPU_* configs is something that kernel config-file is supposed to select. Their availability is enabled by the CONFIG_SYS_HAS_CPU_* configs. CONFIG_CPU_* is supposed to activate CPU-specific features and there is only one CONFIG_CPU_x can be enabled at a time seeing it's a part of the "CPU type" choice kconfig menu. In addition the CPU config also tunes a compiler to activate the arch-specific ISA and optimizations in the arch/mips/Makefile by setting -march=cpu-name (where cpu-name can be p5600, mips32r5, etc).

Quoted 6 lines

Yes, P5600 is based on the MIPS32r5, but it also has got some specific features (see config CPU_P5600 and config MIPS32_R5), which makes it to be different from the ancestor. So In addition to the difficulties described above IMHO converting CPU_P5600 to a set of features activated on top of the CPU_MIPS32_R5 config would contradict the design of the CPU-support configs implemented in the MIPS arch subsystem.

maybe I wasn't clear enough, my suggestion is

use

config CPU_P5600
bool "MIPS Warrior P5600"
depends on SYS_HAS_CPU_P5600
select CPU_MIPS32_R5
select CPU_SUPPORTS_UNCACHED_ACCELERATED
select CPU_SUPPORTS_CPUFREQ
select CPU_MIPSR2_IRQ_VI
select CPU_MIPSR2_IRQ_EI

That way you don't need to any "|| CPU_P5600" where CPU_MIPS32_R5 is already there. Or are there cases, where this would be wrong ?

Thomas.

Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]
#28 lore raw

Re: [PATCH v2 07/20] mips: Add MIPS Warrior P5600 support

Quoted 34 lines · Thomas Bogendoerfer
Quoted 13 lines · Serge Semin
Quoted 2 lines · Thomas Bogendoerfer

P5600 is CPU_MIPS_R5 so can't you select it here and drop all the || CPU_5600 above/below ?

Alas, We can't do this so easy. CONFIG_CPU_MIPS32_{R2,R5,R6} and any other CONFIG_CPU_* configs is something that kernel config-file is supposed to select. Their availability is enabled by the CONFIG_SYS_HAS_CPU_* configs. CONFIG_CPU_* is supposed to activate CPU-specific features and there is only one CONFIG_CPU_x can be enabled at a time seeing it's a part of the "CPU type" choice kconfig menu. In addition the CPU config also tunes a compiler to activate the arch-specific ISA and optimizations in the arch/mips/Makefile by setting -march=cpu-name (where cpu-name can be p5600, mips32r5, etc).

Quoted 6 lines

Yes, P5600 is based on the MIPS32r5, but it also has got some specific features (see config CPU_P5600 and config MIPS32_R5), which makes it to be different from the ancestor. So In addition to the difficulties described above IMHO converting CPU_P5600 to a set of features activated on top of the CPU_MIPS32_R5 config would contradict the design of the CPU-support configs implemented in the MIPS arch subsystem.

maybe I wasn't clear enough, my suggestion is

use

config CPU_P5600
bool "MIPS Warrior P5600"
depends on SYS_HAS_CPU_P5600
select CPU_MIPS32_R5
select CPU_SUPPORTS_UNCACHED_ACCELERATED
select CPU_SUPPORTS_CPUFREQ
select CPU_MIPSR2_IRQ_VI
select CPU_MIPSR2_IRQ_EI

Quoted 2 lines

That way you don't need to any "|| CPU_P5600" where CPU_MIPS32_R5 is already there. Or are there cases, where this would be wrong ?

nevermind, this would also need a select SYS_HAS_CPU_MIPS32_R5, which isn't wanted here. So patch is fine as is.

Thomas.

Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]
#29 lore raw

Re: [PATCH v2 07/20] mips: Add MIPS Warrior P5600 support

Quoted 35 lines · Thomas Bogendoerfer
Quoted 34 lines · Thomas Bogendoerfer
Quoted 13 lines · Serge Semin
Quoted 2 lines · Thomas Bogendoerfer

P5600 is CPU_MIPS_R5 so can't you select it here and drop all the || CPU_5600 above/below ?

Alas, We can't do this so easy. CONFIG_CPU_MIPS32_{R2,R5,R6} and any other CONFIG_CPU_* configs is something that kernel config-file is supposed to select. Their availability is enabled by the CONFIG_SYS_HAS_CPU_* configs. CONFIG_CPU_* is supposed to activate CPU-specific features and there is only one CONFIG_CPU_x can be enabled at a time seeing it's a part of the "CPU type" choice kconfig menu. In addition the CPU config also tunes a compiler to activate the arch-specific ISA and optimizations in the arch/mips/Makefile by setting -march=cpu-name (where cpu-name can be p5600, mips32r5, etc).

Quoted 6 lines

Yes, P5600 is based on the MIPS32r5, but it also has got some specific features (see config CPU_P5600 and config MIPS32_R5), which makes it to be different from the ancestor. So In addition to the difficulties described above IMHO converting CPU_P5600 to a set of features activated on top of the CPU_MIPS32_R5 config would contradict the design of the CPU-support configs implemented in the MIPS arch subsystem.

maybe I wasn't clear enough, my suggestion is

use

config CPU_P5600
bool "MIPS Warrior P5600"
depends on SYS_HAS_CPU_P5600
select CPU_MIPS32_R5
select CPU_SUPPORTS_UNCACHED_ACCELERATED
select CPU_SUPPORTS_CPUFREQ
select CPU_MIPSR2_IRQ_VI
select CPU_MIPSR2_IRQ_EI

Quoted 5 lines
Quoted 2 lines

That way you don't need to any "|| CPU_P5600" where CPU_MIPS32_R5 is already there. Or are there cases, where this would be wrong ?

nevermind, this would also need a select SYS_HAS_CPU_MIPS32_R5, which isn't wanted here. So patch is fine as is.

Ok. Thanks.

-Sergey

Quoted 6 lines

Thomas.

Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]
#30 lore raw

PATCH v2 8/20 mips: Fix cpu_has_mips64r1/2 activation for MIPS32 CPUs

In reply to [email protected]

From: Serge Semin <[email protected]>

Commit 1aeba347b3a9 ("MIPS: Hardcode cpu_has_mips* where target ISA allows") updated the cpu_has_mips* macro to be replaced with a constant expression where it's possible. By mistake it wasn't done correctly for cpu_has_mips64r1/cpu_has_mips64r2 macro. They are defined to be replaced with conditional expression __isa_range_or_flag(), which means either ISA revision being within the range or the corresponding CPU options flag was set at the probe stage or both being true at the same time. But the ISA level value doesn't indicate whether the ISA is MIPS32 or MIPS64. Due to this if we select MIPS32r1 - MIPS32r5 architectures the __isa_range() macro will activate the cpu_has_mips64rX flags, which is incorrect. In order to fix the problem we added a new macro __isa_range_and_flag() and use it to define the cpu_has_mips64r1/cpu_has_mips64r2 flags.

Fixes 1aeba347b3a9 ("MIPS: Hardcode cpu_has_mips* where target ISA allows")
Cc Alexey Malahov <[email protected]>
Cc Thomas Bogendoerfer <[email protected]>
Cc Paul Burton <[email protected]>
Cc Ralf Baechle <[email protected]>
Cc Arnd Bergmann <[email protected]>
Cc Rob Herring <[email protected]>
---
 arch/mips/include/asm/cpu-features.h | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
arch/mips/include/asm/cpu-features.h +4 -2
index e2f31bd6363b..7e22b9c1e279 100644
@@ -64,6 +64,8 @@
 	((MIPS_ISA_REV >= (ge)) && (MIPS_ISA_REV < (lt)))
 #define __isa_range_or_flag(ge, lt, flag) \
 	(__isa_range(ge, lt) || ((MIPS_ISA_REV < (lt)) && __isa(flag)))
+#define __isa_range_and_flag(ge, lt, flag) \
+	(__isa_range(ge, lt) && __isa(flag))
 
 /*
  * SMP assumption: Options of CPU 0 are a superset of all processors.
@@ -291,10 +293,10 @@
 # define cpu_has_mips32r6	__isa_ge_or_flag(6, MIPS_CPU_ISA_M32R6)
 #endif
 #ifndef cpu_has_mips64r1
-# define cpu_has_mips64r1	__isa_range_or_flag(1, 6, MIPS_CPU_ISA_M64R1)
+# define cpu_has_mips64r1	__isa_range_and_flag(1, 6, MIPS_CPU_ISA_M64R1)
 #endif
 #ifndef cpu_has_mips64r2
-# define cpu_has_mips64r2	__isa_range_or_flag(2, 6, MIPS_CPU_ISA_M64R2)
+# define cpu_has_mips64r2	__isa_range_and_flag(2, 6, MIPS_CPU_ISA_M64R2)
 #endif
 #ifndef cpu_has_mips64r6
 # define cpu_has_mips64r6	__isa_ge_and_flag(6, MIPS_CPU_ISA_M64R6)
Signature · 1 line
2.25.1
#31 lore raw

Re: [PATCH v2 08/20] mips: Fix cpu_has_mips64r1/2 activation for MIPS32 CPUs

Quoted 15 lines · On Wed, May 06, 2020 at 08:42:26PM +0300, [email protected]…

From: Serge Semin <[email protected]>

arch/mips/include/asm/cpu-features.h +2 -0
index e2f31bd6363b..7e22b9c1e279 100644
@@ -64,6 +64,8 @@
 	((MIPS_ISA_REV >= (ge)) && (MIPS_ISA_REV < (lt)))
 #define __isa_range_or_flag(ge, lt, flag) \
 	(__isa_range(ge, lt) || ((MIPS_ISA_REV < (lt)) && __isa(flag)))
+#define __isa_range_and_flag(ge, lt, flag) \
+	(__isa_range(ge, lt) && __isa(flag))
 
 /*
  * SMP assumption: Options of CPU 0 are a superset of all processors.

arch/mips/include/asm/cpu-features.h line 296

Quoted 5 lines
arch/mips/include/asm/cpu-features.h +1 -1
@@ -291,10 +293,10 @@
 # define cpu_has_mips32r6	__isa_ge_or_flag(6, MIPS_CPU_ISA_M32R6)
 #endif
 #ifndef cpu_has_mips64r1
-# define cpu_has_mips64r1	__isa_range_or_flag(1, 6, MIPS_CPU_ISA_M64R1)
+# define cpu_has_mips64r1	__isa_range_and_flag(1, 6, MIPS_CPU_ISA_M64R1)

that's not the correct fix. You want to check for cpu_has_64bits here. Something like

# define cpu_has_mips64r1    (cpu_has_64bits && __isa_range_or_flag(1, 6, MIPS_CPU_ISA_M64R1))

should do the trick.

Thomas.

Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]
#32 lore raw

Re: [PATCH v2 08/20] mips: Fix cpu_has_mips64r1/2 activation for MIPS32 CPUs

Quoted 25 lines · Thomas Bogendoerfer
Quoted 15 lines · On Wed, May 06, 2020 at 08:42:26PM +0300, [email protected]…

From: Serge Semin <[email protected]>

arch/mips/include/asm/cpu-features.h +2 -0
index e2f31bd6363b..7e22b9c1e279 100644
@@ -64,6 +64,8 @@
 	((MIPS_ISA_REV >= (ge)) && (MIPS_ISA_REV < (lt)))
 #define __isa_range_or_flag(ge, lt, flag) \
 	(__isa_range(ge, lt) || ((MIPS_ISA_REV < (lt)) && __isa(flag)))
+#define __isa_range_and_flag(ge, lt, flag) \
+	(__isa_range(ge, lt) && __isa(flag))
 
 /*
  * SMP assumption: Options of CPU 0 are a superset of all processors.

arch/mips/include/asm/cpu-features.h line 296

Quoted 5 lines
arch/mips/include/asm/cpu-features.h +1 -1
@@ -291,10 +293,10 @@
 # define cpu_has_mips32r6	__isa_ge_or_flag(6, MIPS_CPU_ISA_M32R6)
 #endif
 #ifndef cpu_has_mips64r1
-# define cpu_has_mips64r1	__isa_range_or_flag(1, 6, MIPS_CPU_ISA_M64R1)
+# define cpu_has_mips64r1	__isa_range_and_flag(1, 6, MIPS_CPU_ISA_M64R1)

that's not the correct fix. You want to check for cpu_has_64bits here. Something like

Quoted 3 lines
# define cpu_has_mips64r1    (cpu_has_64bits && __isa_range_or_flag(1, 6, MIPS_CPU_ISA_M64R1))

should do the trick.

Good point. Thanks. I'll fix it in v3.

-Sergey

Quoted 6 lines

Thomas.

Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]
#33 lore raw

PATCH v2 9/20 mips: Add CP0 Write Merge config support

In reply to [email protected]

From: Serge Semin <[email protected]>

CP0 config register may indicate whether write-through merging is allowed. Currently there are two types of the merging available:
SysAD Valid and Full modes. Whether each of them are supported by the core is implementation dependent. Moreover whether the ability to change the mode also depends on the chip family instance. Taking into account all of this we created a dedicated mm_config() method to detect and enable merging if it's supported. It is called for MIPS-type processors at CPU-probe stage and attempts to detect whether the write merging is available. If it's known to be supported and switchable, then switch on the full mode. Otherwise just perform the CP0.Config.MM field analysis.

In addition there are platforms like InterAptiv/ProAptiv, which do have the MM bit field set by default, but having write-through cacheing unsupported makes write-merging also unsupported. In this case we just ignore the MM field value.

---
 arch/mips/include/asm/cpu-features.h |  8 +++++
 arch/mips/include/asm/cpu.h          |  4 ++-
 arch/mips/include/asm/mipsregs.h     |  3 ++
 arch/mips/kernel/cpu-probe.c         | 48 ++++++++++++++++++++++++++++
 4 files changed, 62 insertions(+), 1 deletion(-)
arch/mips/include/asm/cpu-features.h +8 -0
index 7e22b9c1e279..2b818f2036d0 100644
@@ -630,6 +630,14 @@
 # endif
 #endif
 
+#ifndef cpu_has_mm_sysad
+# define cpu_has_mm_sysad	__opt(MIPS_CPU_MM_SYSAD)
+#endif
+
+#ifndef cpu_has_mm_full
+# define cpu_has_mm_full	__opt(MIPS_CPU_MM_FULL)
+#endif
+
 /*
  * Guest capabilities
  */
arch/mips/include/asm/cpu.h +3 -1
index 9bae99b568c9..191529ef0d05 100644
@@ -417,7 +417,9 @@ enum cpu_type_enum {
 #define MIPS_CPU_MT_PER_TC_PERF_COUNTERS \
 				BIT_ULL(56)	/* CPU has perf counters implemented per TC (MIPSMT ASE) */
 #define MIPS_CPU_MMID		BIT_ULL(57)	/* CPU supports MemoryMapIDs */
-#define MIPS_CPU_MAC_2008_ONLY	BIT_ULL(58)	/* CPU Only support MAC2008 Fused multiply-add instruction */
+#define MIPS_CPU_MM_SYSAD	BIT_ULL(58)	/* CPU supports write-through SysAD Valid merge */
+#define MIPS_CPU_MM_FULL	BIT_ULL(59)	/* CPU supports write-through full merge */
+#define MIPS_CPU_MAC_2008_ONLY	BIT_ULL(60)	/* CPU Only support MAC2008 Fused multiply-add instruction */
 
 /*
  * CPU ASE encodings
arch/mips/include/asm/mipsregs.h +3 -0
index 796fe47cfd17..b1c761279b13 100644
@@ -563,6 +563,9 @@
 #define MIPS_CONF_MT_FTLB	(_ULCAST_(4) <<  7)
 #define MIPS_CONF_AR		(_ULCAST_(7) << 10)
 #define MIPS_CONF_AT		(_ULCAST_(3) << 13)
+#define MIPS_CONF_MM		(_ULCAST_(3) << 17)
+#define MIPS_CONF_MM_SYSAD	(_ULCAST_(1) << 17)
+#define MIPS_CONF_MM_FULL	(_ULCAST_(2) << 17)
 #define MIPS_CONF_M		(_ULCAST_(1) << 31)
 
 /*
arch/mips/kernel/cpu-probe.c +48 -0

60 lines not on this page · raw message

Signature · 1 line
2.25.1
#34 lore raw

PATCH v2 10/20 mips: Add CONFIG/CONFIG6/Cause reg fields macro

In reply to [email protected]

From: Serge Semin <[email protected]>

There are bit fields which persist in the MIPS CONFIG and CONFIG6 registers, but haven't been described in the generic mipsregs.h header so far. In particular, the generic CONFIG bitfields are BE - endian mode, BM - burst mode, SB - SimpleBE, OCP interface mode indicator, UDI - user-defined "CorExtend" instructions, DSP - data scratch pad RAM present, ISP - instruction scratch pad RAM present, etc. The core-specific CONFIG6 bitfields are JRCD - jump register cache prediction disable, R6 - MIPSr6 extensions enable, IFUPerfCtl - IFU performance control, SPCD - sleep state performance counter, DLSB - disable load/store bonding. A new exception code reported in the ExcCode field of the Cause register: 30 - Parity/ECC error exception happened on either fetch, load or cache refill. Lets add them to the mipsregs.h header to be used in future platform code, which have them utilized.

Cc Alexey Malahov <[email protected]>
Cc Thomas Bogendoerfer <[email protected]>
Cc Paul Burton <[email protected]>
Cc Ralf Baechle <[email protected]>
Cc Arnd Bergmann <[email protected]>
Cc Rob Herring <[email protected]>
---
 arch/mips/include/asm/mipsregs.h | 19 +++++++++++++++++++
 arch/mips/kernel/spram.c         |  4 ++--
 2 files changed, 21 insertions(+), 2 deletions(-)
arch/mips/include/asm/mipsregs.h +19 -0
index b1c761279b13..039ebd913f00 100644
@@ -468,6 +468,7 @@
 #define EXCCODE_THREAD		25	/* Thread exceptions (MT) */
 #define EXCCODE_DSPDIS		26	/* DSP disabled exception */
 #define EXCCODE_GE		27	/* Virtualized guest exception (VZ) */
+#define EXCCODE_CACHEERR	30	/* Parity/ECC occured on a core */
 
 /* Implementation specific trap codes used by MIPS cores */
 #define MIPS_EXCCODE_TLBPAR	16	/* TLB parity error exception */
@@ -563,9 +564,17 @@
 #define MIPS_CONF_MT_FTLB	(_ULCAST_(4) <<  7)
 #define MIPS_CONF_AR		(_ULCAST_(7) << 10)
 #define MIPS_CONF_AT		(_ULCAST_(3) << 13)
+#define MIPS_CONF_BE		(_ULCAST_(1) << 15)
+#define MIPS_CONF_BM		(_ULCAST_(1) << 16)
 #define MIPS_CONF_MM		(_ULCAST_(3) << 17)
 #define MIPS_CONF_MM_SYSAD	(_ULCAST_(1) << 17)
 #define MIPS_CONF_MM_FULL	(_ULCAST_(2) << 17)
+#define MIPS_CONF_SB		(_ULCAST_(1) << 21)
+#define MIPS_CONF_UDI		(_ULCAST_(1) << 22)
+#define MIPS_CONF_DSP		(_ULCAST_(1) << 23)
+#define MIPS_CONF_ISP		(_ULCAST_(1) << 24)
+#define MIPS_CONF_KU		(_ULCAST_(3) << 25)
+#define MIPS_CONF_K23		(_ULCAST_(3) << 28)
 #define MIPS_CONF_M		(_ULCAST_(1) << 31)
 
 /*
@@ -677,9 +686,19 @@
 #define MIPS_CONF5_CV		(_ULCAST_(1) << 29)
 #define MIPS_CONF5_K		(_ULCAST_(1) << 30)
 
+/* Jump register cache prediction disable */
+#define MIPS_CONF6_JRCD		(_ULCAST_(1) << 0)
+/* MIPSr6 extensions enable */
+#define MIPS_CONF6_R6		(_ULCAST_(1) << 2)
+/* IFU Performance Control */
+#define MIPS_CONF6_IFUPERFCTL	(_ULCAST_(3) << 10)
 #define MIPS_CONF6_SYND		(_ULCAST_(1) << 13)
+/* Sleep state performance counter disable */
+#define MIPS_CONF6_SPCD		(_ULCAST_(1) << 14)
 /* proAptiv FTLB on/off bit */
 #define MIPS_CONF6_FTLBEN	(_ULCAST_(1) << 15)
+/* Disable load/store bonding */
+#define MIPS_CONF6_DLSB		(_ULCAST_(1) << 21)
 /* Loongson-3 FTLB on/off bit */
 #define MIPS_CONF6_FTLBDIS	(_ULCAST_(1) << 22)
 /* FTLB probability bits */
arch/mips/kernel/spram.c +2 -2
index 26d355462ace..d5d96214cce5 100644
@@ -209,11 +209,11 @@ void spram_config(void)
 	case CPU_P6600:
 		config0 = read_c0_config();
 		/* FIXME: addresses are Malta specific */
-		if (config0 & (1<<24)) {
+		if (config0 & MIPS_CONF_ISP) {
 			probe_spram("ISPRAM", 0x1c000000,
 				    &ispram_load_tag, &ispram_store_tag);
 		}
-		if (config0 & (1<<23))
+		if (config0 & MIPS_CONF_DSP)
 			probe_spram("DSPRAM", 0x1c100000,
 				    &dspram_load_tag, &dspram_store_tag);
 	}
Signature · 1 line
2.25.1
#35 lore raw

PATCH v2 11/20 mips: MAAR: Use more precise address mask

In reply to [email protected]

From: Serge Semin <[email protected]>

Indeed according to the P5600/P6000 manual the MAAR pair register address field either takes [12:31] bits for 32-bits non-XPA systems and [12:35] otherwise. In any case the current address mask is just wrong for 64-bit and 32-bits XPA chips. So lets extend it to 39-bits value. This shall cover the 64-bits architecture and systems with XPA enabled, and won't cause any problem for non-XPA 32-bit systems, since the value will be just truncated when written to the 32-bits register.

---
 arch/mips/include/asm/mipsregs.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/mips/include/asm/mipsregs.h +1 -1
index 039ebd913f00..165f6318d861 100644
@@ -775,7 +775,7 @@
 
 /* MAAR bit definitions */
 #define MIPS_MAAR_VH		(_U64CAST_(1) << 63)
-#define MIPS_MAAR_ADDR		((BIT_ULL(BITS_PER_LONG - 12) - 1) << 12)
+#define MIPS_MAAR_ADDR		GENMASK_ULL(35, 12)
 #define MIPS_MAAR_ADDR_SHIFT	12
 #define MIPS_MAAR_S		(_ULCAST_(1) << 1)
 #define MIPS_MAAR_VL		(_ULCAST_(1) << 0)
Signature · 1 line
2.25.1
#36 lore raw

Re: [PATCH v2 11/20] mips: MAAR: Use more precise address mask

Quoted 9 lines · On Wed, May 06, 2020 at 08:42:29PM +0300, [email protected]…

From: Serge Semin <[email protected]>

Indeed according to the P5600/P6000 manual the MAAR pair register address field either takes [12:31] bits for 32-bits non-XPA systems and [12:35] otherwise. In any case the current address mask is just wrong for 64-bit and 32-bits XPA chips. So lets extend it to 39-bits value. This shall cover the 64-bits architecture and systems with XPA enabled, and won't cause any problem for non-XPA 32-bit systems, since the value will be just truncated when written to the 32-bits register.

according to MIPS32 Priveleged Resoure Architecture Rev. 6.02 ADDR spans from bit 12 to bit 55. So your patch fits only for P5600. Does the wider mask cause any problems ?

Thomas.

Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]
#37 lore raw

Re: [PATCH v2 11/20] mips: MAAR: Use more precise address mask

Quoted 11 lines · Thomas Bogendoerfer
Quoted 9 lines · On Wed, May 06, 2020 at 08:42:29PM +0300, [email protected]…

From: Serge Semin <[email protected]>

Indeed according to the P5600/P6000 manual the MAAR pair register address field either takes [12:31] bits for 32-bits non-XPA systems and [12:35] otherwise. In any case the current address mask is just wrong for 64-bit and 32-bits XPA chips. So lets extend it to 39-bits value. This shall cover the 64-bits architecture and systems with XPA enabled, and won't cause any problem for non-XPA 32-bit systems, since the value will be just truncated when written to the 32-bits register.

Quoted 4 lines

according to MIPS32 Priveleged Resoure Architecture Rev. 6.02 ADDR spans from bit 12 to bit 55. So your patch fits only for P5600.

Does the wider mask cause any problems ?

No, it won't. Bits written to the [40:62] range will be just ignored, while reading from there should return zeros. Setting GENMASK_ULL(55, 12) would also work. Though this solution is a bit workarounding because MIPS_MAAR_ADDR wouldn't reflect the real mask of the ADDR field. Something like the next macro would work better:

+#define MIPS_MAAR_ADDR \
+({ \
+ u64 __mask; \
+ \
+ if (cpu_has_lpa && read_c0_pagegrain() & PG_ELPA) { \
+ __mask = GENMASK_ULL(55, 12); \
+ else \
+ __mask = GENMASK_ULL(31, 12); \
+ \
+ __mask; \
+})

What do you think? What is better: the macro above or setting GENMASK_ULL(55, 12)?

BTW I've just figured out, that since XPA is currently only supported by kernels with CPU_MIPS32x config enabled, then only MIPS32 may have extended physical addressing of 2^60 bytes if CONFIG_XPA is enabled. Generic MIPS64 doesn't support the extended phys addressing so only 2^36 bytes are available on such platforms. (Loongson64 doesn't count, the platform code sets the PG_ELPA bit manually in kernel-entry-init.h)

-Sergey

Quoted 6 lines

Thomas.

Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]
#38 lore raw

Re: [PATCH v2 11/20] mips: MAAR: Use more precise address mask

Quoted 24 lines · Serge Semin
Quoted 11 lines · Thomas Bogendoerfer
Quoted 9 lines · On Wed, May 06, 2020 at 08:42:29PM +0300, [email protected]…

From: Serge Semin <[email protected]>

Indeed according to the P5600/P6000 manual the MAAR pair register address field either takes [12:31] bits for 32-bits non-XPA systems and [12:35] otherwise. In any case the current address mask is just wrong for 64-bit and 32-bits XPA chips. So lets extend it to 39-bits value. This shall cover the 64-bits architecture and systems with XPA enabled, and won't cause any problem for non-XPA 32-bit systems, since the value will be just truncated when written to the 32-bits register.

Quoted 4 lines

according to MIPS32 Priveleged Resoure Architecture Rev. 6.02 ADDR spans from bit 12 to bit 55. So your patch fits only for P5600.

Does the wider mask cause any problems ?

No, it won't. Bits written to the [40:62] range will be just ignored, while reading from there should return zeros. Setting GENMASK_ULL(55, 12) would also work. Though this solution is a bit workarounding because MIPS_MAAR_ADDR wouldn't reflect the real mask of the ADDR field. Something like the next macro would work better:

(quoted diff) +1 -0
+#define MIPS_MAAR_ADDR							\

Review

Quoted 10 lines
(quoted diff) +10 -0
+({									\
+	u64 __mask;							\
+									\
+	if (cpu_has_lpa && read_c0_pagegrain() & PG_ELPA) {		\
+		__mask = GENMASK_ULL(55, 12);				\
+	else								\
+		__mask = GENMASK_ULL(31, 12);				\
+									\
+	__mask;								\
+})

that looks horrible.

Quoted 2 lines

What do you think? What is better: the macro above or setting GENMASK_ULL(55, 12)?

just that one ;-)

Thomas.

Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]
#39 lore raw

Re: [PATCH v2 11/20] mips: MAAR: Use more precise address mask

Quoted 38 lines · Thomas Bogendoerfer
Quoted 24 lines · Serge Semin
Quoted 11 lines · Thomas Bogendoerfer
Quoted 9 lines · On Wed, May 06, 2020 at 08:42:29PM +0300, [email protected]…

From: Serge Semin <[email protected]>

Indeed according to the P5600/P6000 manual the MAAR pair register address field either takes [12:31] bits for 32-bits non-XPA systems and [12:35] otherwise. In any case the current address mask is just wrong for 64-bit and 32-bits XPA chips. So lets extend it to 39-bits value. This shall cover the 64-bits architecture and systems with XPA enabled, and won't cause any problem for non-XPA 32-bit systems, since the value will be just truncated when written to the 32-bits register.

Quoted 4 lines

according to MIPS32 Priveleged Resoure Architecture Rev. 6.02 ADDR spans from bit 12 to bit 55. So your patch fits only for P5600.

Does the wider mask cause any problems ?

No, it won't. Bits written to the [40:62] range will be just ignored, while reading from there should return zeros. Setting GENMASK_ULL(55, 12) would also work. Though this solution is a bit workarounding because MIPS_MAAR_ADDR wouldn't reflect the real mask of the ADDR field. Something like the next macro would work better:

(quoted diff) +1 -0
+#define MIPS_MAAR_ADDR							\

Review

Quoted 10 lines
(quoted diff) +10 -0
+({									\
+	u64 __mask;							\
+									\
+	if (cpu_has_lpa && read_c0_pagegrain() & PG_ELPA) {		\
+		__mask = GENMASK_ULL(55, 12);				\
+	else								\
+		__mask = GENMASK_ULL(31, 12);				\
+									\
+	__mask;								\
+})

that looks horrible.

Quoted 4 lines
Quoted 2 lines

What do you think? What is better: the macro above or setting GENMASK_ULL(55, 12)?

just that one ;-)

Agreed. I'll fix it in v3.

-Sergey

Quoted 6 lines

Thomas.

Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]
#40 lore raw

PATCH v2 12/20 mips: MAAR: Add XPA mode support

In reply to [email protected]

From: Serge Semin <[email protected]>

When XPA mode is enabled the normally 32-bits MAAR pair registers are extended to be of 64-bits width as in pure 64-bits MIPS architecture. In this case the MAAR registers can enable the speculative loads/stores for addresses of up to 39-bits width. But in this case the process of the MAAR initialization changes a bit. The upper 32-bits of the registers are supposed to be accessed by mean of the dedicated instructions mfhc0/mthc0 and there is a CP0.MAAR.VH bit which should be set together with CP0.MAAR.VL as indication of the boundary validity. All of these peculiarities were taken into account in this commit so the speculative loads/stores would work when XPA mode is enabled.

---
 arch/mips/include/asm/maar.h     | 17 +++++++++++++++--
 arch/mips/include/asm/mipsregs.h | 10 ++++++++++
 arch/mips/mm/init.c              |  8 +++++++-
 3 files changed, 32 insertions(+), 3 deletions(-)
arch/mips/include/asm/maar.h +15 -2
index 6908b93c4ff9..99f1c3e4b11f 100644
@@ -32,7 +32,7 @@ unsigned platform_maar_init(unsigned num_pairs);
  * @upper:	The highest address that the MAAR pair will affect. Must be
  *		aligned to one byte before a 2^16 byte boundary.
  * @attrs:	The accessibility attributes to program, eg. MIPS_MAAR_S. The
- *		MIPS_MAAR_VL attribute will automatically be set.
+ *		MIPS_MAAR_VL/MIPS_MAAR_VH attributes will automatically be set.
  *
  * Program the pair of MAAR registers specified by idx to apply the attributes
  * specified by attrs to the range of addresses from lower to higher.
@@ -48,17 +48,30 @@ static inline void write_maar_pair(unsigned idx, phys_addr_t lower,
 	/* Automatically set MIPS_MAAR_VL */
 	attrs |= MIPS_MAAR_VL;
 
-	/* Write the upper address & attributes (only MIPS_MAAR_VL matters) */
+	/*
+	 * Write the upper address & attributes (both MIPS_MAAR_VL and
+	 * MIPS_MAAR_VH matter)
+	 */
 	write_c0_maari(idx << 1);
 	back_to_back_c0_hazard();
 	write_c0_maar(((upper >> 4) & MIPS_MAAR_ADDR) | attrs);
 	back_to_back_c0_hazard();
+#ifdef CONFIG_XPA
+	upper >>= MIPS_MAARX_ADDR_SHIFT;
+	writex_c0_maar(((upper >> 4) & MIPS_MAARX_ADDR) | MIPS_MAARX_VH);
+	back_to_back_c0_hazard();
+#endif
 
 	/* Write the lower address & attributes */
 	write_c0_maari((idx << 1) | 0x1);
 	back_to_back_c0_hazard();
 	write_c0_maar((lower >> 4) | attrs);
 	back_to_back_c0_hazard();
+#ifdef CONFIG_XPA
+	lower >>= MIPS_MAARX_ADDR_SHIFT;
+	writex_c0_maar(((lower >> 4) & MIPS_MAARX_ADDR) | MIPS_MAARX_VH);
+	back_to_back_c0_hazard();
+#endif
 }
 
 /**
arch/mips/include/asm/mipsregs.h +10 -0
index 165f6318d861..aa8599962ea3 100644
@@ -779,6 +779,14 @@
 #define MIPS_MAAR_ADDR_SHIFT	12
 #define MIPS_MAAR_S		(_ULCAST_(1) << 1)
 #define MIPS_MAAR_VL		(_ULCAST_(1) << 0)
+#ifdef CONFIG_XPA
+#define MIPS_MAAR_V		(MIPS_MAAR_VH | MIPS_MAAR_VL)
+#else
+#define MIPS_MAAR_V		MIPS_MAAR_VL
+#endif
+#define MIPS_MAARX_VH		(_ULCAST_(1) << 31)
+#define MIPS_MAARX_ADDR		0xF
+#define MIPS_MAARX_ADDR_SHIFT	32
 
 /* MAARI bit definitions */
 #define MIPS_MAARI_INDEX	(_ULCAST_(0x3f) << 0)
@@ -1739,6 +1747,8 @@ do {									\
 #define write_c0_lladdr(val)	__write_ulong_c0_register($17, 0, val)
 #define read_c0_maar()		__read_ulong_c0_register($17, 1)
 #define write_c0_maar(val)	__write_ulong_c0_register($17, 1, val)
+#define readx_c0_maar()		__readx_32bit_c0_register($17, 1)
+#define writex_c0_maar(val)	__writex_32bit_c0_register($17, 1, val)
 #define read_c0_maari()		__read_32bit_c0_register($17, 2)
 #define write_c0_maari(val)	__write_32bit_c0_register($17, 2, val)
 
arch/mips/mm/init.c +7 -1
index 79684000de0e..620ebfa45ec1 100644
@@ -358,17 +358,23 @@ void maar_init(void)
 		write_c0_maari(i);
 		back_to_back_c0_hazard();
 		upper = read_c0_maar();
+#ifdef CONFIG_XPA
+		upper |= (phys_addr_t)readx_c0_maar() << MIPS_MAARX_ADDR_SHIFT;
+#endif
 
 		write_c0_maari(i + 1);
 		back_to_back_c0_hazard();
 		lower = read_c0_maar();
+#ifdef CONFIG_XPA
+		lower |= (phys_addr_t)readx_c0_maar() << MIPS_MAARX_ADDR_SHIFT;
+#endif
 
 		attr = lower & upper;
 		lower = (lower & MIPS_MAAR_ADDR) << 4;
 		upper = ((upper & MIPS_MAAR_ADDR) << 4) | 0xffff;
 
 		pr_info("  [%d]: ", i / 2);
-		if (!(attr & MIPS_MAAR_VL)) {
+		if ((attr & MIPS_MAAR_V) != MIPS_MAAR_V) {
 			pr_cont("disabled\n");
 			continue;
 		}
Signature · 1 line
2.25.1
maintainer Applied
#41 lore raw

Re: [PATCH v2 12/20] mips: MAAR: Add XPA mode support

Quoted 24 lines · On Wed, May 06, 2020 at 08:42:30PM +0300, [email protected]…

From: Serge Semin <[email protected]>

When XPA mode is enabled the normally 32-bits MAAR pair registers are extended to be of 64-bits width as in pure 64-bits MIPS architecture. In this case the MAAR registers can enable the speculative loads/stores for addresses of up to 39-bits width. But in this case the process of the MAAR initialization changes a bit. The upper 32-bits of the registers are supposed to be accessed by mean of the dedicated instructions mfhc0/mthc0 and there is a CP0.MAAR.VH bit which should be set together with CP0.MAAR.VL as indication of the boundary validity. All of these peculiarities were taken into account in this commit so the speculative loads/stores would work when XPA mode is enabled.

Quoted 5 lines
---
 arch/mips/include/asm/maar.h     | 17 +++++++++++++++--
 arch/mips/include/asm/mipsregs.h | 10 ++++++++++
 arch/mips/mm/init.c              |  8 +++++++-
 3 files changed, 32 insertions(+), 3 deletions(-)

applied to mips-next.

Thomas.

Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]
#42 lore raw

Re: [PATCH v2 12/20] mips: MAAR: Add XPA mode support

Quoted 31 lines · Thomas Bogendoerfer
Quoted 24 lines · On Wed, May 06, 2020 at 08:42:30PM +0300, [email protected]…

From: Serge Semin <[email protected]>

When XPA mode is enabled the normally 32-bits MAAR pair registers are extended to be of 64-bits width as in pure 64-bits MIPS architecture. In this case the MAAR registers can enable the speculative loads/stores for addresses of up to 39-bits width. But in this case the process of the MAAR initialization changes a bit. The upper 32-bits of the registers are supposed to be accessed by mean of the dedicated instructions mfhc0/mthc0 and there is a CP0.MAAR.VH bit which should be set together with CP0.MAAR.VL as indication of the boundary validity. All of these peculiarities were taken into account in this commit so the speculative loads/stores would work when XPA mode is enabled.

Quoted 5 lines
---
 arch/mips/include/asm/maar.h     | 17 +++++++++++++++--
 arch/mips/include/asm/mipsregs.h | 10 ++++++++++
 arch/mips/mm/init.c              |  8 +++++++-
 3 files changed, 32 insertions(+), 3 deletions(-)
Quoted 1 line

applied to mips-next.

Great! Thanks.

-Sergey

Quoted 6 lines

Thomas.

Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]
#43 lore raw

PATCH v2 13/20 mips: early_printk_8250: Use offset-sized IO-mem accessors

In reply to [email protected]

From: Serge Semin <[email protected]>

Some platforms may prohibit to access the IO-memory with instructions of certain memory widths. For instance Bailal-T1 has devices placed behind memory OCP port (which also the reason of DMA accesses being incoherent) and can't be accessed through CCA uncacheable memory with other than 4-bytes aligned (LW/SW) instructions. Ignoring this rule will cause the APB EHB error with 0xFFs returned on read operations. In order to fix the issue for this platform and for others, which may have similar problems, lets recode serial_in()/serial_out() to call a certain memory accessors in accordance with the UART registers shift setting.

Cc Alexey Malahov <[email protected]>
Cc Thomas Bogendoerfer <[email protected]>
Cc Paul Burton <[email protected]>
Cc Ralf Baechle <[email protected]>
Cc Arnd Bergmann <[email protected]>
Cc Rob Herring <[email protected]>
---
 arch/mips/kernel/early_printk_8250.c | 34 ++++++++++++++++++++++++++--
 1 file changed, 32 insertions(+), 2 deletions(-)
arch/mips/kernel/early_printk_8250.c +32 -2
index 567c6ec0cfae..e2c2405cff62 100644
@@ -23,12 +23,42 @@ void setup_8250_early_printk_port(unsigned long base, unsigned int reg_shift,
 
 static inline u8 serial_in(int offset)
 {
-	return readb(serial8250_base + (offset << serial8250_reg_shift));
+	u8 ret = 0xFF;
+
+	offset <<= serial8250_reg_shift;
+	switch (serial8250_reg_shift) {
+	case 0:
+		ret = readb(serial8250_base + offset);
+		break;
+	case 1:
+		ret = readw(serial8250_base + offset);
+		break;
+	case 2:
+		ret = readl(serial8250_base + offset);
+		break;
+	default:
+		break;
+	}
+
+	return ret;
 }
 
 static inline void serial_out(int offset, char value)
 {
-	writeb(value, serial8250_base + (offset << serial8250_reg_shift));
+	offset <<= serial8250_reg_shift;
+	switch (serial8250_reg_shift) {
+	case 0:
+		writeb(value, serial8250_base + offset);
+		break;
+	case 1:
+		writew(value, serial8250_base + offset);
+		break;
+	case 2:
+		writel(value, serial8250_base + offset);
+		break;
+	default:
+		break;
+	}
 }
 
 void prom_putchar(char c)
Signature · 1 line
2.25.1
#44 lore raw

PATCH v2 14/20 mips: Use offset-sized IO-mem accessors in CPS debug printout

In reply to [email protected]

From: Serge Semin <[email protected]>

Similar to commit 8e5c62e38a88 ("mips: early_printk_8250: Use offset-sized IO-mem accessors") the IO-memory might require to use a proper load/store instructions (like Bailal-T1 IO-memory). To fix the cps-vec UART debug printout lets use the memory access instructions in accordance with the UART registers offset config specified at boot time.

Cc Alexey Malahov <[email protected]>
Cc Thomas Bogendoerfer <[email protected]>
Cc Paul Burton <[email protected]>
Cc Ralf Baechle <[email protected]>
Cc Arnd Bergmann <[email protected]>
Cc Rob Herring <[email protected]>

---
There might be another problem in cps-vec-ns16550.S connected with the difference in CPU/devices endinanness on some platforms. But there is no such for Baikal-T1 SoC.

---
 arch/mips/kernel/cps-vec-ns16550.S | 18 ++++++++++++++++--
 1 file changed, 16 insertions(+), 2 deletions(-)
arch/mips/kernel/cps-vec-ns16550.S +16 -2
index d5a67b4ce9f6..2adb1c56c7c5 100644
@@ -14,16 +14,30 @@
 #define UART_TX_OFS	(UART_TX << CONFIG_MIPS_CPS_NS16550_SHIFT)
 #define UART_LSR_OFS	(UART_LSR << CONFIG_MIPS_CPS_NS16550_SHIFT)
 
+#if CONFIG_MIPS_CPS_NS16550_SHIFT == 0
+# define UART_L		lb
+# define UART_S		sb
+#elif CONFIG_MIPS_CPS_NS16550_SHIFT == 1
+# define UART_L		lh
+# define UART_S		sh
+#elif CONFIG_MIPS_CPS_NS16550_SHIFT == 2
+# define UART_L		lw
+# define UART_S		sw
+#else
+# define UART_L		lw
+# define UART_S		sb
+#endif
+
 /**
  * _mips_cps_putc() - write a character to the UART
  * @a0: ASCII character to write
  * @t9: UART base address
  */
 LEAF(_mips_cps_putc)
-1:	lw		t0, UART_LSR_OFS(t9)
+1:	UART_L		t0, UART_LSR_OFS(t9)
 	andi		t0, t0, UART_LSR_TEMT
 	beqz		t0, 1b
-	sb		a0, UART_TX_OFS(t9)
+	UART_S		a0, UART_TX_OFS(t9)
 	jr		ra
 	END(_mips_cps_putc)
 
Signature · 1 line
2.25.1
#45 lore raw

Re: [PATCH v2 14/20] mips: Use offset-sized IO-mem accessors in CPS debug printout

Hello!

Quoted 5 lines · On 05/06/2020 08:42 PM, [email protected] wrote:

From: Serge Semin <[email protected]>

Similar to commit 8e5c62e38a88 ("mips: early_printk_8250: Use offset-sized IO-mem accessors") the IO-memory might require to use a proper load/store instructions (like Bailal-T1 IO-memory). To fix the cps-vec UART debug

Baikal? :-)

Quoted 13 lines

printout lets use the memory access instructions in accordance with the UART registers offset config specified at boot time.

Cc Alexey Malahov <[email protected]>
Cc Thomas Bogendoerfer <[email protected]>
Cc Paul Burton <[email protected]>
Cc Ralf Baechle <[email protected]>
Cc Arnd Bergmann <[email protected]>
Cc Rob Herring <[email protected]>
Quoted 3 lines

---
There might be another problem in cps-vec-ns16550.S connected with the difference in CPU/devices endinanness on some platforms. But there is

Endianness.

Quoted 1 line

no such for Baikal-T1 SoC.

[...]

MBR, Sergei

#46 lore raw

Re: [PATCH v2 14/20] mips: Use offset-sized IO-mem accessors in CPS debug printout

Hi

Quoted 11 lines · Sergei Shtylyov

Hello!

Quoted 5 lines · On 05/06/2020 08:42 PM, [email protected] wrote:

From: Serge Semin <[email protected]>

Similar to commit 8e5c62e38a88 ("mips: early_printk_8250: Use offset-sized IO-mem accessors") the IO-memory might require to use a proper load/store instructions (like Bailal-T1 IO-memory). To fix the cps-vec UART debug

   Baikal? :-)

Yeah, thanks.)

Quoted 14 lines
Quoted 14 lines

printout lets use the memory access instructions in accordance with the UART registers offset config specified at boot time.

Cc Alexey Malahov <[email protected]>
Cc Thomas Bogendoerfer <[email protected]>
Cc Paul Burton <[email protected]>
Cc Ralf Baechle <[email protected]>
Cc Arnd Bergmann <[email protected]>
Cc Rob Herring <[email protected]>
Quoted 5 lines
Quoted 3 lines

---
There might be another problem in cps-vec-ns16550.S connected with the difference in CPU/devices endinanness on some platforms. But there is

Endianness.

Ah, this won't get into the commit message anyway. But thanks for noticing.)

-Sergey

Quoted 5 lines
Quoted 2 lines

no such for Baikal-T1 SoC.

[...]

MBR, Sergei

#47 lore raw

PATCH v2 15/20 mips: cdmm: Add mti,mips-cdmm dtb node support

In reply to [email protected]

From: Serge Semin <[email protected]>

Since having and mapping the CDMM block is platform specific, then instead of just returning a zero-address, lets make the default CDMM base address search method (mips_cdmm_phys_base()) to do something useful. For instance to find the address in a dedicated dtb-node in order to support of-based platforms by default.

Cc Thomas Bogendoerfer <[email protected]>
Cc Alexey Malahov <[email protected]>
Cc Paul Burton <[email protected]>
Cc Ralf Baechle <[email protected]>
Cc Greg Kroah-Hartman <[email protected]>
Cc Arnd Bergmann <[email protected]>
Cc Olof Johansson <[email protected]>
Cc Rob Herring <[email protected]>
---
 drivers/bus/mips_cdmm.c | 15 +++++++++++++++
 1 file changed, 15 insertions(+)
drivers/bus/mips_cdmm.c +15 -0
index 1b14256376d2..7faa8c049f07 100644
@@ -16,6 +16,8 @@
 #include <linux/platform_device.h>
 #include <linux/slab.h>
 #include <linux/smp.h>
+#include <linux/of_address.h>
+#include <linux/of.h>
 #include <asm/cdmm.h>
 #include <asm/hazards.h>
 #include <asm/mipsregs.h>
@@ -337,9 +339,22 @@ static phys_addr_t mips_cdmm_cur_base(void)
  * Picking a suitable physical address at which to map the CDMM region is
  * platform specific, so this weak function can be overridden by platform
  * code to pick a suitable value if none is configured by the bootloader.
+ * By default this method tries to find a CDMM-specific node in the system
+ * dtb. Note that this won't work for early serial console.
  */
 phys_addr_t __weak mips_cdmm_phys_base(void)
 {
+	struct device_node *np;
+	struct resource res;
+	int err;
+
+	np = of_find_compatible_node(NULL, NULL, "mti,mips-cdmm");
+	if (np) {
+		err = of_address_to_resource(np, 0, &res);
+		if (!err)
+			return res.start;
+	}
+
 	return 0;
 }
 
Signature · 1 line
2.25.1
#48 lore raw

PATCH v2 16/20 bus: cdmm: Add MIPS R5 arch support

In reply to [email protected]

From: Serge Semin <[email protected]>

CDMM may be available not only MIPS R2 architectures, but also in newer MIPS R5 chips. For instance our P5600 chip has one. Lets mark the CDMM bus being supported for that MIPS arch too.

Cc Thomas Bogendoerfer <[email protected]>
Cc Paul Burton <[email protected]>
Cc Ralf Baechle <[email protected]>
Cc Greg Kroah-Hartman <[email protected]>
Cc Arnd Bergmann <[email protected]>
Cc Olof Johansson <[email protected]>
Cc Rob Herring <[email protected]>
---
 drivers/bus/Kconfig | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/bus/Kconfig +1 -1
index 6d4e4497b59b..971c07bc92d4 100644
@@ -58,7 +58,7 @@ config IMX_WEIM
 
 config MIPS_CDMM
 	bool "MIPS Common Device Memory Map (CDMM) Driver"
-	depends on CPU_MIPSR2
+	depends on CPU_MIPSR2 || CPU_MIPSR5
 	help
 	  Driver needed for the MIPS Common Device Memory Map bus in MIPS
 	  cores. This bus is for per-CPU tightly coupled devices such as the
Signature · 1 line
2.25.1
#49 lore raw

PATCH v2 17/20 mips: Add udelay lpj numbers adjustment

In reply to [email protected]

From: Serge Semin <[email protected]>

Loops-per-jiffies is a special number which represents a number of noop-loop cycles per CPU-scheduler quantum - jiffies. As you understand aside from CPU-specific implementation it depends on the CPU frequency. So when a platform has the CPU frequency fixed, we have no problem and the current udelay interface will work just fine. But as soon as CPU-freq driver is enabled and the cores frequency changes, we'll end up with distorted udelay's. In order to fix this we have to accordinly adjust the per-CPU udelay_val (the same as the global loops_per_jiffy) number. This can be done in the CPU-freq transition event handler. We subscribe to that event in the MIPS arch time-inititalization method.

---
 arch/mips/kernel/time.c | 70 +++++++++++++++++++++++++++++++++++++++++
 1 file changed, 70 insertions(+)
arch/mips/kernel/time.c +70 -0
index 37e9413a393d..ce89e18af024 100644
@@ -18,12 +18,82 @@
 #include <linux/smp.h>
 #include <linux/spinlock.h>
 #include <linux/export.h>
+#include <linux/cpufreq.h>
+#include <linux/delay.h>
 
 #include <asm/cpu-features.h>
 #include <asm/cpu-type.h>
 #include <asm/div64.h>
 #include <asm/time.h>
 
+#ifdef CONFIG_CPU_FREQ
+
+static DEFINE_PER_CPU(unsigned long, pcp_lpj_ref);
+static DEFINE_PER_CPU(unsigned long, pcp_lpj_ref_freq);
+static unsigned long glb_lpj_ref;
+static unsigned long glb_lpj_ref_freq;
+
+static int cpufreq_callback(struct notifier_block *nb,
+			    unsigned long val, void *data)
+{
+	struct cpufreq_freqs *freq = data;
+	struct cpumask *cpus = freq->policy->cpus;
+	unsigned long lpj;
+	int cpu;
+
+	/*
+	 * Skip lpj numbers adjustment if the CPU-freq transition is safe for
+	 * the loops delay. (Is this possible?)
+	 */
+	if (freq->flags & CPUFREQ_CONST_LOOPS)
+		return NOTIFY_OK;
+
+	/* Save the initial values of the lpjes for future scaling. */
+	if (!glb_lpj_ref) {
+		glb_lpj_ref = boot_cpu_data.udelay_val;
+		glb_lpj_ref_freq = freq->old;
+
+		for_each_online_cpu(cpu) {
+			per_cpu(pcp_lpj_ref, cpu) =
+				cpu_data[cpu].udelay_val;
+			per_cpu(pcp_lpj_ref_freq, cpu) = freq->old;
+		}
+	}
+
+	/*
+	 * Adjust global lpj variable and per-CPU udelay_val number in
+	 * accordance with the new CPU frequency.
+	 */
+	if ((val == CPUFREQ_PRECHANGE  && freq->old < freq->new) ||
+	    (val == CPUFREQ_POSTCHANGE && freq->old > freq->new)) {
+		loops_per_jiffy = cpufreq_scale(glb_lpj_ref,
+						glb_lpj_ref_freq,
+						freq->new);
+
+		for_each_cpu(cpu, cpus) {
+			lpj = cpufreq_scale(per_cpu(pcp_lpj_ref, cpu),
+					    per_cpu(pcp_lpj_ref_freq, cpu),
+					    freq->new);
+			cpu_data[cpu].udelay_val = (unsigned int)lpj;
+		}
+	}
+
+	return NOTIFY_OK;
+}
+
+static struct notifier_block cpufreq_notifier = {
+	.notifier_call  = cpufreq_callback,
+};
+
+static int __init register_cpufreq_notifier(void)
+{
+	return cpufreq_register_notifier(&cpufreq_notifier,
+					 CPUFREQ_TRANSITION_NOTIFIER);
+}
+core_initcall(register_cpufreq_notifier);
+
+#endif /* !CONFIG_CPU_FREQ */
+
 /*
  * forward reference
  */
Signature · 1 line
2.25.1
Reviewed-by on v2 17/20
#50 lore raw

Re: [PATCH v2 17/20] mips: Add udelay lpj numbers adjustment

Quoted 14 lines · On Wed, 6 May 2020 20:42:35 +0300 <[email protected]> wrote:

From: Serge Semin <[email protected]>

Loops-per-jiffies is a special number which represents a number of noop-loop cycles per CPU-scheduler quantum - jiffies. As you understand aside from CPU-specific implementation it depends on the CPU frequency. So when a platform has the CPU frequency fixed, we have no problem and the current udelay interface will work just fine. But as soon as CPU-freq driver is enabled and the cores frequency changes, we'll end up with distorted udelay's. In order to fix this we have to accordinly adjust the per-CPU udelay_val (the same as the global loops_per_jiffy) number. This can be done in the CPU-freq transition event handler. We subscribe to that event in the MIPS arch time-inititalization method.

Quoted 10 lines

That have been absent in MIPS kernel so long!

Thanks.

Quoted 1 line

---

[...]
---
Jiaxun Yang

#51 lore raw

PATCH v2 18/20 mips: csrc-r4k: Decrease r4k-clocksource rating if CPU_FREQ enabled

In reply to [email protected]

From: Serge Semin <[email protected]>

Commit 07d69579e7fe ("MIPS: Don't register r4k sched clock when CPUFREQ enabled") disabled the r4k-clock usage for scheduler ticks counting due to the scheduler being non-tolerant for unstable clocks sources. For the same reason the clock should be used in the system clocksource framework only as a last resort if CPU frequency may change.

Cc Alexey Malahov <[email protected]>
Cc Thomas Bogendoerfer <[email protected]>
Cc Paul Burton <[email protected]>
Cc Ralf Baechle <[email protected]>
Cc Greg Kroah-Hartman <[email protected]>
Cc Arnd Bergmann <[email protected]>
Cc Rob Herring <[email protected]>
---
 arch/mips/kernel/csrc-r4k.c | 4 ++++
 1 file changed, 4 insertions(+)
arch/mips/kernel/csrc-r4k.c +4 -0
index 437dda64fd7a..d81fb374f477 100644
@@ -71,7 +71,11 @@ int __init init_r4k_clocksource(void)
 		return -ENXIO;
 
 	/* Calculate a somewhat reasonable rating value */
+#ifndef CONFIG_CPU_FREQ
 	clocksource_mips.rating = 200 + mips_hpt_frequency / 10000000;
+#else
+	clocksource_mips.rating = 99;
+#endif
 
 	/*
 	 * R2 onwards makes the count accessible to user mode so it can be used
Signature · 1 line
2.25.1
#52 lore raw

Re: [PATCH v2 18/20] mips: csrc-r4k: Decrease r4k-clocksource rating if CPU_FREQ enabled

Quoted 28 lines · On Wed, May 06, 2020 at 08:42:36PM +0300, [email protected]…

From: Serge Semin <[email protected]>

Commit 07d69579e7fe ("MIPS: Don't register r4k sched clock when CPUFREQ enabled") disabled the r4k-clock usage for scheduler ticks counting due to the scheduler being non-tolerant for unstable clocks sources. For the same reason the clock should be used in the system clocksource framework only as a last resort if CPU frequency may change.

Cc Alexey Malahov <[email protected]>
Cc Thomas Bogendoerfer <[email protected]>
Cc Paul Burton <[email protected]>
Cc Ralf Baechle <[email protected]>
Cc Greg Kroah-Hartman <[email protected]>
Cc Arnd Bergmann <[email protected]>
Cc Rob Herring <[email protected]>
---
 arch/mips/kernel/csrc-r4k.c | 4 ++++
 1 file changed, 4 insertions(+)
arch/mips/kernel/csrc-r4k.c +0 -0
index 437dda64fd7a..d81fb374f477 100644

arch/mips/kernel/csrc-r4k.c int __init init_r4k_clocksource(void) · line 78

Quoted 8 lines
arch/mips/kernel/csrc-r4k.c +4 -0
@@ -71,7 +71,11 @@ int __init init_r4k_clocksource(void)
 		return -ENXIO;
 
 	/* Calculate a somewhat reasonable rating value */
+#ifndef CONFIG_CPU_FREQ
 	clocksource_mips.rating = 200 + mips_hpt_frequency / 10000000;
+#else
+	clocksource_mips.rating = 99;
+#endif

I dislike this patch. Assuming you have an other clocksource, why not simply disable csrc-r4k, if CPU_FREQ is enabled ?

Thomas.

Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]
#53 lore raw

Re: [PATCH v2 18/20] mips: csrc-r4k: Decrease r4k-clocksource rating if CPU_FREQ enabled

Quoted 31 lines · Thomas Bogendoerfer
Quoted 30 lines · On Wed, May 06, 2020 at 08:42:36PM +0300, [email protected]…

From: Serge Semin <[email protected]>

Commit 07d69579e7fe ("MIPS: Don't register r4k sched clock when CPUFREQ enabled") disabled the r4k-clock usage for scheduler ticks counting due to the scheduler being non-tolerant for unstable clocks sources. For the same reason the clock should be used in the system clocksource framework only as a last resort if CPU frequency may change.

Cc Alexey Malahov <[email protected]>
Cc Thomas Bogendoerfer <[email protected]>
Cc Paul Burton <[email protected]>
Cc Ralf Baechle <[email protected]>
Cc Greg Kroah-Hartman <[email protected]>
Cc Arnd Bergmann <[email protected]>
Cc Rob Herring <[email protected]>
---
 arch/mips/kernel/csrc-r4k.c | 4 ++++
 1 file changed, 4 insertions(+)
arch/mips/kernel/csrc-r4k.c +0 -0
index 437dda64fd7a..d81fb374f477 100644
@@ -71,7 +71,11 @@ int __init init_r4k_clocksource(void)
 		return -ENXIO;
 
Quoted 9 lines

arch/mips/kernel/csrc-r4k.c int __init init_r4k_clocksource(void) · line 78

Quoted 6 lines
arch/mips/kernel/csrc-r4k.c +4 -0
 	/* Calculate a somewhat reasonable rating value */
+#ifndef CONFIG_CPU_FREQ
 	clocksource_mips.rating = 200 + mips_hpt_frequency / 10000000;
+#else
+	clocksource_mips.rating = 99;
+#endif

I dislike this patch. Assuming you have an other clocksource, why not simply disable csrc-r4k, if CPU_FREQ is enabled ?

Me neither and the best way would be to update the clocksource frequency dynamically the same way it's done for cevt-r4k and MIPS GIC timers. Alas the clocksource doesn't support it. Due to this together with CPU-freq facility enabled we have to use a very slow DW APB Timer instead of the fast embedded into the CPU core r4k and MIPS GIC timers. Just note the difference: it takes 220 ns to read the counter from DW APB Timer in comparison to a few nanoseconds reading from MIPS GIC and R4K. So IMO disabling the timer as you suggest isn't the best option. By making the CPUFREQ and CSRC_R4K mutual exclusive we'd assume a use-case that the system will always use the CPU-freq facility changing the CPU reference frequency. This is obviously not true. Noone prevents the system administrator to leave the default setting of the CPU-freq with fixed frequency and select a faster, more accurate timer like in our case.

My idea was not to try to predict how the system would be used, but to let the system administration to choose which timer is applicable in particular usecase enabling a safest one by default. So if CPUFREQ is available, then we fallback to the external timer as safest one. If the system user wouldn't need to have the CPUFREQ facility utilized, then the system administrator would want to leave the default CPU-freq governor with pre-defined CPU frequency and select either R4K (MIPS) or MIPS GIC timer just by writing its name into /sys/bus/clocksource/devices/clocksource0/current_clocksource .

I should note, that currently CPU_FREQ won't be available if there is no MIPS_EXTERNAL_TIMER available for the platform. It's prohibited by means of the conditional kbuild config inclusion declared in the arch/mips/Kconfig:
+ if CPU_SUPPORTS_CPUFREQ && MIPS_EXTERNAL_TIMER
+ source "drivers/cpufreq/Kconfig"
+ endif
So if there is no external timer working independently from the CPU core clock source, the CPUFREQ won't be available to select for the kernel. Though currently this limitation is supposed to be applicable for the R4K/MIPS GIC clocksource timers only since clockevents must work fine in unstable reference clock conditions.

So what can we do to improve the patch? First one is a solution I suggested in this patch but it could be a bit altered by using IS_ENABLED() macro to:
+ clocksource_mips.rating = !IS_ENABLED(CONFIG_CPU_FREQ) ?
+ 200 + mips_hpt_frequency / 10000000 : 99;

Another idea I discovered when have been searching through the x86 arch code. x86's got the same problem with TSC timer, but it doesn't disable it if CPU-frequency is switched on. Instead it just marks it as unstable by calling the clocksource_mark_unstable() method if CPU frequency changes. I suggest to implement the same approach in our case of MIPS GIC (another patchset I've sent, see "clocksource: Fix MIPS GIC and DW APB Timer for Baikal-T1 SoC support" in your email client) and R4K timers. We'll subscribe to the CPU frequency change and if it changes we'll call clocksource_mark_unstable() with MIPS GIC and R4K clocksource handlers passed. This shall reduce their rating and cause selecting a clocksource with better one. BTW I suppose it won't be necessary to initially lower the rating of the MIPS GIC and R4K clocksource timers if this is implemented.

So, what do you think?

-Sergey

Quoted 6 lines

Thomas.

Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]
#54 lore raw

Re: [PATCH v2 18/20] mips: csrc-r4k: Decrease r4k-clocksource rating if CPU_FREQ enabled

Thomas,
Could you take a look at my comment below so I could proceed with the patchset v3 development?

-Sergey

Quoted 103 lines · Serge Semin
Quoted 31 lines · Thomas Bogendoerfer
Quoted 30 lines · On Wed, May 06, 2020 at 08:42:36PM +0300, [email protected]…

From: Serge Semin <[email protected]>

Commit 07d69579e7fe ("MIPS: Don't register r4k sched clock when CPUFREQ enabled") disabled the r4k-clock usage for scheduler ticks counting due to the scheduler being non-tolerant for unstable clocks sources. For the same reason the clock should be used in the system clocksource framework only as a last resort if CPU frequency may change.

Cc Alexey Malahov <[email protected]>
Cc Thomas Bogendoerfer <[email protected]>
Cc Paul Burton <[email protected]>
Cc Ralf Baechle <[email protected]>
Cc Greg Kroah-Hartman <[email protected]>
Cc Arnd Bergmann <[email protected]>
Cc Rob Herring <[email protected]>
---
 arch/mips/kernel/csrc-r4k.c | 4 ++++
 1 file changed, 4 insertions(+)
arch/mips/kernel/csrc-r4k.c +0 -0
index 437dda64fd7a..d81fb374f477 100644
@@ -71,7 +71,11 @@ int __init init_r4k_clocksource(void)
 		return -ENXIO;
 
Quoted 9 lines

arch/mips/kernel/csrc-r4k.c int __init init_r4k_clocksource(void) · line 78

Quoted 6 lines
arch/mips/kernel/csrc-r4k.c +4 -0
 	/* Calculate a somewhat reasonable rating value */
+#ifndef CONFIG_CPU_FREQ
 	clocksource_mips.rating = 200 + mips_hpt_frequency / 10000000;
+#else
+	clocksource_mips.rating = 99;
+#endif

I dislike this patch. Assuming you have an other clocksource, why not simply disable csrc-r4k, if CPU_FREQ is enabled ?

Me neither and the best way would be to update the clocksource frequency dynamically the same way it's done for cevt-r4k and MIPS GIC timers. Alas the clocksource doesn't support it. Due to this together with CPU-freq facility enabled we have to use a very slow DW APB Timer instead of the fast embedded into the CPU core r4k and MIPS GIC timers. Just note the difference: it takes 220 ns to read the counter from DW APB Timer in comparison to a few nanoseconds reading from MIPS GIC and R4K. So IMO disabling the timer as you suggest isn't the best option. By making the CPUFREQ and CSRC_R4K mutual exclusive we'd assume a use-case that the system will always use the CPU-freq facility changing the CPU reference frequency. This is obviously not true. Noone prevents the system administrator to leave the default setting of the CPU-freq with fixed frequency and select a faster, more accurate timer like in our case.

My idea was not to try to predict how the system would be used, but to let the system administration to choose which timer is applicable in particular usecase enabling a safest one by default. So if CPUFREQ is available, then we fallback to the external timer as safest one. If the system user wouldn't need to have the CPUFREQ facility utilized, then the system administrator would want to leave the default CPU-freq governor with pre-defined CPU frequency and select either R4K (MIPS) or MIPS GIC timer just by writing its name into /sys/bus/clocksource/devices/clocksource0/current_clocksource .

I should note, that currently CPU_FREQ won't be available if there is no MIPS_EXTERNAL_TIMER available for the platform. It's prohibited by means of the conditional kbuild config inclusion declared in the arch/mips/Kconfig:
+ if CPU_SUPPORTS_CPUFREQ && MIPS_EXTERNAL_TIMER
+ source "drivers/cpufreq/Kconfig"
+ endif
So if there is no external timer working independently from the CPU core clock source, the CPUFREQ won't be available to select for the kernel. Though currently this limitation is supposed to be applicable for the R4K/MIPS GIC clocksource timers only since clockevents must work fine in unstable reference clock conditions.

So what can we do to improve the patch? First one is a solution I suggested in this patch but it could be a bit altered by using IS_ENABLED() macro to:
+ clocksource_mips.rating = !IS_ENABLED(CONFIG_CPU_FREQ) ?

(quoted diff) +1 -0
+			    200 + mips_hpt_frequency / 10000000 : 99;

Another idea I discovered when have been searching through the x86 arch code. x86's got the same problem with TSC timer, but it doesn't disable it if CPU-frequency is switched on. Instead it just marks it as unstable by calling the clocksource_mark_unstable() method if CPU frequency changes. I suggest to implement the same approach in our case of MIPS GIC (another patchset I've sent, see "clocksource: Fix MIPS GIC and DW APB Timer for Baikal-T1 SoC support" in your email client) and R4K timers. We'll subscribe to the CPU frequency change and if it changes we'll call clocksource_mark_unstable() with MIPS GIC and R4K clocksource handlers passed. This shall reduce their rating and cause selecting a clocksource with better one. BTW I suppose it won't be necessary to initially lower the rating of the MIPS GIC and R4K clocksource timers if this is implemented.

So, what do you think?

-Sergey

Quoted 6 lines

Thomas.

Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]
#55 lore raw

Re: [PATCH v2 18/20] mips: csrc-r4k: Decrease r4k-clocksource rating if CPU_FREQ enabled

Quoted 3 lines · Serge Semin

Thomas,
Could you take a look at my comment below so I could proceed with the patchset v3 development?

I can't help, but using r4k clocksource with changing frequency is probaly only usefull as a random generator. So IMHO the only two options are disabling it or implement what arch/x86/kernel/tsc.c does.

Thomas.

Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]
#56 lore raw

Re: [PATCH v2 18/20] mips: csrc-r4k: Decrease r4k-clocksource rating if CPU_FREQ enabled

Quoted 8 lines · Thomas Bogendoerfer
Quoted 3 lines · Serge Semin

Thomas,
Could you take a look at my comment below so I could proceed with the patchset v3 development?

I can't help, but using r4k clocksource with changing frequency is probaly only usefull as a random generator. So IMHO the only two options are disabling it or implement what arch/x86/kernel/tsc.c does.

Then it's settled. I'll resend the series with csrc-r4k updated to have the tsc-like design implemented.

-Sergey

Quoted 6 lines

Thomas.

Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]
#57 lore raw

Re: [PATCH v2 18/20] mips: csrc-r4k: Decrease r4k-clocksource rating if CPU_FREQ enabled

In reply to Thomas Bogendoerfer
Quoted 10 lines · Thomas Bogendoerfer
Quoted 3 lines · Serge Semin

Thomas,
Could you take a look at my comment below so I could proceed with the patchset v3 development?

I can't help, but using r4k clocksource with changing frequency is probaly only usefull as a random generator. So IMHO the only two options are disabling it or implement what arch/x86/kernel/tsc.c does.

Thomas.

Thomas, could you proceed with the rest of the patches review?

├─>[PATCH v2 16/20] bus: cdmm: Add MIPS R5 arch support
├─>[PATCH v2 15/20] mips: cdmm: Add mti,mips-cdmm dtb node support
├─>[PATCH v2 13/20] mips: early_printk_8250: Use offset-sized IO-mem accessors
├─>[PATCH v2 12/20] mips: MAAR: Add XPA mode support
├─>[PATCH v2 10/20] mips: Add CONFIG/CONFIG6/Cause reg fields macro
└─>[PATCH v2 09/20] mips: Add CP0 Write Merge config support

It would be great if I fixed more comments in the next patchset version.

-Sergey

Quoted 4 lines
Signature · 3 lines
-- 
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]
#58 lore raw

Re: [PATCH v2 18/20] mips: csrc-r4k: Decrease r4k-clocksource rating if CPU_FREQ enabled

Quoted 10 lines · Serge Semin
Quoted 9 lines · Thomas Bogendoerfer
Quoted 3 lines · Serge Semin

Thomas,
Could you take a look at my comment below so I could proceed with the patchset v3 development?

I can't help, but using r4k clocksource with changing frequency is probaly only usefull as a random generator. So IMHO the only two options are disabling it or implement what arch/x86/kernel/tsc.c does.

Quoted 5 lines
Quoted 1 line

Thomas.

Thomas, could you proceed with the rest of the patches review?

├─>[PATCH v2 16/20] bus: cdmm: Add MIPS R5 arch support
├─>[PATCH v2 15/20] mips: cdmm: Add mti,mips-cdmm dtb node support

both are not my call, but look ok to me.

Quoted 1 line

├─>[PATCH v2 13/20] mips: early_printk_8250: Use offset-sized IO-mem accessors

that's broken. A reg shift of 2 doesn't mean we could use 32bit access to the registers on other platforms. As I don't think adding some ifdefery makes things nicer, just implement the your prom_putchar in board code.

Quoted 1 line

├─>[PATCH v2 12/20] mips: MAAR: Add XPA mode support

looks ok so far.

Quoted 1 line

├─>[PATCH v2 10/20] mips: Add CONFIG/CONFIG6/Cause reg fields macro

that is fine

Quoted 1 line

└─>[PATCH v2 09/20] mips: Add CP0 Write Merge config support

this is IMHO a dangerous change. Enabling write merging for any CPU supporting it might triggers bugs. Do it in your board bringup code and at the moment I don't see a reason for the rest of that patch.

Thomas.

Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]
#59 lore raw

Re: [PATCH v2 18/20] mips: csrc-r4k: Decrease r4k-clocksource rating if CPU_FREQ enabled

Quoted 11 lines · Thomas Bogendoerfer
Quoted 10 lines · Serge Semin
Quoted 9 lines · Thomas Bogendoerfer
Quoted 3 lines · Serge Semin

Thomas,
Could you take a look at my comment below so I could proceed with the patchset v3 development?

I can't help, but using r4k clocksource with changing frequency is probaly only usefull as a random generator. So IMHO the only two options are disabling it or implement what arch/x86/kernel/tsc.c does.

Quoted 7 lines
Quoted 5 lines
Quoted 1 line

Thomas.

Thomas, could you proceed with the rest of the patches review?

├─>[PATCH v2 16/20] bus: cdmm: Add MIPS R5 arch support
├─>[PATCH v2 15/20] mips: cdmm: Add mti,mips-cdmm dtb node support

both are not my call, but look ok to me.

Can I add your Reviewed-by tag there then?

Quoted 6 lines
Quoted 2 lines

├─>[PATCH v2 13/20] mips: early_printk_8250: Use offset-sized IO-mem accessors

that's broken. A reg shift of 2 doesn't mean we could use 32bit access to the registers on other platforms. As I don't think adding some ifdefery makes things nicer, just implement the your prom_putchar in board code.

I thought about that initially, but then I decided to alter the generic early_printk_8250 code instead. My version of prom_putchar() would be almost the same as one implemented in the early_printk_8250 module except minor modification of replacing readb/writeb methods with readl/writel. So I didn't want to duplicate the code, but wanted to provide a general way to fix the problem potentially also for another platforms.

Since you don't like this fix alternatively I'd suggest to add the reg_width parameter passed to the setup_8250_early_printk_port() method like this:
-setup_8250_early_printk_port(unsigned long base, unsigned int reg_shift,
- unsigned int timeout)
+setup_8250_early_printk_port(unsigned long base, unsigned int reg_shift,
+ unsigned int reg_width, unsigned int timeout)

By reg_width parameter we could determine the actual width of the register:
static inline u8 serial_in(int offset)
{
- return readb(serial8250_base + (offset << serial8250_reg_shift));
+ u8 ret = 0xFF;
+

+       offset <<= serial8250_reg_shift;
+       switch (serial8250_reg_width) {
+       case 1:

+ ret = readb(serial8250_base + offset);
+ break;
+ case 2:
+ ret = readw(serial8250_base + offset);
+ break;
+ case 4:
+ ret = readl(serial8250_base + offset);
+ break;
+ default:
+ break;
+ }
+
+ return ret;
}

The similar modification will be implemented for serial_out(). I'll also modify the currently available setup_8250_early_printk_port() calls so they would pass the reg_with as 1 to activate the normal readb/writeb IO methods.

What do you think about this?

Quoted 4 lines
Quoted 2 lines

├─>[PATCH v2 12/20] mips: MAAR: Add XPA mode support

looks ok so far.

Can I add your Reviewed-by tag there then?

Quoted 4 lines
Quoted 2 lines

├─>[PATCH v2 10/20] mips: Add CONFIG/CONFIG6/Cause reg fields macro

that is fine

Can I add your Reviewed-by tag there then?

Quoted 7 lines
Quoted 2 lines

└─>[PATCH v2 09/20] mips: Add CP0 Write Merge config support

this is IMHO a dangerous change. Enabling write merging for any CPU supporting it might triggers bugs. Do it in your board bringup code and at the moment I don't see a reason for the rest of that patch.

Let's at least leave the mm_config() implementation but without the write-merge enabling by default. Providing features availability macro cpu_has_mm_sysad/cpu_has_mm_full and c0 config fields MIPS_CPU_MM_SYSAD/MIPS_CPU_MM_FULL defined will be useful in the platform-specific Write-Merge enable/disable procedure. For instance, in the my prom_init() method I could use them to implement a code pattern like:

+ if (cpu_has_mm_full) {
+ unsigned int config0 = read_c0_config();
+ config0 = (config0 & ~MIPS_CONF_MM) | MIPS_CONF_MM_FULL;
+ write_c0_config(config0);
+ }

By doing so I can manually enable/disable the MM feature in the cpu-feature-overrides.h. Without that I'd have to locally define these macro, which isn't good seeing they are in fact generic and can be useful for other platforms with SYSAD and FULL MM feature available. What do you think?

-Sergey

Quoted 6 lines

Thomas.

Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]
#60 lore raw

Re: [PATCH v2 18/20] mips: csrc-r4k: Decrease r4k-clocksource rating if CPU_FREQ enabled

Quoted 14 lines · Serge Semin
Quoted 13 lines · Thomas Bogendoerfer
Quoted 12 lines · Serge Semin
Quoted 10 lines · Thomas Bogendoerfer
Quoted 3 lines · Serge Semin

Thomas,
Could you take a look at my comment below so I could proceed with the patchset v3 development?

I can't help, but using r4k clocksource with changing frequency is probaly only usefull as a random generator. So IMHO the only two options are disabling it or implement what arch/x86/kernel/tsc.c does.

Thomas.

Quoted 7 lines
Quoted 5 lines
Quoted 3 lines

Thomas, could you proceed with the rest of the patches review?

├─>[PATCH v2 16/20] bus: cdmm: Add MIPS R5 arch support
├─>[PATCH v2 15/20] mips: cdmm: Add mti,mips-cdmm dtb node support

both are not my call, but look ok to me.

Can I add your Reviewed-by tag there then?

only for 16/20. 15/20 looks ok to me, but I have not enough insides on the hardware to say this is good.

Quoted 44 lines
Quoted 5 lines
Quoted 1 line

├─>[PATCH v2 13/20] mips: early_printk_8250: Use offset-sized IO-mem accessors

that's broken. A reg shift of 2 doesn't mean we could use 32bit access to the registers on other platforms. As I don't think adding some ifdefery makes things nicer, just implement the your prom_putchar in board code.

I thought about that initially, but then I decided to alter the generic early_printk_8250 code instead. My version of prom_putchar() would be almost the same as one implemented in the early_printk_8250 module except minor modification of replacing readb/writeb methods with readl/writel. So I didn't want to duplicate the code, but wanted to provide a general way to fix the problem potentially also for another platforms.

Since you don't like this fix alternatively I'd suggest to add the reg_width parameter passed to the setup_8250_early_printk_port() method like this:

(quoted diff) +2 -2
-setup_8250_early_printk_port(unsigned long base, unsigned int reg_shift,
-                             unsigned int timeout)
+setup_8250_early_printk_port(unsigned long base, unsigned int reg_shift,
+                             unsigned int reg_width, unsigned int timeout)

By reg_width parameter we could determine the actual width of the register:
static inline u8 serial_in(int offset)
{

(quoted diff) +18 -1
-       return readb(serial8250_base + (offset << serial8250_reg_shift));
+       u8 ret = 0xFF;
+
+       offset <<= serial8250_reg_shift;
+       switch (serial8250_reg_width) {
+       case 1:
+               ret = readb(serial8250_base + offset);
+               break;
+       case 2:
+               ret = readw(serial8250_base + offset);
+               break;
+       case 4:
+               ret = readl(serial8250_base + offset);
+               break;
+       default:
+               break;
+       }
+
+       return ret;
 }
Quoted 1 line

The similar modification will be implemented for serial_out(). I'll also modify

look at the lines of code you are adding. Doing your own prom_putchar will probably have less lines.

Quoted 1 line

What do you think about this?

please do your own prom_putchar.

Quoted 6 lines
Quoted 4 lines
Quoted 2 lines

├─>[PATCH v2 12/20] mips: MAAR: Add XPA mode support

looks ok so far.

Can I add your Reviewed-by tag there then?

As I'm the maintainer of the part, I've simply applied it.

Quoted 6 lines
Quoted 4 lines
Quoted 2 lines

├─>[PATCH v2 10/20] mips: Add CONFIG/CONFIG6/Cause reg fields macro

that is fine

Can I add your Reviewed-by tag there then?

As this didn't apply cleanly, I'll apply it after you've resent it. IMHO no need for a Reviewed-by.

Quoted 10 lines
Quoted 6 lines
Quoted 1 line

└─>[PATCH v2 09/20] mips: Add CP0 Write Merge config support

this is IMHO a dangerous change. Enabling write merging for any CPU supporting it might triggers bugs. Do it in your board bringup code and at the moment I don't see a reason for the rest of that patch.

Let's at least leave the mm_config() implementation but without the write-merge enabling by default. Providing features availability macro cpu_has_mm_sysad/cpu_has_mm_full and c0 config fields

do you have a user of that ? I'm not introducing code nobody uses.

Review

Quoted 7 lines

I could use them to implement a code pattern like:

(quoted diff) +5 -0
+	if (cpu_has_mm_full) {
+		unsigned int config0 = read_c0_config();
+               config0 = (config0 & ~MIPS_CONF_MM) | MIPS_CONF_MM_FULL;
+               write_c0_config(config0);
+	}

you know you are running on a R5 core, so you know you have MM_FULL. No need to check this.

Quoted 4 lines

By doing so I can manually enable/disable the MM feature in the cpu-feature-overrides.h. Without that I'd have to locally define these macro, which isn't good seeing they are in fact generic and can be useful for other platforms with SYSAD and FULL MM feature available. What do you think?

To me this is a hardware feature I expect to be done by firmware and Linux shouldn't care about it, if it doesn't have any software implications.

Thomas.

Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]
#61 lore raw

Re: [PATCH v2 18/20] mips: csrc-r4k: Decrease r4k-clocksource rating if CPU_FREQ enabled

Quoted 19 lines · Thomas Bogendoerfer
Quoted 18 lines · Serge Semin
Quoted 17 lines · Thomas Bogendoerfer
Quoted 15 lines · Serge Semin
Quoted 10 lines · Thomas Bogendoerfer
Quoted 3 lines · Serge Semin

Thomas,
Could you take a look at my comment below so I could proceed with the patchset v3 development?

I can't help, but using r4k clocksource with changing frequency is probaly only usefull as a random generator. So IMHO the only two options are disabling it or implement what arch/x86/kernel/tsc.c does.

Thomas.

Thomas, could you proceed with the rest of the patches review?

├─>[PATCH v2 16/20] bus: cdmm: Add MIPS R5 arch support
├─>[PATCH v2 15/20] mips: cdmm: Add mti,mips-cdmm dtb node support
Quoted 6 lines
Quoted 3 lines
Quoted 1 line

both are not my call, but look ok to me.

Can I add your Reviewed-by tag there then?

only for 16/20. 15/20 looks ok to me, but I have not enough insides on the hardware to say this is good.

Ok.

Quoted 45 lines
Quoted 45 lines
Quoted 6 lines
Quoted 2 lines

├─>[PATCH v2 13/20] mips: early_printk_8250: Use offset-sized IO-mem accessors

that's broken. A reg shift of 2 doesn't mean we could use 32bit access to the registers on other platforms. As I don't think adding some ifdefery makes things nicer, just implement the your prom_putchar in board code.

I thought about that initially, but then I decided to alter the generic early_printk_8250 code instead. My version of prom_putchar() would be almost the same as one implemented in the early_printk_8250 module except minor modification of replacing readb/writeb methods with readl/writel. So I didn't want to duplicate the code, but wanted to provide a general way to fix the problem potentially also for another platforms.

Since you don't like this fix alternatively I'd suggest to add the reg_width parameter passed to the setup_8250_early_printk_port() method like this:

(quoted diff) +2 -2
-setup_8250_early_printk_port(unsigned long base, unsigned int reg_shift,
-                             unsigned int timeout)
+setup_8250_early_printk_port(unsigned long base, unsigned int reg_shift,
+                             unsigned int reg_width, unsigned int timeout)

By reg_width parameter we could determine the actual width of the register:
static inline u8 serial_in(int offset)
{

(quoted diff) +18 -1
-       return readb(serial8250_base + (offset << serial8250_reg_shift));
+       u8 ret = 0xFF;
+
+       offset <<= serial8250_reg_shift;
+       switch (serial8250_reg_width) {
+       case 1:
+               ret = readb(serial8250_base + offset);
+               break;
+       case 2:
+               ret = readw(serial8250_base + offset);
+               break;
+       case 4:
+               ret = readl(serial8250_base + offset);
+               break;
+       default:
+               break;
+       }
+
+       return ret;
 }
Quoted 9 lines
Quoted 1 line

The similar modification will be implemented for serial_out(). I'll also modify

look at the lines of code you are adding. Doing your own prom_putchar will probably have less lines.

Quoted 1 line

What do you think about this?

please do your own prom_putchar.

Ok.

Quoted 11 lines
Quoted 7 lines
Quoted 5 lines
Quoted 3 lines

├─>[PATCH v2 12/20] mips: MAAR: Add XPA mode support

looks ok so far.

Can I add your Reviewed-by tag there then?

As I'm the maintainer of the part, I've simply applied it.

Quoted 1 line
Quoted 1 line
Quoted 8 lines
Quoted 5 lines
Quoted 3 lines
Quoted 1 line

├─>[PATCH v2 10/20] mips: Add CONFIG/CONFIG6/Cause reg fields macro

that is fine

Can I add your Reviewed-by tag there then?

As this didn't apply cleanly, I'll apply it after you've resent it. IMHO no need for a Reviewed-by.

Ok.

Quoted 8 lines
Quoted 8 lines
Quoted 7 lines
Quoted 2 lines

└─>[PATCH v2 09/20] mips: Add CP0 Write Merge config support

this is IMHO a dangerous change. Enabling write merging for any CPU supporting it might triggers bugs. Do it in your board bringup code and at the moment I don't see a reason for the rest of that patch.

Quoted 6 lines
Quoted 3 lines

Let's at least leave the mm_config() implementation but without the write-merge enabling by default. Providing features availability macro cpu_has_mm_sysad/cpu_has_mm_full and c0 config fields

do you have a user of that ? I'm not introducing code nobody uses.

See my comment below.

Quoted 11 lines

Review

Quoted 7 lines

I could use them to implement a code pattern like:

(quoted diff) +5 -0
+	if (cpu_has_mm_full) {
+		unsigned int config0 = read_c0_config();
+               config0 = (config0 & ~MIPS_CONF_MM) | MIPS_CONF_MM_FULL;
+               write_c0_config(config0);
+	}

you know you are running on a R5 core, so you know you have MM_FULL. No need to check this.

Quoted 8 lines
Quoted 4 lines

By doing so I can manually enable/disable the MM feature in the cpu-feature-overrides.h. Without that I'd have to locally define these macro, which isn't good seeing they are in fact generic and can be useful for other platforms with SYSAD and FULL MM feature available. What do you think?

To me this is a hardware feature I expect to be done by firmware and Linux shouldn't care about it, if it doesn't have any software implications.

I think there is a misunderstanding here. In this patch I am not enabling Write-Merge feature for any memory range. I am enabling the UCA Cache Coherency attribute to be available for utilization. See the user-manual info regarding the CP0.CONFIG.MM field:
Write Merge.This bit indicates whether write-through merging is enabled in the 32-byte collapsing write buffer.
0: No merging allowed
1: Merging allowed

In order to have the Write-merging really enabled for a particular PFN one have to mark its TLB entry with UCA (EntryLoX.C[3:5] = 7) attribute. So in this patch I am attempting to detect whether the feature is either already enabled or if available to enable it for utilization.

If there is no misunderstanding and you said what you said, that even enabling the feature for utilization might be dangerous, let's at least leave the MIPS_CONF_MM, MIPS_CONF_MM_FULL and MIPS_CONF_MM_SYS_SYSAD fields definition in the "arch/mips/include/asm/mipsregs.h" header. I'll use them to enable the write-merge in my platform code.

What do you think?

-Sergey

Quoted 6 lines

Thomas.

Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]
#62 lore raw

Re: [PATCH v2 18/20] mips: csrc-r4k: Decrease r4k-clocksource rating if CPU_FREQ enabled

Quoted 5 lines · Serge Semin
Quoted 4 lines · Thomas Bogendoerfer
Quoted 3 lines · Serge Semin
Quoted 2 lines · Thomas Bogendoerfer
Quoted 1 line · Serge Semin

On Fri, May 15, 2020 at 11:06:47PM +0200, Thomas Bogendoerfer wrote:

[nip]

Quoted 50 lines
Quoted 7 lines
Quoted 7 lines
Quoted 6 lines
Quoted 1 line

└─>[PATCH v2 09/20] mips: Add CP0 Write Merge config support

this is IMHO a dangerous change. Enabling write merging for any CPU supporting it might triggers bugs. Do it in your board bringup code and at the moment I don't see a reason for the rest of that patch.

Quoted 6 lines
Quoted 3 lines

Let's at least leave the mm_config() implementation but without the write-merge enabling by default. Providing features availability macro cpu_has_mm_sysad/cpu_has_mm_full and c0 config fields

do you have a user of that ? I'm not introducing code nobody uses.

See my comment below.

Quoted 11 lines

Review

Quoted 7 lines

I could use them to implement a code pattern like:

(quoted diff) +5 -0
+	if (cpu_has_mm_full) {
+		unsigned int config0 = read_c0_config();
+               config0 = (config0 & ~MIPS_CONF_MM) | MIPS_CONF_MM_FULL;
+               write_c0_config(config0);
+	}

you know you are running on a R5 core, so you know you have MM_FULL. No need to check this.

Quoted 8 lines
Quoted 4 lines

By doing so I can manually enable/disable the MM feature in the cpu-feature-overrides.h. Without that I'd have to locally define these macro, which isn't good seeing they are in fact generic and can be useful for other platforms with SYSAD and FULL MM feature available. What do you think?

To me this is a hardware feature I expect to be done by firmware and Linux shouldn't care about it, if it doesn't have any software implications.

I think there is a misunderstanding here. In this patch I am not enabling Write-Merge feature for any memory range. I am enabling the UCA Cache Coherency attribute to be available for utilization. See the user-manual info regarding the CP0.CONFIG.MM field:
Write Merge.This bit indicates whether write-through merging is enabled in the 32-byte collapsing write buffer.
0: No merging allowed
1: Merging allowed

In order to have the Write-merging really enabled for a particular PFN one have to mark its TLB entry with UCA (EntryLoX.C[3:5] = 7) attribute. So in this patch I am attempting to detect whether the feature is either already enabled or if available to enable it for utilization.

Quoted 8 lines

If there is no misunderstanding and you said what you said, that even enabling the feature for utilization might be dangerous, let's at least leave the MIPS_CONF_MM, MIPS_CONF_MM_FULL and MIPS_CONF_MM_SYS_SYSAD fields definition in the "arch/mips/include/asm/mipsregs.h" header. I'll use them to enable the write-merge in my platform code.

What do you think?

Thomas,
Could you also give me your comment on the above, so to make sure that we understood each other correctly in this question?

-Sergey

#63 lore raw

Re: [PATCH v2 18/20] mips: csrc-r4k: Decrease r4k-clocksource rating if CPU_FREQ enabled

In reply to Serge Semin
Quoted 1 line · Serge Semin

I think there is a misunderstanding here. In this patch I am not enabling

you are right, I've missed the fact, that this also needs to be enabled in TLB entries. Strange that MIPS added the enable bit while R10k simply do uncached acclerated, whenever TLB entry selects it.

Quoted 7 lines

If there is no misunderstanding and you said what you said, that even enabling the feature for utilization might be dangerous, let's at least leave the MIPS_CONF_MM, MIPS_CONF_MM_FULL and MIPS_CONF_MM_SYS_SYSAD fields definition in the "arch/mips/include/asm/mipsregs.h" header. I'll use them to enable the write-merge in my platform code.

What do you think?

I withdraw my concerns and will apply the patch as is.

Thomas.

Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]
#64 lore raw

Re: [PATCH v2 18/20] mips: csrc-r4k: Decrease r4k-clocksource rating if CPU_FREQ enabled

Quoted 13 lines · Thomas Bogendoerfer
Quoted 1 line · Serge Semin

I think there is a misunderstanding here. In this patch I am not enabling

you are right, I've missed the fact, that this also needs to be enabled in TLB entries. Strange that MIPS added the enable bit while R10k simply do uncached acclerated, whenever TLB entry selects it.

Quoted 6 lines

If there is no misunderstanding and you said what you said, that even enabling the feature for utilization might be dangerous, let's at least leave the MIPS_CONF_MM, MIPS_CONF_MM_FULL and MIPS_CONF_MM_SYS_SYSAD fields definition in the "arch/mips/include/asm/mipsregs.h" header. I'll use them to enable the write-merge in my platform code.

Quoted 3 lines
Quoted 1 line

What do you think?

I withdraw my concerns and will apply the patch as is.

Great! Thanks.

-Sergey

Quoted 6 lines

Thomas.

Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]
#65 lore raw

Re: [PATCH v2 18/20] mips: csrc-r4k: Decrease r4k-clocksource rating if CPU_FREQ enabled

In reply to Thomas Bogendoerfer
Quoted 70 lines · Thomas Bogendoerfer
Quoted 14 lines · Serge Semin
Quoted 13 lines · Thomas Bogendoerfer
Quoted 12 lines · Serge Semin
Quoted 10 lines · Thomas Bogendoerfer
Quoted 3 lines · Serge Semin

Thomas,
Could you take a look at my comment below so I could proceed with the patchset v3 development?

I can't help, but using r4k clocksource with changing frequency is probaly only usefull as a random generator. So IMHO the only two options are disabling it or implement what arch/x86/kernel/tsc.c does.

Thomas.

Quoted 7 lines
Quoted 5 lines
Quoted 3 lines

Thomas, could you proceed with the rest of the patches review?

├─>[PATCH v2 16/20] bus: cdmm: Add MIPS R5 arch support
├─>[PATCH v2 15/20] mips: cdmm: Add mti,mips-cdmm dtb node support

both are not my call, but look ok to me.

Can I add your Reviewed-by tag there then?

only for 16/20. 15/20 looks ok to me, but I have not enough insides on the hardware to say this is good.

Quoted 44 lines
Quoted 5 lines
Quoted 1 line

├─>[PATCH v2 13/20] mips: early_printk_8250: Use offset-sized IO-mem accessors

that's broken. A reg shift of 2 doesn't mean we could use 32bit access to the registers on other platforms. As I don't think adding some ifdefery makes things nicer, just implement the your prom_putchar in board code.

I thought about that initially, but then I decided to alter the generic early_printk_8250 code instead. My version of prom_putchar() would be almost the same as one implemented in the early_printk_8250 module except minor modification of replacing readb/writeb methods with readl/writel. So I didn't want to duplicate the code, but wanted to provide a general way to fix the problem potentially also for another platforms.

Since you don't like this fix alternatively I'd suggest to add the reg_width parameter passed to the setup_8250_early_printk_port() method like this:

(quoted diff) +2 -2
-setup_8250_early_printk_port(unsigned long base, unsigned int reg_shift,
-                             unsigned int timeout)
+setup_8250_early_printk_port(unsigned long base, unsigned int reg_shift,
+                             unsigned int reg_width, unsigned int timeout)

By reg_width parameter we could determine the actual width of the register:
static inline u8 serial_in(int offset)
{

(quoted diff) +18 -1
-       return readb(serial8250_base + (offset << serial8250_reg_shift));
+       u8 ret = 0xFF;
+
+       offset <<= serial8250_reg_shift;
+       switch (serial8250_reg_width) {
+       case 1:
+               ret = readb(serial8250_base + offset);
+               break;
+       case 2:
+               ret = readw(serial8250_base + offset);
+               break;
+       case 4:
+               ret = readl(serial8250_base + offset);
+               break;
+       default:
+               break;
+       }
+
+       return ret;
 }
Quoted 8 lines
Quoted 1 line

The similar modification will be implemented for serial_out(). I'll also modify

look at the lines of code you are adding. Doing your own prom_putchar will probably have less lines.

Quoted 1 line

What do you think about this?

please do your own prom_putchar.

One more time regarding this problem but in appliance to another part of the MIPS code. I've missed the patch to draw your attention to:
[PATCH v2 14/20] mips: Use offset-sized IO-mem accessors in CPS debug printout

There I've applied the same fix as in the patch:
[PATCH v2 13/20] mips: early_printk_8250: Use offset-sized IO-mem accessors

Since you don't like the way I initially fixed it, suppose there we don't have another way but to introduce something like CONFIG_MIPS_CPS_NS16550_WIDTH parameter to select a proper accessors, like sw in our case, and sb by defaul). Right?

(Note UART_L is incorrectly created in that patch, I'll remove that macro in v3.)

-Sergey

Quoted 59 lines
Quoted 8 lines
Quoted 6 lines
Quoted 4 lines

├─>[PATCH v2 12/20] mips: MAAR: Add XPA mode support

looks ok so far.

Can I add your Reviewed-by tag there then?

As I'm the maintainer of the part, I've simply applied it.

Quoted 6 lines
Quoted 4 lines
Quoted 2 lines

├─>[PATCH v2 10/20] mips: Add CONFIG/CONFIG6/Cause reg fields macro

that is fine

Can I add your Reviewed-by tag there then?

As this didn't apply cleanly, I'll apply it after you've resent it. IMHO no need for a Reviewed-by.

Quoted 10 lines
Quoted 6 lines
Quoted 1 line

└─>[PATCH v2 09/20] mips: Add CP0 Write Merge config support

this is IMHO a dangerous change. Enabling write merging for any CPU supporting it might triggers bugs. Do it in your board bringup code and at the moment I don't see a reason for the rest of that patch.

Let's at least leave the mm_config() implementation but without the write-merge enabling by default. Providing features availability macro cpu_has_mm_sysad/cpu_has_mm_full and c0 config fields

do you have a user of that ? I'm not introducing code nobody uses.

Review

Quoted 7 lines

I could use them to implement a code pattern like:

(quoted diff) +5 -0
+	if (cpu_has_mm_full) {
+		unsigned int config0 = read_c0_config();
+               config0 = (config0 & ~MIPS_CONF_MM) | MIPS_CONF_MM_FULL;
+               write_c0_config(config0);
+	}

you know you are running on a R5 core, so you know you have MM_FULL. No need to check this.

Quoted 4 lines

By doing so I can manually enable/disable the MM feature in the cpu-feature-overrides.h. Without that I'd have to locally define these macro, which isn't good seeing they are in fact generic and can be useful for other platforms with SYSAD and FULL MM feature available. What do you think?

To me this is a hardware feature I expect to be done by firmware and Linux shouldn't care about it, if it doesn't have any software implications.

Thomas.

Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]
#66 lore raw

Re: [PATCH v2 18/20] mips: csrc-r4k: Decrease r4k-clocksource rating if CPU_FREQ enabled

Quoted 4 lines · Serge Semin
Quoted 3 lines · Thomas Bogendoerfer
Quoted 2 lines · Serge Semin
Quoted 1 line · Thomas Bogendoerfer

On Mon, May 18, 2020 at 04:48:20PM +0300, Serge Semin wrote:

[nip]

Quoted 60 lines
Quoted 44 lines
Quoted 44 lines
Quoted 5 lines
Quoted 1 line

├─>[PATCH v2 13/20] mips: early_printk_8250: Use offset-sized IO-mem accessors

that's broken. A reg shift of 2 doesn't mean we could use 32bit access to the registers on other platforms. As I don't think adding some ifdefery makes things nicer, just implement the your prom_putchar in board code.

I thought about that initially, but then I decided to alter the generic early_printk_8250 code instead. My version of prom_putchar() would be almost the same as one implemented in the early_printk_8250 module except minor modification of replacing readb/writeb methods with readl/writel. So I didn't want to duplicate the code, but wanted to provide a general way to fix the problem potentially also for another platforms.

Since you don't like this fix alternatively I'd suggest to add the reg_width parameter passed to the setup_8250_early_printk_port() method like this:

(quoted diff) +2 -2
-setup_8250_early_printk_port(unsigned long base, unsigned int reg_shift,
-                             unsigned int timeout)
+setup_8250_early_printk_port(unsigned long base, unsigned int reg_shift,
+                             unsigned int reg_width, unsigned int timeout)

By reg_width parameter we could determine the actual width of the register:
static inline u8 serial_in(int offset)
{

(quoted diff) +18 -1
-       return readb(serial8250_base + (offset << serial8250_reg_shift));
+       u8 ret = 0xFF;
+
+       offset <<= serial8250_reg_shift;
+       switch (serial8250_reg_width) {
+       case 1:
+               ret = readb(serial8250_base + offset);
+               break;
+       case 2:
+               ret = readw(serial8250_base + offset);
+               break;
+       case 4:
+               ret = readl(serial8250_base + offset);
+               break;
+       default:
+               break;
+       }
+
+       return ret;
 }
Quoted 8 lines
Quoted 1 line

The similar modification will be implemented for serial_out(). I'll also modify

look at the lines of code you are adding. Doing your own prom_putchar will probably have less lines.

Quoted 1 line

What do you think about this?

please do your own prom_putchar.

One more time regarding this problem but in appliance to another part of the MIPS code. I've missed the patch to draw your attention to:
[PATCH v2 14/20] mips: Use offset-sized IO-mem accessors in CPS debug printout

There I've applied the same fix as in the patch:
[PATCH v2 13/20] mips: early_printk_8250: Use offset-sized IO-mem accessors

Quoted 8 lines

Since you don't like the way I initially fixed it, suppose there we don't have another way but to introduce something like CONFIG_MIPS_CPS_NS16550_WIDTH parameter to select a proper accessors, like sw in our case, and sb by defaul). Right?

(Note UART_L is incorrectly created in that patch, I'll remove that macro in v3.)

Hm, actually no. The macro is correct. According to the code _mips_cps_putc() always perform lw from the UART MMIO registers. This must be a bug. Don't you think?

-Sergey

Quoted 3 lines

-Sergey

#67 lore raw

Re: [PATCH v2 18/20] mips: csrc-r4k: Decrease r4k-clocksource rating if CPU_FREQ enabled

In reply to Serge Semin
Quoted 4 lines · Serge Semin

Since you don't like the way I initially fixed it, suppose there we don't have another way but to introduce something like CONFIG_MIPS_CPS_NS16550_WIDTH parameter to select a proper accessors, like sw in our case, and sb by defaul). Right?

to be on the safe side it's probably the best thing. But I don't know enough about CPS_NS16550 to judge whether shift value correlates with possible access width.

Thomas.

Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]
#68 lore raw

Re: [PATCH v2 18/20] mips: csrc-r4k: Decrease r4k-clocksource rating if CPU_FREQ enabled

Quoted 9 lines · Thomas Bogendoerfer
Quoted 4 lines · Serge Semin

Since you don't like the way I initially fixed it, suppose there we don't have another way but to introduce something like CONFIG_MIPS_CPS_NS16550_WIDTH parameter to select a proper accessors, like sw in our case, and sb by defaul). Right?

to be on the safe side it's probably the best thing. But I don't know enough about CPS_NS16550 to judge whether shift value correlates with possible access width.

The base address passed to the _mips_cps_putc() leaf is UART-base address. It has nothing to do with CPS. See:
/**
* _mips_cps_putc() - write a character to the UART
* @a0: ASCII character to write
* @t9: UART base address
*/
LEAF(_mips_cps_putc)

1:      lw              t0, UART_LSR_OFS(t9)
        andi            t0, t0, UART_LSR_TEMT
        beqz            t0, 1b
        sb              a0, UART_TX_OFS(t9)
        jr              ra
        END(_mips_cps_putc)

So it's base address must be accessed with proper alignment. On our case it's lw/sw instructions. Regarding using lw in the first line of the function. That's must be a bug, since further in the same function they use sb to access the UART Tx register. So reading a data from UART_LSR register should be also byte-sized by using lb.

-Sergey

Quoted 6 lines

Thomas.

Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]
#69 lore raw

Re: [PATCH v2 18/20] mips: csrc-r4k: Decrease r4k-clocksource rating if CPU_FREQ enabled

Quoted 7 lines · Serge Semin
Quoted 6 lines · Thomas Bogendoerfer
Quoted 4 lines · Serge Semin

Since you don't like the way I initially fixed it, suppose there we don't have another way but to introduce something like CONFIG_MIPS_CPS_NS16550_WIDTH parameter to select a proper accessors, like sw in our case, and sb by defaul). Right?

Quoted 6 lines
Quoted 3 lines

to be on the safe side it's probably the best thing. But I don't know enough about CPS_NS16550 to judge whether shift value correlates with possible access width.

The base address passed to the _mips_cps_putc() leaf is UART-base address. It has nothing to do with CPS. See:

ok, I'm confused. So this isn't an uart inside CPS hardware, but an uart used by CPS code for debug output, right ?

To solve the issued please add CONFIG_MIPS_CPS_NS16550_WIDTH to select the correct access width.

Thomas.

Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]
#70 lore raw

Re: [PATCH v2 18/20] mips: csrc-r4k: Decrease r4k-clocksource rating if CPU_FREQ enabled

Quoted 12 lines · Thomas Bogendoerfer
Quoted 11 lines · Serge Semin
Quoted 9 lines · Thomas Bogendoerfer
Quoted 4 lines · Serge Semin

Since you don't like the way I initially fixed it, suppose there we don't have another way but to introduce something like CONFIG_MIPS_CPS_NS16550_WIDTH parameter to select a proper accessors, like sw in our case, and sb by defaul). Right?

to be on the safe side it's probably the best thing. But I don't know enough about CPS_NS16550 to judge whether shift value correlates with possible access width.

Quoted 5 lines
Quoted 2 lines

The base address passed to the _mips_cps_putc() leaf is UART-base address. It has nothing to do with CPS. See:

ok, I'm confused. So this isn't an uart inside CPS hardware, but an uart used by CPS code for debug output, right ?

Right. It's not CPS, but just UART available on the system. See a comment in the arch/mips/kernel/cps-vec-ns16550.S:
/**
* mips_cps_bev_dump() - dump relevant exception state to UART
* @a0: pointer to NULL-terminated ASCII string naming the exception *
* Write information that may be useful in debugging an exception to the
* UART configured by CONFIG_MIPS_CPS_NS16550_*.
*...
*/
LEAF(mips_cps_bev_dump)
move s0, ra
move s1, a0

li t9, CKSEG1ADDR(CONFIG_MIPS_CPS_NS16550_BASE)
...

See the base is just loaded to the t9 register.

Quoted 3 lines

To solve the issued please add CONFIG_MIPS_CPS_NS16550_WIDTH to select the correct access width.

Ok. Thanks.

-Sergey

Quoted 6 lines

Thomas.

Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]
#71 lore raw

PATCH v2 19/20 mips: cevt-r4k: Update the r4k-clockevent frequency in sync with CPU

In reply to [email protected]

From: Serge Semin <[email protected]>

Due to being embedded into the CPU cores MIPS count/compare timer frequency is changed together with the CPU clocks alteration. In case if frequency really changes the kernel clockevent framework must be notified, otherwise the kernel timers won't work correctly. Fix this by calling clockevents_update_freq() for each r4k clockevent handlers registered per available CPUs.

Traditionally MIPS r4k-clock are clocked with CPU frequency divided by 2. But this isn't true for some of the platforms. Due to this we have to save the basic CPU frequency, so then use it to scale the initial timer frequency (mips_hpt_frequency) and pass the updated value further to the clockevent framework.

Cc Alexey Malahov <[email protected]>
Cc Thomas Bogendoerfer <[email protected]>
Cc Paul Burton <[email protected]>
Cc Ralf Baechle <[email protected]>
Cc Greg Kroah-Hartman <[email protected]>
Cc Arnd Bergmann <[email protected]>
Cc Rob Herring <[email protected]>
---
 arch/mips/kernel/cevt-r4k.c | 44 +++++++++++++++++++++++++++++++++++++
 1 file changed, 44 insertions(+)
arch/mips/kernel/cevt-r4k.c +44 -0
index 17a9cbb8b3df..f5b72fb7d5ee 100644
@@ -8,6 +8,7 @@
  */
 #include <linux/clockchips.h>
 #include <linux/interrupt.h>
+#include <linux/cpufreq.h>
 #include <linux/percpu.h>
 #include <linux/smp.h>
 #include <linux/irq.h>
@@ -250,6 +251,49 @@ unsigned int __weak get_c0_compare_int(void)
 	return MIPS_CPU_IRQ_BASE + cp0_compare_irq;
 }
 
+#ifdef CONFIG_CPU_FREQ
+
+static unsigned long mips_ref_freq;
+
+static int cpufreq_callback(struct notifier_block *nb,
+			    unsigned long val, void *data)
+{
+	struct cpufreq_freqs *freq = data;
+	struct clock_event_device *cd;
+	unsigned long rate;
+	int cpu;
+
+	if (!mips_ref_freq)
+		mips_ref_freq = freq->old;
+
+	if (val == CPUFREQ_POSTCHANGE) {
+		rate = cpufreq_scale(mips_hpt_frequency, mips_ref_freq,
+				     freq->new);
+
+		for_each_cpu(cpu, freq->policy->cpus) {
+			cd = &per_cpu(mips_clockevent_device, cpu);
+
+			clockevents_update_freq(cd, rate);
+		}
+	}
+
+	return 0;
+}
+
+static struct notifier_block cpufreq_notifier = {
+	.notifier_call  = cpufreq_callback,
+};
+
+static int __init register_cpufreq_notifier(void)
+{
+	return cpufreq_register_notifier(&cpufreq_notifier,
+					 CPUFREQ_TRANSITION_NOTIFIER);
+
+}
+core_initcall(register_cpufreq_notifier);
+
+#endif /* !CONFIG_CPU_FREQ */
+
 int r4k_clockevent_init(void)
 {
 	unsigned long flags = IRQF_PERCPU | IRQF_TIMER | IRQF_SHARED;
Signature · 1 line
2.25.1
#72 lore raw

Re: [PATCH v2 19/20] mips: cevt-r4k: Update the r4k-clockevent frequency in sync with CPU

Quoted 13 lines · On Wed, May 06, 2020 at 08:42:37PM +0300, [email protected]…
arch/mips/kernel/cevt-r4k.c +1 -0
index 17a9cbb8b3df..f5b72fb7d5ee 100644
@@ -8,6 +8,7 @@
  */
 #include <linux/clockchips.h>
 #include <linux/interrupt.h>
+#include <linux/cpufreq.h>
 #include <linux/percpu.h>
 #include <linux/smp.h>
 #include <linux/irq.h>

arch/mips/kernel/cevt-r4k.c unsigned int __weak get_c0_compare_int(void) · line 259

Quoted 9 lines
arch/mips/kernel/cevt-r4k.c +6 -0
@@ -250,6 +251,49 @@ unsigned int __weak get_c0_compare_int(void)
 	return MIPS_CPU_IRQ_BASE + cp0_compare_irq;
 }
 
+#ifdef CONFIG_CPU_FREQ
+
+static unsigned long mips_ref_freq;
+
+static int cpufreq_callback(struct notifier_block *nb,
+			    unsigned long val, void *data)

please prefix function names with r4k_ to make them different from the other ones you implemented in kernel/time.c. I know they are static, but keeping different names makes looking at crashes easier.

arch/mips/kernel/cevt-r4k.c unsigned int __weak get_c0_compare_int(void) · line 267

Quoted 7 lines
arch/mips/kernel/cevt-r4k.c +7 -0
+	struct cpufreq_freqs *freq = data;
+	struct clock_event_device *cd;
+	unsigned long rate;
+	int cpu;
+
+	if (!mips_ref_freq)
+		mips_ref_freq = freq->old;

isn't this the same as mips_hpt_frequency ?

Thomas.

Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]
#73 lore raw

Re: [PATCH v2 19/20] mips: cevt-r4k: Update the r4k-clockevent frequency in sync with CPU

Quoted 17 lines · Thomas Bogendoerfer
Quoted 16 lines · On Wed, May 06, 2020 at 08:42:37PM +0300, [email protected]…
arch/mips/kernel/cevt-r4k.c +1 -0
index 17a9cbb8b3df..f5b72fb7d5ee 100644
@@ -8,6 +8,7 @@
  */
 #include <linux/clockchips.h>
 #include <linux/interrupt.h>
+#include <linux/cpufreq.h>
 #include <linux/percpu.h>
 #include <linux/smp.h>
 #include <linux/irq.h>
@@ -250,6 +251,49 @@ unsigned int __weak get_c0_compare_int(void)
 	return MIPS_CPU_IRQ_BASE + cp0_compare_irq;
 }
 
Quoted 10 lines

arch/mips/kernel/cevt-r4k.c unsigned int __weak get_c0_compare_int(void) · line 259

Quoted 6 lines
arch/mips/kernel/cevt-r4k.c +6 -0
+#ifdef CONFIG_CPU_FREQ
+
+static unsigned long mips_ref_freq;
+
+static int cpufreq_callback(struct notifier_block *nb,
+			    unsigned long val, void *data)

please prefix function names with r4k_ to make them different from the other ones you implemented in kernel/time.c. I know they are static, but keeping different names makes looking at crashes easier.

Agreed. I'll fix it in v3.

Quoted 10 lines

arch/mips/kernel/cevt-r4k.c unsigned int __weak get_c0_compare_int(void) · line 267

Quoted 8 lines
arch/mips/kernel/cevt-r4k.c +7 -0
+	struct cpufreq_freqs *freq = data;
+	struct clock_event_device *cd;
+	unsigned long rate;
+	int cpu;
+
+	if (!mips_ref_freq)
+		mips_ref_freq = freq->old;

isn't this the same as mips_hpt_frequency ?

No. Here I save the initial CPU frequency so use one then to scale the mips_hpt_frequency value in accordance with the CPU clock rate change. Yes, mips_hpt_frequency value may initially match the CPU frequency on some platforms but normally the r4k timer is clocked with half of it while some systems may have a complicated algorithm of the timer ref clock rate calculation.

-Sergey

Quoted 6 lines

Thomas.

Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]
#74 lore raw

PATCH v2 20/20 cpufreq: Return zero on success in boost sw setting

In reply to [email protected]

From: Serge Semin <[email protected]>

Recent commit e61a41256edf ("cpufreq: dev_pm_qos_update_request() can return 1 on success") fixed a problem when active policies traverse was falsely stopped due to invalidly treating the non-zero return value from freq_qos_update_request() method as an error. Yes, that function can return positive values if the requested update actually took place. The current problem is that the returned value is then passed to the return cell of the cpufreq_boost_set_sw() (set_boost callback) method. This value is then also analyzed for being non-zero, which is also treated as having an error. As a result during the boost activation we'll get an error returned while having the QOS frequency update successfully performed. Fix this by returning a negative value from the cpufreq_boost_set_sw() if actual error was encountered and zero otherwise treating any positive values as the successful operations completion.

Fixes 18c49926c4bf ("cpufreq: Add QoS requests for userspace constraints")
Cc Alexey Malahov <[email protected]>
Cc Thomas Bogendoerfer <[email protected]>
Cc Paul Burton <[email protected]>
Cc Ralf Baechle <[email protected]>
Cc Arnd Bergmann <[email protected]>
Cc Rob Herring <[email protected]>
---
 drivers/cpufreq/cpufreq.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/cpufreq/cpufreq.c +1 -1
index 045f9fe157ce..5870cdca88cf 100644
@@ -2554,7 +2554,7 @@ static int cpufreq_boost_set_sw(int state)
 			break;
 	}
 
-	return ret;
+	return ret < 0 ? ret : 0;
 }
 
 int cpufreq_boost_trigger_state(int state)
Signature · 1 line
2.25.1
#75 lore raw

Re: [PATCH v2 20/20] cpufreq: Return zero on success in boost sw setting

Quoted 38 lines · On 5/6/2020 7:42 PM, [email protected] wrote:

From: Serge Semin <[email protected]>

Recent commit e61a41256edf ("cpufreq: dev_pm_qos_update_request() can return 1 on success") fixed a problem when active policies traverse was falsely stopped due to invalidly treating the non-zero return value from freq_qos_update_request() method as an error. Yes, that function can return positive values if the requested update actually took place. The current problem is that the returned value is then passed to the return cell of the cpufreq_boost_set_sw() (set_boost callback) method. This value is then also analyzed for being non-zero, which is also treated as having an error. As a result during the boost activation we'll get an error returned while having the QOS frequency update successfully performed. Fix this by returning a negative value from the cpufreq_boost_set_sw() if actual error was encountered and zero otherwise treating any positive values as the successful operations completion.

Fixes 18c49926c4bf ("cpufreq: Add QoS requests for userspace constraints")
Cc Alexey Malahov <[email protected]>
Cc Thomas Bogendoerfer <[email protected]>
Cc Paul Burton <[email protected]>
Cc Ralf Baechle <[email protected]>
Cc Arnd Bergmann <[email protected]>
Cc Rob Herring <[email protected]>
---
  drivers/cpufreq/cpufreq.c | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)
drivers/cpufreq/cpufreq.c +0 -0
index 045f9fe157ce..5870cdca88cf 100644

drivers/cpufreq/cpufreq.c static int cpufreq_boost_set_sw(int state) · line 2560

Quoted 8 lines
drivers/cpufreq/cpufreq.c +1 -1
@@ -2554,7 +2554,7 @@ static int cpufreq_boost_set_sw(int state)
  			break;
  	}
 
-	return ret;
+	return ret < 0 ? ret : 0;
  }
 
  int cpufreq_boost_trigger_state(int state)

IMO it is better to update the caller of this function to handle the positive value possibly returned by it correctly.

Thanks!

#76 lore raw

Re: [PATCH v2 20/20] cpufreq: Return zero on success in boost sw setting

Hello Rafael,

Quoted 34 lines · Rafael J. Wysocki
Quoted 33 lines · On 5/6/2020 7:42 PM, [email protected] wrote:

From: Serge Semin <[email protected]>

Recent commit e61a41256edf ("cpufreq: dev_pm_qos_update_request() can return 1 on success") fixed a problem when active policies traverse was falsely stopped due to invalidly treating the non-zero return value from freq_qos_update_request() method as an error. Yes, that function can return positive values if the requested update actually took place. The current problem is that the returned value is then passed to the return cell of the cpufreq_boost_set_sw() (set_boost callback) method. This value is then also analyzed for being non-zero, which is also treated as having an error. As a result during the boost activation we'll get an error returned while having the QOS frequency update successfully performed. Fix this by returning a negative value from the cpufreq_boost_set_sw() if actual error was encountered and zero otherwise treating any positive values as the successful operations completion.

Fixes 18c49926c4bf ("cpufreq: Add QoS requests for userspace constraints")
Cc Alexey Malahov <[email protected]>
Cc Thomas Bogendoerfer <[email protected]>
Cc Paul Burton <[email protected]>
Cc Ralf Baechle <[email protected]>
Cc Arnd Bergmann <[email protected]>
Cc Rob Herring <[email protected]>
---
  drivers/cpufreq/cpufreq.c | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)
Quoted 14 lines

drivers/cpufreq/cpufreq.c static int cpufreq_boost_set_sw(int state) · line 2558

Quoted 11 lines
drivers/cpufreq/cpufreq.c +1 -1
index 045f9fe157ce..5870cdca88cf 100644
@@ -2554,7 +2554,7 @@ static int cpufreq_boost_set_sw(int state)
  			break;
  	}
-	return ret;
+	return ret < 0 ? ret : 0;
  }
  int cpufreq_boost_trigger_state(int state)

IMO it is better to update the caller of this function to handle the positive value possibly returned by it correctly.

Could you elaborate why? Viresh seems to be ok with this solution.

As I see it the caller doesn't expect the positive value returned by the original freq_qos_update_request(). It just doesn't need to know whether the effective policy has been updated or not, it only needs to make sure the operations has been successful. Moreover the positive value is related only to the !last! active policy, which doesn't give the caller a full picture of the policy change anyway. So taking all of these into account I'd leave the fix as is.

Regards,
-Sergey

Quoted 4 lines

Thanks!

maintainer
#77 lore raw

Re: [PATCH v2 20/20] cpufreq: Return zero on success in boost sw setting

Quoted 12 lines · Serge Semin
Quoted 10 lines · Rafael J. Wysocki

drivers/cpufreq/cpufreq.c static int cpufreq_boost_set_sw(int state) · line 2558

Quoted 7 lines
drivers/cpufreq/cpufreq.c +1 -1
@@ -2554,7 +2554,7 @@ static int cpufreq_boost_set_sw(int state)
  			break;
  	}
-	return ret;
+	return ret < 0 ? ret : 0;
  }
  int cpufreq_boost_trigger_state(int state)

IMO it is better to update the caller of this function to handle the positive value possibly returned by it correctly.

Quoted 1 line

Could you elaborate why? Viresh seems to be ok with this solution.

And it is absolutely fine for Rafael to not agree with it :)

Quoted 7 lines

As I see it the caller doesn't expect the positive value returned by the original freq_qos_update_request(). It just doesn't need to know whether the effective policy has been updated or not, it only needs to make sure the operations has been successful. Moreover the positive value is related only to the !last! active policy, which doesn't give the caller a full picture of the policy change anyway. So taking all of these into account I'd leave the fix as is.

Rafael: This function is called via a function pointer, which can call this or a platform dependent routine (like in acpi-cpufreq.c), and it would be reasonable IMO for the return of that callback to only look for 0 or negative values, as is generally done in the kernel. And so this solution looked okay to me as the positive return is coming from the implementation detail here.

Signature · 1 line
viresh
#78 lore raw

Re: [PATCH v2 20/20] cpufreq: Return zero on success in boost sw setting

Quoted 21 lines · Viresh Kumar
Quoted 11 lines · Serge Semin
Quoted 9 lines · Rafael J. Wysocki

drivers/cpufreq/cpufreq.c static int cpufreq_boost_set_sw(int state) · line 2558

Quoted 7 lines
drivers/cpufreq/cpufreq.c +1 -1
@@ -2554,7 +2554,7 @@ static int cpufreq_boost_set_sw(int state)
   			break;
   	}
-	return ret;
+	return ret < 0 ? ret : 0;
   }
   int cpufreq_boost_trigger_state(int state)

IMO it is better to update the caller of this function to handle the positive value possibly returned by it correctly.

Could you elaborate why? Viresh seems to be ok with this solution.

And it is absolutely fine for Rafael to not agree with it :)

Quoted 7 lines

As I see it the caller doesn't expect the positive value returned by the original freq_qos_update_request(). It just doesn't need to know whether the effective policy has been updated or not, it only needs to make sure the operations has been successful. Moreover the positive value is related only to the !last! active policy, which doesn't give the caller a full picture of the policy change anyway. So taking all of these into account I'd leave the fix as is.

Quoted 4 lines

Rafael: This function is called via a function pointer, which can call this or a platform dependent routine (like in acpi-cpufreq.c), and it would be reasonable IMO for the return of that callback to only look for 0 or negative values, as is generally done in the kernel.

But it only has one caller that can easily check ret < 0 instead of just ret, so the extra branch can be saved.

That said if you really only want it to return 0 on success, you may as well add a ret = 0; statement (with a comment explaining why it is needed) after the last break in the loop.

Cheers!

maintainer
#79 lore raw

Re: [PATCH v2 20/20] cpufreq: Return zero on success in boost sw setting

Quoted 3 lines · Rafael J. Wysocki

That said if you really only want it to return 0 on success, you may as well add a ret = 0; statement (with a comment explaining why it is needed) after the last break in the loop.

That can be done as well, but will be a bit less efficient as the loop will execute once for each policy, and so the statement will run multiple times. Though it isn't going to add any significant latency in the code.

Signature · 1 line
viresh
#80 lore raw

Re: [PATCH v2 20/20] cpufreq: Return zero on success in boost sw setting

Quoted 9 lines · CEST Viresh Kumar
Quoted 3 lines · Rafael J. Wysocki

That said if you really only want it to return 0 on success, you may as well add a ret = 0; statement (with a comment explaining why it is needed) after the last break in the loop.

That can be done as well, but will be a bit less efficient as the loop will execute once for each policy, and so the statement will run multiple times. Though it isn't going to add any significant latency in the code.

Right.

However, the logic in this entire function looks somewhat less than straightforward to me, because it looks like it should return an error on the first policy without a frequency table (having a frequency table depends on the driver and that is the same for all policies, so it is pointless to iterate any further in that case).

Also, the error should not be -EINVAL, because that means "invalid argument" which would be the state value.

So I would do something like this:

---
 drivers/cpufreq/cpufreq.c |   11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)
drivers/cpufreq/cpufreq.c +6 -5
Index: linux-pm/drivers/cpufreq/cpufreq.c
===================================================================
@@ -2535,26 +2535,27 @@ EXPORT_SYMBOL_GPL(cpufreq_update_limits)
 static int cpufreq_boost_set_sw(int state)
 {
 	struct cpufreq_policy *policy;
-	int ret = -EINVAL;
 
 	for_each_active_policy(policy) {
+		int ret;
+
 		if (!policy->freq_table)
-			continue;
+			return -ENXIO;
 
 		ret = cpufreq_frequency_table_cpuinfo(policy,
 						      policy->freq_table);
 		if (ret) {
 			pr_err("%s: Policy frequency update failed\n",
 			       __func__);
-			break;
+			return ret;
 		}
 
 		ret = freq_qos_update_request(policy->max_freq_req, policy->max);
 		if (ret < 0)
-			break;
+			return ret;
 	}
 
-	return ret;
+	return 0;
 }
 
 int cpufreq_boost_trigger_state(int state)
maintainer Acked-by
#81 lore raw

Re: [PATCH v2 20/20] cpufreq: Return zero on success in boost sw setting

Quoted 55 lines · Rafael J. Wysocki
Quoted 9 lines · CEST Viresh Kumar
Quoted 3 lines · Rafael J. Wysocki

That said if you really only want it to return 0 on success, you may as well add a ret = 0; statement (with a comment explaining why it is needed) after the last break in the loop.

That can be done as well, but will be a bit less efficient as the loop will execute once for each policy, and so the statement will run multiple times. Though it isn't going to add any significant latency in the code.

Right.

However, the logic in this entire function looks somewhat less than straightforward to me, because it looks like it should return an error on the first policy without a frequency table (having a frequency table depends on the driver and that is the same for all policies, so it is pointless to iterate any further in that case).

Also, the error should not be -EINVAL, because that means "invalid argument" which would be the state value.

So I would do something like this:

---
 drivers/cpufreq/cpufreq.c |   11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)

Index: linux-pm/drivers/cpufreq/cpufreq.c
===================================================================

drivers/cpufreq/cpufreq.c +4 -3
@@ -2535,26 +2535,27 @@ EXPORT_SYMBOL_GPL(cpufreq_update_limits)
 static int cpufreq_boost_set_sw(int state)
 {
 	struct cpufreq_policy *policy;
-	int ret = -EINVAL;
 
 	for_each_active_policy(policy) {
+		int ret;
+
 		if (!policy->freq_table)
-			continue;
+			return -ENXIO;
 
 		ret = cpufreq_frequency_table_cpuinfo(policy,
 						      policy->freq_table);
 		if (ret) {
 			pr_err("%s: Policy frequency update failed\n",
 			       __func__);
-			break;
+			return ret;
 		}
 
 		ret = freq_qos_update_request(policy->max_freq_req, policy->max);

drivers/cpufreq/cpufreq.c EXPORT_SYMBOL_GPL(cpufreq_update_limits) · line 2561

Quoted 10 lines
drivers/cpufreq/cpufreq.c +2 -2
 		if (ret < 0)
-			break;
+			return ret;
 	}
 
-	return ret;
+	return 0;
 }
 
 int cpufreq_boost_trigger_state(int state)
Signature · 1 line
viresh
#82 lore raw

Re: [PATCH v2 20/20] cpufreq: Return zero on success in boost sw setting

Quoted 61 lines · Viresh Kumar
Quoted 60 lines · Rafael J. Wysocki
Quoted 9 lines · CEST Viresh Kumar
Quoted 3 lines · Rafael J. Wysocki

That said if you really only want it to return 0 on success, you may as well add a ret = 0; statement (with a comment explaining why it is needed) after the last break in the loop.

That can be done as well, but will be a bit less efficient as the loop will execute once for each policy, and so the statement will run multiple times. Though it isn't going to add any significant latency in the code.

Right.

However, the logic in this entire function looks somewhat less than straightforward to me, because it looks like it should return an error on the first policy without a frequency table (having a frequency table depends on the driver and that is the same for all policies, so it is pointless to iterate any further in that case).

Also, the error should not be -EINVAL, because that means "invalid argument" which would be the state value.

So I would do something like this:

---
 drivers/cpufreq/cpufreq.c |   11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)

Index: linux-pm/drivers/cpufreq/cpufreq.c
===================================================================

drivers/cpufreq/cpufreq.c +5 -4
@@ -2535,26 +2535,27 @@ EXPORT_SYMBOL_GPL(cpufreq_update_limits)
 static int cpufreq_boost_set_sw(int state)
 {
 	struct cpufreq_policy *policy;
-	int ret = -EINVAL;
 
 	for_each_active_policy(policy) {
+		int ret;
+
 		if (!policy->freq_table)
-			continue;
+			return -ENXIO;
 
 		ret = cpufreq_frequency_table_cpuinfo(policy,
 						      policy->freq_table);
 		if (ret) {
 			pr_err("%s: Policy frequency update failed\n",
 			       __func__);
-			break;
+			return ret;
 		}
 
 		ret = freq_qos_update_request(policy->max_freq_req, policy->max);
 		if (ret < 0)
-			break;
+			return ret;
 	}
 
Quoted 7 lines

drivers/cpufreq/cpufreq.c EXPORT_SYMBOL_GPL(cpufreq_update_limits) · line 2561

Quoted 5 lines
drivers/cpufreq/cpufreq.c +1 -1
-	return ret;
+	return 0;
 }
 
 int cpufreq_boost_trigger_state(int state)

Ok. Thanks for the comments. Shall I resend the patch with update Rafael suggests or you'll merge the Rafael's fix in yourself?

-Sergey

Quoted 3 lines
Signature · 2 lines
-- 
viresh
#83 lore raw

Re: [PATCH v2 20/20] cpufreq: Return zero on success in boost sw setting

Quoted 66 lines · CEST Serge Semin
Quoted 65 lines · Viresh Kumar
Quoted 64 lines · Rafael J. Wysocki
Quoted 9 lines · CEST Viresh Kumar
Quoted 3 lines · Rafael J. Wysocki

That said if you really only want it to return 0 on success, you may as well add a ret = 0; statement (with a comment explaining why it is needed) after the last break in the loop.

That can be done as well, but will be a bit less efficient as the loop will execute once for each policy, and so the statement will run multiple times. Though it isn't going to add any significant latency in the code.

Right.

However, the logic in this entire function looks somewhat less than straightforward to me, because it looks like it should return an error on the first policy without a frequency table (having a frequency table depends on the driver and that is the same for all policies, so it is pointless to iterate any further in that case).

Also, the error should not be -EINVAL, because that means "invalid argument" which would be the state value.

So I would do something like this:

---
 drivers/cpufreq/cpufreq.c |   11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)

Index: linux-pm/drivers/cpufreq/cpufreq.c
===================================================================

drivers/cpufreq/cpufreq.c +6 -5
@@ -2535,26 +2535,27 @@ EXPORT_SYMBOL_GPL(cpufreq_update_limits)
 static int cpufreq_boost_set_sw(int state)
 {
 	struct cpufreq_policy *policy;
-	int ret = -EINVAL;
 
 	for_each_active_policy(policy) {
+		int ret;
+
 		if (!policy->freq_table)
-			continue;
+			return -ENXIO;
 
 		ret = cpufreq_frequency_table_cpuinfo(policy,
 						      policy->freq_table);
 		if (ret) {
 			pr_err("%s: Policy frequency update failed\n",
 			       __func__);
-			break;
+			return ret;
 		}
 
 		ret = freq_qos_update_request(policy->max_freq_req, policy->max);
 		if (ret < 0)
-			break;
+			return ret;
 	}
 
-	return ret;
+	return 0;
 }
 
Quoted 6 lines
Quoted 3 lines

drivers/cpufreq/cpufreq.c EXPORT_SYMBOL_GPL(cpufreq_update_limits) · line 2561

Quoted 1 line
drivers/cpufreq/cpufreq.c +0 -0
 int cpufreq_boost_trigger_state(int state)

Ok. Thanks for the comments. Shall I resend the patch with update Rafael suggests or you'll merge the Rafael's fix in yourself?

I'll apply the fix directly, thanks!

#84 lore raw

Re: [PATCH v2 20/20] cpufreq: Return zero on success in boost sw setting

Quoted 67 lines · Rafael J. Wysocki
Quoted 66 lines · CEST Serge Semin
Quoted 65 lines · Viresh Kumar
Quoted 64 lines · Rafael J. Wysocki
Quoted 9 lines · CEST Viresh Kumar
Quoted 3 lines · Rafael J. Wysocki

That said if you really only want it to return 0 on success, you may as well add a ret = 0; statement (with a comment explaining why it is needed) after the last break in the loop.

That can be done as well, but will be a bit less efficient as the loop will execute once for each policy, and so the statement will run multiple times. Though it isn't going to add any significant latency in the code.

Right.

However, the logic in this entire function looks somewhat less than straightforward to me, because it looks like it should return an error on the first policy without a frequency table (having a frequency table depends on the driver and that is the same for all policies, so it is pointless to iterate any further in that case).

Also, the error should not be -EINVAL, because that means "invalid argument" which would be the state value.

So I would do something like this:

---
 drivers/cpufreq/cpufreq.c |   11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)

Index: linux-pm/drivers/cpufreq/cpufreq.c
===================================================================

drivers/cpufreq/cpufreq.c +6 -5
@@ -2535,26 +2535,27 @@ EXPORT_SYMBOL_GPL(cpufreq_update_limits)
 static int cpufreq_boost_set_sw(int state)
 {
 	struct cpufreq_policy *policy;
-	int ret = -EINVAL;
 
 	for_each_active_policy(policy) {
+		int ret;
+
 		if (!policy->freq_table)
-			continue;
+			return -ENXIO;
 
 		ret = cpufreq_frequency_table_cpuinfo(policy,
 						      policy->freq_table);
 		if (ret) {
 			pr_err("%s: Policy frequency update failed\n",
 			       __func__);
-			break;
+			return ret;
 		}
 
 		ret = freq_qos_update_request(policy->max_freq_req, policy->max);
 		if (ret < 0)
-			break;
+			return ret;
 	}
 
-	return ret;
+	return 0;
 }
 
Quoted 8 lines
Quoted 6 lines
Quoted 3 lines

drivers/cpufreq/cpufreq.c EXPORT_SYMBOL_GPL(cpufreq_update_limits) · line 2561

Quoted 1 line
drivers/cpufreq/cpufreq.c +0 -0
 int cpufreq_boost_trigger_state(int state)

Ok. Thanks for the comments. Shall I resend the patch with update Rafael suggests or you'll merge the Rafael's fix in yourself?

I'll apply the fix directly, thanks!

Great. Is it going to be available in the repo:
https://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm.git/
?
I'll need it to back port into my local kernel tree. Thanks.

-Sergey

Quoted 3 lines
#85 lore raw

Re: [PATCH v2 20/20] cpufreq: Return zero on success in boost sw setting

Quoted 73 lines · Serge Semin
Quoted 71 lines · Rafael J. Wysocki
Quoted 70 lines · CEST Serge Semin
Quoted 68 lines · Viresh Kumar
Quoted 55 lines · Rafael J. Wysocki
Quoted 9 lines · CEST Viresh Kumar
Quoted 3 lines · Rafael J. Wysocki

That said if you really only want it to return 0 on success, you may as well add a ret = 0; statement (with a comment explaining why it is needed) after the last break in the loop.

That can be done as well, but will be a bit less efficient as the loop will execute once for each policy, and so the statement will run multiple times. Though it isn't going to add any significant latency in the code.

Right.

However, the logic in this entire function looks somewhat less than straightforward to me, because it looks like it should return an error on the first policy without a frequency table (having a frequency table depends on the driver and that is the same for all policies, so it is pointless to iterate any further in that case).

Also, the error should not be -EINVAL, because that means "invalid argument" which would be the state value.

So I would do something like this:

---
 drivers/cpufreq/cpufreq.c |   11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)

Index: linux-pm/drivers/cpufreq/cpufreq.c
===================================================================

drivers/cpufreq/cpufreq.c +4 -3
@@ -2535,26 +2535,27 @@ EXPORT_SYMBOL_GPL(cpufreq_update_limits)
 static int cpufreq_boost_set_sw(int state)
 {
        struct cpufreq_policy *policy;
-       int ret = -EINVAL;
 
        for_each_active_policy(policy) {
+               int ret;
+
                if (!policy->freq_table)
-                       continue;
+                       return -ENXIO;
 
                ret = cpufreq_frequency_table_cpuinfo(policy,
                                                      policy->freq_table);
                if (ret) {
                        pr_err("%s: Policy frequency update failed\n",
                               __func__);
-                       break;
+                       return ret;
                }
 
                ret = freq_qos_update_request(policy->max_freq_req, policy->max);

drivers/cpufreq/cpufreq.c EXPORT_SYMBOL_GPL(cpufreq_update_limits) · line 2561

Quoted 10 lines
drivers/cpufreq/cpufreq.c +2 -2
                if (ret < 0)
-                       break;
+                       return ret;
        }
 
-       return ret;
+       return 0;
 }
 
 int cpufreq_boost_trigger_state(int state)
Quoted 8 lines
Quoted 4 lines
Quoted 2 lines

Ok. Thanks for the comments. Shall I resend the patch with update Rafael suggests or you'll merge the Rafael's fix in yourself?

I'll apply the fix directly, thanks!

Great. Is it going to be available in the repo:
https://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm.git/
?

Yes, it is. Please see the bleeding-edge branch in there, thanks!

#86 lore raw

Re: [PATCH v2 20/20] cpufreq: Return zero on success in boost sw setting

Quoted 78 lines · Rafael J. Wysocki
Quoted 76 lines · Serge Semin
Quoted 74 lines · Rafael J. Wysocki
Quoted 72 lines · CEST Serge Semin
Quoted 61 lines · Viresh Kumar
Quoted 60 lines · Rafael J. Wysocki
Quoted 9 lines · CEST Viresh Kumar
Quoted 3 lines · Rafael J. Wysocki

That said if you really only want it to return 0 on success, you may as well add a ret = 0; statement (with a comment explaining why it is needed) after the last break in the loop.

That can be done as well, but will be a bit less efficient as the loop will execute once for each policy, and so the statement will run multiple times. Though it isn't going to add any significant latency in the code.

Right.

However, the logic in this entire function looks somewhat less than straightforward to me, because it looks like it should return an error on the first policy without a frequency table (having a frequency table depends on the driver and that is the same for all policies, so it is pointless to iterate any further in that case).

Also, the error should not be -EINVAL, because that means "invalid argument" which would be the state value.

So I would do something like this:

---
 drivers/cpufreq/cpufreq.c |   11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)

Index: linux-pm/drivers/cpufreq/cpufreq.c
===================================================================

drivers/cpufreq/cpufreq.c +5 -4
@@ -2535,26 +2535,27 @@ EXPORT_SYMBOL_GPL(cpufreq_update_limits)
 static int cpufreq_boost_set_sw(int state)
 {
        struct cpufreq_policy *policy;
-       int ret = -EINVAL;
 
        for_each_active_policy(policy) {
+               int ret;
+
                if (!policy->freq_table)
-                       continue;
+                       return -ENXIO;
 
                ret = cpufreq_frequency_table_cpuinfo(policy,
                                                      policy->freq_table);
                if (ret) {
                        pr_err("%s: Policy frequency update failed\n",
                               __func__);
-                       break;
+                       return ret;
                }
 
                ret = freq_qos_update_request(policy->max_freq_req, policy->max);
                if (ret < 0)
-                       break;
+                       return ret;
        }
 
Quoted 7 lines

drivers/cpufreq/cpufreq.c EXPORT_SYMBOL_GPL(cpufreq_update_limits) · line 2561

Quoted 5 lines
drivers/cpufreq/cpufreq.c +1 -1
-       return ret;
+       return 0;
 }
 
 int cpufreq_boost_trigger_state(int state)

Ok. Thanks for the comments. Shall I resend the patch with update Rafael suggests or you'll merge the Rafael's fix in yourself?

Quoted 7 lines
Quoted 5 lines
Quoted 1 line

I'll apply the fix directly, thanks!

Great. Is it going to be available in the repo:
https://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm.git/
?

Yes, it is. Please see the bleeding-edge branch in there, thanks!

No credits with at least Reported-by tag? That's sad.(

-Sergey

#87 lore raw

Re: [PATCH v2 20/20] cpufreq: Return zero on success in boost sw setting

Quoted 82 lines · Serge Semin
Quoted 80 lines · Rafael J. Wysocki
Quoted 78 lines · Serge Semin
Quoted 75 lines · Rafael J. Wysocki
Quoted 66 lines · CEST Serge Semin
Quoted 65 lines · Viresh Kumar
Quoted 64 lines · Rafael J. Wysocki
Quoted 9 lines · CEST Viresh Kumar
Quoted 3 lines · Rafael J. Wysocki

That said if you really only want it to return 0 on success, you may as well add a ret = 0; statement (with a comment explaining why it is needed) after the last break in the loop.

That can be done as well, but will be a bit less efficient as the loop will execute once for each policy, and so the statement will run multiple times. Though it isn't going to add any significant latency in the code.

Right.

However, the logic in this entire function looks somewhat less than straightforward to me, because it looks like it should return an error on the first policy without a frequency table (having a frequency table depends on the driver and that is the same for all policies, so it is pointless to iterate any further in that case).

Also, the error should not be -EINVAL, because that means "invalid argument" which would be the state value.

So I would do something like this:

---
 drivers/cpufreq/cpufreq.c |   11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)

Index: linux-pm/drivers/cpufreq/cpufreq.c
===================================================================

drivers/cpufreq/cpufreq.c +6 -5
@@ -2535,26 +2535,27 @@ EXPORT_SYMBOL_GPL(cpufreq_update_limits)
 static int cpufreq_boost_set_sw(int state)
 {
        struct cpufreq_policy *policy;
-       int ret = -EINVAL;
 
        for_each_active_policy(policy) {
+               int ret;
+
                if (!policy->freq_table)
-                       continue;
+                       return -ENXIO;
 
                ret = cpufreq_frequency_table_cpuinfo(policy,
                                                      policy->freq_table);
                if (ret) {
                        pr_err("%s: Policy frequency update failed\n",
                               __func__);
-                       break;
+                       return ret;
                }
 
                ret = freq_qos_update_request(policy->max_freq_req, policy->max);
                if (ret < 0)
-                       break;
+                       return ret;
        }
 
-       return ret;
+       return 0;
 }
 
Quoted 6 lines
Quoted 3 lines

drivers/cpufreq/cpufreq.c EXPORT_SYMBOL_GPL(cpufreq_update_limits) · line 2561

Quoted 1 line
drivers/cpufreq/cpufreq.c +0 -0
 int cpufreq_boost_trigger_state(int state)

Ok. Thanks for the comments. Shall I resend the patch with update Rafael suggests or you'll merge the Rafael's fix in yourself?

I'll apply the fix directly, thanks!

Quoted 7 lines
Quoted 5 lines
Quoted 3 lines

Great. Is it going to be available in the repo:
https://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm.git/
?

Yes, it is. Please see the bleeding-edge branch in there, thanks!

No credits with at least Reported-by tag? That's sad.(

OK, done now, but you are not the only reported of it, so I've added the other reporter too.

Thanks!

#88 lore raw

Re: [PATCH v2 20/20] cpufreq: Return zero on success in boost sw setting

Hi Rafael,

Quoted 82 lines · Rafael J. Wysocki
Quoted 80 lines · Serge Semin
Quoted 78 lines · Rafael J. Wysocki
Quoted 76 lines · Serge Semin
Quoted 74 lines · Rafael J. Wysocki
Quoted 72 lines · CEST Serge Semin
Quoted 61 lines · Viresh Kumar
Quoted 60 lines · Rafael J. Wysocki
Quoted 9 lines · CEST Viresh Kumar
Quoted 3 lines · Rafael J. Wysocki

That said if you really only want it to return 0 on success, you may as well add a ret = 0; statement (with a comment explaining why it is needed) after the last break in the loop.

That can be done as well, but will be a bit less efficient as the loop will execute once for each policy, and so the statement will run multiple times. Though it isn't going to add any significant latency in the code.

Right.

However, the logic in this entire function looks somewhat less than straightforward to me, because it looks like it should return an error on the first policy without a frequency table (having a frequency table depends on the driver and that is the same for all policies, so it is pointless to iterate any further in that case).

Also, the error should not be -EINVAL, because that means "invalid argument" which would be the state value.

So I would do something like this:

---
 drivers/cpufreq/cpufreq.c |   11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)

Index: linux-pm/drivers/cpufreq/cpufreq.c
===================================================================

drivers/cpufreq/cpufreq.c +5 -4
@@ -2535,26 +2535,27 @@ EXPORT_SYMBOL_GPL(cpufreq_update_limits)
 static int cpufreq_boost_set_sw(int state)
 {
        struct cpufreq_policy *policy;
-       int ret = -EINVAL;
 
        for_each_active_policy(policy) {
+               int ret;
+
                if (!policy->freq_table)
-                       continue;
+                       return -ENXIO;
 
                ret = cpufreq_frequency_table_cpuinfo(policy,
                                                      policy->freq_table);
                if (ret) {
                        pr_err("%s: Policy frequency update failed\n",
                               __func__);
-                       break;
+                       return ret;
                }
 
                ret = freq_qos_update_request(policy->max_freq_req, policy->max);
                if (ret < 0)
-                       break;
+                       return ret;
        }
 
Quoted 7 lines

drivers/cpufreq/cpufreq.c EXPORT_SYMBOL_GPL(cpufreq_update_limits) · line 2561

Quoted 5 lines
drivers/cpufreq/cpufreq.c +1 -1
-       return ret;
+       return 0;
 }
 
 int cpufreq_boost_trigger_state(int state)

Ok. Thanks for the comments. Shall I resend the patch with update Rafael suggests or you'll merge the Rafael's fix in yourself?

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 5 lines
Quoted 1 line

I'll apply the fix directly, thanks!

Great. Is it going to be available in the repo:
https://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm.git/
?

Yes, it is. Please see the bleeding-edge branch in there, thanks!

Thanks for CCing me. I will write my next version based on this branch.

Thanks,
Xiongfeng

Quoted 10 lines
Quoted 2 lines

No credits with at least Reported-by tag? That's sad.(

OK, done now, but you are not the only reported of it, so I've added the other reporter too.

Thanks!

.