Skip to contentSkip to conversation

CXl 2.0 emulation Support

PATCH Superseded Jonathan Cameron 4 participants 52 messages Apr 29, 2022 – Sep 13, 2023 Arm/Amlogic Meson SoC support

v10 of 11 Superseded Latest is v11 →

lore
Share & tools
lore
#0 lore raw

v10 cover CXl 2.0 emulation Support

Note that due to a bug in the kernel handling of Mem_Enable in the CXL Device DVSEC a fix is needed to sucessfully use these patches with Linux.
https://lore.kernel.org/linux-cxl/CAPcyv4gFWw9s6r9kUoL-_P9qnaZSJBMz+EcmfY+0g1NHrJQx7Q@mail.gmail.com/T/#t

Note the issue was detected using this emulation.

Changes since v9:

Thanks to Tong Zhang and Markus Armbruster for review of v9.
- Fixed return of all 1s on HDM decoder not matching.
- Off by one error in an assert for register writes.
- Write masks for both PCI config space structures and registers in MMIO space. These were partly queued up for a future patch set, but after review comment from Tong, I decided it was best to put these in place from the start. Note I was a bit undecided on whether to drop review tags for this (it's quit a bit of code, but very simple). Alex, and Adam if you want to take another look at patch 2 in particular that would be great.
- Tag from Markus added.
- Fix for wrong setting of Mem_Enable as described in the thread linked above.

The series can be applied in parts if that makes sense:

Possible partial sets:
1-15 (end with the test of the pxb-cxl host bridge)
16-22 (end with the test for root port and type3 device)
23-39 (end with tests on x86 pc for CFMWS including BIOS table updates) 40-41 (arm64 virt support + simple test case)
42 (documentation - we could pull this forwards to before the arm support) 43-45 (switch support)

gitlab tree at gitlab.com/jic23/qemu cxl-v10-draft-2

As before that tree has additional patches on top that will form part of future postings. This time it has SPDM support via a socket connection to DMTF/spdm_emu / libspdm. I'll send an RFC on that topic as it's a discussion that can go on in parallel.

Several people have asked about contributing additional features. As those come in I'll apply them on top of this series and handle rebases etc as necessary whilst we seek to get this first set of patches upstream.

Updated background info:

Looking in particular for:
* Review of the PCI interactions
* x86 and ARM machine interactions (particularly the memory maps)
* Review of the interleaving approach - is the basic idea acceptable?
* Review of the command line interface.
* CXL related review welcome but much of that got reviewed in earlier versions and hasn't changed substantially.

TODOs:

* Volatile memory devices (easy but it's more code so left for now).
* Hotplug? May not need much but it's not tested yet!
* More tests and tighter verification that values written to hardware are actually valid - stuff that real hardware would check.
* Testing, testing and more testing. I have been running a basic set of ARM and x86 tests on this, but there is always room for more tests and greater automation.
* CFMWS flags as requested by Ben.
* Parititioning support - ability to change the balance of volatile and non volatile memory on demand.
* Trace points as suggested my Mark to help with debugging memory interleaving setup.

Why do we want QEMU emulation of CXL?

As Ben stated in V3, QEMU support has been critical to getting OS software written given lack of availability of hardware supporting the latest CXL features (coupled with very high demand for support being ready in a timely fashion). What has become clear since Ben's v3 is that situation is a continuous one. Whilst we can't talk about them yet, CXL 3.0 features and OS support have been prototyped on top of this support and a lot of the ongoing kernel work is being tested against these patches. The kernel CXL mocking code allows some forms of testing, but QEMU provides a more versatile and extensible platform.

Other features on the qemu-list that build on these include PCI-DOE
/CDAT support from the Avery Design team further showing how this
code is useful. Whilst not directly related this is also the test
platform for work on PCI IDE/CMA + related DMTF SPDM as CXL both
utilizes and extends those technologies and is likely to be an early
adopter.
Refs:
CMA Kernel: https://lore.kernel.org/all/[email protected]/
CMA Qemu: https://lore.kernel.org/qemu-devel/[email protected]/
DOE Qemu: https://lore.kernel.org/qemu-devel/[email protected]/

As can be seen there is non trivial interaction with other areas of Qemu, particularly PCI and keeping this set up to date is proving a burden we'd rather do without :)

Ben mentioned a few other good reasons in v3:
https://lore.kernel.org/qemu-devel/[email protected]/

What we have here is about what you need for it to be useful for testing currently kernel code. Note the kernel code is moving fast so since v4, some features have been introduced we don't yet support in QEMU (e.g. use of the PCIe serial number extended capability).

All comments welcome.

Additional info that was here in v5 is now in the documentation patch.

Thanks,

Jonathan

Ben Widawsky (24):
hw/pci/cxl: Add a CXL component type (interface) hw/cxl/component: Introduce CXL components (8.1.x, 8.2.5) hw/cxl/device: Introduce a CXL device (8.2.8) hw/cxl/device: Implement the CAP array (8.2.8.1-2) hw/cxl/device: Implement basic mailbox (8.2.8.4) hw/cxl/device: Add memory device utilities hw/cxl/device: Add cheap EVENTS implementation (8.2.9.1) hw/cxl/device: Timestamp implementation (8.2.9.3) hw/cxl/device: Add log commands (8.2.9.4) + CEL hw/pxb: Use a type for realizing expanders hw/pci/cxl: Create a CXL bus type
hw/pxb: Allow creation of a CXL PXB (host bridge) hw/cxl/rp: Add a root port
hw/cxl/device: Add a memory device (8.2.8.5) hw/cxl/device: Implement MMIO HDM decoding (8.2.5.12) hw/cxl/device: Add some trivial commands hw/cxl/device: Plumb real Label Storage Area (LSA) sizing hw/cxl/device: Implement get/set Label Storage Area (LSA) hw/cxl/component: Implement host bridge MMIO (8.2.5, table 142) acpi/cxl: Add _OSC implementation (9.14.2)
acpi/cxl: Create the CEDT (9.14.1)
acpi/cxl: Introduce CFMWS structures in CEDT hw/cxl/component Add a dumb HDM decoder handler qtest/cxl: Add more complex test cases with CFMWs

Jonathan Cameron (21):
MAINTAINERS: Add entry for Compute Express Link Emulation
cxl: Machine level control on whether CXL support is enabled qtest/cxl: Introduce initial test for pxb-cxl only. qtests/cxl: Add initial root port and CXL type3 tests hw/cxl/component: Add utils for interleave parameter encoding/decoding hw/cxl/host: Add support for CXL Fixed Memory Windows. hw/pci-host/gpex-acpi: Add support for dsdt construction for pxb-cxl pci/pcie_port: Add pci_find_port_by_pn() CXL/cxl_component: Add cxl_get_hb_cstate()
mem/cxl_type3: Add read and write functions for associated hostmem. cxl/cxl-host: Add memops for CFMWS region.
i386/pc: Enable CXL fixed memory windows
tests/acpi: q35: Allow addition of a CXL test. qtests/bios-tables-test: Add a test for CXL emulation. tests/acpi: Add tables for CXL emulation.
hw/arm/virt: Basic CXL enablement on pci_expander_bridge instances
pxb-cxl
qtest/cxl: Add aarch64 virt test for CXL
docs/cxl: Add initial Compute eXpress Link (CXL) documentation. pci-bridge/cxl_upstream: Add a CXL switch upstream port pci-bridge/cxl_downstream: Add a CXL switch downstream port docs/cxl: Add switch documentation

 MAINTAINERS                         |   7 +
 docs/system/device-emulation.rst    |   1 +
 docs/system/devices/cxl.rst         | 386 ++++++++++++++++++++++
 hw/Kconfig                          |   1 +
 hw/acpi/Kconfig                     |   5 +
 hw/acpi/cxl-stub.c                  |  12 +
 hw/acpi/cxl.c                       | 257 +++++++++++++++
 hw/acpi/meson.build                 |   4 +-
 hw/arm/Kconfig                      |   1 +
 hw/arm/virt-acpi-build.c            |  33 ++
 hw/arm/virt.c                       |  40 ++-
 hw/core/machine.c                   |  28 ++
 hw/cxl/Kconfig                      |   3 +
 hw/cxl/cxl-component-utils.c        | 396 +++++++++++++++++++++++
 hw/cxl/cxl-device-utils.c           | 265 +++++++++++++++
 hw/cxl/cxl-host-stubs.c             |  16 +
 hw/cxl/cxl-host.c                   | 261 +++++++++++++++
 hw/cxl/cxl-mailbox-utils.c          | 478 ++++++++++++++++++++++++++++
 hw/cxl/meson.build                  |  12 +
 hw/i386/acpi-build.c                |  57 +++-
 hw/i386/pc.c                        |  57 +++-
 hw/mem/Kconfig                      |   5 +
 hw/mem/cxl_type3.c                  | 371 +++++++++++++++++++++
 hw/mem/meson.build                  |   1 +
 hw/meson.build                      |   1 +
 hw/pci-bridge/Kconfig               |   5 +
 hw/pci-bridge/cxl_downstream.c      | 249 +++++++++++++++
 hw/pci-bridge/cxl_root_port.c       | 236 ++++++++++++++
 hw/pci-bridge/cxl_upstream.c        | 216 +++++++++++++
 hw/pci-bridge/meson.build           |   1 +
 hw/pci-bridge/pci_expander_bridge.c | 168 +++++++++-
 hw/pci-bridge/pcie_root_port.c      |   6 +-
 hw/pci-host/gpex-acpi.c             |  20 +-
 hw/pci/pci.c                        |  21 +-
 hw/pci/pcie_port.c                  |  25 ++
 include/hw/acpi/cxl.h               |  28 ++
 include/hw/arm/virt.h               |   1 +
 include/hw/boards.h                 |   2 +
 include/hw/cxl/cxl.h                |  65 ++++
 include/hw/cxl/cxl_component.h      | 223 +++++++++++++
 include/hw/cxl/cxl_device.h         | 268 ++++++++++++++++
 include/hw/cxl/cxl_pci.h            | 167 ++++++++++
 include/hw/pci/pci.h                |  14 +
 include/hw/pci/pci_bridge.h         |  20 ++
 include/hw/pci/pci_bus.h            |   7 +
 include/hw/pci/pci_ids.h            |   1 +
 include/hw/pci/pcie_port.h          |   2 +
 qapi/machine.json                   |  21 ++
 qemu-options.hx                     |  38 +++
 scripts/device-crash-test           |   1 +
 softmmu/vl.c                        |  47 +++
 tests/data/acpi/q35/CEDT.cxl        | Bin 0 -> 184 bytes
 tests/data/acpi/q35/DSDT.cxl        | Bin 0 -> 9615 bytes
 tests/qtest/bios-tables-test.c      |  44 +++
 tests/qtest/cxl-test.c              | 181 +++++++++++
 tests/qtest/meson.build             |   5 +
 56 files changed, 4753 insertions(+), 27 deletions(-)
 create mode 100644 docs/system/devices/cxl.rst
 create mode 100644 hw/acpi/cxl-stub.c
 create mode 100644 hw/acpi/cxl.c
 create mode 100644 hw/cxl/Kconfig
 create mode 100644 hw/cxl/cxl-component-utils.c
 create mode 100644 hw/cxl/cxl-device-utils.c
 create mode 100644 hw/cxl/cxl-host-stubs.c
 create mode 100644 hw/cxl/cxl-host.c
 create mode 100644 hw/cxl/cxl-mailbox-utils.c
 create mode 100644 hw/cxl/meson.build
 create mode 100644 hw/mem/cxl_type3.c
 create mode 100644 hw/pci-bridge/cxl_downstream.c
 create mode 100644 hw/pci-bridge/cxl_root_port.c
 create mode 100644 hw/pci-bridge/cxl_upstream.c
 create mode 100644 include/hw/acpi/cxl.h
 create mode 100644 include/hw/cxl/cxl.h
 create mode 100644 include/hw/cxl/cxl_component.h
 create mode 100644 include/hw/cxl/cxl_device.h
 create mode 100644 include/hw/cxl/cxl_pci.h
 create mode 100644 tests/data/acpi/q35/CEDT.cxl
 create mode 100644 tests/data/acpi/q35/DSDT.cxl
 create mode 100644 tests/qtest/cxl-test.c
Signature · 1 line
2.32.0
#1 lore raw

PATCH v10 1/45 hw/pci/cxl: Add a CXL component type (interface)

From: Ben Widawsky <[email protected]>

A CXL component is a hardware entity that implements CXL component registers from the CXL 2.0 spec (8.2.3). Currently these represent 3 general types.
1. Host Bridge
2. Ports (root, upstream, downstream)
3. Devices (memory, other)

A CXL component can be conceptually thought of as a PCIe device with extra functionality when enumerated and enabled. For this reason, CXL does here, and will continue to add on to existing PCI code paths.

Host bridges will typically need to be handled specially and so they can implement this newly introduced interface or not. All other components should implement this interface. Implementing this interface allows the core PCI code to treat these devices as special where appropriate.

---
 hw/pci/pci.c         | 10 ++++++++++
 include/hw/pci/pci.h |  8 ++++++++
 2 files changed, 18 insertions(+)
hw/pci/pci.c +10 -0
index e99417e501..08bd2db2ad 100644
@@ -200,6 +200,11 @@ static const TypeInfo pci_bus_info = {
     .class_init = pci_bus_class_init,
 };
 
+static const TypeInfo cxl_interface_info = {
+    .name          = INTERFACE_CXL_DEVICE,
+    .parent        = TYPE_INTERFACE,
+};
+
 static const TypeInfo pcie_interface_info = {
     .name          = INTERFACE_PCIE_DEVICE,
     .parent        = TYPE_INTERFACE,
@@ -2181,6 +2186,10 @@ static void pci_qdev_realize(DeviceState *qdev, Error **errp)
         pci_dev->cap_present |= QEMU_PCI_CAP_EXPRESS;
     }
 
+    if (object_class_dynamic_cast(klass, INTERFACE_CXL_DEVICE)) {
+        pci_dev->cap_present |= QEMU_PCIE_CAP_CXL;
+    }
+
     pci_dev = do_pci_register_device(pci_dev,
                                      object_get_typename(OBJECT(qdev)),
                                      pci_dev->devfn, errp);
@@ -2937,6 +2946,7 @@ static void pci_register_types(void)
     type_register_static(&pci_bus_info);
     type_register_static(&pcie_bus_info);
     type_register_static(&conventional_pci_interface_info);
+    type_register_static(&cxl_interface_info);
     type_register_static(&pcie_interface_info);
     type_register_static(&pci_device_type_info);
 }
include/hw/pci/pci.h +8 -0
index 3a32b8dd40..98f0d1b844 100644
@@ -194,6 +194,8 @@ enum {
     QEMU_PCIE_LNKSTA_DLLLA = (1 << QEMU_PCIE_LNKSTA_DLLLA_BITNR),
 #define QEMU_PCIE_EXTCAP_INIT_BITNR 9
     QEMU_PCIE_EXTCAP_INIT = (1 << QEMU_PCIE_EXTCAP_INIT_BITNR),
+#define QEMU_PCIE_CXL_BITNR 10
+    QEMU_PCIE_CAP_CXL = (1 << QEMU_PCIE_CXL_BITNR),
 };
 
 #define TYPE_PCI_DEVICE "pci-device"
@@ -201,6 +203,12 @@ typedef struct PCIDeviceClass PCIDeviceClass;
 DECLARE_OBJ_CHECKERS(PCIDevice, PCIDeviceClass,
                      PCI_DEVICE, TYPE_PCI_DEVICE)
 
+/*
+ * Implemented by devices that can be plugged on CXL buses. In the spec, this is
+ * actually a "CXL Component, but we name it device to match the PCI naming.
+ */
+#define INTERFACE_CXL_DEVICE "cxl-device"
+
 /* Implemented by devices that can be plugged on PCI Express buses */
 #define INTERFACE_PCIE_DEVICE "pci-express-device"
 
Signature · 1 line
2.32.0
#2 lore raw

PATCH v10 2/45 hw/cxl/component: Introduce CXL components (8.1.x, 8.2.5)

In reply to Jonathan Cameron

From: Ben Widawsky <[email protected]>

A CXL 2.0 component is any entity in the CXL topology. All components have a analogous function in PCIe. Except for the CXL host bridge, all have a PCIe config space that is accessible via the common PCIe mechanisms. CXL components are enumerated via DVSEC fields in the extended PCIe header space. CXL components will minimally implement some subset of CXL.mem and CXL.cache registers defined in 8.2.5 of the CXL 2.0 specification. Two headers and a utility library are introduced to support the minimum functionality needed to enumerate components.

The cxl_pci header manages bits associated with PCI, specifically the DVSEC and related fields. The cxl_component.h variant has data structures and APIs that are useful for drivers implementing any of the CXL 2.0 components. The library takes care of making use of the DVSEC bits and the CXL.[mem|cache] registers. Per spec, the registers are little endian.

None of the mechanisms required to enumerate a CXL capable hostbridge are introduced at this point.

Note that the CXL.mem and CXL.cache registers used are always 4B wide. It's possible in the future that this constraint will not hold.

---
 hw/Kconfig                     |   1 +
 hw/cxl/Kconfig                 |   3 +
 hw/cxl/cxl-component-utils.c   | 313 +++++++++++++++++++++++++++++++++
 hw/cxl/meson.build             |   4 +
 hw/meson.build                 |   1 +
 include/hw/cxl/cxl.h           |  16 ++
 include/hw/cxl/cxl_component.h | 213 ++++++++++++++++++++++
 include/hw/cxl/cxl_pci.h       | 146 +++++++++++++++
 8 files changed, 697 insertions(+)
hw/Kconfig +1 -0
index ad20cce0a9..50e0952889 100644
@@ -6,6 +6,7 @@ source audio/Kconfig
 source block/Kconfig
 source char/Kconfig
 source core/Kconfig
+source cxl/Kconfig
 source display/Kconfig
 source dma/Kconfig
 source gpio/Kconfig
hw/cxl/Kconfig new file +3 -0
new file mode 100644
index 0000000000..8e67519b16
@@ -0,0 +1,3 @@
+config CXL
+    bool
+    default y if PCI_EXPRESS
hw/cxl/cxl-component-utils.c new file +313 -0

313 lines not on this page · raw message

hw/cxl/meson.build new file +4 -0
new file mode 100644
index 0000000000..3231b5de1e
@@ -0,0 +1,4 @@
+softmmu_ss.add(when: 'CONFIG_CXL',
+               if_true: files(
+                   'cxl-component-utils.c',
+               ))
hw/meson.build +1 -0
index b3366c888e..9992c5101e 100644
@@ -6,6 +6,7 @@ subdir('block')
 subdir('char')
 subdir('core')
 subdir('cpu')
+subdir('cxl')
 subdir('display')
 subdir('dma')
 subdir('gpio')
include/hw/cxl/cxl.h new file +16 -0
new file mode 100644
index 0000000000..8c738c7a2b
@@ -0,0 +1,16 @@
+/*
+ * QEMU CXL Support
+ *
+ * Copyright (c) 2020 Intel
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2. See the
+ * COPYING file in the top-level directory.
+ */
+
+#ifndef CXL_H
+#define CXL_H
+
+#include "cxl_pci.h"
+#include "cxl_component.h"
+
+#endif
include/hw/cxl/cxl_component.h new file +213 -0

213 lines not on this page · raw message

include/hw/cxl/cxl_pci.h new file +146 -0

146 lines not on this page · raw message

Signature · 1 line
2.32.0
#3 lore raw

Re: [PATCH v10 02/45] hw/cxl/component: Introduce CXL components (8.1.x, 8.2.5)

Quoted 21 lines · On Fri, 29 Apr 2022 15:40:27 +0100 Jonathan Cameron…

From: Ben Widawsky <[email protected]>

A CXL 2.0 component is any entity in the CXL topology. All components have a analogous function in PCIe. Except for the CXL host bridge, all have a PCIe config space that is accessible via the common PCIe mechanisms. CXL components are enumerated via DVSEC fields in the extended PCIe header space. CXL components will minimally implement some subset of CXL.mem and CXL.cache registers defined in 8.2.5 of the CXL 2.0 specification. Two headers and a utility library are introduced to support the minimum functionality needed to enumerate components.

The cxl_pci header manages bits associated with PCI, specifically the DVSEC and related fields. The cxl_component.h variant has data structures and APIs that are useful for drivers implementing any of the CXL 2.0 components. The library takes care of making use of the DVSEC bits and the CXL.[mem|cache] registers. Per spec, the registers are little endian.

None of the mechanisms required to enumerate a CXL capable hostbridge are introduced at this point.

Quoted 7 lines

Note that the CXL.mem and CXL.cache registers used are always 4B wide. It's possible in the future that this constraint will not hold.

FYI on a bug, in case anyone else hits it.

Quoted 10 lines
hw/cxl/cxl-component-utils.c new file +10 -0
+static void hdm_init_common(uint32_t *reg_state, uint32_t *write_msk)
+{
+    int decoder_count = 1;
+    int i;
+
+    ARRAY_FIELD_DP32(reg_state, CXL_HDM_DECODER_CAPABILITY, DECODER_COUNT,
+                     cxl_decoder_count_enc(decoder_count));
+    ARRAY_FIELD_DP32(reg_state, CXL_HDM_DECODER_CAPABILITY, TARGET_COUNT, 1);
+    ARRAY_FIELD_DP32(reg_state, CXL_HDM_DECODER_CAPABILITY, INTERLEAVE_256B, 1);
+    ARRAY_FIELD_DP32(reg_state, CXL_HDM_DECODER_CAPABILITY, INTERLEAVE_4K, 1);

hw/cxl/cxl-component-utils.c line 144

Quoted 10 lines
hw/cxl/cxl-component-utils.c new file +10 -0
+    ARRAY_FIELD_DP32(reg_state, CXL_HDM_DECODER_CAPABILITY, POISON_ON_ERR_CAP, 0);
+    ARRAY_FIELD_DP32(reg_state, CXL_HDM_DECODER_GLOBAL_CONTROL,
+                     HDM_DECODER_ENABLE, 0);
+    write_msk[R_CXL_HDM_DECODER_GLOBAL_CONTROL] = 0x3;
+    for (i = 0; i < decoder_count; i++) {
+        write_msk[R_CXL_HDM_DECODER0_BASE_LO + i * 0x20] = 0xf0000000;
+        write_msk[R_CXL_HDM_DECODER0_BASE_HI + i * 0x20] = 0xffffffff;
+        write_msk[R_CXL_HDM_DECODER0_SIZE_LO + i * 0x20] = 0xf0000000;
+        write_msk[R_CXL_HDM_DECODER0_SIZE_HI + i * 0x20] = 0xffffffff;
+        write_msk[R_CXL_HDM_DECODER0_CTRL + i * 0x20] = 0x13ff;

For some unknown reason I missed write masks for the target lists in here. It was hidden from superficial testing by a bug in the kernel code I was using that mean these were mostly written to 0.

I'll send a fix out tomorrow (just adds 0xffffffff masks for each of the target registers).

Given issues I'm seeing in testing around HDM decoder programming I'll look to follow that up with a patch adding some more rigorous checking of the values on commit.

Jonathan

#4 lore raw

PATCH v10 3/45 MAINTAINERS: Add entry for Compute Express Link Emulation

In reply to Jonathan Cameron

From: Jonathan Cameron <[email protected]>

The CXL emulation will be jointly maintained by Ben Widawsky and Jonathan Cameron. Broken out as a separate patch to improve visibility.

---
 MAINTAINERS | 7 +++++++
 1 file changed, 7 insertions(+)
MAINTAINERS +7 -0
index 294c88ace9..3fa11bf1f3 100644
@@ -2545,6 +2545,13 @@ F: qapi/block*.json
 F: qapi/transaction.json
 T: git https://repo.or.cz/qemu/armbru.git block-next
 
+Compute Express Link
+M: Ben Widawsky <[email protected]>
+M: Jonathan Cameron <[email protected]>
+S: Supported
+F: hw/cxl/
+F: include/hw/cxl/
+
 Dirty Bitmaps
 M: Eric Blake <[email protected]>
 M: Vladimir Sementsov-Ogievskiy <[email protected]>
Signature · 1 line
2.32.0
#5 lore raw

PATCH v10 4/45 hw/cxl/device: Introduce a CXL device (8.2.8)

In reply to Jonathan Cameron

From: Ben Widawsky <[email protected]>

A CXL device is a type of CXL component. Conceptually, a CXL device would be a leaf node in a CXL topology. From an emulation perspective, CXL devices are the most complex and so the actual implementation is reserved for discrete commits.

This new device type is specifically catered towards the eventual implementation of a Type3 CXL.mem device, 8.2.8.5 in the CXL 2.0 specification.

---
 include/hw/cxl/cxl.h        |   1 +
 include/hw/cxl/cxl_device.h | 166 ++++++++++++++++++++++++++++++++++++
 2 files changed, 167 insertions(+)
include/hw/cxl/cxl.h +1 -0
index 8c738c7a2b..b9d1ac3fad 100644
@@ -12,5 +12,6 @@
 
 #include "cxl_pci.h"
 #include "cxl_component.h"
+#include "cxl_device.h"
 
 #endif
include/hw/cxl/cxl_device.h new file +166 -0

166 lines not on this page · raw message

Signature · 1 line
2.32.0
#6 lore raw

PATCH v10 5/45 hw/cxl/device: Implement the CAP array (8.2.8.1-2)

In reply to Jonathan Cameron

From: Ben Widawsky <[email protected]>

This implements all device MMIO up to the first capability. That includes the CXL Device Capabilities Array Register, as well as all of the CXL Device Capability Header Registers. The latter are filled in as they are implemented in the following patches.

Endianness and alignment are managed by softmmu memory core.

---
 hw/cxl/cxl-device-utils.c   | 109 ++++++++++++++++++++++++++++++++++++
 hw/cxl/meson.build          |   1 +
 include/hw/cxl/cxl_device.h |  31 +++++++++-
 3 files changed, 140 insertions(+), 1 deletion(-)
hw/cxl/cxl-device-utils.c new file +109 -0
new file mode 100644
index 0000000000..241f9f82e3
@@ -0,0 +1,109 @@
+/*
+ * CXL Utility library for devices
+ *
+ * Copyright(C) 2020 Intel Corporation.
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2. See the
+ * COPYING file in the top-level directory.
+ */
+
+#include "qemu/osdep.h"
+#include "qemu/log.h"
+#include "hw/cxl/cxl.h"
+
+/*
+ * Device registers have no restrictions per the spec, and so fall back to the
+ * default memory mapped register rules in 8.2:
+ *   Software shall use CXL.io Memory Read and Write to access memory mapped
+ *   register defined in this section. Unless otherwise specified, software
+ *   shall restrict the accesses width based on the following:
+ *   • A 32 bit register shall be accessed as a 1 Byte, 2 Bytes or 4 Bytes
+ *     quantity.
+ *   • A 64 bit register shall be accessed as a 1 Byte, 2 Bytes, 4 Bytes or 8
+ *     Bytes
+ *   • The address shall be a multiple of the access width, e.g. when
+ *     accessing a register as a 4 Byte quantity, the address shall be
+ *     multiple of 4.
+ *   • The accesses shall map to contiguous bytes.If these rules are not
+ *     followed, the behavior is undefined
+ */
+
+static uint64_t caps_reg_read(void *opaque, hwaddr offset, unsigned size)
+{
+    CXLDeviceState *cxl_dstate = opaque;
+
+    if (size == 4) {
+        return cxl_dstate->caps_reg_state32[offset / sizeof(*cxl_dstate->caps_reg_state32)];
+    } else {
+        return cxl_dstate->caps_reg_state64[offset / sizeof(*cxl_dstate->caps_reg_state64)];
+    }
+}
+
+static uint64_t dev_reg_read(void *opaque, hwaddr offset, unsigned size)
+{
+    return 0;
+}
+
+static const MemoryRegionOps dev_ops = {
+    .read = dev_reg_read,
+    .write = NULL, /* status register is read only */
+    .endianness = DEVICE_LITTLE_ENDIAN,
+    .valid = {
+        .min_access_size = 1,
+        .max_access_size = 8,
+        .unaligned = false,
+    },
+    .impl = {
+        .min_access_size = 1,
+        .max_access_size = 8,
+    },
+};
+
+static const MemoryRegionOps caps_ops = {
+    .read = caps_reg_read,
+    .write = NULL, /* caps registers are read only */
+    .endianness = DEVICE_LITTLE_ENDIAN,
+    .valid = {
+        .min_access_size = 1,
+        .max_access_size = 8,
+        .unaligned = false,
+    },
+    .impl = {
+        .min_access_size = 4,
+        .max_access_size = 8,
+    },
+};
+
+void cxl_device_register_block_init(Object *obj, CXLDeviceState *cxl_dstate)
+{
+    /* This will be a BAR, so needs to be rounded up to pow2 for PCI spec */
+    memory_region_init(&cxl_dstate->device_registers, obj, "device-registers",
+                       pow2ceil(CXL_MMIO_SIZE));
+
+    memory_region_init_io(&cxl_dstate->caps, obj, &caps_ops, cxl_dstate,
+                          "cap-array", CXL_CAPS_SIZE);
+    memory_region_init_io(&cxl_dstate->device, obj, &dev_ops, cxl_dstate,
+                          "device-status", CXL_DEVICE_STATUS_REGISTERS_LENGTH);
+
+    memory_region_add_subregion(&cxl_dstate->device_registers, 0,
+                                &cxl_dstate->caps);
+    memory_region_add_subregion(&cxl_dstate->device_registers,
+                                CXL_DEVICE_STATUS_REGISTERS_OFFSET,
+                                &cxl_dstate->device);
+}
+
+static void device_reg_init_common(CXLDeviceState *cxl_dstate) { }
+
+void cxl_device_register_init_common(CXLDeviceState *cxl_dstate)
+{
+    uint64_t *cap_hdrs = cxl_dstate->caps_reg_state64;
+    const int cap_count = 1;
+
+    /* CXL Device Capabilities Array Register */
+    ARRAY_FIELD_DP64(cap_hdrs, CXL_DEV_CAP_ARRAY, CAP_ID, 0);
+    ARRAY_FIELD_DP64(cap_hdrs, CXL_DEV_CAP_ARRAY, CAP_VERSION, 1);
+    ARRAY_FIELD_DP64(cap_hdrs, CXL_DEV_CAP_ARRAY, CAP_COUNT, cap_count);
+
+    cxl_device_cap_init(cxl_dstate, DEVICE_STATUS, 1);
+    device_reg_init_common(cxl_dstate);
+}
hw/cxl/meson.build +1 -0
index 3231b5de1e..dd7c6f8e5a 100644
@@ -1,4 +1,5 @@
 softmmu_ss.add(when: 'CONFIG_CXL',
                if_true: files(
                    'cxl-component-utils.c',
+                   'cxl-device-utils.c',
                ))
include/hw/cxl/cxl_device.h +30 -1
index 9513aaac77..599c887616 100644
@@ -58,6 +58,8 @@
 #define CXL_DEVICE_CAP_HDR1_OFFSET 0x10 /* Figure 138 */
 #define CXL_DEVICE_CAP_REG_SIZE 0x10 /* 8.2.8.2 */
 #define CXL_DEVICE_CAPS_MAX 4 /* 8.2.8.2.1 + 8.2.8.5 */
+#define CXL_CAPS_SIZE \
+    (CXL_DEVICE_CAP_REG_SIZE * (CXL_DEVICE_CAPS_MAX + 1)) /* +1 for header */
 
 #define CXL_DEVICE_STATUS_REGISTERS_OFFSET 0x80 /* Read comment above */
 #define CXL_DEVICE_STATUS_REGISTERS_LENGTH 0x8 /* 8.2.8.3.1 */
@@ -70,12 +72,22 @@
 #define CXL_MAILBOX_REGISTERS_LENGTH \
     (CXL_MAILBOX_REGISTERS_SIZE + CXL_MAILBOX_MAX_PAYLOAD_SIZE)
 
+#define CXL_MMIO_SIZE                                           \
+    (CXL_DEVICE_CAP_REG_SIZE + CXL_DEVICE_STATUS_REGISTERS_LENGTH + \
+     CXL_MAILBOX_REGISTERS_LENGTH)
+
 typedef struct cxl_device_state {
     MemoryRegion device_registers;
 
     /* mmio for device capabilities array - 8.2.8.2 */
     MemoryRegion device;
-    MemoryRegion caps;
+    struct {
+        MemoryRegion caps;
+        union {
+            uint32_t caps_reg_state32[CXL_CAPS_SIZE / 4];
+            uint64_t caps_reg_state64[CXL_CAPS_SIZE / 8];
+        };
+    };
 
     /* mmio for the mailbox registers 8.2.8.4 */
     MemoryRegion mailbox;
@@ -128,6 +140,23 @@ CXL_DEVICE_CAPABILITY_HEADER_REGISTER(DEVICE_STATUS, CXL_DEVICE_CAP_HDR1_OFFSET)
 CXL_DEVICE_CAPABILITY_HEADER_REGISTER(MAILBOX, CXL_DEVICE_CAP_HDR1_OFFSET + \
                                                CXL_DEVICE_CAP_REG_SIZE)
 
+#define cxl_device_cap_init(dstate, reg, cap_id)                           \
+    do {                                                                   \
+        uint32_t *cap_hdrs = dstate->caps_reg_state32;                     \
+        int which = R_CXL_DEV_##reg##_CAP_HDR0;                            \
+        cap_hdrs[which] =                                                  \
+            FIELD_DP32(cap_hdrs[which], CXL_DEV_##reg##_CAP_HDR0,          \
+                       CAP_ID, cap_id);                                    \
+        cap_hdrs[which] = FIELD_DP32(                                      \
+            cap_hdrs[which], CXL_DEV_##reg##_CAP_HDR0, CAP_VERSION, 1);    \
+        cap_hdrs[which + 1] =                                              \
+            FIELD_DP32(cap_hdrs[which + 1], CXL_DEV_##reg##_CAP_HDR1,      \
+                       CAP_OFFSET, CXL_##reg##_REGISTERS_OFFSET);          \
+        cap_hdrs[which + 2] =                                              \
+            FIELD_DP32(cap_hdrs[which + 2], CXL_DEV_##reg##_CAP_HDR2,      \
+                       CAP_LENGTH, CXL_##reg##_REGISTERS_LENGTH);          \
+    } while (0)
+
 /* CXL 2.0 8.2.8.4.3 Mailbox Capabilities Register */
 REG32(CXL_DEV_MAILBOX_CAP, 0)
     FIELD(CXL_DEV_MAILBOX_CAP, PAYLOAD_SIZE, 0, 5)
Signature · 1 line
2.32.0
#7 lore raw

PATCH v10 6/45 hw/cxl/device: Implement basic mailbox (8.2.8.4)

In reply to Jonathan Cameron

From: Ben Widawsky <[email protected]>

This is the beginning of implementing mailbox support for CXL 2.0 devices. The implementation recognizes when the doorbell is rung, handles the command/payload, clears the doorbell while returning error codes and data.

Generally the mailbox mechanism is designed to permit communication between the host OS and the firmware running on the device. For our purposes, we emulate both the firmware, implemented primarily in cxl-mailbox-utils.c, and the hardware.

No commands are implemented yet.

---
 hw/cxl/cxl-device-utils.c   | 122 ++++++++++++++++++++++++++-
 hw/cxl/cxl-mailbox-utils.c  | 164 ++++++++++++++++++++++++++++++++++++
 hw/cxl/meson.build          |   1 +
 include/hw/cxl/cxl.h        |   3 +
 include/hw/cxl/cxl_device.h |  19 ++++-
 5 files changed, 307 insertions(+), 2 deletions(-)
hw/cxl/cxl-device-utils.c +121 -1
index 241f9f82e3..f6c3e0f095 100644
@@ -44,6 +44,108 @@ static uint64_t dev_reg_read(void *opaque, hwaddr offset, unsigned size)
     return 0;
 }
 
+static uint64_t mailbox_reg_read(void *opaque, hwaddr offset, unsigned size)
+{
+    CXLDeviceState *cxl_dstate = opaque;
+
+    switch (size) {
+    case 1:
+        return cxl_dstate->mbox_reg_state[offset];
+    case 2:
+        return cxl_dstate->mbox_reg_state16[offset / size];
+    case 4:
+        return cxl_dstate->mbox_reg_state32[offset / size];
+    case 8:
+        return cxl_dstate->mbox_reg_state64[offset / size];
+    default:
+        g_assert_not_reached();
+    }
+}
+
+static void mailbox_mem_writel(uint32_t *reg_state, hwaddr offset,
+                               uint64_t value)
+{
+    switch (offset) {
+    case A_CXL_DEV_MAILBOX_CTRL:
+        /* fallthrough */
+    case A_CXL_DEV_MAILBOX_CAP:
+        /* RO register */
+        break;
+    default:
+        qemu_log_mask(LOG_UNIMP,
+                      "%s Unexpected 32-bit access to 0x%" PRIx64 " (WI)\n",
+                      __func__, offset);
+        return;
+    }
+
+    reg_state[offset / sizeof(*reg_state)] = value;
+}
+
+static void mailbox_mem_writeq(uint64_t *reg_state, hwaddr offset,
+                               uint64_t value)
+{
+    switch (offset) {
+    case A_CXL_DEV_MAILBOX_CMD:
+        break;
+    case A_CXL_DEV_BG_CMD_STS:
+        /* BG not supported */
+        /* fallthrough */
+    case A_CXL_DEV_MAILBOX_STS:
+        /* Read only register, will get updated by the state machine */
+        return;
+    default:
+        qemu_log_mask(LOG_UNIMP,
+                      "%s Unexpected 64-bit access to 0x%" PRIx64 " (WI)\n",
+                      __func__, offset);
+        return;
+    }
+
+
+    reg_state[offset / sizeof(*reg_state)] = value;
+}
+
+static void mailbox_reg_write(void *opaque, hwaddr offset, uint64_t value,
+                              unsigned size)
+{
+    CXLDeviceState *cxl_dstate = opaque;
+
+    if (offset >= A_CXL_DEV_CMD_PAYLOAD) {
+        memcpy(cxl_dstate->mbox_reg_state + offset, &value, size);
+        return;
+    }
+
+    switch (size) {
+    case 4:
+        mailbox_mem_writel(cxl_dstate->mbox_reg_state32, offset, value);
+        break;
+    case 8:
+        mailbox_mem_writeq(cxl_dstate->mbox_reg_state64, offset, value);
+        break;
+    default:
+        g_assert_not_reached();
+    }
+
+    if (ARRAY_FIELD_EX32(cxl_dstate->mbox_reg_state32, CXL_DEV_MAILBOX_CTRL,
+                         DOORBELL)) {
+        cxl_process_mailbox(cxl_dstate);
+    }
+}
+
+static const MemoryRegionOps mailbox_ops = {
+    .read = mailbox_reg_read,
+    .write = mailbox_reg_write,
+    .endianness = DEVICE_LITTLE_ENDIAN,
+    .valid = {
+        .min_access_size = 1,
+        .max_access_size = 8,
+        .unaligned = false,
+    },
+    .impl = {
+        .min_access_size = 1,
+        .max_access_size = 8,
+    },
+};
+
 static const MemoryRegionOps dev_ops = {
     .read = dev_reg_read,
     .write = NULL, /* status register is read only */
@@ -84,20 +186,33 @@ void cxl_device_register_block_init(Object *obj, CXLDeviceState *cxl_dstate)
                           "cap-array", CXL_CAPS_SIZE);
     memory_region_init_io(&cxl_dstate->device, obj, &dev_ops, cxl_dstate,
                           "device-status", CXL_DEVICE_STATUS_REGISTERS_LENGTH);
+    memory_region_init_io(&cxl_dstate->mailbox, obj, &mailbox_ops, cxl_dstate,
+                          "mailbox", CXL_MAILBOX_REGISTERS_LENGTH);
 
     memory_region_add_subregion(&cxl_dstate->device_registers, 0,
                                 &cxl_dstate->caps);
     memory_region_add_subregion(&cxl_dstate->device_registers,
                                 CXL_DEVICE_STATUS_REGISTERS_OFFSET,
                                 &cxl_dstate->device);
+    memory_region_add_subregion(&cxl_dstate->device_registers,
+                                CXL_MAILBOX_REGISTERS_OFFSET,
+                                &cxl_dstate->mailbox);
 }
 
 static void device_reg_init_common(CXLDeviceState *cxl_dstate) { }
 
+static void mailbox_reg_init_common(CXLDeviceState *cxl_dstate)
+{
+    /* 2048 payload size, with no interrupt or background support */
+    ARRAY_FIELD_DP32(cxl_dstate->mbox_reg_state32, CXL_DEV_MAILBOX_CAP,
+                     PAYLOAD_SIZE, CXL_MAILBOX_PAYLOAD_SHIFT);
+    cxl_dstate->payload_size = CXL_MAILBOX_MAX_PAYLOAD_SIZE;
+}
+
 void cxl_device_register_init_common(CXLDeviceState *cxl_dstate)
 {
     uint64_t *cap_hdrs = cxl_dstate->caps_reg_state64;
-    const int cap_count = 1;
+    const int cap_count = 2;
 
     /* CXL Device Capabilities Array Register */
     ARRAY_FIELD_DP64(cap_hdrs, CXL_DEV_CAP_ARRAY, CAP_ID, 0);
@@ -106,4 +221,9 @@ void cxl_device_register_init_common(CXLDeviceState *cxl_dstate)
 
     cxl_device_cap_init(cxl_dstate, DEVICE_STATUS, 1);
     device_reg_init_common(cxl_dstate);
+
+    cxl_device_cap_init(cxl_dstate, MAILBOX, 2);
+    mailbox_reg_init_common(cxl_dstate);
+
+    assert(cxl_initialize_mailbox(cxl_dstate) == 0);
 }
hw/cxl/cxl-mailbox-utils.c new file +164 -0

164 lines not on this page · raw message

hw/cxl/meson.build +1 -0
index dd7c6f8e5a..e68eea2358 100644
@@ -2,4 +2,5 @@ softmmu_ss.add(when: 'CONFIG_CXL',
                if_true: files(
                    'cxl-component-utils.c',
                    'cxl-device-utils.c',
+                   'cxl-mailbox-utils.c',
                ))
include/hw/cxl/cxl.h +3 -0
index b9d1ac3fad..554ad93b6b 100644
@@ -14,4 +14,7 @@
 #include "cxl_component.h"
 #include "cxl_device.h"
 
+#define CXL_COMPONENT_REG_BAR_IDX 0
+#define CXL_DEVICE_REG_BAR_IDX 2
+
 #endif
include/hw/cxl/cxl_device.h +18 -1
index 599c887616..35489f635a 100644
@@ -90,7 +90,21 @@ typedef struct cxl_device_state {
     };
 
     /* mmio for the mailbox registers 8.2.8.4 */
-    MemoryRegion mailbox;
+    struct {
+        MemoryRegion mailbox;
+        uint16_t payload_size;
+        union {
+            uint8_t mbox_reg_state[CXL_MAILBOX_REGISTERS_LENGTH];
+            uint16_t mbox_reg_state16[CXL_MAILBOX_REGISTERS_LENGTH / 2];
+            uint32_t mbox_reg_state32[CXL_MAILBOX_REGISTERS_LENGTH / 4];
+            uint64_t mbox_reg_state64[CXL_MAILBOX_REGISTERS_LENGTH / 8];
+        };
+        struct cel_log {
+            uint16_t opcode;
+            uint16_t effect;
+        } cel_log[1 << 16];
+        size_t cel_size;
+    };
 
     /* memory region for persistent memory, HDM */
     uint64_t pmem_size;
@@ -140,6 +154,9 @@ CXL_DEVICE_CAPABILITY_HEADER_REGISTER(DEVICE_STATUS, CXL_DEVICE_CAP_HDR1_OFFSET)
 CXL_DEVICE_CAPABILITY_HEADER_REGISTER(MAILBOX, CXL_DEVICE_CAP_HDR1_OFFSET + \
                                                CXL_DEVICE_CAP_REG_SIZE)
 
+int cxl_initialize_mailbox(CXLDeviceState *cxl_dstate);
+void cxl_process_mailbox(CXLDeviceState *cxl_dstate);
+
 #define cxl_device_cap_init(dstate, reg, cap_id)                           \
     do {                                                                   \
         uint32_t *cap_hdrs = dstate->caps_reg_state32;                     \
Signature · 1 line
2.32.0
#8 lore raw

PATCH v10 7/45 hw/cxl/device: Add memory device utilities

In reply to Jonathan Cameron

From: Ben Widawsky <[email protected]>

Memory devices implement extra capabilities on top of CXL devices. This adds support for that.

A large part of memory devices is the mailbox/command interface. All of the mailbox handling is done in the mailbox-utils library. Longer term, new CXL devices that are being emulated may want to handle commands differently, and therefore would need a mechanism to opt in/out of the specific generic handlers. As such, this is considered sufficient for now, but may need more depth in the future.

---
 hw/cxl/cxl-device-utils.c   | 38 ++++++++++++++++++++++++++++++++++++-
 include/hw/cxl/cxl_device.h | 21 +++++++++++++++++---
 2 files changed, 55 insertions(+), 4 deletions(-)
hw/cxl/cxl-device-utils.c +37 -1
index f6c3e0f095..687759b301 100644
@@ -131,6 +131,31 @@ static void mailbox_reg_write(void *opaque, hwaddr offset, uint64_t value,
     }
 }
 
+static uint64_t mdev_reg_read(void *opaque, hwaddr offset, unsigned size)
+{
+    uint64_t retval = 0;
+
+    retval = FIELD_DP64(retval, CXL_MEM_DEV_STS, MEDIA_STATUS, 1);
+    retval = FIELD_DP64(retval, CXL_MEM_DEV_STS, MBOX_READY, 1);
+
+    return retval;
+}
+
+static const MemoryRegionOps mdev_ops = {
+    .read = mdev_reg_read,
+    .write = NULL, /* memory device register is read only */
+    .endianness = DEVICE_LITTLE_ENDIAN,
+    .valid = {
+        .min_access_size = 1,
+        .max_access_size = 8,
+        .unaligned = false,
+    },
+    .impl = {
+        .min_access_size = 8,
+        .max_access_size = 8,
+    },
+};
+
 static const MemoryRegionOps mailbox_ops = {
     .read = mailbox_reg_read,
     .write = mailbox_reg_write,
@@ -188,6 +213,9 @@ void cxl_device_register_block_init(Object *obj, CXLDeviceState *cxl_dstate)
                           "device-status", CXL_DEVICE_STATUS_REGISTERS_LENGTH);
     memory_region_init_io(&cxl_dstate->mailbox, obj, &mailbox_ops, cxl_dstate,
                           "mailbox", CXL_MAILBOX_REGISTERS_LENGTH);
+    memory_region_init_io(&cxl_dstate->memory_device, obj, &mdev_ops,
+                          cxl_dstate, "memory device caps",
+                          CXL_MEMORY_DEVICE_REGISTERS_LENGTH);
 
     memory_region_add_subregion(&cxl_dstate->device_registers, 0,
                                 &cxl_dstate->caps);
@@ -197,6 +225,9 @@ void cxl_device_register_block_init(Object *obj, CXLDeviceState *cxl_dstate)
     memory_region_add_subregion(&cxl_dstate->device_registers,
                                 CXL_MAILBOX_REGISTERS_OFFSET,
                                 &cxl_dstate->mailbox);
+    memory_region_add_subregion(&cxl_dstate->device_registers,
+                                CXL_MEMORY_DEVICE_REGISTERS_OFFSET,
+                                &cxl_dstate->memory_device);
 }
 
 static void device_reg_init_common(CXLDeviceState *cxl_dstate) { }
@@ -209,10 +240,12 @@ static void mailbox_reg_init_common(CXLDeviceState *cxl_dstate)
     cxl_dstate->payload_size = CXL_MAILBOX_MAX_PAYLOAD_SIZE;
 }
 
+static void memdev_reg_init_common(CXLDeviceState *cxl_dstate) { }
+
 void cxl_device_register_init_common(CXLDeviceState *cxl_dstate)
 {
     uint64_t *cap_hdrs = cxl_dstate->caps_reg_state64;
-    const int cap_count = 2;
+    const int cap_count = 3;
 
     /* CXL Device Capabilities Array Register */
     ARRAY_FIELD_DP64(cap_hdrs, CXL_DEV_CAP_ARRAY, CAP_ID, 0);
@@ -225,5 +258,8 @@ void cxl_device_register_init_common(CXLDeviceState *cxl_dstate)
     cxl_device_cap_init(cxl_dstate, MAILBOX, 2);
     mailbox_reg_init_common(cxl_dstate);
 
+    cxl_device_cap_init(cxl_dstate, MEMORY_DEVICE, 0x4000);
+    memdev_reg_init_common(cxl_dstate);
+
     assert(cxl_initialize_mailbox(cxl_dstate) == 0);
 }
include/hw/cxl/cxl_device.h +18 -3
index 35489f635a..954205653e 100644
@@ -72,15 +72,20 @@
 #define CXL_MAILBOX_REGISTERS_LENGTH \
     (CXL_MAILBOX_REGISTERS_SIZE + CXL_MAILBOX_MAX_PAYLOAD_SIZE)
 
-#define CXL_MMIO_SIZE                                           \
-    (CXL_DEVICE_CAP_REG_SIZE + CXL_DEVICE_STATUS_REGISTERS_LENGTH + \
-     CXL_MAILBOX_REGISTERS_LENGTH)
+#define CXL_MEMORY_DEVICE_REGISTERS_OFFSET \
+    (CXL_MAILBOX_REGISTERS_OFFSET + CXL_MAILBOX_REGISTERS_LENGTH)
+#define CXL_MEMORY_DEVICE_REGISTERS_LENGTH 0x8
+
+#define CXL_MMIO_SIZE                                                   \
+    (CXL_DEVICE_CAP_REG_SIZE + CXL_DEVICE_STATUS_REGISTERS_LENGTH +     \
+     CXL_MAILBOX_REGISTERS_LENGTH + CXL_MEMORY_DEVICE_REGISTERS_LENGTH)
 
 typedef struct cxl_device_state {
     MemoryRegion device_registers;
 
     /* mmio for device capabilities array - 8.2.8.2 */
     MemoryRegion device;
+    MemoryRegion memory_device;
     struct {
         MemoryRegion caps;
         union {
@@ -153,6 +158,9 @@ REG64(CXL_DEV_CAP_ARRAY, 0) /* Documented as 128 bit register but 64 byte access
 CXL_DEVICE_CAPABILITY_HEADER_REGISTER(DEVICE_STATUS, CXL_DEVICE_CAP_HDR1_OFFSET)
 CXL_DEVICE_CAPABILITY_HEADER_REGISTER(MAILBOX, CXL_DEVICE_CAP_HDR1_OFFSET + \
                                                CXL_DEVICE_CAP_REG_SIZE)
+CXL_DEVICE_CAPABILITY_HEADER_REGISTER(MEMORY_DEVICE,
+                                      CXL_DEVICE_CAP_HDR1_OFFSET +
+                                          CXL_DEVICE_CAP_REG_SIZE * 2)
 
 int cxl_initialize_mailbox(CXLDeviceState *cxl_dstate);
 void cxl_process_mailbox(CXLDeviceState *cxl_dstate);
@@ -209,4 +217,11 @@ REG64(CXL_DEV_BG_CMD_STS, 0x18)
 /* CXL 2.0 8.2.8.4.8 Command Payload Registers */
 REG32(CXL_DEV_CMD_PAYLOAD, 0x20)
 
+REG64(CXL_MEM_DEV_STS, 0)
+    FIELD(CXL_MEM_DEV_STS, FATAL, 0, 1)
+    FIELD(CXL_MEM_DEV_STS, FW_HALT, 1, 1)
+    FIELD(CXL_MEM_DEV_STS, MEDIA_STATUS, 2, 2)
+    FIELD(CXL_MEM_DEV_STS, MBOX_READY, 4, 1)
+    FIELD(CXL_MEM_DEV_STS, RESET_NEEDED, 5, 3)
+
 #endif
Signature · 1 line
2.32.0
#9 lore raw

PATCH v10 8/45 hw/cxl/device: Add cheap EVENTS implementation (8.2.9.1)

In reply to Jonathan Cameron

From: Ben Widawsky <[email protected]>

Using the previously implemented stubbed helpers, it is now possible to easily add the missing, required commands to the implementation.

---
 hw/cxl/cxl-mailbox-utils.c | 27 ++++++++++++++++++++++++++-
 1 file changed, 26 insertions(+), 1 deletion(-)
hw/cxl/cxl-mailbox-utils.c +26 -1
index 2557f41f61..fb1f53f48e 100644
@@ -38,6 +38,14 @@
  *  a register interface that already deals with it.
  */
 
+enum {
+    EVENTS      = 0x01,
+        #define GET_RECORDS   0x0
+        #define CLEAR_RECORDS   0x1
+        #define GET_INTERRUPT_POLICY   0x2
+        #define SET_INTERRUPT_POLICY   0x3
+};
+
 /* 8.2.8.4.5.1 Command Return Codes */
 typedef enum {
     CXL_MBOX_SUCCESS = 0x0,
@@ -93,9 +101,26 @@ struct cxl_cmd {
         return CXL_MBOX_SUCCESS;                                          \
     }
 
+DEFINE_MAILBOX_HANDLER_ZEROED(events_get_records, 0x20);
+DEFINE_MAILBOX_HANDLER_NOP(events_clear_records);
+DEFINE_MAILBOX_HANDLER_ZEROED(events_get_interrupt_policy, 4);
+DEFINE_MAILBOX_HANDLER_NOP(events_set_interrupt_policy);
+
 static QemuUUID cel_uuid;
 
-static struct cxl_cmd cxl_cmd_set[256][256] = {};
+#define IMMEDIATE_CONFIG_CHANGE (1 << 1)
+#define IMMEDIATE_LOG_CHANGE (1 << 4)
+
+static struct cxl_cmd cxl_cmd_set[256][256] = {
+    [EVENTS][GET_RECORDS] = { "EVENTS_GET_RECORDS",
+        cmd_events_get_records, 1, 0 },
+    [EVENTS][CLEAR_RECORDS] = { "EVENTS_CLEAR_RECORDS",
+        cmd_events_clear_records, ~0, IMMEDIATE_LOG_CHANGE },
+    [EVENTS][GET_INTERRUPT_POLICY] = { "EVENTS_GET_INTERRUPT_POLICY",
+        cmd_events_get_interrupt_policy, 0, 0 },
+    [EVENTS][SET_INTERRUPT_POLICY] = { "EVENTS_SET_INTERRUPT_POLICY",
+        cmd_events_set_interrupt_policy, 4, IMMEDIATE_CONFIG_CHANGE },
+};
 
 void cxl_process_mailbox(CXLDeviceState *cxl_dstate)
 {
Signature · 1 line
2.32.0
#10 lore raw

PATCH v10 9/45 hw/cxl/device: Timestamp implementation (8.2.9.3)

In reply to Jonathan Cameron

From: Ben Widawsky <[email protected]>

Errata F4 to CXL 2.0 clarified the meaning of the timer as the sum of the value set with the timestamp set command and the number of nano seconds since it was last set.

---
 hw/cxl/cxl-mailbox-utils.c  | 42 +++++++++++++++++++++++++++++++++++++
 include/hw/cxl/cxl_device.h |  6 ++++++
 2 files changed, 48 insertions(+)
hw/cxl/cxl-mailbox-utils.c +42 -0
index fb1f53f48e..4584aa31f7 100644
@@ -44,6 +44,9 @@ enum {
         #define CLEAR_RECORDS   0x1
         #define GET_INTERRUPT_POLICY   0x2
         #define SET_INTERRUPT_POLICY   0x3
+    TIMESTAMP   = 0x03,
+        #define GET           0x0
+        #define SET           0x1
 };
 
 /* 8.2.8.4.5.1 Command Return Codes */
@@ -106,9 +109,46 @@ DEFINE_MAILBOX_HANDLER_NOP(events_clear_records);
 DEFINE_MAILBOX_HANDLER_ZEROED(events_get_interrupt_policy, 4);
 DEFINE_MAILBOX_HANDLER_NOP(events_set_interrupt_policy);
 
+/* 8.2.9.3.1 */
+static ret_code cmd_timestamp_get(struct cxl_cmd *cmd,
+                                  CXLDeviceState *cxl_dstate,
+                                  uint16_t *len)
+{
+    uint64_t time, delta;
+    uint64_t final_time = 0;
+
+    if (cxl_dstate->timestamp.set) {
+        /* First find the delta from the last time the host set the time. */
+        time = qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL);
+        delta = time - cxl_dstate->timestamp.last_set;
+        final_time = cxl_dstate->timestamp.host_set + delta;
+    }
+
+    /* Then adjust the actual time */
+    stq_le_p(cmd->payload, final_time);
+    *len = 8;
+
+    return CXL_MBOX_SUCCESS;
+}
+
+/* 8.2.9.3.2 */
+static ret_code cmd_timestamp_set(struct cxl_cmd *cmd,
+                                  CXLDeviceState *cxl_dstate,
+                                  uint16_t *len)
+{
+    cxl_dstate->timestamp.set = true;
+    cxl_dstate->timestamp.last_set = qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL);
+
+    cxl_dstate->timestamp.host_set = le64_to_cpu(*(uint64_t *)cmd->payload);
+
+    *len = 0;
+    return CXL_MBOX_SUCCESS;
+}
+
 static QemuUUID cel_uuid;
 
 #define IMMEDIATE_CONFIG_CHANGE (1 << 1)
+#define IMMEDIATE_POLICY_CHANGE (1 << 3)
 #define IMMEDIATE_LOG_CHANGE (1 << 4)
 
 static struct cxl_cmd cxl_cmd_set[256][256] = {
@@ -120,6 +160,8 @@ static struct cxl_cmd cxl_cmd_set[256][256] = {
         cmd_events_get_interrupt_policy, 0, 0 },
     [EVENTS][SET_INTERRUPT_POLICY] = { "EVENTS_SET_INTERRUPT_POLICY",
         cmd_events_set_interrupt_policy, 4, IMMEDIATE_CONFIG_CHANGE },
+    [TIMESTAMP][GET] = { "TIMESTAMP_GET", cmd_timestamp_get, 0, 0 },
+    [TIMESTAMP][SET] = { "TIMESTAMP_SET", cmd_timestamp_set, 8, IMMEDIATE_POLICY_CHANGE },
 };
 
 void cxl_process_mailbox(CXLDeviceState *cxl_dstate)
include/hw/cxl/cxl_device.h +6 -0
index 954205653e..797a22ddb4 100644
@@ -111,6 +111,12 @@ typedef struct cxl_device_state {
         size_t cel_size;
     };
 
+    struct {
+        bool set;
+        uint64_t last_set;
+        uint64_t host_set;
+    } timestamp;
+
     /* memory region for persistent memory, HDM */
     uint64_t pmem_size;
 } CXLDeviceState;
Signature · 1 line
2.32.0
#11 lore raw

PATCH v10 10/45 hw/cxl/device: Add log commands (8.2.9.4) + CEL

In reply to Jonathan Cameron

From: Ben Widawsky <[email protected]>

CXL specification provides for the ability to obtain logs from the device. Logs are either spec defined, like the "Command Effects Log" (CEL), or vendor specific. UUIDs are defined for all log types.

The CEL is a mechanism to provide information to the host about which commands are supported. It is useful both to determine which spec'd optional commands are supported, as well as provide a list of vendor specified commands that might be used. The CEL is already created as part of mailbox initialization, but here it is now exported to hosts that use these log commands.

---
 hw/cxl/cxl-mailbox-utils.c | 69 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 69 insertions(+)
hw/cxl/cxl-mailbox-utils.c +69 -0
index 4584aa31f7..db473135c7 100644
@@ -47,6 +47,9 @@ enum {
     TIMESTAMP   = 0x03,
         #define GET           0x0
         #define SET           0x1
+    LOGS        = 0x04,
+        #define GET_SUPPORTED 0x0
+        #define GET_LOG       0x1
 };
 
 /* 8.2.8.4.5.1 Command Return Codes */
@@ -147,6 +150,70 @@ static ret_code cmd_timestamp_set(struct cxl_cmd *cmd,
 
 static QemuUUID cel_uuid;
 
+/* 8.2.9.4.1 */
+static ret_code cmd_logs_get_supported(struct cxl_cmd *cmd,
+                                       CXLDeviceState *cxl_dstate,
+                                       uint16_t *len)
+{
+    struct {
+        uint16_t entries;
+        uint8_t rsvd[6];
+        struct {
+            QemuUUID uuid;
+            uint32_t size;
+        } log_entries[1];
+    } QEMU_PACKED *supported_logs = (void *)cmd->payload;
+    QEMU_BUILD_BUG_ON(sizeof(*supported_logs) != 0x1c);
+
+    supported_logs->entries = 1;
+    supported_logs->log_entries[0].uuid = cel_uuid;
+    supported_logs->log_entries[0].size = 4 * cxl_dstate->cel_size;
+
+    *len = sizeof(*supported_logs);
+    return CXL_MBOX_SUCCESS;
+}
+
+/* 8.2.9.4.2 */
+static ret_code cmd_logs_get_log(struct cxl_cmd *cmd,
+                                 CXLDeviceState *cxl_dstate,
+                                 uint16_t *len)
+{
+    struct {
+        QemuUUID uuid;
+        uint32_t offset;
+        uint32_t length;
+    } QEMU_PACKED QEMU_ALIGNED(16) *get_log = (void *)cmd->payload;
+
+    /*
+     * 8.2.9.4.2
+     *   The device shall return Invalid Parameter if the Offset or Length
+     *   fields attempt to access beyond the size of the log as reported by Get
+     *   Supported Logs.
+     *
+     * XXX: Spec is wrong, "Invalid Parameter" isn't a thing.
+     * XXX: Spec doesn't address incorrect UUID incorrectness.
+     *
+     * The CEL buffer is large enough to fit all commands in the emulation, so
+     * the only possible failure would be if the mailbox itself isn't big
+     * enough.
+     */
+    if (get_log->offset + get_log->length > cxl_dstate->payload_size) {
+        return CXL_MBOX_INVALID_INPUT;
+    }
+
+    if (!qemu_uuid_is_equal(&get_log->uuid, &cel_uuid)) {
+        return CXL_MBOX_UNSUPPORTED;
+    }
+
+    /* Store off everything to local variables so we can wipe out the payload */
+    *len = get_log->length;
+
+    memmove(cmd->payload, cxl_dstate->cel_log + get_log->offset,
+           get_log->length);
+
+    return CXL_MBOX_SUCCESS;
+}
+
 #define IMMEDIATE_CONFIG_CHANGE (1 << 1)
 #define IMMEDIATE_POLICY_CHANGE (1 << 3)
 #define IMMEDIATE_LOG_CHANGE (1 << 4)
@@ -162,6 +229,8 @@ static struct cxl_cmd cxl_cmd_set[256][256] = {
         cmd_events_set_interrupt_policy, 4, IMMEDIATE_CONFIG_CHANGE },
     [TIMESTAMP][GET] = { "TIMESTAMP_GET", cmd_timestamp_get, 0, 0 },
     [TIMESTAMP][SET] = { "TIMESTAMP_SET", cmd_timestamp_set, 8, IMMEDIATE_POLICY_CHANGE },
+    [LOGS][GET_SUPPORTED] = { "LOGS_GET_SUPPORTED", cmd_logs_get_supported, 0, 0 },
+    [LOGS][GET_LOG] = { "LOGS_GET_LOG", cmd_logs_get_log, 0x18, 0 },
 };
 
 void cxl_process_mailbox(CXLDeviceState *cxl_dstate)
Signature · 1 line
2.32.0
#12 lore raw

PATCH v10 11/45 hw/pxb: Use a type for realizing expanders

In reply to Jonathan Cameron

From: Ben Widawsky <[email protected]>

This opens up the possibility for more types of expanders (other than PCI and PCIe). We'll need this to create a CXL expander.

---
 hw/pci-bridge/pci_expander_bridge.c | 11 +++++++----
 1 file changed, 7 insertions(+), 4 deletions(-)
hw/pci-bridge/pci_expander_bridge.c +7 -4
index de932286b5..d4514227a8 100644
@@ -24,6 +24,8 @@
 #include "hw/boards.h"
 #include "qom/object.h"
 
+enum BusType { PCI, PCIE };
+
 #define TYPE_PXB_BUS "pxb-bus"
 typedef struct PXBBus PXBBus;
 DECLARE_INSTANCE_CHECKER(PXBBus, PXB_BUS,
@@ -221,7 +223,8 @@ static gint pxb_compare(gconstpointer a, gconstpointer b)
            0;
 }
 
-static void pxb_dev_realize_common(PCIDevice *dev, bool pcie, Error **errp)
+static void pxb_dev_realize_common(PCIDevice *dev, enum BusType type,
+                                   Error **errp)
 {
     PXBDev *pxb = convert_to_pxb(dev);
     DeviceState *ds, *bds = NULL;
@@ -246,7 +249,7 @@ static void pxb_dev_realize_common(PCIDevice *dev, bool pcie, Error **errp)
     }
 
     ds = qdev_new(TYPE_PXB_HOST);
-    if (pcie) {
+    if (type == PCIE) {
         bus = pci_root_bus_new(ds, dev_name, NULL, NULL, 0, TYPE_PXB_PCIE_BUS);
     } else {
         bus = pci_root_bus_new(ds, "pxb-internal", NULL, NULL, 0, TYPE_PXB_BUS);
@@ -295,7 +298,7 @@ static void pxb_dev_realize(PCIDevice *dev, Error **errp)
         return;
     }
 
-    pxb_dev_realize_common(dev, false, errp);
+    pxb_dev_realize_common(dev, PCI, errp);
 }
 
 static void pxb_dev_exitfn(PCIDevice *pci_dev)
@@ -348,7 +351,7 @@ static void pxb_pcie_dev_realize(PCIDevice *dev, Error **errp)
         return;
     }
 
-    pxb_dev_realize_common(dev, true, errp);
+    pxb_dev_realize_common(dev, PCIE, errp);
 }
 
 static void pxb_pcie_dev_class_init(ObjectClass *klass, void *data)
Signature · 1 line
2.32.0
#13 lore raw

PATCH v10 12/45 hw/pci/cxl: Create a CXL bus type

In reply to Jonathan Cameron

From: Ben Widawsky <[email protected]>

The easiest way to differentiate a CXL bus, and a PCIE bus is using a flag. A CXL bus, in hardware, is backward compatible with PCIE, and therefore the code tries pretty hard to keep them in sync as much as possible.

The other way to implement this would be to try to cast the bus to the correct type. This is less code and useful for debugging via simply looking at the flags.

---
 hw/pci-bridge/pci_expander_bridge.c | 9 ++++++++-
 include/hw/pci/pci_bus.h            | 7 +++++++
 2 files changed, 15 insertions(+), 1 deletion(-)
hw/pci-bridge/pci_expander_bridge.c +8 -1
index d4514227a8..a6caa1e7b5 100644
@@ -24,7 +24,7 @@
 #include "hw/boards.h"
 #include "qom/object.h"
 
-enum BusType { PCI, PCIE };
+enum BusType { PCI, PCIE, CXL };
 
 #define TYPE_PXB_BUS "pxb-bus"
 typedef struct PXBBus PXBBus;
@@ -35,6 +35,10 @@ DECLARE_INSTANCE_CHECKER(PXBBus, PXB_BUS,
 DECLARE_INSTANCE_CHECKER(PXBBus, PXB_PCIE_BUS,
                          TYPE_PXB_PCIE_BUS)
 
+#define TYPE_PXB_CXL_BUS "pxb-cxl-bus"
+DECLARE_INSTANCE_CHECKER(PXBBus, PXB_CXL_BUS,
+                         TYPE_PXB_CXL_BUS)
+
 struct PXBBus {
     /*< private >*/
     PCIBus parent_obj;
@@ -251,6 +255,9 @@ static void pxb_dev_realize_common(PCIDevice *dev, enum BusType type,
     ds = qdev_new(TYPE_PXB_HOST);
     if (type == PCIE) {
         bus = pci_root_bus_new(ds, dev_name, NULL, NULL, 0, TYPE_PXB_PCIE_BUS);
+    } else if (type == CXL) {
+        bus = pci_root_bus_new(ds, dev_name, NULL, NULL, 0, TYPE_PXB_CXL_BUS);
+        bus->flags |= PCI_BUS_CXL;
     } else {
         bus = pci_root_bus_new(ds, "pxb-internal", NULL, NULL, 0, TYPE_PXB_BUS);
         bds = qdev_new("pci-bridge");
include/hw/pci/pci_bus.h +7 -0
index 347440d42c..eb94e7e85c 100644
@@ -24,6 +24,8 @@ enum PCIBusFlags {
     PCI_BUS_IS_ROOT                                         = 0x0001,
     /* PCIe extended configuration space is accessible on this bus */
     PCI_BUS_EXTENDED_CONFIG_SPACE                           = 0x0002,
+    /* This is a CXL Type BUS */
+    PCI_BUS_CXL                                             = 0x0004,
 };
 
 struct PCIBus {
@@ -53,6 +55,11 @@ struct PCIBus {
     Notifier machine_done;
 };
 
+static inline bool pci_bus_is_cxl(PCIBus *bus)
+{
+    return !!(bus->flags & PCI_BUS_CXL);
+}
+
 static inline bool pci_bus_is_root(PCIBus *bus)
 {
     return !!(bus->flags & PCI_BUS_IS_ROOT);
Signature · 1 line
2.32.0
#14 lore raw

PATCH v10 13/45 cxl: Machine level control on whether CXL support is enabled

In reply to Jonathan Cameron

From: Jonathan Cameron <[email protected]>

There are going to be some potential overheads to CXL enablement, for example the host bridge region reserved in memory maps. Add a machine level control so that CXL is disabled by default.

---
 hw/core/machine.c    | 28 ++++++++++++++++++++++++++++
 hw/i386/pc.c         |  1 +
 include/hw/boards.h  |  2 ++
 include/hw/cxl/cxl.h |  4 ++++
 4 files changed, 35 insertions(+)
hw/core/machine.c +28 -0
index cb9bbc844d..6ae2997f16 100644
@@ -31,6 +31,7 @@
 #include "sysemu/qtest.h"
 #include "hw/pci/pci.h"
 #include "hw/mem/nvdimm.h"
+#include "hw/cxl/cxl.h"
 #include "migration/global_state.h"
 #include "migration/vmstate.h"
 #include "exec/confidential-guest-support.h"
@@ -550,6 +551,20 @@ static void machine_set_nvdimm_persistence(Object *obj, const char *value,
     nvdimms_state->persistence_string = g_strdup(value);
 }
 
+static bool machine_get_cxl(Object *obj, Error **errp)
+{
+    MachineState *ms = MACHINE(obj);
+
+    return ms->cxl_devices_state->is_enabled;
+}
+
+static void machine_set_cxl(Object *obj, bool value, Error **errp)
+{
+    MachineState *ms = MACHINE(obj);
+
+    ms->cxl_devices_state->is_enabled = value;
+}
+
 void machine_class_allow_dynamic_sysbus_dev(MachineClass *mc, const char *type)
 {
     QAPI_LIST_PREPEND(mc->allowed_dynamic_sysbus_devices, g_strdup(type));
@@ -782,6 +797,8 @@ static void machine_class_init(ObjectClass *oc, void *data)
     mc->default_ram_size = 128 * MiB;
     mc->rom_file_has_mr = true;
 
+    /* Few machines support CXL, so default to off */
+    mc->cxl_supported = false;
     /* numa node memory size aligned on 8MB by default.
      * On Linux, each node's border has to be 8MB aligned
      */
@@ -927,6 +944,16 @@ static void machine_initfn(Object *obj)
                                         "Valid values are cpu, mem-ctrl");
     }
 
+    if (mc->cxl_supported) {
+        Object *obj = OBJECT(ms);
+
+        ms->cxl_devices_state = g_new0(CXLState, 1);
+        object_property_add_bool(obj, "cxl", machine_get_cxl, machine_set_cxl);
+        object_property_set_description(obj, "cxl",
+                                        "Set on/off to enable/disable "
+                                        "CXL instantiation");
+    }
+
     if (mc->cpu_index_to_instance_props && mc->get_default_cpu_node_id) {
         ms->numa_state = g_new0(NumaState, 1);
         object_property_add_bool(obj, "hmat",
@@ -961,6 +988,7 @@ static void machine_finalize(Object *obj)
     g_free(ms->device_memory);
     g_free(ms->nvdimms_state);
     g_free(ms->numa_state);
+    g_free(ms->cxl_devices_state);
 }
 
 bool machine_usb(MachineState *machine)
hw/i386/pc.c +1 -0
index 23bba9d82c..b752339beb 100644
@@ -1761,6 +1761,7 @@ static void pc_machine_class_init(ObjectClass *oc, void *data)
     mc->default_cpu_type = TARGET_DEFAULT_CPU_TYPE;
     mc->nvdimm_supported = true;
     mc->smp_props.dies_supported = true;
+    mc->cxl_supported = true;
     mc->default_ram_id = "pc.ram";
 
     object_class_property_add(oc, PC_MACHINE_MAX_RAM_BELOW_4G, "size",
include/hw/boards.h +2 -0
index d64b5481e8..f756a1d5fc 100644
@@ -269,6 +269,7 @@ struct MachineClass {
     bool ignore_boot_device_suffixes;
     bool smbus_no_migration_support;
     bool nvdimm_supported;
+    bool cxl_supported;
     bool numa_mem_supported;
     bool auto_enable_numa;
     SMPCompatProps smp_props;
@@ -360,6 +361,7 @@ struct MachineState {
     CPUArchIdList *possible_cpus;
     CpuTopology smp;
     struct NVDIMMState *nvdimms_state;
+    struct CXLState *cxl_devices_state;
     struct NumaState *numa_state;
 };
 
include/hw/cxl/cxl.h +4 -0
index 554ad93b6b..31af92fd5e 100644
@@ -17,4 +17,8 @@
 #define CXL_COMPONENT_REG_BAR_IDX 0
 #define CXL_DEVICE_REG_BAR_IDX 2
 
+typedef struct CXLState {
+    bool is_enabled;
+} CXLState;
+
 #endif
Signature · 1 line
2.32.0
#15 lore raw

Re: [PATCH v10 13/45] cxl: Machine level control on whether CXL support is enabled

Quoted 6 lines · Jonathan Cameron via

From: Jonathan Cameron <[email protected]>

There are going to be some potential overheads to CXL enablement, for example the host bridge region reserved in memory maps. Add a machine level control so that CXL is disabled by default.

Quoted 8 lines
---
  hw/core/machine.c    | 28 ++++++++++++++++++++++++++++
  hw/i386/pc.c         |  1 +
  include/hw/boards.h  |  2 ++
  include/hw/cxl/cxl.h |  4 ++++
  4 files changed, 35 insertions(+)

Another belated review, I think this shouldn't be added to machines that do not support CXL (yes, there are options like -M usb but they are from olden times---and CXL is a little more niche, too :)).

Can you move the CXL code for machines to e.g. hw/cxl/machine.c and have
the various machines call back into hooks to add the properties, resolve
the memory window targets etc.?. A CXLState* like cxl_devices_state can be added to the machines and passed as CXLState** to one of these hooks.

Thanks,

Paolo

Quoted 112 lines
hw/core/machine.c +28 -0
index cb9bbc844d..6ae2997f16 100644
@@ -31,6 +31,7 @@
  #include "sysemu/qtest.h"
  #include "hw/pci/pci.h"
  #include "hw/mem/nvdimm.h"
+#include "hw/cxl/cxl.h"
  #include "migration/global_state.h"
  #include "migration/vmstate.h"
  #include "exec/confidential-guest-support.h"
@@ -550,6 +551,20 @@ static void machine_set_nvdimm_persistence(Object *obj, const char *value,
      nvdimms_state->persistence_string = g_strdup(value);
  }
 
+static bool machine_get_cxl(Object *obj, Error **errp)
+{
+    MachineState *ms = MACHINE(obj);
+
+    return ms->cxl_devices_state->is_enabled;
+}
+
+static void machine_set_cxl(Object *obj, bool value, Error **errp)
+{
+    MachineState *ms = MACHINE(obj);
+
+    ms->cxl_devices_state->is_enabled = value;
+}
+
  void machine_class_allow_dynamic_sysbus_dev(MachineClass *mc, const char *type)
  {
      QAPI_LIST_PREPEND(mc->allowed_dynamic_sysbus_devices, g_strdup(type));
@@ -782,6 +797,8 @@ static void machine_class_init(ObjectClass *oc, void *data)
      mc->default_ram_size = 128 * MiB;
      mc->rom_file_has_mr = true;
 
+    /* Few machines support CXL, so default to off */
+    mc->cxl_supported = false;
      /* numa node memory size aligned on 8MB by default.
       * On Linux, each node's border has to be 8MB aligned
       */
@@ -927,6 +944,16 @@ static void machine_initfn(Object *obj)
                                          "Valid values are cpu, mem-ctrl");
      }
 
+    if (mc->cxl_supported) {
+        Object *obj = OBJECT(ms);
+
+        ms->cxl_devices_state = g_new0(CXLState, 1);
+        object_property_add_bool(obj, "cxl", machine_get_cxl, machine_set_cxl);
+        object_property_set_description(obj, "cxl",
+                                        "Set on/off to enable/disable "
+                                        "CXL instantiation");
+    }
+
      if (mc->cpu_index_to_instance_props && mc->get_default_cpu_node_id) {
          ms->numa_state = g_new0(NumaState, 1);
          object_property_add_bool(obj, "hmat",
@@ -961,6 +988,7 @@ static void machine_finalize(Object *obj)
      g_free(ms->device_memory);
      g_free(ms->nvdimms_state);
      g_free(ms->numa_state);
+    g_free(ms->cxl_devices_state);
  }
 
  bool machine_usb(MachineState *machine)
hw/i386/pc.c +1 -0
index 23bba9d82c..b752339beb 100644
@@ -1761,6 +1761,7 @@ static void pc_machine_class_init(ObjectClass *oc, void *data)
      mc->default_cpu_type = TARGET_DEFAULT_CPU_TYPE;
      mc->nvdimm_supported = true;
      mc->smp_props.dies_supported = true;
+    mc->cxl_supported = true;
      mc->default_ram_id = "pc.ram";
 
      object_class_property_add(oc, PC_MACHINE_MAX_RAM_BELOW_4G, "size",
include/hw/boards.h +2 -0
index d64b5481e8..f756a1d5fc 100644
@@ -269,6 +269,7 @@ struct MachineClass {
      bool ignore_boot_device_suffixes;
      bool smbus_no_migration_support;
      bool nvdimm_supported;
+    bool cxl_supported;
      bool numa_mem_supported;
      bool auto_enable_numa;
      SMPCompatProps smp_props;
@@ -360,6 +361,7 @@ struct MachineState {
      CPUArchIdList *possible_cpus;
      CpuTopology smp;
      struct NVDIMMState *nvdimms_state;
+    struct CXLState *cxl_devices_state;
      struct NumaState *numa_state;
  };
 
include/hw/cxl/cxl.h +4 -0
index 554ad93b6b..31af92fd5e 100644
@@ -17,4 +17,8 @@
  #define CXL_COMPONENT_REG_BAR_IDX 0
  #define CXL_DEVICE_REG_BAR_IDX 2
 
+typedef struct CXLState {
+    bool is_enabled;
+} CXLState;
+
  #endif
#16 lore raw

PATCH v10 14/45 hw/pxb: Allow creation of a CXL PXB (host bridge)

In reply to Jonathan Cameron

From: Ben Widawsky <[email protected]>

This works like adding a typical pxb device, except the name is 'pxb-cxl' instead of 'pxb-pcie'. An example command line would be as follows:
-device pxb-cxl,id=cxl.0,bus="pcie.0",bus_nr=1

A CXL PXB is backward compatible with PCIe. What this means in practice is that an operating system that is unaware of CXL should still be able to enumerate this topology as if it were PCIe.

One can create multiple CXL PXB host bridges, but a host bridge can only be connected to the main root bus. Host bridges cannot appear elsewhere in the topology.

Note that as of this patch, the ACPI tables needed for the host bridge (specifically, an ACPI object in _SB named ACPI0016 and the CEDT) aren't created. So while this patch internally creates it, it cannot be properly used by an operating system or other system software.

Also necessary is to add an exception to scripts/device-crash-test similar to that for exiting pxb as both must created on a PCIexpress host bus.

---
 hw/pci-bridge/pci_expander_bridge.c | 86 ++++++++++++++++++++++++++++-
 hw/pci/pci.c                        |  7 +++
 include/hw/pci/pci.h                |  6 ++
 scripts/device-crash-test           |  1 +
 4 files changed, 98 insertions(+), 2 deletions(-)
hw/pci-bridge/pci_expander_bridge.c +84 -2
index a6caa1e7b5..f762eb4a6e 100644
@@ -17,6 +17,7 @@
 #include "hw/pci/pci_host.h"
 #include "hw/qdev-properties.h"
 #include "hw/pci/pci_bridge.h"
+#include "hw/cxl/cxl.h"
 #include "qemu/range.h"
 #include "qemu/error-report.h"
 #include "qemu/module.h"
@@ -56,6 +57,16 @@ DECLARE_INSTANCE_CHECKER(PXBDev, PXB_DEV,
 DECLARE_INSTANCE_CHECKER(PXBDev, PXB_PCIE_DEV,
                          TYPE_PXB_PCIE_DEVICE)
 
+#define TYPE_PXB_CXL_DEVICE "pxb-cxl"
+DECLARE_INSTANCE_CHECKER(PXBDev, PXB_CXL_DEV,
+                         TYPE_PXB_CXL_DEVICE)
+
+typedef struct CXLHost {
+    PCIHostState parent_obj;
+
+    CXLComponentState cxl_cstate;
+} CXLHost;
+
 struct PXBDev {
     /*< private >*/
     PCIDevice parent_obj;
@@ -68,6 +79,11 @@ struct PXBDev {
 
 static PXBDev *convert_to_pxb(PCIDevice *dev)
 {
+    /* A CXL PXB's parent bus is PCIe, so the normal check won't work */
+    if (object_dynamic_cast(OBJECT(dev), TYPE_PXB_CXL_DEVICE)) {
+        return PXB_CXL_DEV(dev);
+    }
+
     return pci_bus_is_express(pci_get_bus(dev))
         ? PXB_PCIE_DEV(dev) : PXB_DEV(dev);
 }
@@ -112,11 +128,20 @@ static const TypeInfo pxb_pcie_bus_info = {
     .class_init    = pxb_bus_class_init,
 };
 
+static const TypeInfo pxb_cxl_bus_info = {
+    .name          = TYPE_PXB_CXL_BUS,
+    .parent        = TYPE_CXL_BUS,
+    .instance_size = sizeof(PXBBus),
+    .class_init    = pxb_bus_class_init,
+};
+
 static const char *pxb_host_root_bus_path(PCIHostState *host_bridge,
                                           PCIBus *rootbus)
 {
-    PXBBus *bus = pci_bus_is_express(rootbus) ?
-                  PXB_PCIE_BUS(rootbus) : PXB_BUS(rootbus);
+    PXBBus *bus = pci_bus_is_cxl(rootbus) ?
+                      PXB_CXL_BUS(rootbus) :
+                      pci_bus_is_express(rootbus) ? PXB_PCIE_BUS(rootbus) :
+                                                    PXB_BUS(rootbus);
 
     snprintf(bus->bus_path, 8, "0000:%02x", pxb_bus_num(rootbus));
     return bus->bus_path;
@@ -218,6 +243,10 @@ static int pxb_map_irq_fn(PCIDevice *pci_dev, int pin)
     return pin - PCI_SLOT(pxb->devfn);
 }
 
+static void pxb_dev_reset(DeviceState *dev)
+{
+}
+
 static gint pxb_compare(gconstpointer a, gconstpointer b)
 {
     const PXBDev *pxb_a = a, *pxb_b = b;
@@ -389,13 +418,66 @@ static const TypeInfo pxb_pcie_dev_info = {
     },
 };
 
+static void pxb_cxl_dev_realize(PCIDevice *dev, Error **errp)
+{
+    MachineState *ms = MACHINE(qdev_get_machine());
+
+    /* A CXL PXB's parent bus is still PCIe */
+    if (!pci_bus_is_express(pci_get_bus(dev))) {
+        error_setg(errp, "pxb-cxl devices cannot reside on a PCI bus");
+        return;
+    }
+    if (!ms->cxl_devices_state->is_enabled) {
+        error_setg(errp, "Machine does not have cxl=on");
+        return;
+    }
+
+    pxb_dev_realize_common(dev, CXL, errp);
+    pxb_dev_reset(DEVICE(dev));
+}
+
+static void pxb_cxl_dev_class_init(ObjectClass *klass, void *data)
+{
+    DeviceClass *dc   = DEVICE_CLASS(klass);
+    PCIDeviceClass *k = PCI_DEVICE_CLASS(klass);
+
+    k->realize             = pxb_cxl_dev_realize;
+    k->exit                = pxb_dev_exitfn;
+    /*
+     * XXX: These types of bridges don't actually show up in the hierarchy so
+     * vendor, device, class, etc. ids are intentionally left out.
+     */
+
+    dc->desc = "CXL Host Bridge";
+    device_class_set_props(dc, pxb_dev_properties);
+    set_bit(DEVICE_CATEGORY_BRIDGE, dc->categories);
+
+    /* Host bridges aren't hotpluggable. FIXME: spec reference */
+    dc->hotpluggable = false;
+    dc->reset = pxb_dev_reset;
+}
+
+static const TypeInfo pxb_cxl_dev_info = {
+    .name          = TYPE_PXB_CXL_DEVICE,
+    .parent        = TYPE_PCI_DEVICE,
+    .instance_size = sizeof(PXBDev),
+    .class_init    = pxb_cxl_dev_class_init,
+    .interfaces =
+        (InterfaceInfo[]){
+            { INTERFACE_CONVENTIONAL_PCI_DEVICE },
+            {},
+        },
+};
+
 static void pxb_register_types(void)
 {
     type_register_static(&pxb_bus_info);
     type_register_static(&pxb_pcie_bus_info);
+    type_register_static(&pxb_cxl_bus_info);
     type_register_static(&pxb_host_info);
     type_register_static(&pxb_dev_info);
     type_register_static(&pxb_pcie_dev_info);
+    type_register_static(&pxb_cxl_dev_info);
 }
 
 type_init(pxb_register_types)
hw/pci/pci.c +7 -0
index 08bd2db2ad..41ec9ed43d 100644
@@ -228,6 +228,12 @@ static const TypeInfo pcie_bus_info = {
     .class_init = pcie_bus_class_init,
 };
 
+static const TypeInfo cxl_bus_info = {
+    .name       = TYPE_CXL_BUS,
+    .parent     = TYPE_PCIE_BUS,
+    .class_init = pcie_bus_class_init,
+};
+
 static PCIBus *pci_find_bus_nr(PCIBus *bus, int bus_num);
 static void pci_update_mappings(PCIDevice *d);
 static void pci_irq_handler(void *opaque, int irq_num, int level);
@@ -2945,6 +2951,7 @@ static void pci_register_types(void)
 {
     type_register_static(&pci_bus_info);
     type_register_static(&pcie_bus_info);
+    type_register_static(&cxl_bus_info);
     type_register_static(&conventional_pci_interface_info);
     type_register_static(&cxl_interface_info);
     type_register_static(&pcie_interface_info);
include/hw/pci/pci.h +6 -0
index 98f0d1b844..44dacfa224 100644
@@ -408,6 +408,7 @@ typedef PCIINTxRoute (*pci_route_irq_fn)(void *opaque, int pin);
 #define TYPE_PCI_BUS "PCI"
 OBJECT_DECLARE_TYPE(PCIBus, PCIBusClass, PCI_BUS)
 #define TYPE_PCIE_BUS "PCIE"
+#define TYPE_CXL_BUS "CXL"
 
 typedef void (*pci_bus_dev_fn)(PCIBus *b, PCIDevice *d, void *opaque);
 typedef void (*pci_bus_fn)(PCIBus *b, void *opaque);
@@ -770,6 +771,11 @@ static inline void pci_irq_pulse(PCIDevice *pci_dev)
     pci_irq_deassert(pci_dev);
 }
 
+static inline int pci_is_cxl(const PCIDevice *d)
+{
+    return d->cap_present & QEMU_PCIE_CAP_CXL;
+}
+
 static inline int pci_is_express(const PCIDevice *d)
 {
     return d->cap_present & QEMU_PCI_CAP_EXPRESS;
scripts/device-crash-test +1 -0
index 4bfc68c008..a203b3fdea 100755
@@ -93,6 +93,7 @@ ERROR_RULE_LIST = [
     {'device':'pci-bridge', 'expected':True},              # Bridge chassis not specified. Each bridge is required to be assigned a unique chassis id > 0.
     {'device':'pci-bridge-seat', 'expected':True},         # Bridge chassis not specified. Each bridge is required to be assigned a unique chassis id > 0.
     {'device':'pxb', 'expected':True},                     # Bridge chassis not specified. Each bridge is required to be assigned a unique chassis id > 0.
+    {'device':'pxb-cxl', 'expected':True},                 # pxb-cxl devices cannot reside on a PCI bus.
     {'device':'scsi-block', 'expected':True},              # drive property not set
     {'device':'scsi-generic', 'expected':True},            # drive property not set
     {'device':'scsi-hd', 'expected':True},                 # drive property not set
Signature · 1 line
2.32.0
#17 lore raw

PATCH v10 15/45 qtest/cxl: Introduce initial test for pxb-cxl only.

In reply to Jonathan Cameron

Initial test with just pxb-cxl. Other tests will be added alongside functionality.

---
 tests/qtest/cxl-test.c  | 22 ++++++++++++++++++++++
 tests/qtest/meson.build |  4 ++++
 2 files changed, 26 insertions(+)
tests/qtest/cxl-test.c new file +22 -0
new file mode 100644
index 0000000000..c031049930
@@ -0,0 +1,22 @@
+/*
+ * QTest testcase for CXL
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ */
+
+#include "qemu/osdep.h"
+#include "libqtest-single.h"
+
+static void cxl_basic_pxb(void)
+{
+    qtest_start("-machine q35,cxl=on -device pxb-cxl,bus=pcie.0");
+    qtest_end();
+}
+
+int main(int argc, char **argv)
+{
+    g_test_init(&argc, &argv, NULL);
+    qtest_add_func("/pci/cxl/basic_pxb", cxl_basic_pxb);
+    return g_test_run();
+}
tests/qtest/meson.build +4 -0
index 22e1361210..dedb477890 100644
@@ -35,6 +35,9 @@ qtests_pci = \
   (config_all_devices.has_key('CONFIG_VGA') ? ['display-vga-test'] : []) +                  \
   (config_all_devices.has_key('CONFIG_IVSHMEM_DEVICE') ? ['ivshmem-test'] : [])
 
+qtests_cxl = \
+  (config_all_devices.has_key('CONFIG_CXL') ? ['cxl-test'] : [])
+
 qtests_i386 = \
   (slirp.found() ? ['pxe-test', 'test-netfilter'] : []) +             \
   (config_host.has_key('CONFIG_POSIX') ? ['test-filter-mirror'] : []) +                     \
@@ -74,6 +77,7 @@ qtests_i386 = \
    slirp.found() ? ['virtio-net-failover'] : []) +                                          \
   (unpack_edk2_blobs ? ['bios-tables-test'] : []) +                                         \
   qtests_pci +                                                                              \
+  qtests_cxl +                                                                              \
   ['fdc-test',
    'ide-test',
    'hd-geo-test',
Signature · 1 line
2.32.0
#18 lore raw

PATCH v10 16/45 hw/cxl/rp: Add a root port

In reply to Jonathan Cameron

From: Ben Widawsky <[email protected]>

This adds just enough of a root port implementation to be able to enumerate root ports (creating the required DVSEC entries). What's not here yet is the MMIO nor the ability to write some of the DVSEC entries.

This can be added with the qemu commandline by adding a rootport to a specific CXL host bridge. For example:
-device cxl-rp,id=rp0,bus="cxl.0",addr=0.0,chassis=4

Like the host bridge patch, the ACPI tables aren't generated at this point and so system software cannot use it.

---
 hw/pci-bridge/Kconfig          |   5 +
 hw/pci-bridge/cxl_root_port.c  | 236 +++++++++++++++++++++++++++++++++
 hw/pci-bridge/meson.build      |   1 +
 hw/pci-bridge/pcie_root_port.c |   6 +-
 hw/pci/pci.c                   |   4 +-
 5 files changed, 250 insertions(+), 2 deletions(-)
hw/pci-bridge/Kconfig +5 -0
index f8df4315ba..02614f49aa 100644
@@ -27,3 +27,8 @@ config DEC_PCI
 
 config SIMBA
     bool
+
+config CXL
+    bool
+    default y if PCI_EXPRESS && PXB
+    depends on PCI_EXPRESS && MSI_NONBROKEN && PXB
hw/pci-bridge/cxl_root_port.c new file +236 -0

236 lines not on this page · raw message

hw/pci-bridge/meson.build +1 -0
index daab8acf2a..b6d26a03d5 100644
@@ -5,6 +5,7 @@ pci_ss.add(when: 'CONFIG_IOH3420', if_true: files('ioh3420.c'))
 pci_ss.add(when: 'CONFIG_PCIE_PORT', if_true: files('pcie_root_port.c', 'gen_pcie_root_port.c', 'pcie_pci_bridge.c'))
 pci_ss.add(when: 'CONFIG_PXB', if_true: files('pci_expander_bridge.c'))
 pci_ss.add(when: 'CONFIG_XIO3130', if_true: files('xio3130_upstream.c', 'xio3130_downstream.c'))
+pci_ss.add(when: 'CONFIG_CXL', if_true: files('cxl_root_port.c'))
 
 # NewWorld PowerMac
 pci_ss.add(when: 'CONFIG_DEC_PCI', if_true: files('dec.c'))
hw/pci-bridge/pcie_root_port.c +5 -1
index f1cfe9d14a..460e48269d 100644
@@ -67,7 +67,11 @@ static void rp_realize(PCIDevice *d, Error **errp)
     int rc;
 
     pci_config_set_interrupt_pin(d->config, 1);
-    pci_bridge_initfn(d, TYPE_PCIE_BUS);
+    if (d->cap_present & QEMU_PCIE_CAP_CXL) {
+        pci_bridge_initfn(d, TYPE_CXL_BUS);
+    } else {
+        pci_bridge_initfn(d, TYPE_PCIE_BUS);
+    }
     pcie_port_init_reg(d);
 
     rc = pci_bridge_ssvid_init(d, rpc->ssvid_offset, dc->vendor_id,
hw/pci/pci.c +3 -1
index 41ec9ed43d..b888b5201e 100644
@@ -2761,7 +2761,9 @@ static void pci_device_class_base_init(ObjectClass *klass, void *data)
             object_class_dynamic_cast(klass, INTERFACE_CONVENTIONAL_PCI_DEVICE);
         ObjectClass *pcie =
             object_class_dynamic_cast(klass, INTERFACE_PCIE_DEVICE);
-        assert(conventional || pcie);
+        ObjectClass *cxl =
+            object_class_dynamic_cast(klass, INTERFACE_CXL_DEVICE);
+        assert(conventional || pcie || cxl);
     }
 }
 
Signature · 1 line
2.32.0
#19 lore raw

PATCH v10 17/45 hw/cxl/device: Add a memory device (8.2.8.5)

In reply to Jonathan Cameron

From: Ben Widawsky <[email protected]>

A CXL memory device (AKA Type 3) is a CXL component that contains some combination of volatile and persistent memory. It also implements the previously defined mailbox interface as well as the memory device firmware interface.

Although the memory device is configured like a normal PCIe device, the memory traffic is on an entirely separate bus conceptually (using the same physical wires as PCIe, but different protocol).

Once the CXL topology is fully configure and address decoders committed, the guest physical address for the memory device is part of a larger window which is owned by the platform. The creation of these windows is later in this series.

The following example will create a 256M device in a 512M window:
-object "memory-backend-file,id=cxl-mem1,share,mem-path=cxl-type3,size=512M" -device "cxl-type3,bus=rp0,memdev=cxl-mem1,id=cxl-pmem0"

Note: Dropped PCDIMM info interfaces for now. They can be added if appropriate at a later date.

---
 hw/cxl/cxl-component-utils.c |  18 ++++
 hw/cxl/cxl-mailbox-utils.c   |  46 ++++++++++
 hw/mem/Kconfig               |   5 ++
 hw/mem/cxl_type3.c           | 162 +++++++++++++++++++++++++++++++++++
 hw/mem/meson.build           |   1 +
 include/hw/cxl/cxl_device.h  |  15 ++++
 include/hw/cxl/cxl_pci.h     |  21 +++++
 include/hw/pci/pci_ids.h     |   1 +
 8 files changed, 269 insertions(+)
hw/cxl/cxl-component-utils.c +18 -0
index 9457a1b029..afc97b17c2 100644
@@ -252,6 +252,24 @@ void cxl_component_create_dvsec(CXLComponentState *cxl,
     /* Configure write masks */
     switch (type) {
     case PCIE_CXL_DEVICE_DVSEC:
+        /* Cntrl RW Lock - so needs explicit blocking when lock is set */
+        wmask[offset + offsetof(CXLDVSECDevice, ctrl)] = 0xFD;
+        wmask[offset + offsetof(CXLDVSECDevice, ctrl) + 1] = 0x4F;
+        /* Status is RW1CS */
+        wmask[offset + offsetof(CXLDVSECDevice, ctrl2)] = 0x0F;
+       /* Lock is RW Once */
+        wmask[offset + offsetof(CXLDVSECDevice, lock)] = 0x01;
+        /* range1/2_base_high/low is RW Lock */
+        wmask[offset + offsetof(CXLDVSECDevice, range1_base_hi)] = 0xFF;
+        wmask[offset + offsetof(CXLDVSECDevice, range1_base_hi) + 1] = 0xFF;
+        wmask[offset + offsetof(CXLDVSECDevice, range1_base_hi) + 2] = 0xFF;
+        wmask[offset + offsetof(CXLDVSECDevice, range1_base_hi) + 3] = 0xFF;
+        wmask[offset + offsetof(CXLDVSECDevice, range1_base_lo) + 3] = 0xF0;
+        wmask[offset + offsetof(CXLDVSECDevice, range2_base_hi)] = 0xFF;
+        wmask[offset + offsetof(CXLDVSECDevice, range2_base_hi) + 1] = 0xFF;
+        wmask[offset + offsetof(CXLDVSECDevice, range2_base_hi) + 2] = 0xFF;
+        wmask[offset + offsetof(CXLDVSECDevice, range2_base_hi) + 3] = 0xFF;
+        wmask[offset + offsetof(CXLDVSECDevice, range2_base_lo) + 3] = 0xF0;
         break;
     case NON_CXL_FUNCTION_MAP_DVSEC:
         break; /* Not yet implemented */
hw/cxl/cxl-mailbox-utils.c +46 -0

64 lines not on this page · raw message

hw/mem/Kconfig +5 -0
index 03dbb3c7df..73c5ae8ad9 100644
@@ -11,3 +11,8 @@ config NVDIMM
 
 config SPARSE_MEM
     bool
+
+config CXL_MEM_DEVICE
+    bool
+    default y if CXL
+    select MEM_DEVICE
hw/mem/cxl_type3.c new file +162 -0

162 lines not on this page · raw message

hw/mem/meson.build +1 -0
index 82f86d117e..609b2b36fc 100644
@@ -3,6 +3,7 @@ mem_ss.add(files('memory-device.c'))
 mem_ss.add(when: 'CONFIG_DIMM', if_true: files('pc-dimm.c'))
 mem_ss.add(when: 'CONFIG_NPCM7XX', if_true: files('npcm7xx_mc.c'))
 mem_ss.add(when: 'CONFIG_NVDIMM', if_true: files('nvdimm.c'))
+mem_ss.add(when: 'CONFIG_CXL_MEM_DEVICE', if_true: files('cxl_type3.c'))
 
 softmmu_ss.add_all(when: 'CONFIG_MEM_DEVICE', if_true: mem_ss)
 
include/hw/cxl/cxl_device.h +15 -0
index 797a22ddb4..d8da2c7b68 100644
@@ -230,4 +230,19 @@ REG64(CXL_MEM_DEV_STS, 0)
     FIELD(CXL_MEM_DEV_STS, MBOX_READY, 4, 1)
     FIELD(CXL_MEM_DEV_STS, RESET_NEEDED, 5, 3)
 
+struct CXLType3Dev {
+    /* Private */
+    PCIDevice parent_obj;
+
+    /* Properties */
+    HostMemoryBackend *hostmem;
+
+    /* State */
+    CXLComponentState cxl_cstate;
+    CXLDeviceState cxl_dstate;
+};
+
+#define TYPE_CXL_TYPE3 "cxl-type3"
+OBJECT_DECLARE_SIMPLE_TYPE(CXLType3Dev, CXL_TYPE3)
+
 #endif
include/hw/cxl/cxl_pci.h +21 -0
index 3f7e44daac..01cf002096 100644
@@ -67,6 +67,27 @@ QEMU_BUILD_BUG_ON(sizeof(DVSECHeader) != 10);
  * CXL 2.0 Downstream Port: 3, 4, 7, 8
  */
 
+/* CXL 2.0 - 8.1.3 (ID 0001) */
+typedef struct CXLDVSECDevice {
+    DVSECHeader hdr;
+    uint16_t cap;
+    uint16_t ctrl;
+    uint16_t status;
+    uint16_t ctrl2;
+    uint16_t status2;
+    uint16_t lock;
+    uint16_t cap2;
+    uint32_t range1_size_hi;
+    uint32_t range1_size_lo;
+    uint32_t range1_base_hi;
+    uint32_t range1_base_lo;
+    uint32_t range2_size_hi;
+    uint32_t range2_size_lo;
+    uint32_t range2_base_hi;
+    uint32_t range2_base_lo;
+} CXLDVSECDevice;
+QEMU_BUILD_BUG_ON(sizeof(CXLDVSECDevice) != 0x38);
+
 /* CXL 2.0 - 8.1.5 (ID 0003) */
 typedef struct CXLDVSECPortExtensions {
     DVSECHeader hdr;
include/hw/pci/pci_ids.h +1 -0
index 11abe22d46..898083b86f 100644
@@ -53,6 +53,7 @@
 #define PCI_BASE_CLASS_MEMORY            0x05
 #define PCI_CLASS_MEMORY_RAM             0x0500
 #define PCI_CLASS_MEMORY_FLASH           0x0501
+#define PCI_CLASS_MEMORY_CXL             0x0502
 #define PCI_CLASS_MEMORY_OTHER           0x0580
 
 #define PCI_BASE_CLASS_BRIDGE            0x06
Signature · 1 line
2.32.0
#20 lore raw

Re: [PATCH v10 17/45] hw/cxl/device: Add a memory device (8.2.8.5)

Hi Ben, Jonathan,

hw/mem/cxl_type3.c

Quoted 43 lines · Jonathan Cameron via

From: Ben Widawsky <[email protected]>

A CXL memory device (AKA Type 3) is a CXL component that contains some combination of volatile and persistent memory. It also implements the previously defined mailbox interface as well as the memory device firmware interface.

Although the memory device is configured like a normal PCIe device, the memory traffic is on an entirely separate bus conceptually (using the same physical wires as PCIe, but different protocol).

Once the CXL topology is fully configure and address decoders committed, the guest physical address for the memory device is part of a larger window which is owned by the platform. The creation of these windows is later in this series.

The following example will create a 256M device in a 512M window:

(quoted diff) +0 -2
-object "memory-backend-file,id=cxl-mem1,share,mem-path=cxl-type3,size=512M"
-device "cxl-type3,bus=rp0,memdev=cxl-mem1,id=cxl-pmem0"

Note: Dropped PCDIMM info interfaces for now. They can be added if appropriate at a later date.

---
  hw/cxl/cxl-component-utils.c |  18 ++++
  hw/cxl/cxl-mailbox-utils.c   |  46 ++++++++++
  hw/mem/Kconfig               |   5 ++
  hw/mem/cxl_type3.c           | 162 +++++++++++++++++++++++++++++++++++
  hw/mem/meson.build           |   1 +
  include/hw/cxl/cxl_device.h  |  15 ++++
  include/hw/cxl/cxl_pci.h     |  21 +++++
  include/hw/pci/pci_ids.h     |   1 +
  8 files changed, 269 insertions(+)
hw/mem/cxl_type3.c new file +0 -0
new file mode 100644
index 0000000000..290c088dd6
@@ -0,0 +1,162 @@

This file lacks a license, do you mind adding one?

Thanks!

Quoted 162 lines
hw/mem/cxl_type3.c new file +162 -0
+#include "qemu/osdep.h"
+#include "qemu/units.h"
+#include "qemu/error-report.h"
+#include "hw/mem/memory-device.h"
+#include "hw/mem/pc-dimm.h"
+#include "hw/pci/pci.h"
+#include "hw/qdev-properties.h"
+#include "qapi/error.h"
+#include "qemu/log.h"
+#include "qemu/module.h"
+#include "qemu/range.h"
+#include "qemu/rcu.h"
+#include "sysemu/hostmem.h"
+#include "hw/cxl/cxl.h"
+
+static void build_dvsecs(CXLType3Dev *ct3d)
+{
+    CXLComponentState *cxl_cstate = &ct3d->cxl_cstate;
+    uint8_t *dvsec;
+
+    dvsec = (uint8_t *)&(CXLDVSECDevice){
+        .cap = 0x1e,
+        .ctrl = 0x2,
+        .status2 = 0x2,
+        .range1_size_hi = ct3d->hostmem->size >> 32,
+        .range1_size_lo = (2 << 5) | (2 << 2) | 0x3 |
+        (ct3d->hostmem->size & 0xF0000000),
+        .range1_base_hi = 0,
+        .range1_base_lo = 0,
+    };
+    cxl_component_create_dvsec(cxl_cstate, CXL2_TYPE3_DEVICE,
+                               PCIE_CXL_DEVICE_DVSEC_LENGTH,
+                               PCIE_CXL_DEVICE_DVSEC,
+                               PCIE_CXL2_DEVICE_DVSEC_REVID, dvsec);
+
+    dvsec = (uint8_t *)&(CXLDVSECRegisterLocator){
+        .rsvd         = 0,
+        .reg0_base_lo = RBI_COMPONENT_REG | CXL_COMPONENT_REG_BAR_IDX,
+        .reg0_base_hi = 0,
+        .reg1_base_lo = RBI_CXL_DEVICE_REG | CXL_DEVICE_REG_BAR_IDX,
+        .reg1_base_hi = 0,
+    };
+    cxl_component_create_dvsec(cxl_cstate, CXL2_TYPE3_DEVICE,
+                               REG_LOC_DVSEC_LENGTH, REG_LOC_DVSEC,
+                               REG_LOC_DVSEC_REVID, dvsec);
+    dvsec = (uint8_t *)&(CXLDVSECDeviceGPF){
+        .phase2_duration = 0x603, /* 3 seconds */
+        .phase2_power = 0x33, /* 0x33 miliwatts */
+    };
+    cxl_component_create_dvsec(cxl_cstate, CXL2_TYPE3_DEVICE,
+                               GPF_DEVICE_DVSEC_LENGTH, GPF_PORT_DVSEC,
+                               GPF_DEVICE_DVSEC_REVID, dvsec);
+}
+
+static bool cxl_setup_memory(CXLType3Dev *ct3d, Error **errp)
+{
+    MemoryRegion *mr;
+
+    if (!ct3d->hostmem) {
+        error_setg(errp, "memdev property must be set");
+        return false;
+    }
+
+    mr = host_memory_backend_get_memory(ct3d->hostmem);
+    if (!mr) {
+        error_setg(errp, "memdev property must be set");
+        return false;
+    }
+    memory_region_set_nonvolatile(mr, true);
+    memory_region_set_enabled(mr, true);
+    host_memory_backend_set_mapped(ct3d->hostmem, true);
+    ct3d->cxl_dstate.pmem_size = ct3d->hostmem->size;
+
+    return true;
+}
+
+static void ct3_realize(PCIDevice *pci_dev, Error **errp)
+{
+    CXLType3Dev *ct3d = CXL_TYPE3(pci_dev);
+    CXLComponentState *cxl_cstate = &ct3d->cxl_cstate;
+    ComponentRegisters *regs = &cxl_cstate->crb;
+    MemoryRegion *mr = &regs->component_registers;
+    uint8_t *pci_conf = pci_dev->config;
+
+    if (!cxl_setup_memory(ct3d, errp)) {
+        return;
+    }
+
+    pci_config_set_prog_interface(pci_conf, 0x10);
+    pci_config_set_class(pci_conf, PCI_CLASS_MEMORY_CXL);
+
+    pcie_endpoint_cap_init(pci_dev, 0x80);
+    cxl_cstate->dvsec_offset = 0x100;
+
+    ct3d->cxl_cstate.pdev = pci_dev;
+    build_dvsecs(ct3d);
+
+    cxl_component_register_block_init(OBJECT(pci_dev), cxl_cstate,
+                                      TYPE_CXL_TYPE3);
+
+    pci_register_bar(
+        pci_dev, CXL_COMPONENT_REG_BAR_IDX,
+        PCI_BASE_ADDRESS_SPACE_MEMORY | PCI_BASE_ADDRESS_MEM_TYPE_64, mr);
+
+    cxl_device_register_block_init(OBJECT(pci_dev), &ct3d->cxl_dstate);
+    pci_register_bar(pci_dev, CXL_DEVICE_REG_BAR_IDX,
+                     PCI_BASE_ADDRESS_SPACE_MEMORY |
+                         PCI_BASE_ADDRESS_MEM_TYPE_64,
+                     &ct3d->cxl_dstate.device_registers);
+}
+
+static void ct3d_reset(DeviceState *dev)
+{
+    CXLType3Dev *ct3d = CXL_TYPE3(dev);
+    uint32_t *reg_state = ct3d->cxl_cstate.crb.cache_mem_registers;
+    uint32_t *write_msk = ct3d->cxl_cstate.crb.cache_mem_regs_write_mask;
+
+    cxl_component_register_init_common(reg_state, write_msk, CXL2_TYPE3_DEVICE);
+    cxl_device_register_init_common(&ct3d->cxl_dstate);
+}
+
+static Property ct3_props[] = {
+    DEFINE_PROP_LINK("memdev", CXLType3Dev, hostmem, TYPE_MEMORY_BACKEND,
+                     HostMemoryBackend *),
+    DEFINE_PROP_END_OF_LIST(),
+};
+
+static void ct3_class_init(ObjectClass *oc, void *data)
+{
+    DeviceClass *dc = DEVICE_CLASS(oc);
+    PCIDeviceClass *pc = PCI_DEVICE_CLASS(oc);
+
+    pc->realize = ct3_realize;
+    pc->class_id = PCI_CLASS_STORAGE_EXPRESS;
+    pc->vendor_id = PCI_VENDOR_ID_INTEL;
+    pc->device_id = 0xd93; /* LVF for now */
+    pc->revision = 1;
+
+    set_bit(DEVICE_CATEGORY_STORAGE, dc->categories);
+    dc->desc = "CXL PMEM Device (Type 3)";
+    dc->reset = ct3d_reset;
+    device_class_set_props(dc, ct3_props);
+}
+
+static const TypeInfo ct3d_info = {
+    .name = TYPE_CXL_TYPE3,
+    .parent = TYPE_PCI_DEVICE,
+    .class_init = ct3_class_init,
+    .instance_size = sizeof(CXLType3Dev),
+    .interfaces = (InterfaceInfo[]) {
+        { INTERFACE_CXL_DEVICE },
+        { INTERFACE_PCIE_DEVICE },
+        {}
+    },
+};
+
+static void ct3d_registers(void)
+{
+    type_register_static(&ct3d_info);
+}
+
+type_init(ct3d_registers);
#21 lore raw

PATCH v10 18/45 hw/cxl/device: Implement MMIO HDM decoding (8.2.5.12)

In reply to Jonathan Cameron

From: Ben Widawsky <[email protected]>

A device's volatile and persistent memory are known Host Defined Memory (HDM) regions. The mechanism by which the device is programmed to claim the addresses associated with those regions is through dedicated logic known as the HDM decoder. In order to allow the OS to properly program the HDMs, the HDM decoders must be modeled.

There are two ways the HDM decoders can be implemented, the legacy mechanism is through the PCIe DVSEC programming from CXL 1.1 (8.1.3.8), and MMIO is found in 8.2.5.12 of the spec. For now, 8.1.3.8 is not implemented.

Much of CXL device logic is implemented in cxl-utils. The HDM decoder however is implemented directly by the device implementation. Whilst the implementation currently does no validity checks on the encoder set up, future work will add sanity checking specific to the type of cxl component.

---
 hw/mem/cxl_type3.c | 55 ++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 55 insertions(+)
hw/mem/cxl_type3.c +55 -0
index 290c088dd6..318b6f9fbc 100644
@@ -52,6 +52,48 @@ static void build_dvsecs(CXLType3Dev *ct3d)
                                GPF_DEVICE_DVSEC_REVID, dvsec);
 }
 
+static void hdm_decoder_commit(CXLType3Dev *ct3d, int which)
+{
+    ComponentRegisters *cregs = &ct3d->cxl_cstate.crb;
+    uint32_t *cache_mem = cregs->cache_mem_registers;
+
+    assert(which == 0);
+
+    /* TODO: Sanity checks that the decoder is possible */
+    ARRAY_FIELD_DP32(cache_mem, CXL_HDM_DECODER0_CTRL, COMMIT, 0);
+    ARRAY_FIELD_DP32(cache_mem, CXL_HDM_DECODER0_CTRL, ERR, 0);
+
+    ARRAY_FIELD_DP32(cache_mem, CXL_HDM_DECODER0_CTRL, COMMITTED, 1);
+}
+
+static void ct3d_reg_write(void *opaque, hwaddr offset, uint64_t value,
+                           unsigned size)
+{
+    CXLComponentState *cxl_cstate = opaque;
+    ComponentRegisters *cregs = &cxl_cstate->crb;
+    CXLType3Dev *ct3d = container_of(cxl_cstate, CXLType3Dev, cxl_cstate);
+    uint32_t *cache_mem = cregs->cache_mem_registers;
+    bool should_commit = false;
+    int which_hdm = -1;
+
+    assert(size == 4);
+    g_assert(offset < CXL2_COMPONENT_CM_REGION_SIZE);
+
+    switch (offset) {
+    case A_CXL_HDM_DECODER0_CTRL:
+        should_commit = FIELD_EX32(value, CXL_HDM_DECODER0_CTRL, COMMIT);
+        which_hdm = 0;
+        break;
+    default:
+        break;
+    }
+
+    stl_le_p((uint8_t *)cache_mem + offset, value);
+    if (should_commit) {
+        hdm_decoder_commit(ct3d, which_hdm);
+    }
+}
+
 static bool cxl_setup_memory(CXLType3Dev *ct3d, Error **errp)
 {
     MemoryRegion *mr;
@@ -95,6 +137,9 @@ static void ct3_realize(PCIDevice *pci_dev, Error **errp)
     ct3d->cxl_cstate.pdev = pci_dev;
     build_dvsecs(ct3d);
 
+    regs->special_ops = g_new0(MemoryRegionOps, 1);
+    regs->special_ops->write = ct3d_reg_write;
+
     cxl_component_register_block_init(OBJECT(pci_dev), cxl_cstate,
                                       TYPE_CXL_TYPE3);
 
@@ -109,6 +154,15 @@ static void ct3_realize(PCIDevice *pci_dev, Error **errp)
                      &ct3d->cxl_dstate.device_registers);
 }
 
+static void ct3_exit(PCIDevice *pci_dev)
+{
+    CXLType3Dev *ct3d = CXL_TYPE3(pci_dev);
+    CXLComponentState *cxl_cstate = &ct3d->cxl_cstate;
+    ComponentRegisters *regs = &cxl_cstate->crb;
+
+    g_free(regs->special_ops);
+}
+
 static void ct3d_reset(DeviceState *dev)
 {
     CXLType3Dev *ct3d = CXL_TYPE3(dev);
@@ -131,6 +185,7 @@ static void ct3_class_init(ObjectClass *oc, void *data)
     PCIDeviceClass *pc = PCI_DEVICE_CLASS(oc);
 
     pc->realize = ct3_realize;
+    pc->exit = ct3_exit;
     pc->class_id = PCI_CLASS_STORAGE_EXPRESS;
     pc->vendor_id = PCI_VENDOR_ID_INTEL;
     pc->device_id = 0xd93; /* LVF for now */
Signature · 1 line
2.32.0
#22 lore raw

PATCH v10 19/45 hw/cxl/device: Add some trivial commands

In reply to Jonathan Cameron

From: Ben Widawsky <[email protected]>

GET_FW_INFO and GET_PARTITION_INFO, for this emulation, is equivalent to info already returned in the IDENTIFY command. To have a more robust implementation, add those.

---
 hw/cxl/cxl-mailbox-utils.c | 69 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 69 insertions(+)
hw/cxl/cxl-mailbox-utils.c +69 -0
index 4ae0561dfc..c8188d7087 100644
@@ -10,6 +10,7 @@
 #include "qemu/osdep.h"
 #include "hw/cxl/cxl.h"
 #include "hw/pci/pci.h"
+#include "qemu/cutils.h"
 #include "qemu/log.h"
 #include "qemu/uuid.h"
 
@@ -44,6 +45,8 @@ enum {
         #define CLEAR_RECORDS   0x1
         #define GET_INTERRUPT_POLICY   0x2
         #define SET_INTERRUPT_POLICY   0x3
+    FIRMWARE_UPDATE = 0x02,
+        #define GET_INFO      0x0
     TIMESTAMP   = 0x03,
         #define GET           0x0
         #define SET           0x1
@@ -52,6 +55,8 @@ enum {
         #define GET_LOG       0x1
     IDENTIFY    = 0x40,
         #define MEMORY_DEVICE 0x0
+    CCLS        = 0x41,
+        #define GET_PARTITION_INFO     0x0
 };
 
 /* 8.2.8.4.5.1 Command Return Codes */
@@ -114,6 +119,39 @@ DEFINE_MAILBOX_HANDLER_NOP(events_clear_records);
 DEFINE_MAILBOX_HANDLER_ZEROED(events_get_interrupt_policy, 4);
 DEFINE_MAILBOX_HANDLER_NOP(events_set_interrupt_policy);
 
+/* 8.2.9.2.1 */
+static ret_code cmd_firmware_update_get_info(struct cxl_cmd *cmd,
+                                             CXLDeviceState *cxl_dstate,
+                                             uint16_t *len)
+{
+    struct {
+        uint8_t slots_supported;
+        uint8_t slot_info;
+        uint8_t caps;
+        uint8_t rsvd[0xd];
+        char fw_rev1[0x10];
+        char fw_rev2[0x10];
+        char fw_rev3[0x10];
+        char fw_rev4[0x10];
+    } QEMU_PACKED *fw_info;
+    QEMU_BUILD_BUG_ON(sizeof(*fw_info) != 0x50);
+
+    if (cxl_dstate->pmem_size < (256 << 20)) {
+        return CXL_MBOX_INTERNAL_ERROR;
+    }
+
+    fw_info = (void *)cmd->payload;
+    memset(fw_info, 0, sizeof(*fw_info));
+
+    fw_info->slots_supported = 2;
+    fw_info->slot_info = BIT(0) | BIT(3);
+    fw_info->caps = 0;
+    pstrcpy(fw_info->fw_rev1, sizeof(fw_info->fw_rev1), "BWFW VERSION 0");
+
+    *len = sizeof(*fw_info);
+    return CXL_MBOX_SUCCESS;
+}
+
 /* 8.2.9.3.1 */
 static ret_code cmd_timestamp_get(struct cxl_cmd *cmd,
                                   CXLDeviceState *cxl_dstate,
@@ -258,6 +296,33 @@ static ret_code cmd_identify_memory_device(struct cxl_cmd *cmd,
     return CXL_MBOX_SUCCESS;
 }
 
+static ret_code cmd_ccls_get_partition_info(struct cxl_cmd *cmd,
+                                           CXLDeviceState *cxl_dstate,
+                                           uint16_t *len)
+{
+    struct {
+        uint64_t active_vmem;
+        uint64_t active_pmem;
+        uint64_t next_vmem;
+        uint64_t next_pmem;
+    } QEMU_PACKED *part_info = (void *)cmd->payload;
+    QEMU_BUILD_BUG_ON(sizeof(*part_info) != 0x20);
+    uint64_t size = cxl_dstate->pmem_size;
+
+    if (!QEMU_IS_ALIGNED(size, 256 << 20)) {
+        return CXL_MBOX_INTERNAL_ERROR;
+    }
+
+    /* PMEM only */
+    part_info->active_vmem = 0;
+    part_info->next_vmem = 0;
+    part_info->active_pmem = size / (256 << 20);
+    part_info->next_pmem = 0;
+
+    *len = sizeof(*part_info);
+    return CXL_MBOX_SUCCESS;
+}
+
 #define IMMEDIATE_CONFIG_CHANGE (1 << 1)
 #define IMMEDIATE_POLICY_CHANGE (1 << 3)
 #define IMMEDIATE_LOG_CHANGE (1 << 4)
@@ -271,12 +336,16 @@ static struct cxl_cmd cxl_cmd_set[256][256] = {
         cmd_events_get_interrupt_policy, 0, 0 },
     [EVENTS][SET_INTERRUPT_POLICY] = { "EVENTS_SET_INTERRUPT_POLICY",
         cmd_events_set_interrupt_policy, 4, IMMEDIATE_CONFIG_CHANGE },
+    [FIRMWARE_UPDATE][GET_INFO] = { "FIRMWARE_UPDATE_GET_INFO",
+        cmd_firmware_update_get_info, 0, 0 },
     [TIMESTAMP][GET] = { "TIMESTAMP_GET", cmd_timestamp_get, 0, 0 },
     [TIMESTAMP][SET] = { "TIMESTAMP_SET", cmd_timestamp_set, 8, IMMEDIATE_POLICY_CHANGE },
     [LOGS][GET_SUPPORTED] = { "LOGS_GET_SUPPORTED", cmd_logs_get_supported, 0, 0 },
     [LOGS][GET_LOG] = { "LOGS_GET_LOG", cmd_logs_get_log, 0x18, 0 },
     [IDENTIFY][MEMORY_DEVICE] = { "IDENTIFY_MEMORY_DEVICE",
         cmd_identify_memory_device, 0, 0 },
+    [CCLS][GET_PARTITION_INFO] = { "CCLS_GET_PARTITION_INFO",
+        cmd_ccls_get_partition_info, 0, 0 },
 };
 
 void cxl_process_mailbox(CXLDeviceState *cxl_dstate)
Signature · 1 line
2.32.0
#23 lore raw

PATCH v10 20/45 hw/cxl/device: Plumb real Label Storage Area (LSA) sizing

In reply to Jonathan Cameron

From: Ben Widawsky <[email protected]>

This should introduce no change. Subsequent work will make use of this new class member.

---
 hw/cxl/cxl-mailbox-utils.c  |  3 +++
 hw/mem/cxl_type3.c          |  9 +++++++++
 include/hw/cxl/cxl_device.h | 11 ++++++++++-
 3 files changed, 22 insertions(+), 1 deletion(-)
hw/cxl/cxl-mailbox-utils.c +3 -0
index c8188d7087..492739aef3 100644
@@ -277,6 +277,8 @@ static ret_code cmd_identify_memory_device(struct cxl_cmd *cmd,
     } QEMU_PACKED *id;
     QEMU_BUILD_BUG_ON(sizeof(*id) != 0x43);
 
+    CXLType3Dev *ct3d = container_of(cxl_dstate, CXLType3Dev, cxl_dstate);
+    CXLType3Class *cvc = CXL_TYPE3_GET_CLASS(ct3d);
     uint64_t size = cxl_dstate->pmem_size;
 
     if (!QEMU_IS_ALIGNED(size, 256 << 20)) {
@@ -291,6 +293,7 @@ static ret_code cmd_identify_memory_device(struct cxl_cmd *cmd,
 
     id->total_capacity = size / (256 << 20);
     id->persistent_capacity = size / (256 << 20);
+    id->lsa_size = cvc->get_lsa_size(ct3d);
 
     *len = sizeof(*id);
     return CXL_MBOX_SUCCESS;
hw/mem/cxl_type3.c +9 -0
index 318b6f9fbc..b5aa1067df 100644
@@ -179,10 +179,16 @@ static Property ct3_props[] = {
     DEFINE_PROP_END_OF_LIST(),
 };
 
+static uint64_t get_lsa_size(CXLType3Dev *ct3d)
+{
+    return 0;
+}
+
 static void ct3_class_init(ObjectClass *oc, void *data)
 {
     DeviceClass *dc = DEVICE_CLASS(oc);
     PCIDeviceClass *pc = PCI_DEVICE_CLASS(oc);
+    CXLType3Class *cvc = CXL_TYPE3_CLASS(oc);
 
     pc->realize = ct3_realize;
     pc->exit = ct3_exit;
@@ -195,11 +201,14 @@ static void ct3_class_init(ObjectClass *oc, void *data)
     dc->desc = "CXL PMEM Device (Type 3)";
     dc->reset = ct3d_reset;
     device_class_set_props(dc, ct3_props);
+
+    cvc->get_lsa_size = get_lsa_size;
 }
 
 static const TypeInfo ct3d_info = {
     .name = TYPE_CXL_TYPE3,
     .parent = TYPE_PCI_DEVICE,
+    .class_size = sizeof(struct CXLType3Class),
     .class_init = ct3_class_init,
     .instance_size = sizeof(CXLType3Dev),
     .interfaces = (InterfaceInfo[]) {
include/hw/cxl/cxl_device.h +10 -1
index d8da2c7b68..ea2571a69b 100644
@@ -236,6 +236,7 @@ struct CXLType3Dev {
 
     /* Properties */
     HostMemoryBackend *hostmem;
+    HostMemoryBackend *lsa;
 
     /* State */
     CXLComponentState cxl_cstate;
@@ -243,6 +244,14 @@ struct CXLType3Dev {
 };
 
 #define TYPE_CXL_TYPE3 "cxl-type3"
-OBJECT_DECLARE_SIMPLE_TYPE(CXLType3Dev, CXL_TYPE3)
+OBJECT_DECLARE_TYPE(CXLType3Dev, CXLType3Class, CXL_TYPE3)
+
+struct CXLType3Class {
+    /* Private */
+    PCIDeviceClass parent_class;
+
+    /* public */
+    uint64_t (*get_lsa_size)(CXLType3Dev *ct3d);
+};
 
 #endif
Signature · 1 line
2.32.0
#24 lore raw

PATCH v10 21/45 hw/cxl/device: Implement get/set Label Storage Area (LSA)

In reply to Jonathan Cameron

From: Ben Widawsky <[email protected]>

Implement get and set handlers for the Label Storage Area used to hold data describing persistent memory configuration so that it can be ensured it is seen in the same configuration after reboot.

---
 hw/cxl/cxl-mailbox-utils.c  | 60 +++++++++++++++++++++++++++++++++++++
 hw/mem/cxl_type3.c          | 56 +++++++++++++++++++++++++++++++++-
 include/hw/cxl/cxl_device.h |  5 ++++
 3 files changed, 120 insertions(+), 1 deletion(-)
hw/cxl/cxl-mailbox-utils.c +60 -0
index 492739aef3..bb66c765a5 100644
@@ -57,6 +57,8 @@ enum {
         #define MEMORY_DEVICE 0x0
     CCLS        = 0x41,
         #define GET_PARTITION_INFO     0x0
+        #define GET_LSA       0x2
+        #define SET_LSA       0x3
 };
 
 /* 8.2.8.4.5.1 Command Return Codes */
@@ -326,7 +328,62 @@ static ret_code cmd_ccls_get_partition_info(struct cxl_cmd *cmd,
     return CXL_MBOX_SUCCESS;
 }
 
+static ret_code cmd_ccls_get_lsa(struct cxl_cmd *cmd,
+                                 CXLDeviceState *cxl_dstate,
+                                 uint16_t *len)
+{
+    struct {
+        uint32_t offset;
+        uint32_t length;
+    } QEMU_PACKED *get_lsa;
+    CXLType3Dev *ct3d = container_of(cxl_dstate, CXLType3Dev, cxl_dstate);
+    CXLType3Class *cvc = CXL_TYPE3_GET_CLASS(ct3d);
+    uint32_t offset, length;
+
+    get_lsa = (void *)cmd->payload;
+    offset = get_lsa->offset;
+    length = get_lsa->length;
+
+    if (offset + length > cvc->get_lsa_size(ct3d)) {
+        *len = 0;
+        return CXL_MBOX_INVALID_INPUT;
+    }
+
+    *len = cvc->get_lsa(ct3d, get_lsa, length, offset);
+    return CXL_MBOX_SUCCESS;
+}
+
+static ret_code cmd_ccls_set_lsa(struct cxl_cmd *cmd,
+                                 CXLDeviceState *cxl_dstate,
+                                 uint16_t *len)
+{
+    struct set_lsa_pl {
+        uint32_t offset;
+        uint32_t rsvd;
+        uint8_t data[];
+    } QEMU_PACKED;
+    struct set_lsa_pl *set_lsa_payload = (void *)cmd->payload;
+    CXLType3Dev *ct3d = container_of(cxl_dstate, CXLType3Dev, cxl_dstate);
+    CXLType3Class *cvc = CXL_TYPE3_GET_CLASS(ct3d);
+    const size_t hdr_len = offsetof(struct set_lsa_pl, data);
+    uint16_t plen = *len;
+
+    *len = 0;
+    if (!plen) {
+        return CXL_MBOX_SUCCESS;
+    }
+
+    if (set_lsa_payload->offset + plen > cvc->get_lsa_size(ct3d) + hdr_len) {
+        return CXL_MBOX_INVALID_INPUT;
+    }
+    plen -= hdr_len;
+
+    cvc->set_lsa(ct3d, set_lsa_payload->data, plen, set_lsa_payload->offset);
+    return CXL_MBOX_SUCCESS;
+}
+
 #define IMMEDIATE_CONFIG_CHANGE (1 << 1)
+#define IMMEDIATE_DATA_CHANGE (1 << 2)
 #define IMMEDIATE_POLICY_CHANGE (1 << 3)
 #define IMMEDIATE_LOG_CHANGE (1 << 4)
 
@@ -349,6 +406,9 @@ static struct cxl_cmd cxl_cmd_set[256][256] = {
         cmd_identify_memory_device, 0, 0 },
     [CCLS][GET_PARTITION_INFO] = { "CCLS_GET_PARTITION_INFO",
         cmd_ccls_get_partition_info, 0, 0 },
+    [CCLS][GET_LSA] = { "CCLS_GET_LSA", cmd_ccls_get_lsa, 0, 0 },
+    [CCLS][SET_LSA] = { "CCLS_SET_LSA", cmd_ccls_set_lsa,
+        ~0, IMMEDIATE_CONFIG_CHANGE | IMMEDIATE_DATA_CHANGE },
 };
 
 void cxl_process_mailbox(CXLDeviceState *cxl_dstate)
hw/mem/cxl_type3.c +55 -1

85 lines not on this page · raw message

include/hw/cxl/cxl_device.h +5 -0
index ea2571a69b..4285fbda08 100644
@@ -252,6 +252,11 @@ struct CXLType3Class {
 
     /* public */
     uint64_t (*get_lsa_size)(CXLType3Dev *ct3d);
+
+    uint64_t (*get_lsa)(CXLType3Dev *ct3d, void *buf, uint64_t size,
+                        uint64_t offset);
+    void (*set_lsa)(CXLType3Dev *ct3d, const void *buf, uint64_t size,
+                    uint64_t offset);
 };
 
 #endif
Signature · 1 line
2.32.0
#25 lore raw

PATCH v10 22/45 qtests/cxl: Add initial root port and CXL type3 tests

In reply to Jonathan Cameron

At this stage we can boot configurations with host bridges, root ports and type 3 memory devices, so add appropriate tests.

---
 tests/qtest/cxl-test.c | 127 +++++++++++++++++++++++++++++++++++++++++
 1 file changed, 127 insertions(+)
tests/qtest/cxl-test.c +127 -0
index c031049930..5f0794e816 100644
@@ -8,15 +8,142 @@
 #include "qemu/osdep.h"
 #include "libqtest-single.h"
 
+#define QEMU_PXB_CMD "-machine q35,cxl=on " \
+                     "-device pxb-cxl,id=cxl.0,bus=pcie.0,bus_nr=52 "
+
+#define QEMU_2PXB_CMD "-machine q35,cxl=on " \
+                      "-device pxb-cxl,id=cxl.0,bus=pcie.0,bus_nr=52 "  \
+                      "-device pxb-cxl,id=cxl.1,bus=pcie.0,bus_nr=53 "
+
+#define QEMU_RP "-device cxl-rp,id=rp0,bus=cxl.0,chassis=0,slot=0 "
+
+/* Dual ports on first pxb */
+#define QEMU_2RP "-device cxl-rp,id=rp0,bus=cxl.0,chassis=0,slot=0 " \
+                 "-device cxl-rp,id=rp1,bus=cxl.0,chassis=0,slot=1 "
+
+/* Dual ports on each of the pxb instances */
+#define QEMU_4RP "-device cxl-rp,id=rp0,bus=cxl.0,chassis=0,slot=0 " \
+                 "-device cxl-rp,id=rp1,bus=cxl.0,chassis=0,slot=1 " \
+                 "-device cxl-rp,id=rp2,bus=cxl.1,chassis=0,slot=2 " \
+                 "-device cxl-rp,id=rp3,bus=cxl.1,chassis=0,slot=3 "
+
+#define QEMU_T3D "-object memory-backend-file,id=cxl-mem0,mem-path=%s,size=256M " \
+                 "-object memory-backend-file,id=lsa0,mem-path=%s,size=256M "    \
+                 "-device cxl-type3,bus=rp0,memdev=cxl-mem0,lsa=lsa0,id=cxl-pmem0 "
+
+#define QEMU_2T3D "-object memory-backend-file,id=cxl-mem0,mem-path=%s,size=256M "    \
+                  "-object memory-backend-file,id=lsa0,mem-path=%s,size=256M "    \
+                  "-device cxl-type3,bus=rp0,memdev=cxl-mem0,lsa=lsa0,id=cxl-pmem0 " \
+                  "-object memory-backend-file,id=cxl-mem1,mem-path=%s,size=256M "    \
+                  "-object memory-backend-file,id=lsa1,mem-path=%s,size=256M "    \
+                  "-device cxl-type3,bus=rp1,memdev=cxl-mem1,lsa=lsa1,id=cxl-pmem1 "
+
+#define QEMU_4T3D "-object memory-backend-file,id=cxl-mem0,mem-path=%s,size=256M " \
+                  "-object memory-backend-file,id=lsa0,mem-path=%s,size=256M "    \
+                  "-device cxl-type3,bus=rp0,memdev=cxl-mem0,lsa=lsa0,id=cxl-pmem0 " \
+                  "-object memory-backend-file,id=cxl-mem1,mem-path=%s,size=256M "    \
+                  "-object memory-backend-file,id=lsa1,mem-path=%s,size=256M "    \
+                  "-device cxl-type3,bus=rp1,memdev=cxl-mem1,lsa=lsa1,id=cxl-pmem1 " \
+                  "-object memory-backend-file,id=cxl-mem2,mem-path=%s,size=256M "    \
+                  "-object memory-backend-file,id=lsa2,mem-path=%s,size=256M "    \
+                  "-device cxl-type3,bus=rp2,memdev=cxl-mem2,lsa=lsa2,id=cxl-pmem2 " \
+                  "-object memory-backend-file,id=cxl-mem3,mem-path=%s,size=256M "    \
+                  "-object memory-backend-file,id=lsa3,mem-path=%s,size=256M "    \
+                  "-device cxl-type3,bus=rp3,memdev=cxl-mem3,lsa=lsa3,id=cxl-pmem3 "
+
+static void cxl_basic_hb(void)
+{
+    qtest_start("-machine q35,cxl=on");
+    qtest_end();
+}
+
 static void cxl_basic_pxb(void)
 {
     qtest_start("-machine q35,cxl=on -device pxb-cxl,bus=pcie.0");
     qtest_end();
 }
 
+static void cxl_pxb_with_window(void)
+{
+    qtest_start(QEMU_PXB_CMD);
+    qtest_end();
+}
+
+static void cxl_2pxb_with_window(void)
+{
+    qtest_start(QEMU_2PXB_CMD);
+    qtest_end();
+}
+
+static void cxl_root_port(void)
+{
+    qtest_start(QEMU_PXB_CMD QEMU_RP);
+    qtest_end();
+}
+
+static void cxl_2root_port(void)
+{
+    qtest_start(QEMU_PXB_CMD QEMU_2RP);
+    qtest_end();
+}
+
+static void cxl_t3d(void)
+{
+    g_autoptr(GString) cmdline = g_string_new(NULL);
+    char template[] = "/tmp/cxl-test-XXXXXX";
+    const char *tmpfs;
+
+    tmpfs = mkdtemp(template);
+
+    g_string_printf(cmdline, QEMU_PXB_CMD QEMU_RP QEMU_T3D, tmpfs, tmpfs);
+
+    qtest_start(cmdline->str);
+    qtest_end();
+}
+
+static void cxl_1pxb_2rp_2t3d(void)
+{
+    g_autoptr(GString) cmdline = g_string_new(NULL);
+    char template[] = "/tmp/cxl-test-XXXXXX";
+    const char *tmpfs;
+
+    tmpfs = mkdtemp(template);
+
+    g_string_printf(cmdline, QEMU_PXB_CMD QEMU_2RP QEMU_2T3D,
+                    tmpfs, tmpfs, tmpfs, tmpfs);
+
+    qtest_start(cmdline->str);
+    qtest_end();
+}
+
+static void cxl_2pxb_4rp_4t3d(void)
+{
+    g_autoptr(GString) cmdline = g_string_new(NULL);
+    char template[] = "/tmp/cxl-test-XXXXXX";
+    const char *tmpfs;
+
+    tmpfs = mkdtemp(template);
+
+    g_string_printf(cmdline, QEMU_2PXB_CMD QEMU_4RP QEMU_4T3D,
+                    tmpfs, tmpfs, tmpfs, tmpfs, tmpfs, tmpfs,
+                    tmpfs, tmpfs);
+
+    qtest_start(cmdline->str);
+    qtest_end();
+}
+
 int main(int argc, char **argv)
 {
     g_test_init(&argc, &argv, NULL);
+
+    qtest_add_func("/pci/cxl/basic_hostbridge", cxl_basic_hb);
     qtest_add_func("/pci/cxl/basic_pxb", cxl_basic_pxb);
+    qtest_add_func("/pci/cxl/pxb_with_window", cxl_pxb_with_window);
+    qtest_add_func("/pci/cxl/pxb_x2_with_window", cxl_2pxb_with_window);
+    qtest_add_func("/pci/cxl/rp", cxl_root_port);
+    qtest_add_func("/pci/cxl/rp_x2", cxl_2root_port);
+    qtest_add_func("/pci/cxl/type3_device", cxl_t3d);
+    qtest_add_func("/pci/cxl/rp_x2_type3_x2", cxl_1pxb_2rp_2t3d);
+    qtest_add_func("/pci/cxl/pxb_x2_root_port_x4_type3_x4", cxl_2pxb_4rp_4t3d);
     return g_test_run();
 }
Signature · 1 line
2.32.0
#26 lore raw

PATCH v10 23/45 hw/cxl/component: Implement host bridge MMIO (8.2.5, table 142)

In reply to Jonathan Cameron

From: Ben Widawsky <[email protected]>

CXL host bridges themselves may have MMIO. Since host bridges don't have a BAR they are treated as special for MMIO. This patch includes i386/pc support.
Also hook up the device reset now that we have have the MMIO space in which the results are visible.

Note that we duplicate the PCI express case for the aml_build but the implementations will diverge when the CXL specific _OSC is introduced.

---
 hw/i386/acpi-build.c                | 25 ++++++++++-
 hw/i386/pc.c                        | 27 +++++++++++-
 hw/pci-bridge/pci_expander_bridge.c | 66 ++++++++++++++++++++++++++---
 include/hw/cxl/cxl.h                | 14 ++++++
 4 files changed, 122 insertions(+), 10 deletions(-)
hw/i386/acpi-build.c +23 -2
index dcf6ece3d0..2d81b0f40c 100644
@@ -28,6 +28,7 @@
 #include "qemu/bitmap.h"
 #include "qemu/error-report.h"
 #include "hw/pci/pci.h"
+#include "hw/cxl/cxl.h"
 #include "hw/core/cpu.h"
 #include "target/i386/cpu.h"
 #include "hw/misc/pvpanic.h"
@@ -1572,10 +1573,21 @@ build_dsdt(GArray *table_data, BIOSLinker *linker,
             }
 
             scope = aml_scope("\\_SB");
-            dev = aml_device("PC%.02X", bus_num);
+
+            if (pci_bus_is_cxl(bus)) {
+                dev = aml_device("CL%.02X", bus_num);
+            } else {
+                dev = aml_device("PC%.02X", bus_num);
+            }
             aml_append(dev, aml_name_decl("_UID", aml_int(bus_num)));
             aml_append(dev, aml_name_decl("_BBN", aml_int(bus_num)));
-            if (pci_bus_is_express(bus)) {
+            if (pci_bus_is_cxl(bus)) {
+                aml_append(dev, aml_name_decl("_HID", aml_eisaid("PNP0A08")));
+                aml_append(dev, aml_name_decl("_CID", aml_eisaid("PNP0A03")));
+
+                /* Expander bridges do not have ACPI PCI Hot-plug enabled */
+                aml_append(dev, build_q35_osc_method(true));
+            } else if (pci_bus_is_express(bus)) {
                 aml_append(dev, aml_name_decl("_HID", aml_eisaid("PNP0A08")));
                 aml_append(dev, aml_name_decl("_CID", aml_eisaid("PNP0A03")));
 
@@ -1595,6 +1607,15 @@ build_dsdt(GArray *table_data, BIOSLinker *linker,
             aml_append(dev, aml_name_decl("_CRS", crs));
             aml_append(scope, dev);
             aml_append(dsdt, scope);
+
+            /* Handle the ranges for the PXB expanders */
+            if (pci_bus_is_cxl(bus)) {
+                MemoryRegion *mr = &machine->cxl_devices_state->host_mr;
+                uint64_t base = mr->addr;
+
+                crs_range_insert(crs_range_set.mem_ranges, base,
+                                 base + memory_region_size(mr) - 1);
+            }
         }
     }
 
hw/i386/pc.c +26 -1
index b752339beb..b56af1e4e7 100644
@@ -75,6 +75,7 @@
 #include "acpi-build.h"
 #include "hw/mem/pc-dimm.h"
 #include "hw/mem/nvdimm.h"
+#include "hw/cxl/cxl.h"
 #include "qapi/error.h"
 #include "qapi/qapi-visit-common.h"
 #include "qapi/qapi-visit-machine.h"
@@ -816,6 +817,7 @@ void pc_memory_init(PCMachineState *pcms,
     MachineClass *mc = MACHINE_GET_CLASS(machine);
     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
     X86MachineState *x86ms = X86_MACHINE(pcms);
+    hwaddr cxl_base;
 
     assert(machine->ram_size == x86ms->below_4g_mem_size +
                                 x86ms->above_4g_mem_size);
@@ -905,6 +907,26 @@ void pc_memory_init(PCMachineState *pcms,
                                     &machine->device_memory->mr);
     }
 
+    if (machine->cxl_devices_state->is_enabled) {
+        MemoryRegion *mr = &machine->cxl_devices_state->host_mr;
+        hwaddr cxl_size = MiB;
+
+        if (pcmc->has_reserved_memory && machine->device_memory->base) {
+            cxl_base = machine->device_memory->base;
+            if (!pcmc->broken_reserved_end) {
+                cxl_base += memory_region_size(&machine->device_memory->mr);
+            }
+        } else if (pcms->sgx_epc.size != 0) {
+            cxl_base = sgx_epc_above_4g_end(&pcms->sgx_epc);
+        } else {
+            cxl_base = 0x100000000ULL + x86ms->above_4g_mem_size;
+        }
+
+        e820_add_entry(cxl_base, cxl_size, E820_RESERVED);
+        memory_region_init(mr, OBJECT(machine), "cxl_host_reg", cxl_size);
+        memory_region_add_subregion(system_memory, cxl_base, mr);
+    }
+
     /* Initialize PC system firmware */
     pc_system_firmware_init(pcms, rom_memory);
 
@@ -965,7 +987,10 @@ uint64_t pc_pci_hole64_start(void)
     X86MachineState *x86ms = X86_MACHINE(pcms);
     uint64_t hole64_start = 0;
 
-    if (pcmc->has_reserved_memory && ms->device_memory->base) {
+    if (ms->cxl_devices_state->host_mr.addr) {
+        hole64_start = ms->cxl_devices_state->host_mr.addr +
+            memory_region_size(&ms->cxl_devices_state->host_mr);
+    } else if (pcmc->has_reserved_memory && ms->device_memory->base) {
         hole64_start = ms->device_memory->base;
         if (!pcmc->broken_reserved_end) {
             hole64_start += memory_region_size(&ms->device_memory->mr);
hw/pci-bridge/pci_expander_bridge.c +59 -7

107 lines not on this page · raw message

include/hw/cxl/cxl.h +14 -0
index 31af92fd5e..8d1a7245d0 100644
@@ -10,6 +10,7 @@
 #ifndef CXL_H
 #define CXL_H
 
+#include "hw/pci/pci_host.h"
 #include "cxl_pci.h"
 #include "cxl_component.h"
 #include "cxl_device.h"
@@ -17,8 +18,21 @@
 #define CXL_COMPONENT_REG_BAR_IDX 0
 #define CXL_DEVICE_REG_BAR_IDX 2
 
+#define CXL_WINDOW_MAX 10
+
 typedef struct CXLState {
     bool is_enabled;
+    MemoryRegion host_mr;
+    unsigned int next_mr_idx;
 } CXLState;
 
+struct CXLHost {
+    PCIHostState parent_obj;
+
+    CXLComponentState cxl_cstate;
+};
+
+#define TYPE_PXB_CXL_HOST "pxb-cxl-host"
+OBJECT_DECLARE_SIMPLE_TYPE(CXLHost, PXB_CXL_HOST)
+
 #endif
Signature · 1 line
2.32.0
#27 lore raw

PATCH v10 24/45 acpi/cxl: Add _OSC implementation (9.14.2)

In reply to Jonathan Cameron

From: Ben Widawsky <[email protected]>

CXL 2.0 specification adds 2 new dwords to the existing _OSC definition from PCIe. The new dwords are accessed with a new uuid. This implementation supports what is in the specification.

iasl -d decodes the result of this patch as:

Name (SUPP, Zero)
Name (CTRL, Zero)
Name (SUPC, Zero)
Name (CTRC, Zero)
Method (_OSC, 4, NotSerialized) // _OSC: Operating System Capabilities

{
    CreateDWordField (Arg3, Zero, CDW1)
    If (((Arg0 == ToUUID ("33db4d5b-1ff7-401c-9657-7441c03dd766") /* PCI Host Bridge Device */) || (Arg0 == ToUUID ("68f2d50b-c469-4d8a-bd3d-941a103fd3fc") /* Unknown UUID */)))
    {
        CreateDWordField (Arg3, 0x04, CDW2)
        CreateDWordField (Arg3, 0x08, CDW3)
        Local0 = CDW3 /* \_SB_.PC0C._OSC.CDW3 */
        Local0 &= 0x1F
        If ((Arg1 != One))
        {
            CDW1 |= 0x08
        }

        If ((CDW3 != Local0))
        {
            CDW1 |= 0x10
        }

        SUPP = CDW2 /* \_SB_.PC0C._OSC.CDW2 */
        CTRL = CDW3 /* \_SB_.PC0C._OSC.CDW3 */
        CDW3 = Local0
        If ((Arg0 == ToUUID ("68f2d50b-c469-4d8a-bd3d-941a103fd3fc") /* Unknown UUID */))
        {
            CreateDWordField (Arg3, 0x0C, CDW4)
            CreateDWordField (Arg3, 0x10, CDW5)
            SUPC = CDW4 /* \_SB_.PC0C._OSC.CDW4 */
            CTRC = CDW5 /* \_SB_.PC0C._OSC.CDW5 */
            CDW5 |= One
        }

        Return (Arg3)
    }
    Else
    {
        CDW1 |= 0x04
        Return (Arg3)
    }
---
 hw/acpi/Kconfig       |   5 ++
 hw/acpi/cxl-stub.c    |  12 ++++
 hw/acpi/cxl.c         | 130 ++++++++++++++++++++++++++++++++++++++++++
 hw/acpi/meson.build   |   4 +-
 hw/i386/acpi-build.c  |  15 +++--
 include/hw/acpi/cxl.h |  23 ++++++++
 6 files changed, 183 insertions(+), 6 deletions(-)
hw/acpi/Kconfig +5 -0
index 19caebde6c..3703aca212 100644
@@ -5,6 +5,7 @@ config ACPI_X86
     bool
     select ACPI
     select ACPI_NVDIMM
+    select ACPI_CXL
     select ACPI_CPU_HOTPLUG
     select ACPI_MEMORY_HOTPLUG
     select ACPI_HMAT
@@ -66,3 +67,7 @@ config ACPI_ERST
     bool
     default y
     depends on ACPI && PCI
+
+config ACPI_CXL
+    bool
+    depends on ACPI
hw/acpi/cxl-stub.c new file +12 -0
new file mode 100644
index 0000000000..15bc21076b
@@ -0,0 +1,12 @@
+
+/*
+ * Stubs for ACPI platforms that don't support CXl
+ */
+#include "qemu/osdep.h"
+#include "hw/acpi/aml-build.h"
+#include "hw/acpi/cxl.h"
+
+void build_cxl_osc_method(Aml *dev)
+{
+    g_assert_not_reached();
+}
hw/acpi/cxl.c new file +130 -0

130 lines not on this page · raw message

hw/acpi/meson.build +3 -1
index 8bea2e6933..cea2f5f93a 100644
@@ -13,6 +13,7 @@ acpi_ss.add(when: 'CONFIG_ACPI_MEMORY_HOTPLUG', if_false: files('acpi-mem-hotplu
 acpi_ss.add(when: 'CONFIG_ACPI_NVDIMM', if_true: files('nvdimm.c'))
 acpi_ss.add(when: 'CONFIG_ACPI_NVDIMM', if_false: files('acpi-nvdimm-stub.c'))
 acpi_ss.add(when: 'CONFIG_ACPI_PCI', if_true: files('pci.c'))
+acpi_ss.add(when: 'CONFIG_ACPI_CXL', if_true: files('cxl.c'), if_false: files('cxl-stub.c'))
 acpi_ss.add(when: 'CONFIG_ACPI_VMGENID', if_true: files('vmgenid.c'))
 acpi_ss.add(when: 'CONFIG_ACPI_HW_REDUCED', if_true: files('generic_event_device.c'))
 acpi_ss.add(when: 'CONFIG_ACPI_HMAT', if_true: files('hmat.c'))
@@ -33,4 +34,5 @@ softmmu_ss.add_all(when: 'CONFIG_ACPI', if_true: acpi_ss)
 softmmu_ss.add(when: 'CONFIG_ALL', if_true: files('acpi-stub.c', 'aml-build-stub.c',
                                                   'acpi-x86-stub.c', 'ipmi-stub.c', 'ghes-stub.c',
                                                   'acpi-mem-hotplug-stub.c', 'acpi-cpu-hotplug-stub.c',
-                                                  'acpi-pci-hotplug-stub.c', 'acpi-nvdimm-stub.c'))
+                                                  'acpi-pci-hotplug-stub.c', 'acpi-nvdimm-stub.c',
+                                                  'cxl-stub.c'))
hw/i386/acpi-build.c +10 -5
index 2d81b0f40c..59ede8b2e9 100644
@@ -67,6 +67,7 @@
 #include "hw/acpi/aml-build.h"
 #include "hw/acpi/utils.h"
 #include "hw/acpi/pci.h"
+#include "hw/acpi/cxl.h"
 
 #include "qom/qom-qobject.h"
 #include "hw/i386/amd_iommu.h"
@@ -1582,11 +1583,15 @@ build_dsdt(GArray *table_data, BIOSLinker *linker,
             aml_append(dev, aml_name_decl("_UID", aml_int(bus_num)));
             aml_append(dev, aml_name_decl("_BBN", aml_int(bus_num)));
             if (pci_bus_is_cxl(bus)) {
-                aml_append(dev, aml_name_decl("_HID", aml_eisaid("PNP0A08")));
-                aml_append(dev, aml_name_decl("_CID", aml_eisaid("PNP0A03")));
-
-                /* Expander bridges do not have ACPI PCI Hot-plug enabled */
-                aml_append(dev, build_q35_osc_method(true));
+                struct Aml *pkg = aml_package(2);
+
+                aml_append(dev, aml_name_decl("_HID", aml_string("ACPI0016")));
+                aml_append(pkg, aml_eisaid("PNP0A08"));
+                aml_append(pkg, aml_eisaid("PNP0A03"));
+                aml_append(dev, aml_name_decl("_CID", pkg));
+                aml_append(dev, aml_name_decl("_ADR", aml_int(0)));
+                aml_append(dev, aml_name_decl("_UID", aml_int(bus_num)));
+                build_cxl_osc_method(dev);
             } else if (pci_bus_is_express(bus)) {
                 aml_append(dev, aml_name_decl("_HID", aml_eisaid("PNP0A08")));
                 aml_append(dev, aml_name_decl("_CID", aml_eisaid("PNP0A03")));
include/hw/acpi/cxl.h new file +23 -0
new file mode 100644
index 0000000000..7b8f3b8a2e
@@ -0,0 +1,23 @@
+/*
+ * Copyright (C) 2020 Intel Corporation
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+
+ * You should have received a copy of the GNU General Public License along
+ * with this program; if not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef HW_ACPI_CXL_H
+#define HW_ACPI_CXL_H
+
+void build_cxl_osc_method(Aml *dev);
+
+#endif
Signature · 1 line
2.32.0
#28 lore raw

PATCH v10 25/45 acpi/cxl: Create the CEDT (9.14.1)

In reply to Jonathan Cameron

From: Ben Widawsky <[email protected]>

The CXL Early Discovery Table is defined in the CXL 2.0 specification as a way for the OS to get CXL specific information from the system firmware.

CXL 2.0 specification adds an _HID, ACPI0016, for CXL capable host bridges, with a _CID of PNP0A08 (PCIe host bridge). CXL aware software is able to use this initiate the proper _OSC method, and get the _UID which is referenced by the CEDT. Therefore the existence of an ACPI0016 device allows a CXL aware driver perform the necessary actions. For a CXL capable OS, this works. For a CXL unaware OS, this works.

CEDT awaremess requires more. The motivation for ACPI0017 is to provide the possibility of having a Linux CXL module that can work on a legacy Linux kernel. Linux core PCI/ACPI which won't be built as a module, will see the _CID of PNP0A08 and bind a driver to it. If we later loaded a driver for ACPI0016, Linux won't be able to bind it to the hardware because it has already bound the PNP0A08 driver. The ACPI0017 device is an opportunity to have an object to bind a driver will be used by a Linux driver to walk the CXL topology and do everything that we would have preferred to do with ACPI0016.

There is another motivation for an ACPI0017 device which isn't implemented here. An operating system needs an attach point for a non-volatile region provider that understands cross-hostbridge interleaving. Since QEMU emulation doesn't support interleaving yet, this is more important on the OS side, for now.

As of CXL 2.0 spec, only 1 sub structure is defined, the CXL Host Bridge Structure (CHBS) which is primarily useful for telling the OS exactly where the MMIO for the host bridge is.

---
 hw/acpi/cxl.c                       | 68 +++++++++++++++++++++++++++++
 hw/i386/acpi-build.c                | 27 ++++++++++++
 hw/pci-bridge/pci_expander_bridge.c | 17 --------
 include/hw/acpi/cxl.h               |  5 +++
 include/hw/pci/pci_bridge.h         | 20 +++++++++
 5 files changed, 120 insertions(+), 17 deletions(-)
hw/acpi/cxl.c +68 -0
index ca1f04f359..aa4af86a4c 100644
@@ -18,7 +18,11 @@
  */
 
 #include "qemu/osdep.h"
+#include "hw/sysbus.h"
+#include "hw/pci/pci_bridge.h"
+#include "hw/pci/pci_host.h"
 #include "hw/cxl/cxl.h"
+#include "hw/mem/memory-device.h"
 #include "hw/acpi/acpi.h"
 #include "hw/acpi/aml-build.h"
 #include "hw/acpi/bios-linker-loader.h"
@@ -26,6 +30,70 @@
 #include "qapi/error.h"
 #include "qemu/uuid.h"
 
+static void cedt_build_chbs(GArray *table_data, PXBDev *cxl)
+{
+    SysBusDevice *sbd = SYS_BUS_DEVICE(cxl->cxl.cxl_host_bridge);
+    struct MemoryRegion *mr = sbd->mmio[0].memory;
+
+    /* Type */
+    build_append_int_noprefix(table_data, 0, 1);
+
+    /* Reserved */
+    build_append_int_noprefix(table_data, 0, 1);
+
+    /* Record Length */
+    build_append_int_noprefix(table_data, 32, 2);
+
+    /* UID - currently equal to bus number */
+    build_append_int_noprefix(table_data, cxl->bus_nr, 4);
+
+    /* Version */
+    build_append_int_noprefix(table_data, 1, 4);
+
+    /* Reserved */
+    build_append_int_noprefix(table_data, 0, 4);
+
+    /* Base - subregion within a container that is in PA space */
+    build_append_int_noprefix(table_data, mr->container->addr + mr->addr, 8);
+
+    /* Length */
+    build_append_int_noprefix(table_data, memory_region_size(mr), 8);
+}
+
+static int cxl_foreach_pxb_hb(Object *obj, void *opaque)
+{
+    Aml *cedt = opaque;
+
+    if (object_dynamic_cast(obj, TYPE_PXB_CXL_DEVICE)) {
+        cedt_build_chbs(cedt->buf, PXB_CXL_DEV(obj));
+    }
+
+    return 0;
+}
+
+void cxl_build_cedt(MachineState *ms, GArray *table_offsets, GArray *table_data,
+                    BIOSLinker *linker, const char *oem_id,
+                    const char *oem_table_id)
+{
+    Aml *cedt;
+    AcpiTable table = { .sig = "CEDT", .rev = 1, .oem_id = oem_id,
+                        .oem_table_id = oem_table_id };
+
+    acpi_add_table(table_offsets, table_data);
+    acpi_table_begin(&table, table_data);
+    cedt = init_aml_allocator();
+
+    /* reserve space for CEDT header */
+
+    object_child_foreach_recursive(object_get_root(), cxl_foreach_pxb_hb, cedt);
+
+    /* copy AML table into ACPI tables blob and patch header there */
+    g_array_append_vals(table_data, cedt->buf->data, cedt->buf->len);
+    free_aml_allocator();
+
+    acpi_table_end(linker, &table);
+}
+
 static Aml *__build_cxl_osc_method(void)
 {
     Aml *method, *if_uuid, *else_uuid, *if_arg1_not_1, *if_cxl, *if_caps_masked;
hw/i386/acpi-build.c +27 -0
index 59ede8b2e9..c125939ed6 100644
@@ -77,6 +77,7 @@
 #include "hw/acpi/ipmi.h"
 #include "hw/acpi/hmat.h"
 #include "hw/acpi/viot.h"
+#include "hw/acpi/cxl.h"
 
 #include CONFIG_DEVICES
 
@@ -1411,6 +1412,22 @@ static void build_smb0(Aml *table, I2CBus *smbus, int devnr, int func)
     aml_append(table, scope);
 }
 
+static void build_acpi0017(Aml *table)
+{
+    Aml *dev, *scope, *method;
+
+    scope =  aml_scope("_SB");
+    dev = aml_device("CXLM");
+    aml_append(dev, aml_name_decl("_HID", aml_string("ACPI0017")));
+
+    method = aml_method("_STA", 0, AML_NOTSERIALIZED);
+    aml_append(method, aml_return(aml_int(0x01)));
+    aml_append(dev, method);
+
+    aml_append(scope, dev);
+    aml_append(table, scope);
+}
+
 static void
 build_dsdt(GArray *table_data, BIOSLinker *linker,
            AcpiPmInfo *pm, AcpiMiscInfo *misc,
@@ -1430,6 +1447,7 @@ build_dsdt(GArray *table_data, BIOSLinker *linker,
 #ifdef CONFIG_TPM
     TPMIf *tpm = tpm_find();
 #endif
+    bool cxl_present = false;
     int i;
     VMBusBridge *vmbus_bridge = vmbus_bridge_find();
     AcpiTable table = { .sig = "DSDT", .rev = 1, .oem_id = x86ms->oem_id,
@@ -1618,12 +1636,17 @@ build_dsdt(GArray *table_data, BIOSLinker *linker,
                 MemoryRegion *mr = &machine->cxl_devices_state->host_mr;
                 uint64_t base = mr->addr;
 
+                cxl_present = true;
                 crs_range_insert(crs_range_set.mem_ranges, base,
                                  base + memory_region_size(mr) - 1);
             }
         }
     }
 
+    if (cxl_present) {
+        build_acpi0017(dsdt);
+    }
+
     /*
      * At this point crs_range_set has all the ranges used by pci
      * busses *other* than PCI0.  These ranges will be excluded from
@@ -2688,6 +2711,10 @@ void acpi_build(AcpiBuildTables *tables, MachineState *machine)
                           machine->nvdimms_state, machine->ram_slots,
                           x86ms->oem_id, x86ms->oem_table_id);
     }
+    if (machine->cxl_devices_state->is_enabled) {
+        cxl_build_cedt(machine, table_offsets, tables_blob, tables->linker,
+                       x86ms->oem_id, x86ms->oem_table_id);
+    }
 
     acpi_add_table(table_offsets, tables_blob);
     build_waet(tables_blob, tables->linker, x86ms->oem_id, x86ms->oem_table_id);
hw/pci-bridge/pci_expander_bridge.c +0 -17
index 8fb4f2ea91..22feda1ff0 100644
@@ -57,23 +57,6 @@ DECLARE_INSTANCE_CHECKER(PXBDev, PXB_DEV,
 DECLARE_INSTANCE_CHECKER(PXBDev, PXB_PCIE_DEV,
                          TYPE_PXB_PCIE_DEVICE)
 
-#define TYPE_PXB_CXL_DEVICE "pxb-cxl"
-DECLARE_INSTANCE_CHECKER(PXBDev, PXB_CXL_DEV,
-                         TYPE_PXB_CXL_DEVICE)
-
-struct PXBDev {
-    /*< private >*/
-    PCIDevice parent_obj;
-    /*< public >*/
-
-    uint8_t bus_nr;
-    uint16_t numa_node;
-    bool bypass_iommu;
-    struct cxl_dev {
-        CXLHost *cxl_host_bridge;
-    } cxl;
-};
-
 static PXBDev *convert_to_pxb(PCIDevice *dev)
 {
     /* A CXL PXB's parent bus is PCIe, so the normal check won't work */
include/hw/acpi/cxl.h +5 -0
index 7b8f3b8a2e..0c496538c0 100644
@@ -18,6 +18,11 @@
 #ifndef HW_ACPI_CXL_H
 #define HW_ACPI_CXL_H
 
+#include "hw/acpi/bios-linker-loader.h"
+
+void cxl_build_cedt(MachineState *ms, GArray *table_offsets, GArray *table_data,
+                    BIOSLinker *linker, const char *oem_id,
+                    const char *oem_table_id);
 void build_cxl_osc_method(Aml *dev);
 
 #endif
include/hw/pci/pci_bridge.h +20 -0
index 30691a6e57..ba4bafac7c 100644
@@ -28,6 +28,7 @@
 
 #include "hw/pci/pci.h"
 #include "hw/pci/pci_bus.h"
+#include "hw/cxl/cxl.h"
 #include "qom/object.h"
 
 typedef struct PCIBridgeWindows PCIBridgeWindows;
@@ -80,6 +81,25 @@ struct PCIBridge {
 #define PCI_BRIDGE_DEV_PROP_CHASSIS_NR "chassis_nr"
 #define PCI_BRIDGE_DEV_PROP_MSI        "msi"
 #define PCI_BRIDGE_DEV_PROP_SHPC       "shpc"
+typedef struct CXLHost CXLHost;
+
+struct PXBDev {
+    /*< private >*/
+    PCIDevice parent_obj;
+    /*< public >*/
+
+    uint8_t bus_nr;
+    uint16_t numa_node;
+    bool bypass_iommu;
+    struct cxl_dev {
+        CXLHost *cxl_host_bridge; /* Pointer to a CXLHost */
+    } cxl;
+};
+
+typedef struct PXBDev PXBDev;
+#define TYPE_PXB_CXL_DEVICE "pxb-cxl"
+DECLARE_INSTANCE_CHECKER(PXBDev, PXB_CXL_DEV,
+                         TYPE_PXB_CXL_DEVICE)
 
 int pci_bridge_ssvid_init(PCIDevice *dev, uint8_t offset,
                           uint16_t svid, uint16_t ssid,
Signature · 1 line
2.32.0
#29 lore raw

PATCH v10 26/45 hw/cxl/component: Add utils for interleave parameter encoding/decoding

In reply to Jonathan Cameron

From: Jonathan Cameron <[email protected]>

Both registers and the CFMWS entries in CDAT use simple encodings for the number of interleave ways and the interleave granularity. Introduce simple conversion functions to/from the unencoded number / size. So far the iw decode has not been needed so is it not implemented.

---
 hw/cxl/cxl-component-utils.c   | 34 ++++++++++++++++++++++++++++++++++
 include/hw/cxl/cxl_component.h |  8 ++++++++
 2 files changed, 42 insertions(+)
hw/cxl/cxl-component-utils.c +34 -0
index afc97b17c2..69cb07171c 100644
@@ -9,6 +9,7 @@
 
 #include "qemu/osdep.h"
 #include "qemu/log.h"
+#include "qapi/error.h"
 #include "hw/pci/pci.h"
 #include "hw/cxl/cxl.h"
 
@@ -329,3 +330,36 @@ void cxl_component_create_dvsec(CXLComponentState *cxl,
     range_init_nofail(&cxl->dvsecs[type], cxl->dvsec_offset, length);
     cxl->dvsec_offset += length;
 }
+
+uint8_t cxl_interleave_ways_enc(int iw, Error **errp)
+{
+    switch (iw) {
+    case 1: return 0x0;
+    case 2: return 0x1;
+    case 4: return 0x2;
+    case 8: return 0x3;
+    case 16: return 0x4;
+    case 3: return 0x8;
+    case 6: return 0x9;
+    case 12: return 0xa;
+    default:
+        error_setg(errp, "Interleave ways: %d not supported", iw);
+        return 0;
+    }
+}
+
+uint8_t cxl_interleave_granularity_enc(uint64_t gran, Error **errp)
+{
+    switch (gran) {
+    case 256: return 0;
+    case 512: return 1;
+    case 1024: return 2;
+    case 2048: return 3;
+    case 4096: return 4;
+    case 8192: return 5;
+    case 16384: return 6;
+    default:
+        error_setg(errp, "Interleave granularity: %" PRIu64 " invalid", gran);
+        return 0;
+    }
+}
include/hw/cxl/cxl_component.h +8 -0
index 7d8f395cbe..4f69688c47 100644
@@ -210,4 +210,12 @@ static inline int cxl_decoder_count_enc(int count)
     return 0;
 }
 
+uint8_t cxl_interleave_ways_enc(int iw, Error **errp);
+uint8_t cxl_interleave_granularity_enc(uint64_t gran, Error **errp);
+
+static inline hwaddr cxl_decode_ig(int ig)
+{
+    return 1 << (ig + 8);
+}
+
 #endif
Signature · 1 line
2.32.0
#30 lore raw

PATCH v10 27/45 hw/cxl/host: Add support for CXL Fixed Memory Windows.

In reply to Jonathan Cameron

From: Jonathan Cameron <[email protected]>

The concept of these is introduced in [1] in terms of the description the CEDT ACPI table. The principal is more general. Unlike once traffic hits the CXL root bridges, the host system memory address routing is implementation defined and effectively static once observable by standard / generic system software. Each CXL Fixed Memory Windows (CFMW) is a region of PA space which has fixed system dependent routing configured so that accesses can be routed to the CXL devices below a set of target root bridges. The accesses may be interleaved across multiple root bridges.

For QEMU we could have fully specified these regions in terms of a base PA + size, but as the absolute address does not matter it is simpler to let individual platforms place the memory regions.

ExampleS:
-cxl-fixed-memory-window targets.0=cxl.0,size=128G
-cxl-fixed-memory-window targets.0=cxl.1,size=128G
-cxl-fixed-memory-window targets.0=cxl0,targets.1=cxl.1,size=256G,interleave-granularity=2k

Specifies
* 2x 128G regions not interleaved across root bridges, one for each of the root bridges with ids cxl.0 and cxl.1
* 256G region interleaved across root bridges with ids cxl.0 and cxl.1 with a 2k interleave granularity.

When system software enumerates the devices below a given root bridge it can then decide which CFMW to use. If non interleave is desired (or possible) it can use the appropriate CFMW for the root bridge in question. If there are suitable devices to interleave across the two root bridges then it may use the 3rd CFMS.

A number of other designs were considered but the following constraints made it hard to adapt existing QEMU approaches to this particular problem.
1) The size must be known before a specific architecture / board brings up it's PA memory map. We need to set up an appropriate region.
2) Using links to the host bridges provides a clean command line interface but these links cannot be established until command line devices have been added.

Hence the two step process used here of first establishing the size, interleave-ways and granularity + caching the ids of the host bridges and then, once available finding the actual host bridges so they can be used later to support interleave decoding.

[1] CXL 2.0 ECN: CEDT CFMWS & QTG DSM (computeexpresslink.org / specifications)

---
 hw/cxl/cxl-host-stubs.c | 14 ++++++
 hw/cxl/cxl-host.c       | 94 +++++++++++++++++++++++++++++++++++++++++
 hw/cxl/meson.build      |  6 +++
 include/hw/cxl/cxl.h    | 21 +++++++++
 qapi/machine.json       | 21 +++++++++
 qemu-options.hx         | 38 +++++++++++++++++
 softmmu/vl.c            | 47 +++++++++++++++++++++
 7 files changed, 241 insertions(+)
hw/cxl/cxl-host-stubs.c new file +14 -0
new file mode 100644
index 0000000000..f8fd278d5d
@@ -0,0 +1,14 @@
+/*
+ * CXL host parameter parsing routine stubs
+ *
+ * Copyright (c) 2022 Huawei
+ */
+#include "qemu/osdep.h"
+#include "qapi/error.h"
+#include "hw/cxl/cxl.h"
+
+void cxl_fixed_memory_window_config(MachineState *ms,
+                                    CXLFixedMemoryWindowOptions *object,
+                                    Error **errp) {};
+
+void cxl_fixed_memory_window_link_targets(Error **errp) {};
hw/cxl/cxl-host.c new file +94 -0

94 lines not on this page · raw message

hw/cxl/meson.build +6 -0
index e68eea2358..f117b99949 100644
@@ -3,4 +3,10 @@ softmmu_ss.add(when: 'CONFIG_CXL',
                    'cxl-component-utils.c',
                    'cxl-device-utils.c',
                    'cxl-mailbox-utils.c',
+                   'cxl-host.c',
+               ),
+               if_false: files(
+                   'cxl-host-stubs.c',
                ))
+
+softmmu_ss.add(when: 'CONFIG_ALL', if_true: files('cxl-host-stubs.c'))
include/hw/cxl/cxl.h +21 -0
index 8d1a7245d0..dce38124db 100644
@@ -10,6 +10,9 @@
 #ifndef CXL_H
 #define CXL_H
 
+
+#include "qapi/qapi-types-machine.h"
+#include "hw/pci/pci_bridge.h"
 #include "hw/pci/pci_host.h"
 #include "cxl_pci.h"
 #include "cxl_component.h"
@@ -20,10 +23,23 @@
 
 #define CXL_WINDOW_MAX 10
 
+typedef struct CXLFixedWindow {
+    uint64_t size;
+    char **targets;
+    struct PXBDev *target_hbs[8];
+    uint8_t num_targets;
+    uint8_t enc_int_ways;
+    uint8_t enc_int_gran;
+    /* Todo: XOR based interleaving */
+    MemoryRegion mr;
+    hwaddr base;
+} CXLFixedWindow;
+
 typedef struct CXLState {
     bool is_enabled;
     MemoryRegion host_mr;
     unsigned int next_mr_idx;
+    GList *fixed_windows;
 } CXLState;
 
 struct CXLHost {
@@ -35,4 +51,9 @@ struct CXLHost {
 #define TYPE_PXB_CXL_HOST "pxb-cxl-host"
 OBJECT_DECLARE_SIMPLE_TYPE(CXLHost, PXB_CXL_HOST)
 
+void cxl_fixed_memory_window_config(MachineState *ms,
+                                    CXLFixedMemoryWindowOptions *object,
+                                    Error **errp);
+void cxl_fixed_memory_window_link_targets(Error **errp);
+
 #endif
qapi/machine.json +21 -0
index d25a481ce4..b22c27f196 100644
@@ -502,6 +502,27 @@
    'dst': 'uint16',
    'val': 'uint8' }}
 
+##
+# @CXLFixedMemoryWindowOptions:
+#
+# Create a CXL Fixed Memory Window
+#
+# @size: Size of the Fixed Memory Window in bytes. Must be a multiple
+#        of 256MiB.
+# @interleave-granularity: Number of contiguous bytes for which
+#                          accesses will go to a given interleave target.
+#                          Accepted values [256, 512, 1k, 2k, 4k, 8k, 16k]
+# @targets: Target root bridge IDs from -device ...,id=<ID> for each root
+#           bridge.
+#
+# Since 7.1
+##
+{ 'struct': 'CXLFixedMemoryWindowOptions',
+  'data': {
+      'size': 'size',
+      '*interleave-granularity': 'size',
+      'targets': ['str'] }}
+
 ##
 # @X86CPURegister32:
 #
qemu-options.hx +38 -0
index 1764eebfaf..841896380e 100644
@@ -467,6 +467,44 @@ SRST
         -numa hmat-cache,node-id=1,size=10K,level=1,associativity=direct,policy=write-back,line=8
 ERST
 
+DEF("cxl-fixed-memory-window", HAS_ARG, QEMU_OPTION_cxl_fixed_memory_window,
+    "-cxl-fixed-memory-window targets.0=firsttarget,targets.1=secondtarget,size=size[,interleave-granularity=granularity]\n",
+    QEMU_ARCH_ALL)
+SRST
+``-cxl-fixed-memory-window targets.0=firsttarget,targets.1=secondtarget,size=size[,interleave-granularity=granularity]``
+    Define a CXL Fixed Memory Window (CFMW).
+
+    Described in the CXL 2.0 ECN: CEDT CFMWS & QTG _DSM.
+
+    They are regions of Host Physical Addresses (HPA) on a system which
+    may be interleaved across one or more CXL host bridges.  The system
+    software will assign particular devices into these windows and
+    configure the downstream Host-managed Device Memory (HDM) decoders
+    in root ports, switch ports and devices appropriately to meet the
+    interleave requirements before enabling the memory devices.
+
+    ``targets.X=firsttarget`` provides the mapping to CXL host bridges
+    which may be identified by the id provied in the -device entry.
+    Multiple entries are needed to specify all the targets when
+    the fixed memory window represents interleaved memory. X is the
+    target index from 0.
+
+    ``size=size`` sets the size of the CFMW. This must be a multiple of
+    256MiB. The region will be aligned to 256MiB but the location is
+    platform and configuration dependent.
+
+    ``interleave-granularity=granularity`` sets the granularity of
+    interleave. Default 256KiB. Only 256KiB, 512KiB, 1024KiB, 2048KiB
+    4096KiB, 8192KiB and 16384KiB granularities supported.
+
+    Example:
+
+    ::
+
+        -cxl-fixed-memory-window targets.0=cxl.0,targets.1=cxl.1,size=128G,interleave-granularity=512k
+
+ERST
+
 DEF("add-fd", HAS_ARG, QEMU_OPTION_add_fd,
     "-add-fd fd=fd,set=set[,opaque=opaque]\n"
     "                Add 'fd' to fd 'set'\n", QEMU_ARCH_ALL)
softmmu/vl.c +47 -0

101 lines not on this page · raw message

Signature · 1 line
2.32.0
#31 lore raw

Re: [PATCH v10 27/45] hw/cxl/host: Add support for CXL Fixed Memory Windows.

Quoted 18 lines · Jonathan Cameron via

From: Jonathan Cameron <[email protected]>

The concept of these is introduced in [1] in terms of the description the CEDT ACPI table. The principal is more general. Unlike once traffic hits the CXL root bridges, the host system memory address routing is implementation defined and effectively static once observable by standard / generic system software. Each CXL Fixed Memory Windows (CFMW) is a region of PA space which has fixed system dependent routing configured so that accesses can be routed to the CXL devices below a set of target root bridges. The accesses may be interleaved across multiple root bridges.

For QEMU we could have fully specified these regions in terms of a base PA + size, but as the absolute address does not matter it is simpler to let individual platforms place the memory regions.

ExampleS:

Review

Quoted 3 lines
(quoted diff) +0 -3
-cxl-fixed-memory-window targets.0=cxl.0,size=128G
-cxl-fixed-memory-window targets.0=cxl.1,size=128G
-cxl-fixed-memory-window targets.0=cxl0,targets.1=cxl.1,size=256G,interleave-granularity=2k

Sorry for the late review, but: no more command line options should be added to QEMU. This should be:

-M cxl-fmw.0.targets.0=cxl.0,cxl-fmw.0.size=128G \
-M cxl-fmw.1.targets.0=cxl.1,cxl-fmw.1.size=128G \
-M cxl-fmw.2.targets.0=cxl.0,cxl-fmw.2.targets.1=cxl.1,cxl-fmw.2.size=128G,cxl-fmw.2.interleave-granularity=2k

I'm not against adding a shortcut as above, but the implementation should be entirely in MachineState using a QOM property.

The reason for this is that we're looking into doing machine setup entirely via RPC, and any extra option is a new command to be implemented.

Would you be able to do the change? Since you are already using QAPI to deserialize the option it should not be hard. I can promise a quick review, and I can also provide with an example conversion for -boot at
https://lore.kernel.org/r/[email protected]/.

Paolo

#32 lore raw

Re: [PATCH v10 27/45] hw/cxl/host: Add support for CXL Fixed Memory Windows.

Quoted 29 lines · On Sat, 21 May 2022 11:07:13 +0200 Paolo Bonzini <[email protected]> wrote:
Quoted 18 lines · Jonathan Cameron via

From: Jonathan Cameron <[email protected]>

The concept of these is introduced in [1] in terms of the description the CEDT ACPI table. The principal is more general. Unlike once traffic hits the CXL root bridges, the host system memory address routing is implementation defined and effectively static once observable by standard / generic system software. Each CXL Fixed Memory Windows (CFMW) is a region of PA space which has fixed system dependent routing configured so that accesses can be routed to the CXL devices below a set of target root bridges. The accesses may be interleaved across multiple root bridges.

For QEMU we could have fully specified these regions in terms of a base PA + size, but as the absolute address does not matter it is simpler to let individual platforms place the memory regions.

ExampleS:

Review

Quoted 3 lines
(quoted diff) +0 -3
-cxl-fixed-memory-window targets.0=cxl.0,size=128G
-cxl-fixed-memory-window targets.0=cxl.1,size=128G
-cxl-fixed-memory-window targets.0=cxl0,targets.1=cxl.1,size=256G,interleave-granularity=2k  

Sorry for the late review, but: no more command line options should be added to QEMU. This should be:

Review

Quoted 3 lines
(quoted diff) +0 -3
-M cxl-fmw.0.targets.0=cxl.0,cxl-fmw.0.size=128G \
-M cxl-fmw.1.targets.0=cxl.1,cxl-fmw.1.size=128G \
-M cxl-fmw.2.targets.0=cxl.0,cxl-fmw.2.targets.1=cxl.1,cxl-fmw.2.size=128G,cxl-fmw.2.interleave-granularity=2k

Hi Paolo,

I thought I had this working but have run into an issue where the parser seems not to like multiple parts of cxl-fmw as below being in different -M entries. I get

qemu-system-aarch64: -M cxl-fmw.1.targets.0=cxl.1,cxl-fmw.1.size=4G,cxl-fmw.1.interleave-granularity=8k: Parameters 'cxl-fmw.*' used inconsistently

Seems it doesn't let me have multiple elements of a list in different -M entries.

Not a disaster but it's going to lead to some unreadable command lines.

Very lightly tested patch below. I've not yet looked at movingstuff to machine specific callbacks as per your other comment. I recall that wasn't easy to do with an earlier version of the series, but given a lot has changed since then it might now easier to accomplish.

Thanks,

Jonathan

Quoted 10 lines

I'm not against adding a shortcut as above, but the implementation should be entirely in MachineState using a QOM property.

The reason for this is that we're looking into doing machine setup entirely via RPC, and any extra option is a new command to be implemented.

Would you be able to do the change? Since you are already using QAPI to deserialize the option it should not be hard. I can promise a quick review, and I can also provide with an example conversion for -boot at
https://lore.kernel.org/r/[email protected]/.

From d856959a08513b098fda22472f0b870750910543 Mon Sep 17 00:00:00 2001
From: Jonathan Cameron <[email protected]>
Date: Mon, 23 May 2022 16:07:45 +0100
Subject: [PATCH] hw/cxl: Make the CXL fixed memory window setup a machine
parameter.

Paolo Bonzini requested this change to simplify the ongoing effort to allow machine setup entirely via RPC.

Includes shortening the command line form cxl-fixed-memory-window to cxl-fmw as the command lines are extremely long even with this change.

The json change is needed to ensure that there is a CXLFixedMemoryWindowOptionsList even though the actual element in the json is never used.

---
 hw/core/machine.c   | 33 +++++++++++++++++++++++++++++++
 include/hw/boards.h |  1 +
 qapi/machine.json   | 13 +++++++++++++
 softmmu/vl.c        | 47 ++++-----------------------------------------
 4 files changed, 51 insertions(+), 43 deletions(-)
hw/core/machine.c +33 -0
index bb0dc8f6a9..93b3b8d34a 100644
@@ -920,6 +920,33 @@ out_free:
     qapi_free_BootConfiguration(config);
 }
 
+static void machine_get_cfmw(Object *obj, Visitor *v, const char *name,
+                             void *opaque, Error **errp)
+{
+    MachineState *ms = MACHINE(obj);
+    CXLFixedMemoryWindowOptionsList *list = ms->cfmws_list;
+
+    visit_type_CXLFixedMemoryWindowOptionsList(v, name, &list, errp);
+}
+
+static void machine_set_cfmw(Object *obj, Visitor *v, const char *name,
+                             void *opaque, Error **errp)
+{
+    MachineState *ms = MACHINE(obj);
+    CXLFixedMemoryWindowOptionsList *cfmw_list = NULL;
+    CXLFixedMemoryWindowOptionsList *it;
+
+    visit_type_CXLFixedMemoryWindowOptionsList(v, name, &cfmw_list, errp);
+    if (!cfmw_list) {
+        return;
+    }
+
+    for (it = cfmw_list; it; it = it->next) {
+        cxl_fixed_memory_window_config(current_machine, it->value, errp);
+    }
+    ms->cfmws_list = cfmw_list;
+}
+
 static void machine_class_init(ObjectClass *oc, void *data)
 {
     MachineClass *mc = MACHINE_CLASS(oc);
@@ -966,6 +993,12 @@ static void machine_class_init(ObjectClass *oc, void *data)
     object_class_property_set_description(oc, "boot",
         "Boot configuration");
 
+    object_class_property_add(oc, "cxl-fmw", "CXLFixedMemoryWindow",
+        machine_get_cfmw, machine_set_cfmw,
+        NULL, NULL);
+    object_class_property_set_description(oc, "cxl-fmw",
+        "CXL Fixed Memory Window");
+
     object_class_property_add(oc, "smp", "SMPConfiguration",
         machine_get_smp, machine_set_smp,
         NULL, NULL);
include/hw/boards.h +1 -0
index fa57bac4fb..dd9fc56df2 100644
@@ -362,6 +362,7 @@ struct MachineState {
     struct NVDIMMState *nvdimms_state;
     struct CXLState *cxl_devices_state;
     struct NumaState *numa_state;
+    CXLFixedMemoryWindowOptionsList *cfmws_list;
 };
 
 #define DEFINE_MACHINE(namestr, machine_initfn) \
qapi/machine.json +13 -0
index 883ce3f9ea..b17ac79494 100644
@@ -523,6 +523,19 @@
       '*interleave-granularity': 'size',
       'targets': ['str'] }}
 
+##
+# @CXLFMWProperties:
+#
+# List of CXL Fixed Memory Windows.
+#
+# @cxl-fmw: List of CXLFixedMemoryWindowOptions
+#
+# Since 7.1
+##
+{ 'struct' : 'CXLFMWProperties',
+  'data': { 'cxl-fmw': ['CXLFixedMemoryWindowOptions'] }
+}
+
 ##
 # @X86CPURegister32:
 #
softmmu/vl.c +4 -43

95 lines not on this page · raw message

Signature · 1 line
2.32.0
#33 lore raw

PATCH v10 28/45 acpi/cxl: Introduce CFMWS structures in CEDT

In reply to Jonathan Cameron

From: Ben Widawsky <[email protected]>

The CEDT CXL Fixed Window Memory Window Structures (CFMWs) define regions of the host phyiscal address map which (via an impdef means) are configured such that they have a particular interleave setup across one or more CXL Host Bridges.

---
 hw/acpi/cxl.c | 59 +++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 59 insertions(+)
hw/acpi/cxl.c +59 -0
index aa4af86a4c..31d5235136 100644
@@ -60,6 +60,64 @@ static void cedt_build_chbs(GArray *table_data, PXBDev *cxl)
     build_append_int_noprefix(table_data, memory_region_size(mr), 8);
 }
 
+/*
+ * CFMWS entries in CXL 2.0 ECN: CEDT CFMWS & QTG _DSM.
+ * Interleave ways encoding in CXL 2.0 ECN: 3, 6, 12 and 16-way memory
+ * interleaving.
+ */
+static void cedt_build_cfmws(GArray *table_data, MachineState *ms)
+{
+    CXLState *cxls = ms->cxl_devices_state;
+    GList *it;
+
+    for (it = cxls->fixed_windows; it; it = it->next) {
+        CXLFixedWindow *fw = it->data;
+        int i;
+
+        /* Type */
+        build_append_int_noprefix(table_data, 1, 1);
+
+        /* Reserved */
+        build_append_int_noprefix(table_data, 0, 1);
+
+        /* Record Length */
+        build_append_int_noprefix(table_data, 36 + 4 * fw->num_targets, 2);
+
+        /* Reserved */
+        build_append_int_noprefix(table_data, 0, 4);
+
+        /* Base HPA */
+        build_append_int_noprefix(table_data, fw->mr.addr, 8);
+
+        /* Window Size */
+        build_append_int_noprefix(table_data, fw->size, 8);
+
+        /* Host Bridge Interleave Ways */
+        build_append_int_noprefix(table_data, fw->enc_int_ways, 1);
+
+        /* Host Bridge Interleave Arithmetic */
+        build_append_int_noprefix(table_data, 0, 1);
+
+        /* Reserved */
+        build_append_int_noprefix(table_data, 0, 2);
+
+        /* Host Bridge Interleave Granularity */
+        build_append_int_noprefix(table_data, fw->enc_int_gran, 4);
+
+        /* Window Restrictions */
+        build_append_int_noprefix(table_data, 0x0f, 2); /* No restrictions */
+
+        /* QTG ID */
+        build_append_int_noprefix(table_data, 0, 2);
+
+        /* Host Bridge List (list of UIDs - currently bus_nr) */
+        for (i = 0; i < fw->num_targets; i++) {
+            g_assert(fw->target_hbs[i]);
+            build_append_int_noprefix(table_data, fw->target_hbs[i]->bus_nr, 4);
+        }
+    }
+}
+
 static int cxl_foreach_pxb_hb(Object *obj, void *opaque)
 {
     Aml *cedt = opaque;
@@ -86,6 +144,7 @@ void cxl_build_cedt(MachineState *ms, GArray *table_offsets, GArray *table_data,
     /* reserve space for CEDT header */
 
     object_child_foreach_recursive(object_get_root(), cxl_foreach_pxb_hb, cedt);
+    cedt_build_cfmws(cedt->buf, ms);
 
     /* copy AML table into ACPI tables blob and patch header there */
     g_array_append_vals(table_data, cedt->buf->data, cedt->buf->len);
Signature · 1 line
2.32.0
#34 lore raw

PATCH v10 29/45 hw/pci-host/gpex-acpi: Add support for dsdt construction for pxb-cxl

In reply to Jonathan Cameron

This adds code to instantiate the slightly extended ACPI root port description in DSDT as per the CXL 2.0 specification.

Basically a cut and paste job from the i386/pc code.

---
 hw/arm/Kconfig          |  1 +
 hw/pci-host/gpex-acpi.c | 20 +++++++++++++++++---
 2 files changed, 18 insertions(+), 3 deletions(-)
hw/arm/Kconfig +1 -0
index 97f3b38019..219262a8da 100644
@@ -29,6 +29,7 @@ config ARM_VIRT
     select ACPI_APEI
     select ACPI_VIOT
     select VIRTIO_MEM_SUPPORTED
+    select ACPI_CXL
 
 config CHEETAH
     bool
hw/pci-host/gpex-acpi.c +17 -3
index e7e162a00a..7c7316bc96 100644
@@ -5,6 +5,7 @@
 #include "hw/pci/pci_bus.h"
 #include "hw/pci/pci_bridge.h"
 #include "hw/pci/pcie_host.h"
+#include "hw/acpi/cxl.h"
 
 static void acpi_dsdt_add_pci_route_table(Aml *dev, uint32_t irq)
 {
@@ -139,6 +140,7 @@ void acpi_dsdt_add_gpex(Aml *scope, struct GPEXConfig *cfg)
         QLIST_FOREACH(bus, &bus->child, sibling) {
             uint8_t bus_num = pci_bus_num(bus);
             uint8_t numa_node = pci_bus_numa_node(bus);
+            bool is_cxl = pci_bus_is_cxl(bus);
 
             if (!pci_bus_is_root(bus)) {
                 continue;
@@ -154,8 +156,16 @@ void acpi_dsdt_add_gpex(Aml *scope, struct GPEXConfig *cfg)
             }
 
             dev = aml_device("PC%.02X", bus_num);
-            aml_append(dev, aml_name_decl("_HID", aml_string("PNP0A08")));
-            aml_append(dev, aml_name_decl("_CID", aml_string("PNP0A03")));
+            if (is_cxl) {
+                struct Aml *pkg = aml_package(2);
+                aml_append(dev, aml_name_decl("_HID", aml_string("ACPI0016")));
+                aml_append(pkg, aml_eisaid("PNP0A08"));
+                aml_append(pkg, aml_eisaid("PNP0A03"));
+                aml_append(dev, aml_name_decl("_CID", pkg));
+            } else {
+                aml_append(dev, aml_name_decl("_HID", aml_string("PNP0A08")));
+                aml_append(dev, aml_name_decl("_CID", aml_string("PNP0A03")));
+            }
             aml_append(dev, aml_name_decl("_BBN", aml_int(bus_num)));
             aml_append(dev, aml_name_decl("_UID", aml_int(bus_num)));
             aml_append(dev, aml_name_decl("_STR", aml_unicode("pxb Device")));
@@ -175,7 +185,11 @@ void acpi_dsdt_add_gpex(Aml *scope, struct GPEXConfig *cfg)
                             cfg->pio.base, 0, 0, 0);
             aml_append(dev, aml_name_decl("_CRS", crs));
 
-            acpi_dsdt_add_pci_osc(dev);
+            if (is_cxl) {
+                build_cxl_osc_method(dev);
+            } else {
+                acpi_dsdt_add_pci_osc(dev);
+            }
 
             aml_append(scope, dev);
         }
Signature · 1 line
2.32.0
#35 lore raw

PATCH v10 30/45 pci/pcie_port: Add pci_find_port_by_pn()

In reply to Jonathan Cameron

From: Jonathan Cameron <[email protected]>

Simple function to search a PCIBus to find a port by it's port number.

CXL interleave decoding uses the port number as a target so it is necessary to locate the port when doing interleave decoding.

---
 hw/pci/pcie_port.c         | 25 +++++++++++++++++++++++++
 include/hw/pci/pcie_port.h |  2 ++
 2 files changed, 27 insertions(+)
hw/pci/pcie_port.c +25 -0
index e95c1e5519..687e4e763a 100644
@@ -136,6 +136,31 @@ static void pcie_port_class_init(ObjectClass *oc, void *data)
     device_class_set_props(dc, pcie_port_props);
 }
 
+PCIDevice *pcie_find_port_by_pn(PCIBus *bus, uint8_t pn)
+{
+    int devfn;
+
+    for (devfn = 0; devfn < ARRAY_SIZE(bus->devices); devfn++) {
+        PCIDevice *d = bus->devices[devfn];
+        PCIEPort *port;
+
+        if (!d || !pci_is_express(d) || !d->exp.exp_cap) {
+            continue;
+        }
+
+        if (!object_dynamic_cast(OBJECT(d), TYPE_PCIE_PORT)) {
+            continue;
+        }
+
+        port = PCIE_PORT(d);
+        if (port->port == pn) {
+            return d;
+        }
+    }
+
+    return NULL;
+}
+
 static const TypeInfo pcie_port_type_info = {
     .name = TYPE_PCIE_PORT,
     .parent = TYPE_PCI_BRIDGE,
include/hw/pci/pcie_port.h +2 -0
index e25b289ce8..7b8193061a 100644
@@ -39,6 +39,8 @@ struct PCIEPort {
 
 void pcie_port_init_reg(PCIDevice *d);
 
+PCIDevice *pcie_find_port_by_pn(PCIBus *bus, uint8_t pn);
+
 #define TYPE_PCIE_SLOT "pcie-slot"
 OBJECT_DECLARE_SIMPLE_TYPE(PCIESlot, PCIE_SLOT)
 
Signature · 1 line
2.32.0
#36 lore raw

PATCH v10 31/45 CXL/cxl_component: Add cxl_get_hb_cstate()

In reply to Jonathan Cameron

From: Jonathan Cameron <[email protected]>

Accessor to get hold of the cxl state for a CXL host bridge without exposing the internals of the implementation.

---
 hw/pci-bridge/pci_expander_bridge.c | 7 +++++++
 include/hw/cxl/cxl_component.h      | 2 ++
 2 files changed, 9 insertions(+)
hw/pci-bridge/pci_expander_bridge.c +7 -0
index 22feda1ff0..69244decdb 100644
@@ -72,6 +72,13 @@ static GList *pxb_dev_list;
 
 #define TYPE_PXB_HOST "pxb-host"
 
+CXLComponentState *cxl_get_hb_cstate(PCIHostState *hb)
+{
+    CXLHost *host = PXB_CXL_HOST(hb);
+
+    return &host->cxl_cstate;
+}
+
 static int pxb_bus_num(PCIBus *bus)
 {
     PXBDev *pxb = convert_to_pxb(bus->parent_dev);
include/hw/cxl/cxl_component.h +2 -0
index 4f69688c47..70b5018156 100644
@@ -218,4 +218,6 @@ static inline hwaddr cxl_decode_ig(int ig)
     return 1 << (ig + 8);
 }
 
+CXLComponentState *cxl_get_hb_cstate(PCIHostState *hb);
+
 #endif
Signature · 1 line
2.32.0
#37 lore raw

PATCH v10 32/45 mem/cxl_type3: Add read and write functions for associated hostmem.

In reply to Jonathan Cameron

From: Jonathan Cameron <[email protected]>

Once a read or write reaches a CXL type 3 device, the HDM decoders on the device are used to establish the Device Physical Address which should be accessed. These functions peform the required maths and then use a device specific address space to access the hostmem->mr to fullfil the actual operation. Note that failed writes are silent, but failed reads return poison. Note this is based loosely on:

https://lore.kernel.org/qemu-devel/[email protected]/
[RFC PATCH 0/9] hw/misc: Add support for interleaved memory accesses

Only lightly tested so far. More complex test cases yet to be written.

---
 hw/mem/cxl_type3.c          | 91 +++++++++++++++++++++++++++++++++++++
 include/hw/cxl/cxl_device.h |  6 +++
 2 files changed, 97 insertions(+)
hw/mem/cxl_type3.c +91 -0
index 6c6ed9a776..3bf2869573 100644
@@ -97,7 +97,9 @@ static void ct3d_reg_write(void *opaque, hwaddr offset, uint64_t value,
 
 static bool cxl_setup_memory(CXLType3Dev *ct3d, Error **errp)
 {
+    DeviceState *ds = DEVICE(ct3d);
     MemoryRegion *mr;
+    char *name;
 
     if (!ct3d->hostmem) {
         error_setg(errp, "memdev property must be set");
@@ -112,6 +114,15 @@ static bool cxl_setup_memory(CXLType3Dev *ct3d, Error **errp)
     memory_region_set_nonvolatile(mr, true);
     memory_region_set_enabled(mr, true);
     host_memory_backend_set_mapped(ct3d->hostmem, true);
+
+    if (ds->id) {
+        name = g_strdup_printf("cxl-type3-dpa-space:%s", ds->id);
+    } else {
+        name = g_strdup("cxl-type3-dpa-space");
+    }
+    address_space_init(&ct3d->hostmem_as, mr, name);
+    g_free(name);
+
     ct3d->cxl_dstate.pmem_size = ct3d->hostmem->size;
 
     if (!ct3d->lsa) {
@@ -167,6 +178,86 @@ static void ct3_exit(PCIDevice *pci_dev)
     ComponentRegisters *regs = &cxl_cstate->crb;
 
     g_free(regs->special_ops);
+    address_space_destroy(&ct3d->hostmem_as);
+}
+
+/* TODO: Support multiple HDM decoders and DPA skip */
+static bool cxl_type3_dpa(CXLType3Dev *ct3d, hwaddr host_addr, uint64_t *dpa)
+{
+    uint32_t *cache_mem = ct3d->cxl_cstate.crb.cache_mem_registers;
+    uint64_t decoder_base, decoder_size, hpa_offset;
+    uint32_t hdm0_ctrl;
+    int ig, iw;
+
+    decoder_base = (((uint64_t)cache_mem[R_CXL_HDM_DECODER0_BASE_HI] << 32) |
+                    cache_mem[R_CXL_HDM_DECODER0_BASE_LO]);
+    if ((uint64_t)host_addr < decoder_base) {
+        return false;
+    }
+
+    hpa_offset = (uint64_t)host_addr - decoder_base;
+
+    decoder_size = ((uint64_t)cache_mem[R_CXL_HDM_DECODER0_SIZE_HI] << 32) |
+        cache_mem[R_CXL_HDM_DECODER0_SIZE_LO];
+    if (hpa_offset >= decoder_size) {
+        return false;
+    }
+
+    hdm0_ctrl = cache_mem[R_CXL_HDM_DECODER0_CTRL];
+    iw = FIELD_EX32(hdm0_ctrl, CXL_HDM_DECODER0_CTRL, IW);
+    ig = FIELD_EX32(hdm0_ctrl, CXL_HDM_DECODER0_CTRL, IG);
+
+    *dpa = (MAKE_64BIT_MASK(0, 8 + ig) & hpa_offset) |
+        ((MAKE_64BIT_MASK(8 + ig + iw, 64 - 8 - ig - iw) & hpa_offset) >> iw);
+
+    return true;
+}
+
+MemTxResult cxl_type3_read(PCIDevice *d, hwaddr host_addr, uint64_t *data,
+                           unsigned size, MemTxAttrs attrs)
+{
+    CXLType3Dev *ct3d = CXL_TYPE3(d);
+    uint64_t dpa_offset;
+    MemoryRegion *mr;
+
+    /* TODO support volatile region */
+    mr = host_memory_backend_get_memory(ct3d->hostmem);
+    if (!mr) {
+        return MEMTX_ERROR;
+    }
+
+    if (!cxl_type3_dpa(ct3d, host_addr, &dpa_offset)) {
+        return MEMTX_ERROR;
+    }
+
+    if (dpa_offset > int128_get64(mr->size)) {
+        return MEMTX_ERROR;
+    }
+
+    return address_space_read(&ct3d->hostmem_as, dpa_offset, attrs, data, size);
+}
+
+MemTxResult cxl_type3_write(PCIDevice *d, hwaddr host_addr, uint64_t data,
+                            unsigned size, MemTxAttrs attrs)
+{
+    CXLType3Dev *ct3d = CXL_TYPE3(d);
+    uint64_t dpa_offset;
+    MemoryRegion *mr;
+
+    mr = host_memory_backend_get_memory(ct3d->hostmem);
+    if (!mr) {
+        return MEMTX_OK;
+    }
+
+    if (!cxl_type3_dpa(ct3d, host_addr, &dpa_offset)) {
+        return MEMTX_OK;
+    }
+
+    if (dpa_offset > int128_get64(mr->size)) {
+        return MEMTX_OK;
+    }
+    return address_space_write(&ct3d->hostmem_as, dpa_offset, attrs,
+                               &data, size);
 }
 
 static void ct3d_reset(DeviceState *dev)
include/hw/cxl/cxl_device.h +6 -0
index 4285fbda08..1e141b6621 100644
@@ -239,6 +239,7 @@ struct CXLType3Dev {
     HostMemoryBackend *lsa;
 
     /* State */
+    AddressSpace hostmem_as;
     CXLComponentState cxl_cstate;
     CXLDeviceState cxl_dstate;
 };
@@ -259,4 +260,9 @@ struct CXLType3Class {
                     uint64_t offset);
 };
 
+MemTxResult cxl_type3_read(PCIDevice *d, hwaddr host_addr, uint64_t *data,
+                           unsigned size, MemTxAttrs attrs);
+MemTxResult cxl_type3_write(PCIDevice *d, hwaddr host_addr, uint64_t data,
+                            unsigned size, MemTxAttrs attrs);
+
 #endif
Signature · 1 line
2.32.0
#38 lore raw

PATCH v10 33/45 cxl/cxl-host: Add memops for CFMWS region.

In reply to Jonathan Cameron

From: Jonathan Cameron <[email protected]>

These memops perform interleave decoding, walking down the CXL topology from CFMWS described host interleave decoder via CXL host bridge HDM decoders, through the CXL root ports and finally call CXL type 3 specific read and write functions.

Note that, whilst functional the current implementation does not support:
* switches
* multiple HDM decoders at a given level.
* unaligned accesses across the interleave boundaries

---
 hw/cxl/cxl-host-stubs.c |   2 +
 hw/cxl/cxl-host.c       | 128 ++++++++++++++++++++++++++++++++++++++++
 include/hw/cxl/cxl.h    |   2 +
 3 files changed, 132 insertions(+)
hw/cxl/cxl-host-stubs.c +2 -0
index f8fd278d5d..24465a52ab 100644
@@ -12,3 +12,5 @@ void cxl_fixed_memory_window_config(MachineState *ms,
                                     Error **errp) {};
 
 void cxl_fixed_memory_window_link_targets(Error **errp) {};
+
+const MemoryRegionOps cfmws_ops;
hw/cxl/cxl-host.c +128 -0

137 lines not on this page · raw message

include/hw/cxl/cxl.h +2 -0
index dce38124db..21d28ca110 100644
@@ -56,4 +56,6 @@ void cxl_fixed_memory_window_config(MachineState *ms,
                                     Error **errp);
 void cxl_fixed_memory_window_link_targets(Error **errp);
 
+extern const MemoryRegionOps cfmws_ops;
+
 #endif
Signature · 1 line
2.32.0
#39 lore raw

PATCH v10 34/45 hw/cxl/component Add a dumb HDM decoder handler

In reply to Jonathan Cameron

From: Ben Widawsky <[email protected]>

Add a trivial handler for now to cover the root bridge where we could do some error checking in future.

---
 hw/cxl/cxl-component-utils.c | 31 +++++++++++++++++++++++++++++++
 1 file changed, 31 insertions(+)
hw/cxl/cxl-component-utils.c +31 -0
index 69cb07171c..7985c9bfca 100644
@@ -32,6 +32,31 @@ static uint64_t cxl_cache_mem_read_reg(void *opaque, hwaddr offset,
     }
 }
 
+static void dumb_hdm_handler(CXLComponentState *cxl_cstate, hwaddr offset,
+                             uint32_t value)
+{
+    ComponentRegisters *cregs = &cxl_cstate->crb;
+    uint32_t *cache_mem = cregs->cache_mem_registers;
+    bool should_commit = false;
+
+    switch (offset) {
+    case A_CXL_HDM_DECODER0_CTRL:
+        should_commit = FIELD_EX32(value, CXL_HDM_DECODER0_CTRL, COMMIT);
+        break;
+    default:
+        break;
+    }
+
+    memory_region_transaction_begin();
+    stl_le_p((uint8_t *)cache_mem + offset, value);
+    if (should_commit) {
+        ARRAY_FIELD_DP32(cache_mem, CXL_HDM_DECODER0_CTRL, COMMIT, 0);
+        ARRAY_FIELD_DP32(cache_mem, CXL_HDM_DECODER0_CTRL, ERR, 0);
+        ARRAY_FIELD_DP32(cache_mem, CXL_HDM_DECODER0_CTRL, COMMITTED, 1);
+    }
+    memory_region_transaction_commit();
+}
+
 static void cxl_cache_mem_write_reg(void *opaque, hwaddr offset, uint64_t value,
                                     unsigned size)
 {
@@ -50,6 +75,12 @@ static void cxl_cache_mem_write_reg(void *opaque, hwaddr offset, uint64_t value,
     value |= ~mask & cregs->cache_mem_registers[offset / sizeof(*cregs->cache_mem_registers)];
     if (cregs->special_ops && cregs->special_ops->write) {
         cregs->special_ops->write(cxl_cstate, offset, value, size);
+        return;
+    }
+
+    if (offset >= A_CXL_HDM_DECODER_CAPABILITY &&
+        offset <= A_CXL_HDM_DECODER0_TARGET_LIST_HI) {
+        dumb_hdm_handler(cxl_cstate, offset, value);
     } else {
         cregs->cache_mem_registers[offset / sizeof(*cregs->cache_mem_registers)] = value;
     }
Signature · 1 line
2.32.0
#40 lore raw

PATCH v10 35/45 i386/pc: Enable CXL fixed memory windows

In reply to Jonathan Cameron

From: Jonathan Cameron <[email protected]>

Add the CFMWs memory regions to the memorymap and adjust the PCI window to avoid hitting the same memory.

---
 hw/i386/pc.c | 31 ++++++++++++++++++++++++++++++-
 1 file changed, 30 insertions(+), 1 deletion(-)
hw/i386/pc.c +30 -1
index b56af1e4e7..dd7cffd340 100644
@@ -817,7 +817,7 @@ void pc_memory_init(PCMachineState *pcms,
     MachineClass *mc = MACHINE_GET_CLASS(machine);
     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
     X86MachineState *x86ms = X86_MACHINE(pcms);
-    hwaddr cxl_base;
+    hwaddr cxl_base, cxl_resv_end = 0;
 
     assert(machine->ram_size == x86ms->below_4g_mem_size +
                                 x86ms->above_4g_mem_size);
@@ -925,6 +925,24 @@ void pc_memory_init(PCMachineState *pcms,
         e820_add_entry(cxl_base, cxl_size, E820_RESERVED);
         memory_region_init(mr, OBJECT(machine), "cxl_host_reg", cxl_size);
         memory_region_add_subregion(system_memory, cxl_base, mr);
+        cxl_resv_end = cxl_base + cxl_size;
+        if (machine->cxl_devices_state->fixed_windows) {
+            hwaddr cxl_fmw_base;
+            GList *it;
+
+            cxl_fmw_base = ROUND_UP(cxl_base + cxl_size, 256 * MiB);
+            for (it = machine->cxl_devices_state->fixed_windows; it; it = it->next) {
+                CXLFixedWindow *fw = it->data;
+
+                fw->base = cxl_fmw_base;
+                memory_region_init_io(&fw->mr, OBJECT(machine), &cfmws_ops, fw,
+                                      "cxl-fixed-memory-region", fw->size);
+                memory_region_add_subregion(system_memory, fw->base, &fw->mr);
+                e820_add_entry(fw->base, fw->size, E820_RESERVED);
+                cxl_fmw_base += fw->size;
+                cxl_resv_end = cxl_fmw_base;
+            }
+        }
     }
 
     /* Initialize PC system firmware */
@@ -954,6 +972,10 @@ void pc_memory_init(PCMachineState *pcms,
         if (!pcmc->broken_reserved_end) {
             res_mem_end += memory_region_size(&machine->device_memory->mr);
         }
+
+        if (machine->cxl_devices_state->is_enabled) {
+            res_mem_end = cxl_resv_end;
+        }
         *val = cpu_to_le64(ROUND_UP(res_mem_end, 1 * GiB));
         fw_cfg_add_file(fw_cfg, "etc/reserved-memory-end", val, sizeof(*val));
     }
@@ -990,6 +1012,13 @@ uint64_t pc_pci_hole64_start(void)
     if (ms->cxl_devices_state->host_mr.addr) {
         hole64_start = ms->cxl_devices_state->host_mr.addr +
             memory_region_size(&ms->cxl_devices_state->host_mr);
+        if (ms->cxl_devices_state->fixed_windows) {
+            GList *it;
+            for (it = ms->cxl_devices_state->fixed_windows; it; it = it->next) {
+                CXLFixedWindow *fw = it->data;
+                hole64_start = fw->mr.addr + memory_region_size(&fw->mr);
+            }
+        }
     } else if (pcmc->has_reserved_memory && ms->device_memory->base) {
         hole64_start = ms->device_memory->base;
         if (!pcmc->broken_reserved_end) {
Signature · 1 line
2.32.0
#41 lore raw

PATCH v10 36/45 tests/acpi: q35: Allow addition of a CXL test.

In reply to Jonathan Cameron

Add exceptions for the DSDT and the new CEDT tables specific to a new CXL test in the following patch.

---
 tests/data/acpi/q35/CEDT.cxl                | 0
 tests/data/acpi/q35/DSDT.cxl                | 0
 tests/qtest/bios-tables-test-allowed-diff.h | 2 ++
 3 files changed, 2 insertions(+)
tests/data/acpi/q35/CEDT.cxl new file +0 -0
new file mode 100644
index 0000000000..e69de29bb2
tests/data/acpi/q35/DSDT.cxl new file +0 -0
new file mode 100644
index 0000000000..e69de29bb2
tests/qtest/bios-tables-test-allowed-diff.h +2 -0
index dfb8523c8b..7c7f9fbc44 100644
@@ -1 +1,3 @@
 /* List of comma-separated changed AML files to ignore */
+"tests/data/acpi/q35/DSDT.cxl",
+"tests/data/acpi/q35/CEDT.cxl",
Signature · 1 line
2.32.0
#42 lore raw

PATCH v10 37/45 qtests/bios-tables-test: Add a test for CXL emulation.

In reply to Jonathan Cameron

The DSDT includes several CXL specific elements and the CEDT table is only present if we enable CXL.

The test exercises all current functionality with several CFMWS, CHBS structures in CEDT and ACPI0016/ACPI00017 and _OSC entries in DSDT.

---
 tests/qtest/bios-tables-test.c | 44 ++++++++++++++++++++++++++++++++++
 1 file changed, 44 insertions(+)
tests/qtest/bios-tables-test.c +44 -0
index 5dddedabcd..a4a46e97f0 100644
@@ -1536,6 +1536,49 @@ static void test_acpi_q35_viot(void)
     free_test_data(&data);
 }
 
+static void test_acpi_q35_cxl(void)
+{
+    gchar *tmp_path = g_dir_make_tmp("qemu-test-cxl.XXXXXX", NULL);
+    gchar *params;
+
+    test_data data = {
+        .machine = MACHINE_Q35,
+        .variant = ".cxl",
+    };
+    /*
+     * A complex CXL setup.
+     */
+    params = g_strdup_printf(" -machine cxl=on"
+                             " -object memory-backend-file,id=cxl-mem1,mem-path=%s,size=256M"
+                             " -object memory-backend-file,id=cxl-mem2,mem-path=%s,size=256M"
+                             " -object memory-backend-file,id=cxl-mem3,mem-path=%s,size=256M"
+                             " -object memory-backend-file,id=cxl-mem4,mem-path=%s,size=256M"
+                             " -object memory-backend-file,id=lsa1,mem-path=%s,size=256M"
+                             " -object memory-backend-file,id=lsa2,mem-path=%s,size=256M"
+                             " -object memory-backend-file,id=lsa3,mem-path=%s,size=256M"
+                             " -object memory-backend-file,id=lsa4,mem-path=%s,size=256M"
+                             " -device pxb-cxl,bus_nr=12,bus=pcie.0,id=cxl.1"
+                             " -device pxb-cxl,bus_nr=222,bus=pcie.0,id=cxl.2"
+                             " -device cxl-rp,port=0,bus=cxl.1,id=rp1,chassis=0,slot=2"
+                             " -device cxl-type3,bus=rp1,memdev=cxl-mem1,lsa=lsa1"
+                             " -device cxl-rp,port=1,bus=cxl.1,id=rp2,chassis=0,slot=3"
+                             " -device cxl-type3,bus=rp2,memdev=cxl-mem2,lsa=lsa2"
+                             " -device cxl-rp,port=0,bus=cxl.2,id=rp3,chassis=0,slot=5"
+                             " -device cxl-type3,bus=rp3,memdev=cxl-mem3,lsa=lsa3"
+                             " -device cxl-rp,port=1,bus=cxl.2,id=rp4,chassis=0,slot=6"
+                             " -device cxl-type3,bus=rp4,memdev=cxl-mem4,lsa=lsa4"
+                             " -cxl-fixed-memory-window targets.0=cxl.1,size=4G,interleave-granularity=8k"
+                             " -cxl-fixed-memory-window targets.0=cxl.1,targets.1=cxl.2,size=4G,interleave-granularity=8k",
+                             tmp_path, tmp_path, tmp_path, tmp_path,
+                             tmp_path, tmp_path, tmp_path, tmp_path);
+    test_acpi_one(params, &data);
+
+    g_free(params);
+    g_assert(g_rmdir(tmp_path) == 0);
+    g_free(tmp_path);
+    free_test_data(&data);
+}
+
 static void test_acpi_virt_viot(void)
 {
     test_data data = {
@@ -1741,6 +1784,7 @@ int main(int argc, char *argv[])
             qtest_add_func("acpi/q35/kvm/dmar", test_acpi_q35_kvm_dmar);
         }
         qtest_add_func("acpi/q35/viot", test_acpi_q35_viot);
+        qtest_add_func("acpi/q35/cxl", test_acpi_q35_cxl);
         qtest_add_func("acpi/q35/slic", test_acpi_q35_slic);
     } else if (strcmp(arch, "aarch64") == 0) {
         if (has_tcg) {
Signature · 1 line
2.32.0
#43 lore raw

PATCH v10 38/45 tests/acpi: Add tables for CXL emulation.

In reply to Jonathan Cameron

Tables that differ from normal Q35 tables when running the CXL test.

---
 tests/data/acpi/q35/CEDT.cxl                | Bin 0 -> 184 bytes
 tests/data/acpi/q35/DSDT.cxl                | Bin 0 -> 9615 bytes
 tests/qtest/bios-tables-test-allowed-diff.h |   2 --
 3 files changed, 2 deletions(-)
tests/data/acpi/q35/CEDT.cxl binary +0 -0
index e69de29bb2d1d6434b8b29ae775ad8c2e48c5391..b8fa06b00e65712e91e0a5ea0d9277e0146d1c00 100644
GIT binary patch
tests/data/acpi/q35/DSDT.cxl binary +0 -0
index e69de29bb2d1d6434b8b29ae775ad8c2e48c5391..c1206defed0154e9024702bba88453b2790a306d 100644
GIT binary patch
tests/qtest/bios-tables-test-allowed-diff.h +0 -2
index 7c7f9fbc44..dfb8523c8b 100644
@@ -1,3 +1 @@
 /* List of comma-separated changed AML files to ignore */
-"tests/data/acpi/q35/DSDT.cxl",
-"tests/data/acpi/q35/CEDT.cxl",
Signature · 1 line
2.32.0
#44 lore raw

PATCH v10 39/45 qtest/cxl: Add more complex test cases with CFMWs

In reply to Jonathan Cameron

From: Ben Widawsky <[email protected]>

Add CXL Fixed Memory Windows to the CXL tests.

---
 tests/qtest/cxl-test.c | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)
tests/qtest/cxl-test.c +5 -3
index 5f0794e816..079011af6a 100644
@@ -9,11 +9,13 @@
 #include "libqtest-single.h"
 
 #define QEMU_PXB_CMD "-machine q35,cxl=on " \
-                     "-device pxb-cxl,id=cxl.0,bus=pcie.0,bus_nr=52 "
+                     "-device pxb-cxl,id=cxl.0,bus=pcie.0,bus_nr=52 "  \
+                     "-cxl-fixed-memory-window targets.0=cxl.0,size=4G "
 
-#define QEMU_2PXB_CMD "-machine q35,cxl=on " \
+#define QEMU_2PXB_CMD "-machine q35,cxl=on "                            \
                       "-device pxb-cxl,id=cxl.0,bus=pcie.0,bus_nr=52 "  \
-                      "-device pxb-cxl,id=cxl.1,bus=pcie.0,bus_nr=53 "
+                      "-device pxb-cxl,id=cxl.1,bus=pcie.0,bus_nr=53 " \
+                      "-cxl-fixed-memory-window targets.0=cxl.0,targets.1=cxl.1,size=4G "
 
 #define QEMU_RP "-device cxl-rp,id=rp0,bus=cxl.0,chassis=0,slot=0 "
 
Signature · 1 line
2.32.0
#45 lore raw

PATCH v10 40/45 hw/arm/virt: Basic CXL enablement on pci_expander_bridge instances pxb-cxl

In reply to Jonathan Cameron

Code based on i386/pc enablement.
The memory layout places space for 16 host bridge register regions after the GIC_REDIST2 in the extended memmap.
The CFMWs are placed above the extended memmap.

Only create the CEDT table if cxl=on set for the machine.

---
 hw/arm/virt-acpi-build.c | 33 +++++++++++++++++++++++++++++++++
 hw/arm/virt.c            | 40 +++++++++++++++++++++++++++++++++++++++-
 include/hw/arm/virt.h    |  1 +
 3 files changed, 73 insertions(+), 1 deletion(-)
hw/arm/virt-acpi-build.c +33 -0
index 449fab0080..86a2f40437 100644
@@ -39,9 +39,11 @@
 #include "hw/acpi/aml-build.h"
 #include "hw/acpi/utils.h"
 #include "hw/acpi/pci.h"
+#include "hw/acpi/cxl.h"
 #include "hw/acpi/memory_hotplug.h"
 #include "hw/acpi/generic_event_device.h"
 #include "hw/acpi/tpm.h"
+#include "hw/cxl/cxl.h"
 #include "hw/pci/pcie_host.h"
 #include "hw/pci/pci.h"
 #include "hw/pci/pci_bus.h"
@@ -157,10 +159,29 @@ static void acpi_dsdt_add_virtio(Aml *scope,
     }
 }
 
+/* Uses local definition of AcpiBuildState so can't easily be common code */
+static void build_acpi0017(Aml *table)
+{
+    Aml *dev, *scope, *method;
+
+    scope =  aml_scope("_SB");
+    dev = aml_device("CXLM");
+    aml_append(dev, aml_name_decl("_HID", aml_string("ACPI0017")));
+
+    method = aml_method("_STA", 0, AML_NOTSERIALIZED);
+    aml_append(method, aml_return(aml_int(0x01)));
+    aml_append(dev, method);
+
+    aml_append(scope, dev);
+    aml_append(table, scope);
+}
+
 static void acpi_dsdt_add_pci(Aml *scope, const MemMapEntry *memmap,
                               uint32_t irq, VirtMachineState *vms)
 {
     int ecam_id = VIRT_ECAM_ID(vms->highmem_ecam);
+    bool cxl_present = false;
+    PCIBus *bus = vms->bus;
     struct GPEXConfig cfg = {
         .mmio32 = memmap[VIRT_PCIE_MMIO],
         .pio    = memmap[VIRT_PCIE_PIO],
@@ -174,6 +195,14 @@ static void acpi_dsdt_add_pci(Aml *scope, const MemMapEntry *memmap,
     }
 
     acpi_dsdt_add_gpex(scope, &cfg);
+    QLIST_FOREACH(bus, &vms->bus->child, sibling) {
+        if (pci_bus_is_cxl(bus)) {
+            cxl_present = true;
+        }
+    }
+    if (cxl_present) {
+        build_acpi0017(scope);
+    }
 }
 
 static void acpi_dsdt_add_gpio(Aml *scope, const MemMapEntry *gpio_memmap,
@@ -991,6 +1020,10 @@ void virt_acpi_build(VirtMachineState *vms, AcpiBuildTables *tables)
                        vms->oem_table_id);
         }
     }
+    if (ms->cxl_devices_state->is_enabled) {
+        cxl_build_cedt(ms, table_offsets, tables_blob, tables->linker,
+                       vms->oem_id, vms->oem_table_id);
+    }
 
     if (ms->nvdimms_state->is_enabled) {
         nvdimm_build_acpi(table_offsets, tables_blob, tables->linker,
hw/arm/virt.c +39 -1
index 955223984c..5927d741a3 100644
@@ -78,6 +78,7 @@
 #include "hw/virtio/virtio-mem-pci.h"
 #include "hw/virtio/virtio-iommu.h"
 #include "hw/char/pl011.h"
+#include "hw/cxl/cxl.h"
 #include "qemu/guest-random.h"
 
 #define DEFINE_VIRT_MACHINE_LATEST(major, minor, latest) \
@@ -178,6 +179,7 @@ static const MemMapEntry base_memmap[] = {
 static MemMapEntry extended_memmap[] = {
     /* Additional 64 MB redist region (can contain up to 512 redistributors) */
     [VIRT_HIGH_GIC_REDIST2] =   { 0x0, 64 * MiB },
+    [VIRT_CXL_HOST] =           { 0x0, 64 * KiB * 16 }, /* 16 UID */
     [VIRT_HIGH_PCIE_ECAM] =     { 0x0, 256 * MiB },
     /* Second PCIe window */
     [VIRT_HIGH_PCIE_MMIO] =     { 0x0, 512 * GiB },
@@ -1525,6 +1527,17 @@ static void create_pcie(VirtMachineState *vms)
     }
 }
 
+static void create_cxl_host_reg_region(VirtMachineState *vms)
+{
+    MemoryRegion *sysmem = get_system_memory();
+    MachineState *ms = MACHINE(vms);
+    MemoryRegion *mr = &ms->cxl_devices_state->host_mr;
+
+    memory_region_init(mr, OBJECT(ms), "cxl_host_reg",
+                       vms->memmap[VIRT_CXL_HOST].size);
+    memory_region_add_subregion(sysmem, vms->memmap[VIRT_CXL_HOST].base, mr);
+}
+
 static void create_platform_bus(VirtMachineState *vms)
 {
     DeviceState *dev;
@@ -1687,7 +1700,7 @@ static uint64_t virt_cpu_mp_affinity(VirtMachineState *vms, int idx)
 static void virt_set_memmap(VirtMachineState *vms, int pa_bits)
 {
     MachineState *ms = MACHINE(vms);
-    hwaddr base, device_memory_base, device_memory_size, memtop;
+    hwaddr base, device_memory_base, device_memory_size, memtop, cxl_fmw_base;
     int i;
 
     vms->memmap = extended_memmap;
@@ -1779,6 +1792,20 @@ static void virt_set_memmap(VirtMachineState *vms, int pa_bits)
         memory_region_init(&ms->device_memory->mr, OBJECT(vms),
                            "device-memory", device_memory_size);
     }
+
+    if (ms->cxl_devices_state->fixed_windows) {
+        GList *it;
+
+        cxl_fmw_base = ROUND_UP(base, 256 * MiB);
+        for (it = ms->cxl_devices_state->fixed_windows; it; it = it->next) {
+            CXLFixedWindow *fw = it->data;
+
+            fw->base = cxl_fmw_base;
+            memory_region_init_io(&fw->mr, OBJECT(vms), &cfmws_ops, fw,
+                                  "cxl-fixed-memory-region", fw->size);
+            cxl_fmw_base += fw->size;
+        }
+    }
 }
 
 /*
@@ -2215,6 +2242,15 @@ static void machvirt_init(MachineState *machine)
         memory_region_add_subregion(sysmem, machine->device_memory->base,
                                     &machine->device_memory->mr);
     }
+    if (machine->cxl_devices_state->fixed_windows) {
+        GList *it;
+        for (it = machine->cxl_devices_state->fixed_windows; it;
+             it = it->next) {
+            CXLFixedWindow *fw = it->data;
+
+            memory_region_add_subregion(sysmem, fw->base, &fw->mr);
+        }
+    }
 
     virt_flash_fdt(vms, sysmem, secure_sysmem ?: sysmem);
 
@@ -2241,6 +2277,7 @@ static void machvirt_init(MachineState *machine)
     create_rtc(vms);
 
     create_pcie(vms);
+    create_cxl_host_reg_region(vms);
 
     if (has_ged && aarch64 && firmware_loaded && virt_is_acpi_enabled(vms)) {
         vms->acpi_dev = create_acpi_ged(vms);
@@ -2909,6 +2946,7 @@ static void virt_machine_class_init(ObjectClass *oc, void *data)
     hc->unplug = virt_machine_device_unplug_cb;
     mc->nvdimm_supported = true;
     mc->smp_props.clusters_supported = true;
+    mc->cxl_supported = true;
     mc->auto_enable_numa_with_memhp = true;
     mc->auto_enable_numa_with_memdev = true;
     mc->default_ram_id = "mach-virt.ram";
include/hw/arm/virt.h +1 -0
index 15feabac63..67c08a62af 100644
@@ -92,6 +92,7 @@ enum {
 /* indices of IO regions located after the RAM */
 enum {
     VIRT_HIGH_GIC_REDIST2 =  VIRT_LOWMEMMAP_LAST,
+    VIRT_CXL_HOST,
     VIRT_HIGH_PCIE_ECAM,
     VIRT_HIGH_PCIE_MMIO,
 };
Signature · 1 line
2.32.0
#46 lore raw

PATCH v10 41/45 qtest/cxl: Add aarch64 virt test for CXL

In reply to Jonathan Cameron

Add a single complex case for aarch64 virt machine.

---
 tests/qtest/cxl-test.c  | 48 +++++++++++++++++++++++++++++++++--------
 tests/qtest/meson.build |  1 +
 2 files changed, 40 insertions(+), 9 deletions(-)
tests/qtest/cxl-test.c +39 -9
index 079011af6a..ac7d71fd74 100644
@@ -17,6 +17,11 @@
                       "-device pxb-cxl,id=cxl.1,bus=pcie.0,bus_nr=53 " \
                       "-cxl-fixed-memory-window targets.0=cxl.0,targets.1=cxl.1,size=4G "
 
+#define QEMU_VIRT_2PXB_CMD "-machine virt,cxl=on "                      \
+                      "-device pxb-cxl,id=cxl.0,bus=pcie.0,bus_nr=52 "  \
+                      "-device pxb-cxl,id=cxl.1,bus=pcie.0,bus_nr=53 "  \
+                      "-cxl-fixed-memory-window targets.0=cxl.0,targets.1=cxl.1,size=4G "
+
 #define QEMU_RP "-device cxl-rp,id=rp0,bus=cxl.0,chassis=0,slot=0 "
 
 /* Dual ports on first pxb */
@@ -134,18 +139,43 @@ static void cxl_2pxb_4rp_4t3d(void)
     qtest_end();
 }
 
+static void cxl_virt_2pxb_4rp_4t3d(void)
+{
+    g_autoptr(GString) cmdline = g_string_new(NULL);
+    char template[] = "/tmp/cxl-test-XXXXXX";
+    const char *tmpfs;
+
+    tmpfs = mkdtemp(template);
+
+    g_string_printf(cmdline, QEMU_VIRT_2PXB_CMD QEMU_4RP QEMU_4T3D,
+                    tmpfs, tmpfs, tmpfs, tmpfs, tmpfs, tmpfs,
+                    tmpfs, tmpfs);
+
+    qtest_start(cmdline->str);
+    qtest_end();
+}
+
 int main(int argc, char **argv)
 {
+    const char *arch = qtest_get_arch();
+
     g_test_init(&argc, &argv, NULL);
 
-    qtest_add_func("/pci/cxl/basic_hostbridge", cxl_basic_hb);
-    qtest_add_func("/pci/cxl/basic_pxb", cxl_basic_pxb);
-    qtest_add_func("/pci/cxl/pxb_with_window", cxl_pxb_with_window);
-    qtest_add_func("/pci/cxl/pxb_x2_with_window", cxl_2pxb_with_window);
-    qtest_add_func("/pci/cxl/rp", cxl_root_port);
-    qtest_add_func("/pci/cxl/rp_x2", cxl_2root_port);
-    qtest_add_func("/pci/cxl/type3_device", cxl_t3d);
-    qtest_add_func("/pci/cxl/rp_x2_type3_x2", cxl_1pxb_2rp_2t3d);
-    qtest_add_func("/pci/cxl/pxb_x2_root_port_x4_type3_x4", cxl_2pxb_4rp_4t3d);
+    if (strcmp(arch, "i386") == 0 || strcmp(arch, "x86_64") == 0) {
+        qtest_add_func("/pci/cxl/basic_hostbridge", cxl_basic_hb);
+        qtest_add_func("/pci/cxl/basic_pxb", cxl_basic_pxb);
+        qtest_add_func("/pci/cxl/pxb_with_window", cxl_pxb_with_window);
+        qtest_add_func("/pci/cxl/pxb_x2_with_window", cxl_2pxb_with_window);
+        qtest_add_func("/pci/cxl/rp", cxl_root_port);
+        qtest_add_func("/pci/cxl/rp_x2", cxl_2root_port);
+        qtest_add_func("/pci/cxl/type3_device", cxl_t3d);
+        qtest_add_func("/pci/cxl/rp_x2_type3_x2", cxl_1pxb_2rp_2t3d);
+        qtest_add_func("/pci/cxl/pxb_x2_root_port_x4_type3_x4",
+                       cxl_2pxb_4rp_4t3d);
+    } else if (strcmp(arch, "aarch64") == 0) {
+        qtest_add_func("/pci/cxl/virt/pxb_x2_root_port_x4_type3_x4",
+                       cxl_virt_2pxb_4rp_4t3d);
+    }
+
     return g_test_run();
 }
tests/qtest/meson.build +1 -0
index dedb477890..aa78d2d537 100644
@@ -215,6 +215,7 @@ qtests_aarch64 = \
   (config_all_devices.has_key('CONFIG_TPM_TIS_SYSBUS') ? ['tpm-tis-device-test'] : []) +        \
   (config_all_devices.has_key('CONFIG_TPM_TIS_SYSBUS') ? ['tpm-tis-device-swtpm-test'] : []) +  \
   (config_all_devices.has_key('CONFIG_XLNX_ZYNQMP_ARM') ? ['xlnx-can-test', 'fuzz-xlnx-dp-test'] : []) + \
+  qtests_cxl +                                                                                  \
   ['arm-cpu-features',
    'numa-test',
    'boot-serial-test',
Signature · 1 line
2.32.0
#47 lore raw

PATCH v10 42/45 docs/cxl: Add initial Compute eXpress Link (CXL) documentation.

In reply to Jonathan Cameron

Provide an introduction to the main components of a CXL system, with detailed explanation of memory interleaving, example command lines and kernel configuration.

This was a challenging document to write due to the need to extract only that subset of CXL information which is relevant to either users of QEMU emulation of CXL or to those interested in the implementation. Much of CXL is concerned with specific elements of the protocol, management of memory pooling etc which is simply not relevant to what is currently planned for CXL emulation in QEMU. All comments welcome

---
 docs/system/device-emulation.rst |   1 +
 docs/system/devices/cxl.rst      | 302 +++++++++++++++++++++++++++++++
 2 files changed, 303 insertions(+)
docs/system/device-emulation.rst +1 -0
index ae8dd233e8..3b729b920d 100644
@@ -84,6 +84,7 @@ Emulated Devices
 
    devices/can.rst
    devices/ccid.rst
+   devices/cxl.rst
    devices/ivshmem.rst
    devices/net.rst
    devices/nvme.rst
docs/system/devices/cxl.rst new file +302 -0

302 lines not on this page · raw message

Signature · 1 line
2.32.0
#48 lore raw

PATCH v10 43/45 pci-bridge/cxl_upstream: Add a CXL switch upstream port

In reply to Jonathan Cameron

An initial simple upstream port emulation to allow the creation of CXL switches. The Device ID has been allocated for this use.

---
 hw/pci-bridge/cxl_upstream.c | 216 +++++++++++++++++++++++++++++++++++
 hw/pci-bridge/meson.build    |   2 +-
 include/hw/cxl/cxl.h         |   4 +
 3 files changed, 221 insertions(+), 1 deletion(-)
hw/pci-bridge/cxl_upstream.c new file +216 -0
new file mode 100644
index 0000000000..6d98f9537a
@@ -0,0 +1,216 @@
+/*
+ * Emulated CXL Switch Upstream Port
+ *
+ * Copyright (c) 2022 Huawei Technologies.
+ *
+ * Based on xio31130_upstream.c
+ *
+ * SPDX-License-Identifier: GPL-2.0-or-later
+ */
+
+#include "qemu/osdep.h"
+#include "qemu/log.h"
+#include "hw/pci/msi.h"
+#include "hw/pci/pcie.h"
+#include "hw/pci/pcie_port.h"
+
+#define CXL_UPSTREAM_PORT_MSI_NR_VECTOR 1
+
+#define CXL_UPSTREAM_PORT_MSI_OFFSET 0x70
+#define CXL_UPSTREAM_PORT_PCIE_CAP_OFFSET 0x90
+#define CXL_UPSTREAM_PORT_AER_OFFSET 0x100
+#define CXL_UPSTREAM_PORT_DVSEC_OFFSET \
+    (CXL_UPSTREAM_PORT_AER_OFFSET + PCI_ERR_SIZEOF)
+
+typedef struct CXLUpstreamPort {
+    /*< private >*/
+    PCIEPort parent_obj;
+
+    /*< public >*/
+    CXLComponentState cxl_cstate;
+} CXLUpstreamPort;
+
+CXLComponentState *cxl_usp_to_cstate(CXLUpstreamPort *usp)
+{
+    return &usp->cxl_cstate;
+}
+
+static void cxl_usp_dvsec_write_config(PCIDevice *dev, uint32_t addr,
+                                       uint32_t val, int len)
+{
+    CXLUpstreamPort *usp = CXL_USP(dev);
+
+    if (range_contains(&usp->cxl_cstate.dvsecs[EXTENSIONS_PORT_DVSEC], addr)) {
+        uint8_t *reg = &dev->config[addr];
+        addr -= usp->cxl_cstate.dvsecs[EXTENSIONS_PORT_DVSEC].lob;
+        if (addr == PORT_CONTROL_OFFSET) {
+            if (pci_get_word(reg) & PORT_CONTROL_UNMASK_SBR) {
+                /* unmask SBR */
+                qemu_log_mask(LOG_UNIMP, "SBR mask control is not supported\n");
+            }
+            if (pci_get_word(reg) & PORT_CONTROL_ALT_MEMID_EN) {
+                /* Alt Memory & ID Space Enable */
+                qemu_log_mask(LOG_UNIMP,
+                              "Alt Memory & ID space is not supported\n");
+            }
+        }
+    }
+}
+
+static void cxl_usp_write_config(PCIDevice *d, uint32_t address,
+                                 uint32_t val, int len)
+{
+    pci_bridge_write_config(d, address, val, len);
+    pcie_cap_flr_write_config(d, address, val, len);
+    pcie_aer_write_config(d, address, val, len);
+
+    cxl_usp_dvsec_write_config(d, address, val, len);
+}
+
+static void latch_registers(CXLUpstreamPort *usp)
+{
+    uint32_t *reg_state = usp->cxl_cstate.crb.cache_mem_registers;
+    uint32_t *write_msk = usp->cxl_cstate.crb.cache_mem_regs_write_mask;
+
+    cxl_component_register_init_common(reg_state, write_msk,
+                                       CXL2_UPSTREAM_PORT);
+    ARRAY_FIELD_DP32(reg_state, CXL_HDM_DECODER_CAPABILITY, TARGET_COUNT, 8);
+}
+
+static void cxl_usp_reset(DeviceState *qdev)
+{
+    PCIDevice *d = PCI_DEVICE(qdev);
+    CXLUpstreamPort *usp = CXL_USP(qdev);
+
+    pci_bridge_reset(qdev);
+    pcie_cap_deverr_reset(d);
+    latch_registers(usp);
+}
+
+static void build_dvsecs(CXLComponentState *cxl)
+{
+    uint8_t *dvsec;
+
+    dvsec = (uint8_t *)&(CXLDVSECPortExtensions){
+        .status = 0x1, /* Port Power Management Init Complete */
+    };
+    cxl_component_create_dvsec(cxl, CXL2_UPSTREAM_PORT,
+                               EXTENSIONS_PORT_DVSEC_LENGTH,
+                               EXTENSIONS_PORT_DVSEC,
+                               EXTENSIONS_PORT_DVSEC_REVID, dvsec);
+    dvsec = (uint8_t *)&(CXLDVSECPortFlexBus){
+        .cap                     = 0x27, /* Cache, IO, Mem, non-MLD */
+        .ctrl                    = 0x27, /* Cache, IO, Mem */
+        .status                  = 0x26, /* same */
+        .rcvd_mod_ts_data_phase1 = 0xef, /* WTF? */
+    };
+    cxl_component_create_dvsec(cxl, CXL2_UPSTREAM_PORT,
+                               PCIE_FLEXBUS_PORT_DVSEC_LENGTH_2_0,
+                               PCIE_FLEXBUS_PORT_DVSEC,
+                               PCIE_FLEXBUS_PORT_DVSEC_REVID_2_0, dvsec);
+
+    dvsec = (uint8_t *)&(CXLDVSECRegisterLocator){
+        .rsvd         = 0,
+        .reg0_base_lo = RBI_COMPONENT_REG | CXL_COMPONENT_REG_BAR_IDX,
+        .reg0_base_hi = 0,
+    };
+    cxl_component_create_dvsec(cxl, CXL2_UPSTREAM_PORT,
+                               REG_LOC_DVSEC_LENGTH, REG_LOC_DVSEC,
+                               REG_LOC_DVSEC_REVID, dvsec);
+}
+
+static void cxl_usp_realize(PCIDevice *d, Error **errp)
+{
+    PCIEPort *p = PCIE_PORT(d);
+    CXLUpstreamPort *usp = CXL_USP(d);
+    CXLComponentState *cxl_cstate = &usp->cxl_cstate;
+    ComponentRegisters *cregs = &cxl_cstate->crb;
+    MemoryRegion *component_bar = &cregs->component_registers;
+    int rc;
+
+    pci_bridge_initfn(d, TYPE_PCIE_BUS);
+    pcie_port_init_reg(d);
+
+    rc = msi_init(d, CXL_UPSTREAM_PORT_MSI_OFFSET,
+                  CXL_UPSTREAM_PORT_MSI_NR_VECTOR, true, true, errp);
+    if (rc) {
+        assert(rc == -ENOTSUP);
+        goto err_bridge;
+    }
+
+    rc = pcie_cap_init(d, CXL_UPSTREAM_PORT_PCIE_CAP_OFFSET,
+                       PCI_EXP_TYPE_UPSTREAM, p->port, errp);
+    if (rc < 0) {
+        goto err_msi;
+    }
+
+    pcie_cap_flr_init(d);
+    pcie_cap_deverr_init(d);
+    rc = pcie_aer_init(d, PCI_ERR_VER, CXL_UPSTREAM_PORT_AER_OFFSET,
+                       PCI_ERR_SIZEOF, errp);
+    if (rc) {
+        goto err_cap;
+    }
+
+    cxl_cstate->dvsec_offset = CXL_UPSTREAM_PORT_DVSEC_OFFSET;
+    cxl_cstate->pdev = d;
+    build_dvsecs(cxl_cstate);
+    cxl_component_register_block_init(OBJECT(d), cxl_cstate, TYPE_CXL_USP);
+    pci_register_bar(d, CXL_COMPONENT_REG_BAR_IDX,
+                     PCI_BASE_ADDRESS_SPACE_MEMORY |
+                     PCI_BASE_ADDRESS_MEM_TYPE_64,
+                     component_bar);
+
+    return;
+
+err_cap:
+    pcie_cap_exit(d);
+err_msi:
+    msi_uninit(d);
+err_bridge:
+    pci_bridge_exitfn(d);
+}
+
+static void cxl_usp_exitfn(PCIDevice *d)
+{
+    pcie_aer_exit(d);
+    pcie_cap_exit(d);
+    msi_uninit(d);
+    pci_bridge_exitfn(d);
+}
+
+static void cxl_upstream_class_init(ObjectClass *oc, void *data)
+{
+    DeviceClass *dc = DEVICE_CLASS(oc);
+    PCIDeviceClass *k = PCI_DEVICE_CLASS(oc);
+
+    k->is_bridge = true;
+    k->config_write = cxl_usp_write_config;
+    k->realize = cxl_usp_realize;
+    k->exit = cxl_usp_exitfn;
+    k->vendor_id = 0x19e5; /* Huawei */
+    k->device_id = 0xa128; /* Emulated CXL Switch Upstream Port */
+    k->revision = 0;
+    set_bit(DEVICE_CATEGORY_BRIDGE, dc->categories);
+    dc->desc = "CXL Switch Upstream Port";
+    dc->reset = cxl_usp_reset;
+}
+
+static const TypeInfo cxl_usp_info = {
+    .name = TYPE_CXL_USP,
+    .parent = TYPE_PCIE_PORT,
+    .instance_size = sizeof(CXLUpstreamPort),
+    .class_init = cxl_upstream_class_init,
+    .interfaces = (InterfaceInfo[]) {
+        { INTERFACE_PCIE_DEVICE },
+        { INTERFACE_CXL_DEVICE },
+        { }
+    },
+};
+
+static void cxl_usp_register_type(void)
+{
+    type_register_static(&cxl_usp_info);
+}
+
+type_init(cxl_usp_register_type);
hw/pci-bridge/meson.build +1 -1
index b6d26a03d5..4a8c26b5a1 100644
@@ -5,7 +5,7 @@ pci_ss.add(when: 'CONFIG_IOH3420', if_true: files('ioh3420.c'))
 pci_ss.add(when: 'CONFIG_PCIE_PORT', if_true: files('pcie_root_port.c', 'gen_pcie_root_port.c', 'pcie_pci_bridge.c'))
 pci_ss.add(when: 'CONFIG_PXB', if_true: files('pci_expander_bridge.c'))
 pci_ss.add(when: 'CONFIG_XIO3130', if_true: files('xio3130_upstream.c', 'xio3130_downstream.c'))
-pci_ss.add(when: 'CONFIG_CXL', if_true: files('cxl_root_port.c'))
+pci_ss.add(when: 'CONFIG_CXL', if_true: files('cxl_root_port.c', 'cxl_upstream.c'))
 
 # NewWorld PowerMac
 pci_ss.add(when: 'CONFIG_DEC_PCI', if_true: files('dec.c'))
include/hw/cxl/cxl.h +4 -0
index 21d28ca110..ff7b296b9f 100644
@@ -57,5 +57,9 @@ void cxl_fixed_memory_window_config(MachineState *ms,
 void cxl_fixed_memory_window_link_targets(Error **errp);
 
 extern const MemoryRegionOps cfmws_ops;
+#define TYPE_CXL_USP "cxl-upstream"
 
+typedef struct CXLUpstreamPort CXLUpstreamPort;
+DECLARE_INSTANCE_CHECKER(CXLUpstreamPort, CXL_USP, TYPE_CXL_USP)
+CXLComponentState *cxl_usp_to_cstate(CXLUpstreamPort *usp);
 #endif
Signature · 1 line
2.32.0
#49 lore raw

PATCH v10 44/45 pci-bridge/cxl_downstream: Add a CXL switch downstream port

In reply to Jonathan Cameron

Emulation of a simple CXL Switch downstream port.
The Device ID has been allocated for this use.

---
 hw/cxl/cxl-host.c              |  43 +++++-
 hw/pci-bridge/cxl_downstream.c | 249 +++++++++++++++++++++++++++++++++
 hw/pci-bridge/meson.build      |   2 +-
 3 files changed, 291 insertions(+), 3 deletions(-)
hw/cxl/cxl-host.c +41 -2
index 469b3c4ced..317f5a37ca 100644
@@ -130,8 +130,9 @@ static bool cxl_hdm_find_target(uint32_t *cache_mem, hwaddr addr,
 
 static PCIDevice *cxl_cfmws_find_device(CXLFixedWindow *fw, hwaddr addr)
 {
-    CXLComponentState *hb_cstate;
+    CXLComponentState *hb_cstate, *usp_cstate;
     PCIHostState *hb;
+    CXLUpstreamPort *usp;
     int rb_index;
     uint32_t *cache_mem;
     uint8_t target;
@@ -165,8 +166,46 @@ static PCIDevice *cxl_cfmws_find_device(CXLFixedWindow *fw, hwaddr addr)
     }
 
     d = pci_bridge_get_sec_bus(PCI_BRIDGE(rp))->devices[0];
+    if (!d) {
+        return NULL;
+    }
+
+    if (object_dynamic_cast(OBJECT(d), TYPE_CXL_TYPE3)) {
+        return d;
+    }
+
+    /*
+     * Could also be a switch.  Note only one level of switching currently
+     * supported.
+     */
+    if (!object_dynamic_cast(OBJECT(d), TYPE_CXL_USP)) {
+        return NULL;
+    }
+    usp = CXL_USP(d);
+
+    usp_cstate = cxl_usp_to_cstate(usp);
+    if (!usp_cstate) {
+        return NULL;
+    }
+
+    cache_mem = usp_cstate->crb.cache_mem_registers;
+
+    target_found = cxl_hdm_find_target(cache_mem, addr, &target);
+    if (!target_found) {
+        return NULL;
+    }
+
+    d = pcie_find_port_by_pn(&PCI_BRIDGE(d)->sec_bus, target);
+    if (!d) {
+        return NULL;
+    }
+
+    d = pci_bridge_get_sec_bus(PCI_BRIDGE(d))->devices[0];
+    if (!d) {
+        return NULL;
+    }
 
-    if (!d || !object_dynamic_cast(OBJECT(d), TYPE_CXL_TYPE3)) {
+    if (!object_dynamic_cast(OBJECT(d), TYPE_CXL_TYPE3)) {
         return NULL;
     }
 
hw/pci-bridge/cxl_downstream.c new file +249 -0

249 lines not on this page · raw message

hw/pci-bridge/meson.build +1 -1
index 4a8c26b5a1..53cc53314a 100644
@@ -5,7 +5,7 @@ pci_ss.add(when: 'CONFIG_IOH3420', if_true: files('ioh3420.c'))
 pci_ss.add(when: 'CONFIG_PCIE_PORT', if_true: files('pcie_root_port.c', 'gen_pcie_root_port.c', 'pcie_pci_bridge.c'))
 pci_ss.add(when: 'CONFIG_PXB', if_true: files('pci_expander_bridge.c'))
 pci_ss.add(when: 'CONFIG_XIO3130', if_true: files('xio3130_upstream.c', 'xio3130_downstream.c'))
-pci_ss.add(when: 'CONFIG_CXL', if_true: files('cxl_root_port.c', 'cxl_upstream.c'))
+pci_ss.add(when: 'CONFIG_CXL', if_true: files('cxl_root_port.c', 'cxl_upstream.c', 'cxl_downstream.c'))
 
 # NewWorld PowerMac
 pci_ss.add(when: 'CONFIG_DEC_PCI', if_true: files('dec.c'))
Signature · 1 line
2.32.0
#50 lore raw

PATCH v10 45/45 docs/cxl: Add switch documentation

In reply to Jonathan Cameron

Switches were already introduced, but now we support them update the documentation to provide an example in diagram and qemu command line parameter forms.

---
 docs/system/devices/cxl.rst | 88 ++++++++++++++++++++++++++++++++++++-
 1 file changed, 86 insertions(+), 2 deletions(-)
docs/system/devices/cxl.rst +86 -2
index 9293cbf01a..abf7c1f243 100644
@@ -118,8 +118,6 @@ and associated component register access via PCI bars.
 
 CXL Switch
 ~~~~~~~~~~
-Not yet implemented in QEMU.
-
 Here we consider a simple CXL switch with only a single
 virtual hierarchy. Whilst more complex devices exist, their
 visibility to a particular host is generally the same as for
@@ -137,6 +135,10 @@ BARs.  The Upstream Port has the configuration interfaces for
 the HDM decoders which route incoming memory accesses to the
 appropriate downstream port.
 
+A CXL switch is created in a similar fashion to PCI switches
+by creating an upstream port (cxl-upstream) and a number of
+downstream ports on the internal switch bus (cxl-downstream).
+
 CXL Memory Devices - Type 3
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~
 CXL type 3 devices use a PCI class code and are intended to be supported
@@ -240,6 +242,62 @@ Notes:
     they will take the Host Physical Addresses of accesses and map
     them to their own local Device Physical Address Space (DPA).
 
+Example topology involving a switch::
+
+  |<------------------SYSTEM PHYSICAL ADDRESS MAP (1)----------------->|
+  |    __________   __________________________________   __________    |
+  |   |          | |                                  | |          |   |
+  |   | CFMW 0   | |  CXL Fixed Memory Window 1       | | CFMW 1   |   |
+  |   | HB0 only | |  Configured to interleave memory | | HB1 only |   |
+  |   |          | |  memory accesses across HB0/HB1  | |          |   |
+  |   |____x_____| |__________________________________| |__________|   |
+           |             |                     |             |
+           |             |                     |             |
+           |             |                     |
+  Interleave Decoder     |                     |             |
+   Matches this HB       |                     |             |
+           \_____________|                     |_____________/
+               __________|__________      _____|_______________
+              |                     |    |                     |
+              | CXL HB 0            |    | CXL HB 1            |
+              | HB IntLv Decoders   |    | HB IntLv Decoders   |
+              | PCI/CXL Root Bus 0c |    | PCI/CXL Root Bus 0d |
+              |                     |    |                     |
+              |___x_________________|    |_____________________|
+                  |              |          |               |
+                  |
+       A HB 0 HDM Decoder
+       matches this Port
+       ___________|___
+      |  Root Port 0  |
+      |  Appears in   |
+      |  PCI topology |
+      |  As 0c:00.0   |
+      |___________x___|
+                  |
+                  |
+                  \_____________________
+                                        |
+                                        |
+            ---------------------------------------------------
+           |    Switch 0  USP as PCI 0d:00.0                   |
+           |    USP has HDM decoder which direct traffic to    |
+           |    appropiate downstream port                     |
+           |    Switch BUS appears as 0e                       |
+           |x__________________________________________________|
+            |                  |               |              |
+            |                  |               |              |
+       _____|_________   ______|______   ______|_____   ______|_______
+   (4)|     x         | |             | |            | |              |
+      | CXL Type3 0   | | CXL Type3 1 | | CXL type3 2| | CLX Type 3 3 |
+      |               | |             | |            | |              |
+      | PMEM0(Vol LSA)| | PMEM1 (...) | | PMEM2 (...)| | PMEM3 (...)  |
+      | Decoder to go | |             | |            | |              |
+      | from host PA  | | PCI 10:00.0 | | PCI 11:00.0| | PCI 12:00.0  |
+      | to device PA  | |             | |            | |              |
+      | PCI as 0f:00.0| |             | |            | |              |
+      |_______________| |_____________| |____________| |______________|
+
 Example command lines
 ---------------------
 A very simple setup with just one directly attached CXL Type 3 device::
@@ -279,6 +337,32 @@ the CXL Type3 device directly attached (no switches).::
   -device cxl-type3,bus=root_port16,memdev=cxl-mem4,lsa=cxl-lsa4,id=cxl-pmem3 \
   -cxl-fixed-memory-window targets.0=cxl.1,targets.1=cxl.2,size=4G,interleave-granularity=8k
 
+An example of 4 devices below a switch suitable for 1, 2 or 4 way interleave::
+
+  qemu-system-aarch64 -M virt,gic-version=3,cxl=on -m 4g,maxmem=8G,slots=8 -cpu max \
+  ...
+  -object memory-backend-file,id=cxl-mem0,share=on,mem-path=/tmp/cxltest.raw,size=256M \
+  -object memory-backend-file,id=cxl-mem1,share=on,mem-path=/tmp/cxltest1.raw,size=256M \
+  -object memory-backend-file,id=cxl-mem2,share=on,mem-path=/tmp/cxltest2.raw,size=256M \
+  -object memory-backend-file,id=cxl-mem3,share=on,mem-path=/tmp/cxltest3.raw,size=256M \
+  -object memory-backend-file,id=cxl-lsa0,share=on,mem-path=/tmp/lsa0.raw,size=256M \
+  -object memory-backend-file,id=cxl-lsa1,share=on,mem-path=/tmp/lsa1.raw,size=256M \
+  -object memory-backend-file,id=cxl-lsa2,share=on,mem-path=/tmp/lsa2.raw,size=256M \
+  -object memory-backend-file,id=cxl-lsa3,share=on,mem-path=/tmp/lsa3.raw,size=256M \
+  -device pxb-cxl,bus_nr=12,bus=pcie.0,id=cxl.1 \
+  -device cxl-rp,port=0,bus=cxl.1,id=root_port0,chassis=0,slot=0 \
+  -device cxl-rp,port=1,bus=cxl.1,id=root_port1,chassis=0,slot=1 \
+  -device cxl-upstream,bus=root_port0,id=us0 \
+  -device cxl-downstream,port=0,bus=us0,id=swport0,chassis=0,slot=4 \
+  -device cxl-type3,bus=swport0,memdev=cxl-mem0,lsa=cxl-lsa0,id=cxl-pmem0,size=256M \
+  -device cxl-downstream,port=1,bus=us0,id=swport1,chassis=0,slot=5 \
+  -device cxl-type3,bus=swport1,memdev=cxl-mem1,lsa=cxl-lsa1,id=cxl-pmem1,size=256M \
+  -device cxl-downstream,port=2,bus=us0,id=swport2,chassis=0,slot=6 \
+  -device cxl-type3,bus=swport2,memdev=cxl-mem2,lsa=cxl-lsa2,id=cxl-pmem2,size=256M \
+  -device cxl-downstream,port=3,bus=us0,id=swport3,chassis=0,slot=7 \
+  -device cxl-type3,bus=swport3,memdev=cxl-mem3,lsa=cxl-lsa3,id=cxl-pmem3,size=256M \
+  -cxl-fixed-memory-window targets.0=cxl.1,size=4G,interleave-granularity=4k
+
 Kernel Configuration Options
 ----------------------------
 
Signature · 1 line
2.32.0
#51 lore raw
In reply to Jonathan Cameron
Quoted 26 lines · Jonathan Cameron

Note that due to a bug in the kernel handling of Mem_Enable in the CXL Device DVSEC a fix is needed to sucessfully use these patches with Linux.
https://lore.kernel.org/linux-cxl/CAPcyv4gFWw9s6r9kUoL-_P9qnaZSJBMz+EcmfY+0g1NHrJQx7Q@mail.gmail.com/T/#t

Note the issue was detected using this emulation.

Changes since v9:

Thanks to Tong Zhang and Markus Armbruster for review of v9.
- Fixed return of all 1s on HDM decoder not matching.
- Off by one error in an assert for register writes.
- Write masks for both PCI config space structures and registers in MMIO space. These were partly queued up for a future patch set, but after review comment from Tong, I decided it was best to put these in place from the start. Note I was a bit undecided on whether to drop review tags for this (it's quit a bit of code, but very simple). Alex, and Adam if you want to take another look at patch 2 in particular that would be great.
- Tag from Markus added.
- Fix for wrong setting of Mem_Enable as described in the thread linked above.

The series can be applied in parts if that makes sense:

Quoted 7 lines

Possible partial sets:
1-15 (end with the test of the pxb-cxl host bridge)
16-22 (end with the test for root port and type3 device)
23-39 (end with tests on x86 pc for CFMWS including BIOS table updates) 40-41 (arm64 virt support + simple test case)
42 (documentation - we could pull this forwards to before the arm support) 43-45 (switch support)

OK, I picked up 1-39 and 42. Let's see how it goes, then arm people can apply the arm bits.
I did my best with review but my knowledge is limited.

Quoted 214 lines

gitlab tree at gitlab.com/jic23/qemu cxl-v10-draft-2

As before that tree has additional patches on top that will form part of future postings. This time it has SPDM support via a socket connection to DMTF/spdm_emu / libspdm. I'll send an RFC on that topic as it's a discussion that can go on in parallel.

Several people have asked about contributing additional features. As those come in I'll apply them on top of this series and handle rebases etc as necessary whilst we seek to get this first set of patches upstream.

Updated background info:

Looking in particular for:
* Review of the PCI interactions
* x86 and ARM machine interactions (particularly the memory maps)
* Review of the interleaving approach - is the basic idea acceptable?
* Review of the command line interface.
* CXL related review welcome but much of that got reviewed in earlier versions and hasn't changed substantially.

TODOs:

* Volatile memory devices (easy but it's more code so left for now).
* Hotplug? May not need much but it's not tested yet!
* More tests and tighter verification that values written to hardware are actually valid - stuff that real hardware would check.
* Testing, testing and more testing. I have been running a basic set of ARM and x86 tests on this, but there is always room for more tests and greater automation.
* CFMWS flags as requested by Ben.
* Parititioning support - ability to change the balance of volatile and non volatile memory on demand.
* Trace points as suggested my Mark to help with debugging memory interleaving setup.

Why do we want QEMU emulation of CXL?

As Ben stated in V3, QEMU support has been critical to getting OS software written given lack of availability of hardware supporting the latest CXL features (coupled with very high demand for support being ready in a timely fashion). What has become clear since Ben's v3 is that situation is a continuous one. Whilst we can't talk about them yet, CXL 3.0 features and OS support have been prototyped on top of this support and a lot of the ongoing kernel work is being tested against these patches. The kernel CXL mocking code allows some forms of testing, but QEMU provides a more versatile and extensible platform.

Other features on the qemu-list that build on these include PCI-DOE
/CDAT support from the Avery Design team further showing how this
code is useful. Whilst not directly related this is also the test
platform for work on PCI IDE/CMA + related DMTF SPDM as CXL both
utilizes and extends those technologies and is likely to be an early
adopter.
Refs:
CMA Kernel: https://lore.kernel.org/all/[email protected]/
CMA Qemu: https://lore.kernel.org/qemu-devel/[email protected]/
DOE Qemu: https://lore.kernel.org/qemu-devel/[email protected]/

As can be seen there is non trivial interaction with other areas of Qemu, particularly PCI and keeping this set up to date is proving a burden we'd rather do without :)

Ben mentioned a few other good reasons in v3:
https://lore.kernel.org/qemu-devel/[email protected]/

What we have here is about what you need for it to be useful for testing currently kernel code. Note the kernel code is moving fast so since v4, some features have been introduced we don't yet support in QEMU (e.g. use of the PCIe serial number extended capability).

All comments welcome.

Additional info that was here in v5 is now in the documentation patch.

Thanks,

Jonathan

Ben Widawsky (24):
hw/pci/cxl: Add a CXL component type (interface) hw/cxl/component: Introduce CXL components (8.1.x, 8.2.5) hw/cxl/device: Introduce a CXL device (8.2.8) hw/cxl/device: Implement the CAP array (8.2.8.1-2) hw/cxl/device: Implement basic mailbox (8.2.8.4) hw/cxl/device: Add memory device utilities hw/cxl/device: Add cheap EVENTS implementation (8.2.9.1) hw/cxl/device: Timestamp implementation (8.2.9.3) hw/cxl/device: Add log commands (8.2.9.4) + CEL hw/pxb: Use a type for realizing expanders hw/pci/cxl: Create a CXL bus type
hw/pxb: Allow creation of a CXL PXB (host bridge) hw/cxl/rp: Add a root port
hw/cxl/device: Add a memory device (8.2.8.5) hw/cxl/device: Implement MMIO HDM decoding (8.2.5.12) hw/cxl/device: Add some trivial commands hw/cxl/device: Plumb real Label Storage Area (LSA) sizing hw/cxl/device: Implement get/set Label Storage Area (LSA) hw/cxl/component: Implement host bridge MMIO (8.2.5, table 142) acpi/cxl: Add _OSC implementation (9.14.2)
acpi/cxl: Create the CEDT (9.14.1)
acpi/cxl: Introduce CFMWS structures in CEDT hw/cxl/component Add a dumb HDM decoder handler qtest/cxl: Add more complex test cases with CFMWs

Jonathan Cameron (21):
MAINTAINERS: Add entry for Compute Express Link Emulation
cxl: Machine level control on whether CXL support is enabled qtest/cxl: Introduce initial test for pxb-cxl only. qtests/cxl: Add initial root port and CXL type3 tests hw/cxl/component: Add utils for interleave parameter encoding/decoding hw/cxl/host: Add support for CXL Fixed Memory Windows. hw/pci-host/gpex-acpi: Add support for dsdt construction for pxb-cxl pci/pcie_port: Add pci_find_port_by_pn() CXL/cxl_component: Add cxl_get_hb_cstate()
mem/cxl_type3: Add read and write functions for associated hostmem. cxl/cxl-host: Add memops for CFMWS region.
i386/pc: Enable CXL fixed memory windows
tests/acpi: q35: Allow addition of a CXL test. qtests/bios-tables-test: Add a test for CXL emulation. tests/acpi: Add tables for CXL emulation.
hw/arm/virt: Basic CXL enablement on pci_expander_bridge instances
pxb-cxl
qtest/cxl: Add aarch64 virt test for CXL
docs/cxl: Add initial Compute eXpress Link (CXL) documentation. pci-bridge/cxl_upstream: Add a CXL switch upstream port pci-bridge/cxl_downstream: Add a CXL switch downstream port docs/cxl: Add switch documentation

 MAINTAINERS                         |   7 +
 docs/system/device-emulation.rst    |   1 +
 docs/system/devices/cxl.rst         | 386 ++++++++++++++++++++++
 hw/Kconfig                          |   1 +
 hw/acpi/Kconfig                     |   5 +
 hw/acpi/cxl-stub.c                  |  12 +
 hw/acpi/cxl.c                       | 257 +++++++++++++++
 hw/acpi/meson.build                 |   4 +-
 hw/arm/Kconfig                      |   1 +
 hw/arm/virt-acpi-build.c            |  33 ++
 hw/arm/virt.c                       |  40 ++-
 hw/core/machine.c                   |  28 ++
 hw/cxl/Kconfig                      |   3 +
 hw/cxl/cxl-component-utils.c        | 396 +++++++++++++++++++++++
 hw/cxl/cxl-device-utils.c           | 265 +++++++++++++++
 hw/cxl/cxl-host-stubs.c             |  16 +
 hw/cxl/cxl-host.c                   | 261 +++++++++++++++
 hw/cxl/cxl-mailbox-utils.c          | 478 ++++++++++++++++++++++++++++
 hw/cxl/meson.build                  |  12 +
 hw/i386/acpi-build.c                |  57 +++-
 hw/i386/pc.c                        |  57 +++-
 hw/mem/Kconfig                      |   5 +
 hw/mem/cxl_type3.c                  | 371 +++++++++++++++++++++
 hw/mem/meson.build                  |   1 +
 hw/meson.build                      |   1 +
 hw/pci-bridge/Kconfig               |   5 +
 hw/pci-bridge/cxl_downstream.c      | 249 +++++++++++++++
 hw/pci-bridge/cxl_root_port.c       | 236 ++++++++++++++
 hw/pci-bridge/cxl_upstream.c        | 216 +++++++++++++
 hw/pci-bridge/meson.build           |   1 +
 hw/pci-bridge/pci_expander_bridge.c | 168 +++++++++-
 hw/pci-bridge/pcie_root_port.c      |   6 +-
 hw/pci-host/gpex-acpi.c             |  20 +-
 hw/pci/pci.c                        |  21 +-
 hw/pci/pcie_port.c                  |  25 ++
 include/hw/acpi/cxl.h               |  28 ++
 include/hw/arm/virt.h               |   1 +
 include/hw/boards.h                 |   2 +
 include/hw/cxl/cxl.h                |  65 ++++
 include/hw/cxl/cxl_component.h      | 223 +++++++++++++
 include/hw/cxl/cxl_device.h         | 268 ++++++++++++++++
 include/hw/cxl/cxl_pci.h            | 167 ++++++++++
 include/hw/pci/pci.h                |  14 +
 include/hw/pci/pci_bridge.h         |  20 ++
 include/hw/pci/pci_bus.h            |   7 +
 include/hw/pci/pci_ids.h            |   1 +
 include/hw/pci/pcie_port.h          |   2 +
 qapi/machine.json                   |  21 ++
 qemu-options.hx                     |  38 +++
 scripts/device-crash-test           |   1 +
 softmmu/vl.c                        |  47 +++
 tests/data/acpi/q35/CEDT.cxl        | Bin 0 -> 184 bytes
 tests/data/acpi/q35/DSDT.cxl        | Bin 0 -> 9615 bytes
 tests/qtest/bios-tables-test.c      |  44 +++
 tests/qtest/cxl-test.c              | 181 +++++++++++
 tests/qtest/meson.build             |   5 +
 56 files changed, 4753 insertions(+), 27 deletions(-)
 create mode 100644 docs/system/devices/cxl.rst
 create mode 100644 hw/acpi/cxl-stub.c
 create mode 100644 hw/acpi/cxl.c
 create mode 100644 hw/cxl/Kconfig
 create mode 100644 hw/cxl/cxl-component-utils.c
 create mode 100644 hw/cxl/cxl-device-utils.c
 create mode 100644 hw/cxl/cxl-host-stubs.c
 create mode 100644 hw/cxl/cxl-host.c
 create mode 100644 hw/cxl/cxl-mailbox-utils.c
 create mode 100644 hw/cxl/meson.build
 create mode 100644 hw/mem/cxl_type3.c
 create mode 100644 hw/pci-bridge/cxl_downstream.c
 create mode 100644 hw/pci-bridge/cxl_root_port.c
 create mode 100644 hw/pci-bridge/cxl_upstream.c
 create mode 100644 include/hw/acpi/cxl.h
 create mode 100644 include/hw/cxl/cxl.h
 create mode 100644 include/hw/cxl/cxl_component.h
 create mode 100644 include/hw/cxl/cxl_device.h
 create mode 100644 include/hw/cxl/cxl_pci.h
 create mode 100644 tests/data/acpi/q35/CEDT.cxl
 create mode 100644 tests/data/acpi/q35/DSDT.cxl
 create mode 100644 tests/qtest/cxl-test.c
Signature · 1 line
2.32.0