arm/virt: CXL support via pxb_cxl
v14 cover arm/virt: CXL support via pxb_cxl
v14: Simplifications suggeseted by Itaru (and some extra simplifications
that became apparent) and gather tags.
See individual patches for more information.
Updated cover letter
Richard Henderson has posted a pull request with a fix for the TCG TLB issue which will hopefully merge shortly (Thanks Richard!).
Back in 2022, this series stalled on the absence of a solution to device tree support for PCI Expander Bridges (PXB) and we ended up only having x86 support upstream. I've been carrying the arm64 support out of tree since then, with occasional nasty surprises (e.g. UNIMP + DT issue seen a few weeks ago) and a fair number of fiddly rebases. gitlab.com/jic23/qemu cxl-<latest date>. Will update shortly with this series.
A recent discussion with Peter Maydell indicated that there are various other ACPI only features now, so in general he might be more relaxed about DT support being necessary. The upcoming vSMMUv3 support would run into this problem as well.
I presented the background to the PXB issue at Linaro connect 2022. In short the issue is that PXBs steal MMIO space from the main PCI root bridge. The challenge is knowing how much to steal.
On ACPI platforms, we can rely on EDK2 to perform an enumeration and configuration of the PCI topology and QEMU can update the ACPI tables after EDK2 has done this when it can simply read the space used by the root ports. On device tree, there is no entity to figure out that enumeration so we don't know how to size the stolen region.
Three approaches were discussed:
1) Enumerating in QEMU. Horribly complex and the last thing we want is a 3rd enumeration implementation that ends up out of sync with EDK2 and the kernel (there are frequent issues because of how those existing implementations differ.
2) Figure out how to enumerate in kernel. I never put a huge amount of work into this, but it seemed likely to involve a nasty dance with similar very specific code to that EDK2 is carrying and would very challenging to upstream (given the lack of clarity on real use cases for PXBs and DT).
3) Hack it based on the control we have which is bus numbers. No one liked this but it worked :)
The other little wrinkle would be the need to define full bindings for CXL on DT + implement a fairly complex kernel stack as equivalent in ACPI involves a static table, CEDT, new runtime queries via _DSM and a description of various components. Doable, but so far there is no interest on physical platforms. Worth noting that for now, the QEMU CXL emulation is all about testing and developing the OS stack, not about virtualization (performance is terrible except in some very contrived situations!)
There is only a very simple test in here, because my intent is not to duplicate what we have on x86, but just to do a smoke test that everything is hooked up. In general we need much more comprehensive end to end CXL tests but that requires a reaonsably stable guest software stack. A few people have expressed interest in working on that, but we aren't there yet.
Note that this series has a very different use case to that in the proposed SBSA-ref support:
https://lore.kernel.org/qemu-devel/[email protected]/
SBSA-ref is a good choice if you want a relatively simple mostly fixed configuration. That works well with the limited host system discoverability etc as EDK2 can be build against a known configuration.
My interest with this support in arm/virt is support host software stack development (we have a wide range of contributors, most of whom are working on emulation + the kernel support). I care about the weird corners. As such I need to be able to bring up variable numbers of host bridges, multiple CXL Fixed Memory Windows with varying characteristics (interleave etc), complex NUMA topologies with wierd performance characteristics etc. We can do that on x86 upstream today, or my gitlab tree. Note that we need arm support for some arch specific features in the near future (cache flushing). Doing kernel development with this need for flexibility on SBSA-ref is not currently practical. SBSA-ref CXL support is an excellent thing, just not much use to me for this work.
Also, we are kicking off some work on DCD virtualization, particularly to support inter-host shared memory being presented up into a VM. That will need upstream support on arm64 as it is built on top of the existing CXL emulation to avoid the need for a separate guest software stack.
Note this is TCG only - it is possible to support limited use with KVM but that needs additional patches not yet ready for upstream. The challenge is interleave - and the solution is don't interleave if you want to run with KVM.
Jonathan Cameron (5):
hw/cxl-host: Add an index field to CXLFixedMemoryWindow hw/cxl: Make the CXL fixed memory windows devices. hw/cxl-host: Allow split of establishing memory address and mmio
setup.
hw/arm/virt: Basic CXL enablement on pci_expander_bridge instances
pxb-cxl
qtest/cxl: Add aarch64 virt test for CXL
include/hw/arm/virt.h | 4 + include/hw/cxl/cxl.h | 4 + include/hw/cxl/cxl_host.h | 6 +- hw/acpi/cxl.c | 76 +++++++-------- hw/arm/virt-acpi-build.c | 34 +++++++ hw/arm/virt.c | 29 ++++++ hw/cxl/cxl-host-stubs.c | 8 +- hw/cxl/cxl-host.c | 190 ++++++++++++++++++++++++++++++++------ hw/i386/pc.c | 51 +++++----- tests/qtest/cxl-test.c | 59 +++++++++--- tests/qtest/meson.build | 1 + 11 files changed, 353 insertions(+), 109 deletions(-)
Signature · 1 line
2.48.1
PATCH v14 1/5 hw/cxl-host: Add an index field to CXLFixedMemoryWindow
To enable these to be found in a fixed order, that order needs to be known. This will later be used to sort a list of these structures that address map and ACPI table entries are predictable.
---
v14: Picked up tags. Thanks!
--- include/hw/cxl/cxl.h | 1 + hw/cxl/cxl-host.c | 9 ++++++--- 2 files changed, 7 insertions(+), 3 deletions(-)
include/hw/cxl/cxl.h +1 -0
@@ -27,6 +27,7 @@typedef struct PXBCXLDev PXBCXLDev ; typedef struct CXLFixedWindow { + int index; uint64_t size; char **targets; PXBCXLDev *target_hbs[16];
hw/cxl/cxl-host.c +6 -3
@@ -24,13 +24,15 @@static void cxl_fixed_memory_window_config (CXLState *cxl_state,CXLFixedMemoryWindowOptions *object, -Error **errp) +int index,Error **errp) {ERRP_GUARD (); g_autofree CXLFixedWindow *fw = g_malloc0(sizeof(*fw));strList *target;int i; + fw->index = index; +for (target = object->targets; target; target = target->next) { fw->num_targets++; }
@@ -325,14 +327,15 @@ static void machine_set_cfmw(Object *obj, Visitor *v, const char *name,CXLState *state = opaque;CXLFixedMemoryWindowOptionsList *cfmw_list =NULL ;CXLFixedMemoryWindowOptionsList *it; +int index;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 (state, it->value, errp); +for (it = cfmw_list, index =0 ; it; it = it->next, index++) { +cxl_fixed_memory_window_config (state, it->value, index, errp); } state->cfmw_list = cfmw_list; }
Signature · 1 line
2.48.1
PATCH v14 2/5 hw/cxl: Make the CXL fixed memory windows devices.
In reply to Jonathan CameronPreviously these somewhat device like structures were tracked using a list in the CXLState in each machine. This is proving restrictive in a few cases where we need to iterate through these without being aware of the machine type. Just make them sysbus devices.
Restrict them to not user created as they need to be visible to early stages of machine init given effects on the memory map.
This change both simplifies state tracking and enables features needed for performance optimization and hotness tracking by making it possible to retrieve the fixed memory window on actions elsewhere in the topology.
In some cases the ordering of the Fixed Memory Windows matters. For those utility functions provide a GSList sorted by the window index. This ensures that we get consistency across:
- ordering in the command line
- ordering of the host PA ranges
- ordering of ACPI CEDT structures describing the CFMWS.
Other aspects don't have this constraint. For those direct iteration of the underlying hash structures is fine.
---
v14: Drop some unnecessary checks on device type. (Zhijian)
Use explicit pointer types given we know what they are at the caller.
- cedt_build_cfmws
- cxl_fmws_update
Cleanup a few things that are left overs from earlier approaches:
- Don't return unnecessarily at end of functions.
- Drop return values that are unused and always zero.
I think Peter Maydell suggested this a long time back when the original CXL support series was under review but not 100% sure.
--- include/hw/cxl/cxl.h | 3 + include/hw/cxl/cxl_host.h | 4 +- hw/acpi/cxl.c | 76 ++++++++++---------- hw/cxl/cxl-host-stubs.c | 6 +- hw/cxl/cxl-host.c | 148 ++++++++++++++++++++++++++++++-------- hw/i386/pc.c | 51 ++++++------- 6 files changed, 194 insertions(+), 94 deletions(-)
include/hw/cxl/cxl.h +3 -0
@@ -27,6 +27,7 @@typedef struct PXBCXLDev PXBCXLDev ; typedef struct CXLFixedWindow { + SysBusDevice parent_obj; int index; uint64_t size; char **targets;
@@ -38,6 +39,8 @@ typedef struct CXLFixedWindow {MemoryRegion mr;hwaddr base; } CXLFixedWindow; +#define TYPE_CXL_FMW "cxl-fmw" +OBJECT_DECLARE_SIMPLE_TYPE(CXLFixedWindow, CXL_FMW) typedef struct CXLState { bool is_enabled;
include/hw/cxl/cxl_host.h +3 -1
@@ -14,8 +14,10 @@#define CXL_HOST_H void cxl_machine_init (Object *obj,CXLState *state); -void cxl_fmws_link_targets (CXLState *stat,Error **errp); +void cxl_fmws_link_targets (Error **errp);void cxl_hook_up_pxb_registers (PCIBus *bus,CXLState *state,Error **errp); +hwaddr cxl_fmws_set_memmap_and_update_mmio (hwaddr base,hwaddr max_addr); +GSList *cxl_fmws_get_all_sorted (void );extern const MemoryRegionOps cfmws_ops;
hw/acpi/cxl.c +40 -36
114 lines not on this page · raw message
hw/cxl/cxl-host-stubs.c +5 -1
@@ -8,8 +8,12 @@#include "hw/cxl/cxl.h" #include "hw/cxl/cxl_host.h" -void cxl_fmws_link_targets (CXLState *stat,Error **errp) {}; +void cxl_fmws_link_targets (Error **errp) {};void cxl_machine_init (Object *obj,CXLState *state) {};void cxl_hook_up_pxb_registers (PCIBus *bus,CXLState *state,Error **errp) {}; +hwaddr cxl_fmws_set_memmap_and_update_mmio (hwaddr base,hwaddr max_addr) +{ +return base; +};const MemoryRegionOps cfmws_ops;
hw/cxl/cxl-host.c +120 -28
177 lines not on this page · raw message
hw/i386/pc.c +23 -28
73 lines not on this page · raw message
Signature · 1 line
2.48.1
Re: [PATCH v14 2/5] hw/cxl: Make the CXL fixed memory windows devices.
Quoted 19 lines · On Wed, 28 May 2025 12:07:23 +0100 Jonathan Cameron…
Previously these somewhat device like structures were tracked using a list in the CXLState in each machine. This is proving restrictive in a few cases where we need to iterate through these without being aware of the machine type. Just make them sysbus devices.
Restrict them to not user created as they need to be visible to early stages of machine init given effects on the memory map.
This change both simplifies state tracking and enables features needed for performance optimization and hotness tracking by making it possible to retrieve the fixed memory window on actions elsewhere in the topology.
In some cases the ordering of the Fixed Memory Windows matters. For those utility functions provide a GSList sorted by the window index. This ensures that we get consistency across:
- ordering in the command line
- ordering of the host PA ranges
- ordering of ACPI CEDT structures describing the CFMWS.
Quoted 4 lines
Other aspects don't have this constraint. For those direct iteration of the underlying hash structures is fine.
Signed-off-by Jonathan Cameron <[email protected]>
I'll not post v15 for a while to give time for review, but I just realized this snippet was in a patch I was carrying on top of this and should have been in this patch.
include/hw/cxl/cxl.h +0 -1
@@ -46,7 +46,6 @@ typedef struct CXLState {bool is_enabled;MemoryRegion host_mr;unsigned int next_mr_idx; -GList *fixed_windows;CXLFixedMemoryWindowOptionsList *cfmw_list; } CXLState;
Itaru Kitayama
Tested-byRe: [PATCH v14 2/5] hw/cxl: Make the CXL fixed memory windows devices.
Quoted 36 lines · Jonathan Cameron
Quoted 19 lines · On Wed, 28 May 2025 12:07:23 +0100 Jonathan Cameron…
Previously these somewhat device like structures were tracked using a list in the CXLState in each machine. This is proving restrictive in a few cases where we need to iterate through these without being aware of the machine type. Just make them sysbus devices.
Restrict them to not user created as they need to be visible to early stages of machine init given effects on the memory map.
This change both simplifies state tracking and enables features needed for performance optimization and hotness tracking by making it possible to retrieve the fixed memory window on actions elsewhere in the topology.
In some cases the ordering of the Fixed Memory Windows matters. For those utility functions provide a GSList sorted by the window index. This ensures that we get consistency across:
- ordering in the command line
- ordering of the host PA ranges
- ordering of ACPI CEDT structures describing the CFMWS.Quoted 4 lines
Other aspects don't have this constraint. For those direct iteration of the underlying hash structures is fine.
Signed-off-by Jonathan Cameron <[email protected]>I'll not post v15 for a while to give time for review, but I just realized this snippet was in a patch I was carrying on top of this and should have been in this patch.
include/hw/cxl/cxl.h +0
-0
include/hw/cxl/cxl.h typedef struct CXLState { · line 50
Quoted 6 lines
include/hw/cxl/cxl.h +0
-1@@ -46,7 +46,6 @@ typedef struct CXLState {bool is_enabled;MemoryRegion host_mr;unsigned int next_mr_idx; -GList *fixed_windows;CXLFixedMemoryWindowOptionsList *cfmw_list; } CXLState;
With this one line removed on top of v14, today's Dave's cxl/next kernel makes cxl test suite ran through without a single failure.
Jonathan, Zhi jian of Fujitsu gave you feedback on the QEMU core CXL emulation code, are you still waiting on any other reviewers to take a look at the series v14 (or v14-ish)?
Thanks,
Itaru.
Quoted 1 line
Re: [PATCH v14 2/5] hw/cxl: Make the CXL fixed memory windows devices.
Quoted 47 lines · On Tue, 10 Jun 2025 07:53:31 +0900 Itaru Kitayama <[email protected]>…
Quoted 36 lines · Jonathan Cameron
Quoted 19 lines · On Wed, 28 May 2025 12:07:23 +0100 Jonathan Cameron…
Previously these somewhat device like structures were tracked using a list in the CXLState in each machine. This is proving restrictive in a few cases where we need to iterate through these without being aware of the machine type. Just make them sysbus devices.
Restrict them to not user created as they need to be visible to early stages of machine init given effects on the memory map.
This change both simplifies state tracking and enables features needed for performance optimization and hotness tracking by making it possible to retrieve the fixed memory window on actions elsewhere in the topology.
In some cases the ordering of the Fixed Memory Windows matters. For those utility functions provide a GSList sorted by the window index. This ensures that we get consistency across:
- ordering in the command line
- ordering of the host PA ranges
- ordering of ACPI CEDT structures describing the CFMWS.Quoted 4 lines
Other aspects don't have this constraint. For those direct iteration of the underlying hash structures is fine.
Signed-off-by Jonathan Cameron <[email protected]>I'll not post v15 for a while to give time for review, but I just realized this snippet was in a patch I was carrying on top of this and should have been in this patch.
include/hw/cxl/cxl.h +0
-0include/hw/cxl/cxl.h typedef struct CXLState { · line 50
Quoted 6 lines
include/hw/cxl/cxl.h +0
-1@@ -46,7 +46,6 @@ typedef struct CXLState {bool is_enabled;MemoryRegion host_mr;unsigned int next_mr_idx; -GList *fixed_windows;CXLFixedMemoryWindowOptionsList *cfmw_list; } CXLState;With this one line removed on top of v14, today's Dave's cxl/next kernel makes cxl test suite ran through without a single failure.
Quoted 5 lines
Tested-by Itaru Kitayama <[email protected]>Jonathan, Zhi jian of Fujitsu gave you feedback on the QEMU core CXL emulation code, are you still waiting on any other reviewers to take a look at the series v14 (or v14-ish)?
No - obviously extra review is always good but I wasn't waiting on any. Just been distracted so not sent it out yet. Should get it out in next day or two.
Jonathan
Quoted 6 lines
Thanks,
Itaru.Quoted 1 line
Re: [PATCH v14 2/5] hw/cxl: Make the CXL fixed memory windows devices.
In reply to Jonathan CameronQuoted 19 lines · Jonathan Cameron via
Previously these somewhat device like structures were tracked using a list in the CXLState in each machine. This is proving restrictive in a few cases where we need to iterate through these without being aware of the machine type. Just make them sysbus devices.
Restrict them to not user created as they need to be visible to early stages of machine init given effects on the memory map.
This change both simplifies state tracking and enables features needed for performance optimization and hotness tracking by making it possible to retrieve the fixed memory window on actions elsewhere in the topology.
In some cases the ordering of the Fixed Memory Windows matters. For those utility functions provide a GSList sorted by the window index. This ensures that we get consistency across:
- ordering in the command line
- ordering of the host PA ranges
- ordering of ACPI CEDT structures describing the CFMWS.
Quoted 4 lines
Other aspects don't have this constraint. For those direct iteration of the underlying hash structures is fine.
Signed-off-by Jonathan Cameron<[email protected]>
LGTM,
PATCH v14 3/5 hw/cxl-host: Allow split of establishing memory address and mmio setup.
In reply to Jonathan CameronOn arm/virt the memory map is set up before any devices are brought up. To enable this provide split functions to establish the fw->base and later to actually map it.
---
v14: Update wrt to changes in previous patch.
Add a do_cfwms_set_memmap_and_update_mmio() utility function to reduce code duplication. (Zhijian)
--- include/hw/cxl/cxl_host.h | 2 ++ hw/cxl/cxl-host-stubs.c | 2 ++ hw/cxl/cxl-host.c | 43 +++++++++++++++++++++++++++++++++++---- 3 files changed, 43 insertions(+), 4 deletions(-)
include/hw/cxl/cxl_host.h +2 -0
@@ -16,6 +16,8 @@void cxl_machine_init (Object *obj,CXLState *state);void cxl_fmws_link_targets (Error **errp);void cxl_hook_up_pxb_registers (PCIBus *bus,CXLState *state,Error **errp); +hwaddr cxl_fmws_set_memmap (hwaddr base,hwaddr max_addr); +void cxl_fmws_update_mmio (void );hwaddr cxl_fmws_set_memmap_and_update_mmio (hwaddr base,hwaddr max_addr);GSList *cxl_fmws_get_all_sorted (void );
hw/cxl/cxl-host-stubs.c +2 -0
@@ -11,6 +11,8 @@void cxl_fmws_link_targets (Error **errp) {};void cxl_machine_init (Object *obj,CXLState *state) {};void cxl_hook_up_pxb_registers (PCIBus *bus,CXLState *state,Error **errp) {}; +hwaddr cxl_fmws_set_memmap (hwaddr base,hwaddr max_addr) {return base; }; +void cxl_fmws_update_mmio (void ) {};hwaddr cxl_fmws_set_memmap_and_update_mmio (hwaddr base,hwaddr max_addr) {return base;
hw/cxl/cxl-host.c +39 -4
69 lines not on this page · raw message
Signature · 1 line
2.48.1
Re: [PATCH v14 3/5] hw/cxl-host: Allow split of establishing memory address and mmio setup.
In patch 2/5, we introduced cxl_fmws_set_memmap_and_update_mmio().
Initially, I assumed patch 3/5 would split cxl_fmws_set_memmap_and_update_mmio() into two steps:
1. Traverse CXLFixedWindow and update fw->base.
2. Call sysbus_mmio_map(SYS_BUS_DEVICE(fw), 0, fw->base).
For example (my personal preference):
```c
hwaddr cxl_fmws_set_memmap_and_update_mmio(hwaddr base, hwaddr max_addr)
{
hwaddr end = cxl_fmws_set_memmap(base, max_addr);
cxl_fmws_update_mmio();
return end;
}If we had implemented this design in patch 2/5, patch 3/5 might not be necessary. The only potential benefit I see in the current patch 3/5 is efficiency improvements in cxl_fmws_set_memmap_and_update_mmio(), but since the function is typically called only once and the GLib list (glist) is small, the practical impact should be minimal.
I'm interested in others' perspectives on this.
Thanks
Zhijian
Quoted 116 lines · Jonathan Cameron via
On arm/virt the memory map is set up before any devices are brought up. To enable this provide split functions to establish the
fw->baseand later to actually map it.Signed-off-by Jonathan Cameron <[email protected]>---
v14: Update wrt to changes in previous patch.
Add ado_cfwms_set_memmap_and_update_mmio()utility function to reduce code duplication. (Zhijian)--- include/hw/cxl/cxl_host.h | 2 ++ hw/cxl/cxl-host-stubs.c | 2 ++ hw/cxl/cxl-host.c | 43 +++++++++++++++++++++++++++++++++++---- 3 files changed, 43 insertions(+), 4 deletions(-)include/hw/cxl/cxl_host.h +2
-0@@ -16,6 +16,8 @@void cxl_machine_init (Object *obj,CXLState *state);void cxl_fmws_link_targets (Error **errp);void cxl_hook_up_pxb_registers (PCIBus *bus,CXLState *state,Error **errp); +hwaddr cxl_fmws_set_memmap (hwaddr base,hwaddr max_addr); +void cxl_fmws_update_mmio (void );hwaddr cxl_fmws_set_memmap_and_update_mmio (hwaddr base,hwaddr max_addr);GSList *cxl_fmws_get_all_sorted (void );hw/cxl/cxl-host-stubs.c +2
-0@@ -11,6 +11,8 @@void cxl_fmws_link_targets (Error **errp) {};void cxl_machine_init (Object *obj,CXLState *state) {};void cxl_hook_up_pxb_registers (PCIBus *bus,CXLState *state,Error **errp) {}; +hwaddr cxl_fmws_set_memmap (hwaddr base,hwaddr max_addr) {return base; }; +void cxl_fmws_update_mmio (void ) {};hwaddr cxl_fmws_set_memmap_and_update_mmio (hwaddr base,hwaddr max_addr) {return base;hw/cxl/cxl-host.c +39
-4@@ -378,11 +378,14 @@ void cxl_hook_up_pxb_registers(PCIBus *bus, CXLState *state, Error **errp) } } -static void cxl_fmws_update (CXLFixedWindow *fw,hwaddr *base,hwaddr max_addr) +static void cxl_fmws_update (CXLFixedWindow *fw,hwaddr *base,hwaddr max_addr, +bool update_mmio) {if (*base + fw->size <= max_addr) { fw->base = *base; -sysbus_mmio_map (SYS_BUS_DEVICE (fw),0 , fw->base); +if (update_mmio) { +sysbus_mmio_map (SYS_BUS_DEVICE (fw),0 , fw->base); + } *base += fw->size; } }@@ -421,19 +424,51 @@ GSList *cxl_fmws_get_all_sorted(void)return g_slist_sort_with_data (cxl_fmws_get_all (), cfmws_cmp,NULL ); } -hwaddr cxl_fmws_set_memmap_and_update_mmio (hwaddr base,hwaddr max_addr) +static hwaddr do_cxl_fmws_set_memmap_and_update_mmio (hwaddr base, +hwaddr max_addr, +bool update_mmio) {GSList *cfmws_list, *iter; cfmws_list =cxl_fmws_get_all_sorted ();for (iter = cfmws_list; iter; iter = iter->next) { -cxl_fmws_update (CXL_FMW (iter->data), &base, max_addr); +cxl_fmws_update (CXL_FMW (iter->data), &base, max_addr, update_mmio); }g_slist_free (cfmws_list);return base; } +hwaddr cxl_fmws_set_memmap (hwaddr base,hwaddr max_addr) +{ +return do_cxl_fmws_set_memmap_and_update_mmio (base, max_addr, false); +} + +hwaddr cxl_fmws_set_memmap_and_update_mmio (hwaddr base,hwaddr max_addr) +{ +return do_cxl_fmws_set_memmap_and_update_mmio (base, max_addr, true); +} + +static int cxl_fmws_mmio_map (Object *obj,void *opaque) +{ +struct CXLFixedWindow *fw; + +if (!object_dynamic_cast (obj,TYPE_CXL_FMW )) { +return 0 ; + } + fw =CXL_FMW (obj); +sysbus_mmio_map (SYS_BUS_DEVICE (fw),0 , fw->base); + +return 0 ; +} + +void cxl_fmws_update_mmio (void ) +{ +/* Ordering is not required for this */ +object_child_foreach_recursive (object_get_root (), cxl_fmws_mmio_map, +NULL ); +} +static void cxl_fmw_realize (DeviceState *dev,Error **errp) {CXLFixedWindow *fw =CXL_FMW (dev);
Re: [PATCH v14 3/5] hw/cxl-host: Allow split of establishing memory address and mmio setup.
Quoted 8 lines · On Mon, 9 Jun 2025 01:15:10 +0000 "Zhijian Li (Fujitsu)"…
In patch 2/5, we introduced
cxl_fmws_set_memmap_and_update_mmio().Initially, I assumed patch 3/5 would split
cxl_fmws_set_memmap_and_update_mmio()into two steps:
1. Traverse CXLFixedWindow and updatefw->base.
2. Callsysbus_mmio_map(SYS_BUS_DEVICE(fw), 0, fw->base).
For example (my personal preference):
```c
hwaddrcxl_fmws_set_memmap_and_update_mmio(hwaddr base, hwaddr max_addr)
Quoted 8 lines
{ hwaddr end = cxl_fmws_set_memmap(base, max_addr); cxl_fmws_update_mmio(); return end; }If we had implemented this design in patch 2/5, patch 3/5 might not be necessary.
At the time of patch 2 we had no justification for the split as for x86 that would just look like a pointless double loop.
However you are right that this is too complex given it's not a performance path and perhaps some commentary in the patch description will be enough that no one minds.
I'll go a little further than you suggest and push the two calls in your cxl_fmws_set_memmap_and_mmio() into pc.c (patch 2) as that wrapper isn't adding much value.
I think it is a big enough change that I'll drop tags given on patch 2.
Thanks,
Jonathan
Quoted 128 lines
The only potential benefit I see in the current patch 3/5 is efficiency improvements in
cxl_fmws_set_memmap_and_update_mmio(), but since the function is typically called only once and the GLib list (glist) is small, the practical impact should be minimal.I'm interested in others' perspectives on this.
Thanks
ZhijianQuoted 116 lines · Jonathan Cameron via
On arm/virt the memory map is set up before any devices are brought up. To enable this provide split functions to establish the
fw->baseand later to actually map it.Signed-off-by Jonathan Cameron <[email protected]>---
v14: Update wrt to changes in previous patch.
Add ado_cfwms_set_memmap_and_update_mmio()utility function to reduce code duplication. (Zhijian)--- include/hw/cxl/cxl_host.h | 2 ++ hw/cxl/cxl-host-stubs.c | 2 ++ hw/cxl/cxl-host.c | 43 +++++++++++++++++++++++++++++++++++---- 3 files changed, 43 insertions(+), 4 deletions(-)include/hw/cxl/cxl_host.h +2
-0@@ -16,6 +16,8 @@void cxl_machine_init (Object *obj,CXLState *state);void cxl_fmws_link_targets (Error **errp);void cxl_hook_up_pxb_registers (PCIBus *bus,CXLState *state,Error **errp); +hwaddr cxl_fmws_set_memmap (hwaddr base,hwaddr max_addr); +void cxl_fmws_update_mmio (void );hwaddr cxl_fmws_set_memmap_and_update_mmio (hwaddr base,hwaddr max_addr);GSList *cxl_fmws_get_all_sorted (void );hw/cxl/cxl-host-stubs.c +2
-0@@ -11,6 +11,8 @@void cxl_fmws_link_targets (Error **errp) {};void cxl_machine_init (Object *obj,CXLState *state) {};void cxl_hook_up_pxb_registers (PCIBus *bus,CXLState *state,Error **errp) {}; +hwaddr cxl_fmws_set_memmap (hwaddr base,hwaddr max_addr) {return base; }; +void cxl_fmws_update_mmio (void ) {};hwaddr cxl_fmws_set_memmap_and_update_mmio (hwaddr base,hwaddr max_addr) {return base;hw/cxl/cxl-host.c +39
-4@@ -378,11 +378,14 @@ void cxl_hook_up_pxb_registers(PCIBus *bus, CXLState *state, Error **errp) } } -static void cxl_fmws_update (CXLFixedWindow *fw,hwaddr *base,hwaddr max_addr) +static void cxl_fmws_update (CXLFixedWindow *fw,hwaddr *base,hwaddr max_addr, +bool update_mmio) {if (*base + fw->size <= max_addr) { fw->base = *base; -sysbus_mmio_map (SYS_BUS_DEVICE (fw),0 , fw->base); +if (update_mmio) { +sysbus_mmio_map (SYS_BUS_DEVICE (fw),0 , fw->base); + } *base += fw->size; } }@@ -421,19 +424,51 @@ GSList *cxl_fmws_get_all_sorted(void)return g_slist_sort_with_data (cxl_fmws_get_all (), cfmws_cmp,NULL ); } -hwaddr cxl_fmws_set_memmap_and_update_mmio (hwaddr base,hwaddr max_addr) +static hwaddr do_cxl_fmws_set_memmap_and_update_mmio (hwaddr base, +hwaddr max_addr, +bool update_mmio) {GSList *cfmws_list, *iter; cfmws_list =cxl_fmws_get_all_sorted ();for (iter = cfmws_list; iter; iter = iter->next) { -cxl_fmws_update (CXL_FMW (iter->data), &base, max_addr); +cxl_fmws_update (CXL_FMW (iter->data), &base, max_addr, update_mmio); }g_slist_free (cfmws_list);return base; } +hwaddr cxl_fmws_set_memmap (hwaddr base,hwaddr max_addr) +{ +return do_cxl_fmws_set_memmap_and_update_mmio (base, max_addr, false); +} + +hwaddr cxl_fmws_set_memmap_and_update_mmio (hwaddr base,hwaddr max_addr) +{ +return do_cxl_fmws_set_memmap_and_update_mmio (base, max_addr, true); +} + +static int cxl_fmws_mmio_map (Object *obj,void *opaque) +{ +struct CXLFixedWindow *fw; + +if (!object_dynamic_cast (obj,TYPE_CXL_FMW )) { +return 0 ; + } + fw =CXL_FMW (obj); +sysbus_mmio_map (SYS_BUS_DEVICE (fw),0 , fw->base); + +return 0 ; +} + +void cxl_fmws_update_mmio (void ) +{ +/* Ordering is not required for this */ +object_child_foreach_recursive (object_get_root (), cxl_fmws_mmio_map, +NULL ); +} +static void cxl_fmw_realize (DeviceState *dev,Error **errp) {CXLFixedWindow *fw =CXL_FMW (dev)
PATCH v14 4/5 hw/arm/virt: Basic CXL enablement on pci_expander_bridge instances pxb-cxl
In reply to Jonathan CameronCode 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.
--- include/hw/arm/virt.h | 4 ++++ hw/arm/virt-acpi-build.c | 34 ++++++++++++++++++++++++++++++++++ hw/arm/virt.c | 29 +++++++++++++++++++++++++++++ 3 files changed, 67 insertions(+)
include/hw/arm/virt.h +4 -0
@@ -36,6 +36,7 @@#include "hw/arm/boot.h" #include "hw/arm/bsa.h" #include "hw/block/flash.h" +#include "hw/cxl/cxl.h" #include "system/kvm.h" #include "hw/intc/arm_gicv3_common.h" #include "qom/object.h"
@@ -85,6 +86,7 @@ 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 , };
@@ -140,6 +142,7 @@ struct VirtMachineState {bool secure;bool highmem;bool highmem_compact; +bool highmem_cxl;bool highmem_ecam;bool highmem_mmio;bool highmem_redists;
@@ -174,6 +177,7 @@ struct VirtMachineState {char *oem_id;char *oem_table_id;bool ns_el2_virt_timer_irq; +CXLState cxl_devices_state; };#define VIRT_ECAM_ID (high) (high ? VIRT_HIGH_PCIE_ECAM : VIRT_PCIE_ECAM)
hw/arm/virt-acpi-build.c +34 -0
@@ -39,10 +39,12 @@#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/acpi/hmat.h" +#include "hw/cxl/cxl.h" #include "hw/pci/pcie_host.h" #include "hw/pci/pci.h" #include "hw/pci/pci_bus.h"
@@ -119,10 +121,29 @@ static void acpi_dsdt_add_flash(Aml *scope, const MemMapEntry *flash_memmap)aml_append (scope, dev); } +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 (0x0B ))); +aml_append (dev, method); +build_cxl_dsm_method (dev); + +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 ],
@@ -136,6 +157,14 @@ 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,
@@ -963,6 +992,11 @@ void virt_acpi_build(VirtMachineState *vms, AcpiBuildTables *tables) } } +if (vms->cxl_devices_state.is_enabled) { +cxl_build_cedt (table_offsets, tables_blob, tables->linker, + vms->oem_id, vms->oem_table_id, &vms->cxl_devices_state); + } +if (ms->nvdimms_state->is_enabled) { nvdimm_build_acpi(table_offsets, tables_blob, tables->linker, ms->nvdimms_state, ms->ram_slots, vms->oem_id,
hw/arm/virt.c +29 -0
@@ -57,6 +57,7 @@#include "qemu/error-report.h" #include "qemu/module.h" #include "hw/pci-host/gpex.h" +#include "hw/pci-bridge/pci_expander_bridge.h" #include "hw/virtio/virtio-pci.h" #include "hw/core/sysbus-fdt.h" #include "hw/platform-bus.h"
@@ -86,6 +87,8 @@#include "hw/virtio/virtio-md-pci.h" #include "hw/virtio/virtio-iommu.h" #include "hw/char/pl011.h" +#include "hw/cxl/cxl.h" +#include "hw/cxl/cxl_host.h" #include "qemu/guest-random.h" static GlobalProperty arm_virt_compat[] = {
@@ -220,6 +223,7 @@ 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 ,DEFAULT_HIGH_PCIE_MMIO_SIZE },
@@ -1621,6 +1625,17 @@ static void create_pcie(VirtMachineState *vms) } } +static void create_cxl_host_reg_region (VirtMachineState *vms) +{ +MemoryRegion *sysmem =get_system_memory (); +MemoryRegion *mr = &vms->cxl_devices_state.host_mr; + +memory_region_init (mr,OBJECT (vms),"cxl_host_reg" , + vms->memmap[VIRT_CXL_HOST ].size); +memory_region_add_subregion (sysmem, vms->memmap[VIRT_CXL_HOST ].base, mr); + vms->highmem_cxl = true; +} +static void create_platform_bus (VirtMachineState *vms) {DeviceState *dev;
@@ -1737,6 +1752,12 @@ void virt_machine_done(Notifier *notifier, void *data)struct arm_boot_info *info = &vms->bootinfo;AddressSpace *as =arm_boot_address_space (cpu, info); +cxl_hook_up_pxb_registers (vms->bus, &vms->cxl_devices_state, + &error_fatal); + +if (vms->cxl_devices_state.is_enabled) { +cxl_fmws_link_targets (&error_fatal); + }/* * If the user provided a dtb, we assume the dynamic sysbus nodes * already are integrated there. This corresponds to a use case where
@@ -1783,6 +1804,7 @@ static inline bool *virt_get_high_memmap_enabled(VirtMachineState *vms, {bool *enabled_array[] = { &vms->highmem_redists, + &vms->highmem_cxl, &vms->highmem_ecam, &vms->highmem_mmio, };
@@ -1890,6 +1912,9 @@ static void virt_set_memmap(VirtMachineState *vms, int pa_bits)if (device_memory_size >0 ) {machine_memory_devices_init (ms, device_memory_base, device_memory_size); } + +cxl_fmws_set_memmap (ROUND_UP (vms->highest_gpa +1 ,256 * MiB), +BIT_ULL (pa_bits)); } static VirtGICType finalize_gic_version_do(const char *accel_name,
@@ -2340,6 +2365,8 @@ static void machvirt_init(MachineState *machine)memory_region_add_subregion (sysmem, vms->memmap[VIRT_MEM ].base, machine->ram); +cxl_fmws_update_mmio (); +virt_flash_fdt (vms, sysmem, secure_sysmem ?: sysmem);create_gic (vms, sysmem);
@@ -2395,6 +2422,7 @@ 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);
@@ -3365,6 +3393,7 @@ static void virt_instance_init(Object *obj) vms->oem_id =g_strndup (ACPI_BUILD_APPNAME6 ,6 ); vms->oem_table_id =g_strndup (ACPI_BUILD_APPNAME8 ,8 ); +cxl_machine_init (obj, &vms->cxl_devices_state); }static const TypeInfo virt_machine_info = {
Signature · 1 line
2.48.1
PATCH v14 5/5 qtest/cxl: Add aarch64 virt test for CXL
In reply to Jonathan CameronAdd a single complex case for aarch64 virt machine. Given existing much more comprehensive tests for x86 cover the common functionality, a single test should be enough to verify that the aarch64 part continue to work.
---
v14: Tags only.
--- tests/qtest/cxl-test.c | 59 ++++++++++++++++++++++++++++++++--------- tests/qtest/meson.build | 1 + 2 files changed, 47 insertions(+), 13 deletions(-)
tests/qtest/cxl-test.c +46 -13
@@ -19,6 +19,12 @@"-device pxb-cxl,id=cxl.1,bus=pcie.0,bus_nr=53 " \ "-M cxl-fmw.0.targets.0=cxl.0,cxl-fmw.0.targets.1=cxl.1,cxl-fmw.0.size=4G " +#define QEMU_VIRT_2PXB_CMD \ + "-machine virt,cxl=on -cpu max " \ + "-device pxb-cxl,id=cxl.0,bus=pcie.0,bus_nr=52 " \ + "-device pxb-cxl,id=cxl.1,bus=pcie.0,bus_nr=53 " \ + "-M cxl-fmw.0.targets.0=cxl.0,cxl-fmw.0.targets.1=cxl.1,cxl-fmw.0.size=4G " +#define QEMU_RP \ "-device cxl-rp,id=rp0,bus=cxl.0,chassis=0,slot=0 "
@@ -197,25 +203,52 @@ static void cxl_2pxb_4rp_4t3d(void)qtest_end ();rmdir (tmpfs); } + +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 (); +rmdir (tmpfs); +}#endif /* CONFIG_POSIX */ int main (int argc,char **argv) { -g_test_init (&argc, &argv,NULL ); +const char *arch =qtest_get_arch (); -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); +g_test_init (&argc, &argv,NULL ); +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);#ifdef CONFIG_POSIX -qtest_add_func ("/pci/cxl/type3_device" , cxl_t3d_deprecated); -qtest_add_func ("/pci/cxl/type3_device_pmem" , cxl_t3d_persistent); -qtest_add_func ("/pci/cxl/type3_device_vmem" , cxl_t3d_volatile); -qtest_add_func ("/pci/cxl/type3_device_vmem_lsa" , cxl_t3d_volatile_lsa); -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); +qtest_add_func ("/pci/cxl/type3_device" , cxl_t3d_deprecated); +qtest_add_func ("/pci/cxl/type3_device_pmem" , cxl_t3d_persistent); +qtest_add_func ("/pci/cxl/type3_device_vmem" , cxl_t3d_volatile); +qtest_add_func ("/pci/cxl/type3_device_vmem_lsa" , cxl_t3d_volatile_lsa); +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);#endif + }else if (strcmp (arch,"aarch64" ) ==0 ) { +#ifdef CONFIG_POSIX +qtest_add_func ("/pci/cxl/virt/pxb_x2_root_port_x4_type3_x4" , + cxl_virt_2pxb_4rp_4t3d); +#endif + } +return g_test_run (); }
tests/qtest/meson.build +1 -0
@@ -259,6 +259,7 @@ qtests_aarch64 = \ (config_all_accel.has_key('CONFIG_TCG') and \ config_all_devices.has_key('CONFIG_TPM_TIS_I2C') ? ['tpm-tis-i2c-test'] : []) + \ (config_all_devices.has_key('CONFIG_ASPEED_SOC') ? qtests_aspeed64 : []) + \ + qtests_cxl + \ ['arm-cpu-features', 'numa-test', 'boot-serial-test',
Signature · 1 line
2.48.1
Re: [PATCH v14 5/5] qtest/cxl: Add aarch64 virt test for CXL
Quoted 118 lines · On Wed, 28 May 2025 12:07:26 +0100 Jonathan Cameron…
Add a single complex case for aarch64 virt machine. Given existing much more comprehensive tests for x86 cover the common functionality, a single test should be enough to verify that the aarch64 part continue to work.
Tested-by Itaru Kitayama <[email protected]>Signed-off-by Jonathan Cameron <[email protected]>---
v14: Tags only.--- tests/qtest/cxl-test.c | 59++++++++++++++++++++++++++++++++--------- tests/qtest/meson.build | 1 + 2 files changed, 47 insertions(+), 13 deletions(-)
tests/qtest/cxl-test.c +0
-0@@ -19,6 +19,12 @@"-device pxb-cxl,id=cxl.1,bus=pcie.0,bus_nr=53 " \ "-Mcxl-fmw.0.targets.0=cxl.0,cxl-fmw.0.targets.1=cxl.1,cxl-fmw.0.size=4G " +#define
QEMU_VIRT_2PXB_CMD\tests/qtest/cxl-test.c +4
-0+"-machine virt,cxl=on -cpu max " \ +"-device pxb-cxl,id=cxl.0,bus=pcie.0,bus_nr=52 " \ +"-device pxb-cxl,id=cxl.1,bus=pcie.0,bus_nr=53 " \ + "-Mcxl-fmw.0.targets.0=cxl.0,cxl-fmw.0.targets.1=cxl.1,cxl-fmw.0.size=4G
" + #define QEMU_RP \ "-device cxl-rp,id=rp0,bus=cxl.0,chassis=0,slot=0 "tests/qtest/cxl-test.c +26
-13@@ -197,25 +203,52 @@ static void cxl_2pxb_4rp_4t3d(void)qtest_end ();rmdir (tmpfs); } + +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 (); +rmdir (tmpfs); +}#endif /* CONFIG_POSIX */ int main (int argc,char **argv) { -g_test_init (&argc, &argv,NULL ); +const char *arch =qtest_get_arch (); -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); +g_test_init (&argc, &argv,NULL ); +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);#ifdef CONFIG_POSIX -qtest_add_func ("/pci/cxl/type3_device" , cxl_t3d_deprecated); -qtest_add_func ("/pci/cxl/type3_device_pmem" , cxl_t3d_persistent); -qtest_add_func ("/pci/cxl/type3_device_vmem" , cxl_t3d_volatile); -qtest_add_func ("/pci/cxl/type3_device_vmem_lsa" , cxl_t3d_volatile_lsa); -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);tests/qtest/cxl-test.c +2
-0+qtest_add_func ("/pci/cxl/type3_device" , cxl_t3d_deprecated); + qtest_add_func("/pci/cxl/type3_device_pmem",
cxl_t3d_persistent);
+ qtest_add_func("/pci/cxl/type3_device_vmem",cxl_t3d_volatile);
+ qtest_add_func("/pci/cxl/type3_device_vmem_lsa",cxl_t3d_volatile_lsa);tests/qtest/cxl-test.c +10
-0+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);#endif + }else if (strcmp (arch,"aarch64" ) ==0 ) { +#ifdef CONFIG_POSIX +qtest_add_func ("/pci/cxl/virt/pxb_x2_root_port_x4_type3_x4" , + cxl_virt_2pxb_4rp_4t3d); +#endif + } +return g_test_run (); }tests/qtest/meson.build +0
-0@@ -259,6 +259,7 @@ qtests_aarch64 = \ (config_all_accel.has_key('CONFIG_TCG') and \
config_all_devices.has_key('CONFIG_TPM_TIS_I2C')? ['tpm-tis-i2c-test'] : []) + \
(config_all_devices.has_key('CONFIG_ASPEED_SOC')?qtests_aspeed64:
[]) + \
+qtests_cxl+
Quoted 3 lines
\ ['arm-cpu-features',
'numa-test',
'boot-serial-test',Hi Jonathan,
This patch does not apply on the latest master anymore. I think did a few days ago though. Not sure what's wrong.
Thanks,
Alireza
Hi Jonathan,
Quoted 3 lines · Jonathan Cameron
v14: Simplifications suggeseted by Itaru (and some extra simplifications
that became apparent) and gather tags.
See individual patches for more information.
I think the suggestion was made by Zhi jian or Fan? who enaged in the rewview of your proposed series v13.
Itaru.
Quoted 112 lines
Updated cover letter
Richard Henderson has posted a pull request with a fix for the TCG TLB issue which will hopefully merge shortly (Thanks Richard!).
Back in 2022, this series stalled on the absence of a solution to device tree support for PCI Expander Bridges (PXB) and we ended up only having x86 support upstream. I've been carrying the arm64 support out of tree since then, with occasional nasty surprises (e.g. UNIMP + DT issue seen a few weeks ago) and a fair number of fiddly rebases. gitlab.com/jic23/qemu cxl-<latest date>. Will update shortly with this series.
A recent discussion with Peter Maydell indicated that there are various other ACPI only features now, so in general he might be more relaxed about DT support being necessary. The upcoming vSMMUv3 support would run into this problem as well.
I presented the background to the PXB issue at Linaro connect 2022. In short the issue is that PXBs steal MMIO space from the main PCI root bridge. The challenge is knowing how much to steal.
On ACPI platforms, we can rely on EDK2 to perform an enumeration and configuration of the PCI topology and QEMU can update the ACPI tables after EDK2 has done this when it can simply read the space used by the root ports. On device tree, there is no entity to figure out that enumeration so we don't know how to size the stolen region.
Three approaches were discussed:
1) Enumerating in QEMU. Horribly complex and the last thing we want is a 3rd enumeration implementation that ends up out of sync with EDK2 and the kernel (there are frequent issues because of how those existing implementations differ.
2) Figure out how to enumerate in kernel. I never put a huge amount of work into this, but it seemed likely to involve a nasty dance with similar very specific code to that EDK2 is carrying and would very challenging to upstream (given the lack of clarity on real use cases for PXBs and DT).
3) Hack it based on the control we have which is bus numbers. No one liked this but it worked :)The other little wrinkle would be the need to define full bindings for CXL on DT + implement a fairly complex kernel stack as equivalent in ACPI involves a static table, CEDT, new runtime queries via
_DSMand a description of various components. Doable, but so far there is no interest on physical platforms. Worth noting that for now, the QEMU CXL emulation is all about testing and developing the OS stack, not about virtualization (performance is terrible except in some very contrived situations!)There is only a very simple test in here, because my intent is not to duplicate what we have on x86, but just to do a smoke test that everything is hooked up. In general we need much more comprehensive end to end CXL tests but that requires a reaonsably stable guest software stack. A few people have expressed interest in working on that, but we aren't there yet.
Note that this series has a very different use case to that in the proposed SBSA-ref support:
https://lore.kernel.org/qemu-devel/[email protected]/SBSA-ref is a good choice if you want a relatively simple mostly fixed configuration. That works well with the limited host system discoverability etc as EDK2 can be build against a known configuration.
My interest with this support in arm/virt is support host software stack development (we have a wide range of contributors, most of whom are working on emulation + the kernel support). I care about the weird corners. As such I need to be able to bring up variable numbers of host bridges, multiple CXL Fixed Memory Windows with varying characteristics (interleave etc), complex NUMA topologies with wierd performance characteristics etc. We can do that on x86 upstream today, or my gitlab tree. Note that we need arm support for some arch specific features in the near future (cache flushing). Doing kernel development with this need for flexibility on SBSA-ref is not currently practical. SBSA-ref CXL support is an excellent thing, just not much use to me for this work.
Also, we are kicking off some work on DCD virtualization, particularly to support inter-host shared memory being presented up into a VM. That will need upstream support on arm64 as it is built on top of the existing CXL emulation to avoid the need for a separate guest software stack.
Note this is TCG only - it is possible to support limited use with KVM but that needs additional patches not yet ready for upstream. The challenge is interleave - and the solution is don't interleave if you want to run with KVM.
Jonathan Cameron (5):
hw/cxl-host: Add an index field to CXLFixedMemoryWindow hw/cxl: Make the CXL fixed memory windows devices. hw/cxl-host: Allow split of establishing memory address and mmio
setup.
hw/arm/virt: Basic CXL enablement onpci_expander_bridgeinstances
pxb-cxl
qtest/cxl: Add aarch64 virt test for CXLinclude/hw/arm/virt.h | 4 + include/hw/cxl/cxl.h | 4 + include/hw/cxl/cxl_host.h | 6 +- hw/acpi/cxl.c | 76 +++++++-------- hw/arm/virt-acpi-build.c | 34 +++++++ hw/arm/virt.c | 29 ++++++ hw/cxl/cxl-host-stubs.c | 8 +- hw/cxl/cxl-host.c | 190 ++++++++++++++++++++++++++++++++------ hw/i386/pc.c | 51 +++++----- tests/qtest/cxl-test.c | 59 +++++++++--- tests/qtest/meson.build | 1 + 11 files changed, 353 insertions(+), 109 deletions(-)Signature · 1 line
2.48.1
Quoted 9 lines · On Thu, 29 May 2025 06:57:29 +0900 Itaru Kitayama <[email protected]>…
Hi Jonathan,
Quoted 3 lines · Jonathan Cameron
v14: Simplifications suggeseted by Itaru (and some extra simplifications
that became apparent) and gather tags.
See individual patches for more information.I think the suggestion was made by Zhi jian or Fan? who enaged in the rewview of your proposed series v13.
Ah. You are absolutely correct! It was Zhijian Li (with some additional comments on the comments from Fan)
Sorry about that and thanks for pointing out my mistake.
I also failed to +CC Zhijian. Again my apologies.
Jonathan
Quoted 115 lines
Itaru.
Quoted 112 lines
Updated cover letter
Richard Henderson has posted a pull request with a fix for the TCG TLB issue which will hopefully merge shortly (Thanks Richard!).
Back in 2022, this series stalled on the absence of a solution to device tree support for PCI Expander Bridges (PXB) and we ended up only having x86 support upstream. I've been carrying the arm64 support out of tree since then, with occasional nasty surprises (e.g. UNIMP + DT issue seen a few weeks ago) and a fair number of fiddly rebases. gitlab.com/jic23/qemu cxl-<latest date>. Will update shortly with this series.
A recent discussion with Peter Maydell indicated that there are various other ACPI only features now, so in general he might be more relaxed about DT support being necessary. The upcoming vSMMUv3 support would run into this problem as well.
I presented the background to the PXB issue at Linaro connect 2022. In short the issue is that PXBs steal MMIO space from the main PCI root bridge. The challenge is knowing how much to steal.
On ACPI platforms, we can rely on EDK2 to perform an enumeration and configuration of the PCI topology and QEMU can update the ACPI tables after EDK2 has done this when it can simply read the space used by the root ports. On device tree, there is no entity to figure out that enumeration so we don't know how to size the stolen region.
Three approaches were discussed:
1) Enumerating in QEMU. Horribly complex and the last thing we want is a 3rd enumeration implementation that ends up out of sync with EDK2 and the kernel (there are frequent issues because of how those existing implementations differ.
2) Figure out how to enumerate in kernel. I never put a huge amount of work into this, but it seemed likely to involve a nasty dance with similar very specific code to that EDK2 is carrying and would very challenging to upstream (given the lack of clarity on real use cases for PXBs and DT).
3) Hack it based on the control we have which is bus numbers. No one liked this but it worked :)The other little wrinkle would be the need to define full bindings for CXL on DT + implement a fairly complex kernel stack as equivalent in ACPI involves a static table, CEDT, new runtime queries via
_DSMand a description of various components. Doable, but so far there is no interest on physical platforms. Worth noting that for now, the QEMU CXL emulation is all about testing and developing the OS stack, not about virtualization (performance is terrible except in some very contrived situations!)There is only a very simple test in here, because my intent is not to duplicate what we have on x86, but just to do a smoke test that everything is hooked up. In general we need much more comprehensive end to end CXL tests but that requires a reaonsably stable guest software stack. A few people have expressed interest in working on that, but we aren't there yet.
Note that this series has a very different use case to that in the proposed SBSA-ref support:
https://lore.kernel.org/qemu-devel/[email protected]/SBSA-ref is a good choice if you want a relatively simple mostly fixed configuration. That works well with the limited host system discoverability etc as EDK2 can be build against a known configuration.
My interest with this support in arm/virt is support host software stack development (we have a wide range of contributors, most of whom are working on emulation + the kernel support). I care about the weird corners. As such I need to be able to bring up variable numbers of host bridges, multiple CXL Fixed Memory Windows with varying characteristics (interleave etc), complex NUMA topologies with wierd performance characteristics etc. We can do that on x86 upstream today, or my gitlab tree. Note that we need arm support for some arch specific features in the near future (cache flushing). Doing kernel development with this need for flexibility on SBSA-ref is not currently practical. SBSA-ref CXL support is an excellent thing, just not much use to me for this work.
Also, we are kicking off some work on DCD virtualization, particularly to support inter-host shared memory being presented up into a VM. That will need upstream support on arm64 as it is built on top of the existing CXL emulation to avoid the need for a separate guest software stack.
Note this is TCG only - it is possible to support limited use with KVM but that needs additional patches not yet ready for upstream. The challenge is interleave - and the solution is don't interleave if you want to run with KVM.
Jonathan Cameron (5):
hw/cxl-host: Add an index field to CXLFixedMemoryWindow hw/cxl: Make the CXL fixed memory windows devices. hw/cxl-host: Allow split of establishing memory address and mmio
setup.
hw/arm/virt: Basic CXL enablement onpci_expander_bridgeinstances
pxb-cxl
qtest/cxl: Add aarch64 virt test for CXLinclude/hw/arm/virt.h | 4 + include/hw/cxl/cxl.h | 4 + include/hw/cxl/cxl_host.h | 6 +- hw/acpi/cxl.c | 76 +++++++-------- hw/arm/virt-acpi-build.c | 34 +++++++ hw/arm/virt.c | 29 ++++++ hw/cxl/cxl-host-stubs.c | 8 +- hw/cxl/cxl-host.c | 190 ++++++++++++++++++++++++++++++++------ hw/i386/pc.c | 51 +++++----- tests/qtest/cxl-test.c | 59 +++++++++--- tests/qtest/meson.build | 1 + 11 files changed, 353 insertions(+), 109 deletions(-)Signature · 1 line
2.48.1
Quoted 84 lines · Jonathan Cameron
v14: Simplifications suggeseted by Itaru (and some extra simplifications
that became apparent) and gather tags.
See individual patches for more information.Updated cover letter
Richard Henderson has posted a pull request with a fix for the TCG TLB issue which will hopefully merge shortly (Thanks Richard!).
Back in 2022, this series stalled on the absence of a solution to device tree support for PCI Expander Bridges (PXB) and we ended up only having x86 support upstream. I've been carrying the arm64 support out of tree since then, with occasional nasty surprises (e.g. UNIMP + DT issue seen a few weeks ago) and a fair number of fiddly rebases. gitlab.com/jic23/qemu cxl-<latest date>. Will update shortly with this series.
A recent discussion with Peter Maydell indicated that there are various other ACPI only features now, so in general he might be more relaxed about DT support being necessary. The upcoming vSMMUv3 support would run into this problem as well.
I presented the background to the PXB issue at Linaro connect 2022. In short the issue is that PXBs steal MMIO space from the main PCI root bridge. The challenge is knowing how much to steal.
On ACPI platforms, we can rely on EDK2 to perform an enumeration and configuration of the PCI topology and QEMU can update the ACPI tables after EDK2 has done this when it can simply read the space used by the root ports. On device tree, there is no entity to figure out that enumeration so we don't know how to size the stolen region.
Three approaches were discussed:
1) Enumerating in QEMU. Horribly complex and the last thing we want is a 3rd enumeration implementation that ends up out of sync with EDK2 and the kernel (there are frequent issues because of how those existing implementations differ.
2) Figure out how to enumerate in kernel. I never put a huge amount of work into this, but it seemed likely to involve a nasty dance with similar very specific code to that EDK2 is carrying and would very challenging to upstream (given the lack of clarity on real use cases for PXBs and DT).
3) Hack it based on the control we have which is bus numbers. No one liked this but it worked :)The other little wrinkle would be the need to define full bindings for CXL on DT + implement a fairly complex kernel stack as equivalent in ACPI involves a static table, CEDT, new runtime queries via
_DSMand a description of various components. Doable, but so far there is no interest on physical platforms. Worth noting that for now, the QEMU CXL emulation is all about testing and developing the OS stack, not about virtualization (performance is terrible except in some very contrived situations!)There is only a very simple test in here, because my intent is not to duplicate what we have on x86, but just to do a smoke test that everything is hooked up. In general we need much more comprehensive end to end CXL tests but that requires a reaonsably stable guest software stack. A few people have expressed interest in working on that, but we aren't there yet.
Note that this series has a very different use case to that in the proposed SBSA-ref support:
https://lore.kernel.org/qemu-devel/[email protected]/SBSA-ref is a good choice if you want a relatively simple mostly fixed configuration. That works well with the limited host system discoverability etc as EDK2 can be build against a known configuration.
My interest with this support in arm/virt is support host software stack development (we have a wide range of contributors, most of whom are working on emulation + the kernel support). I care about the weird corners. As such I need to be able to bring up variable numbers of host bridges, multiple CXL Fixed Memory Windows with varying characteristics (interleave etc), complex NUMA topologies with wierd performance characteristics etc. We can do that on x86 upstream today, or my gitlab tree. Note that we need arm support for some arch specific features in the near future (cache flushing). Doing kernel development with this need for flexibility on SBSA-ref is not currently practical. SBSA-ref CXL support is an excellent thing, just not much use to me for this work.
Also, we are kicking off some work on DCD virtualization, particularly to support inter-host shared memory being presented up into a VM. That will need upstream support on arm64 as it is built on top of the existing CXL emulation to avoid the need for a separate guest software stack.
Quoted 4 lines
Note this is TCG only - it is possible to support limited use with KVM but that needs additional patches not yet ready for upstream. The challenge is interleave - and the solution is don't interleave if you want to run with KVM.
Series applied cleanly on top of today's QEMU. And I confirm that qtest-aarch64/cxl-test passes the test as it should and ndctl cxl test suite ran fine 11 out of 12 (1 SKIP) on this series again this time tracing subsystem enabled as Alison suggested. Used the Intel folk cxl/next kernel.
Thanks,
Itaru.
Quoted 26 lines
Jonathan Cameron (5):
hw/cxl-host: Add an index field to CXLFixedMemoryWindow hw/cxl: Make the CXL fixed memory windows devices. hw/cxl-host: Allow split of establishing memory address and mmio
setup.
hw/arm/virt: Basic CXL enablement onpci_expander_bridgeinstances
pxb-cxl
qtest/cxl: Add aarch64 virt test for CXLinclude/hw/arm/virt.h | 4 + include/hw/cxl/cxl.h | 4 + include/hw/cxl/cxl_host.h | 6 +- hw/acpi/cxl.c | 76 +++++++-------- hw/arm/virt-acpi-build.c | 34 +++++++ hw/arm/virt.c | 29 ++++++ hw/cxl/cxl-host-stubs.c | 8 +- hw/cxl/cxl-host.c | 190 ++++++++++++++++++++++++++++++++------ hw/i386/pc.c | 51 +++++----- tests/qtest/cxl-test.c | 59 +++++++++--- tests/qtest/meson.build | 1 + 11 files changed, 353 insertions(+), 109 deletions(-)Signature · 1 line
2.48.1
Quoted 90 lines · On Thu, 29 May 2025 14:13:30 +0900 Itaru Kitayama <[email protected]>…
Quoted 84 lines · Jonathan Cameron
v14: Simplifications suggeseted by Itaru (and some extra simplifications
that became apparent) and gather tags.
See individual patches for more information.Updated cover letter
Richard Henderson has posted a pull request with a fix for the TCG TLB issue which will hopefully merge shortly (Thanks Richard!).
Back in 2022, this series stalled on the absence of a solution to device tree support for PCI Expander Bridges (PXB) and we ended up only having x86 support upstream. I've been carrying the arm64 support out of tree since then, with occasional nasty surprises (e.g. UNIMP + DT issue seen a few weeks ago) and a fair number of fiddly rebases. gitlab.com/jic23/qemu cxl-<latest date>. Will update shortly with this series.
A recent discussion with Peter Maydell indicated that there are various other ACPI only features now, so in general he might be more relaxed about DT support being necessary. The upcoming vSMMUv3 support would run into this problem as well.
I presented the background to the PXB issue at Linaro connect 2022. In short the issue is that PXBs steal MMIO space from the main PCI root bridge. The challenge is knowing how much to steal.
On ACPI platforms, we can rely on EDK2 to perform an enumeration and configuration of the PCI topology and QEMU can update the ACPI tables after EDK2 has done this when it can simply read the space used by the root ports. On device tree, there is no entity to figure out that enumeration so we don't know how to size the stolen region.
Three approaches were discussed:
1) Enumerating in QEMU. Horribly complex and the last thing we want is a 3rd enumeration implementation that ends up out of sync with EDK2 and the kernel (there are frequent issues because of how those existing implementations differ.
2) Figure out how to enumerate in kernel. I never put a huge amount of work into this, but it seemed likely to involve a nasty dance with similar very specific code to that EDK2 is carrying and would very challenging to upstream (given the lack of clarity on real use cases for PXBs and DT).
3) Hack it based on the control we have which is bus numbers. No one liked this but it worked :)The other little wrinkle would be the need to define full bindings for CXL on DT + implement a fairly complex kernel stack as equivalent in ACPI involves a static table, CEDT, new runtime queries via
_DSMand a description of various components. Doable, but so far there is no interest on physical platforms. Worth noting that for now, the QEMU CXL emulation is all about testing and developing the OS stack, not about virtualization (performance is terrible except in some very contrived situations!)There is only a very simple test in here, because my intent is not to duplicate what we have on x86, but just to do a smoke test that everything is hooked up. In general we need much more comprehensive end to end CXL tests but that requires a reaonsably stable guest software stack. A few people have expressed interest in working on that, but we aren't there yet.
Note that this series has a very different use case to that in the proposed SBSA-ref support:
https://lore.kernel.org/qemu-devel/[email protected]/SBSA-ref is a good choice if you want a relatively simple mostly fixed configuration. That works well with the limited host system discoverability etc as EDK2 can be build against a known configuration.
My interest with this support in arm/virt is support host software stack development (we have a wide range of contributors, most of whom are working on emulation + the kernel support). I care about the weird corners. As such I need to be able to bring up variable numbers of host bridges, multiple CXL Fixed Memory Windows with varying characteristics (interleave etc), complex NUMA topologies with wierd performance characteristics etc. We can do that on x86 upstream today, or my gitlab tree. Note that we need arm support for some arch specific features in the near future (cache flushing). Doing kernel development with this need for flexibility on SBSA-ref is not currently practical. SBSA-ref CXL support is an excellent thing, just not much use to me for this work.
Also, we are kicking off some work on DCD virtualization, particularly to support inter-host shared memory being presented up into a VM. That will need upstream support on arm64 as it is built on top of the existing CXL emulation to avoid the need for a separate guest software stack.
Quoted 4 lines
Note this is TCG only - it is possible to support limited use with KVM but that needs additional patches not yet ready for upstream. The challenge is interleave - and the solution is don't interleave if you want to run with KVM.
Quoted 7 lines
Series applied cleanly on top of today's QEMU. And I confirm that qtest-aarch64/cxl-test passes the test as it should and ndctl cxl test suite ran fine 11 out of 12 (1 SKIP) on this series again this time tracing subsystem enabled as Alison suggested. Used the Intel folk cxl/next kernel.
Tested-by Itaru Kitayama <[email protected]>
Thanks Itaru.
I'm not going to pick this tag up purely because of the changes in patch 2 and related dropping of patch 3. Whilst not huge that is in fairly critical bit of the code, so feels inappropriate to carryn tags forwards.
Jonathan
Quoted 30 lines
Thanks,
Itaru.Quoted 26 lines
Jonathan Cameron (5):
hw/cxl-host: Add an index field to CXLFixedMemoryWindow hw/cxl: Make the CXL fixed memory windows devices. hw/cxl-host: Allow split of establishing memory address and mmio
setup.
hw/arm/virt: Basic CXL enablement onpci_expander_bridgeinstances
pxb-cxl
qtest/cxl: Add aarch64 virt test for CXLinclude/hw/arm/virt.h | 4 + include/hw/cxl/cxl.h | 4 + include/hw/cxl/cxl_host.h | 6 +- hw/acpi/cxl.c | 76 +++++++-------- hw/arm/virt-acpi-build.c | 34 +++++++ hw/arm/virt.c | 29 ++++++ hw/cxl/cxl-host-stubs.c | 8 +- hw/cxl/cxl-host.c | 190 ++++++++++++++++++++++++++++++++------ hw/i386/pc.c | 51 +++++----- tests/qtest/cxl-test.c | 59 +++++++++--- tests/qtest/meson.build | 1 + 11 files changed, 353 insertions(+), 109 deletions(-)Signature · 1 line
2.48.1
Quoted 5 lines · Jonathan Cameron via
Jonathan Cameron (5):
hw/cxl-host: Add an index field to CXLFixedMemoryWindow hw/cxl: Make the CXL fixed memory windows devices. hw/cxl-host: Allow split of establishing memory address and mmio
setup.
With above 3 patches + x86_64 platform,
Quoted 3 lines
hw/arm/virt: Basic CXL enablement on
pci_expander_bridgeinstances
pxb-cxl
qtest/cxl: Add aarch64 virt test for CXL
Please forgive me for not reviewing the last two patches, as I lack access to an aarch64 machine nor sufficient aarch64 architecture knowledge.
Thanks
Zhijian