Skip to contentSkip to conversation

hw/cxl: Misc minor improvements

PATCH In review Jonathan Cameron 2 participants 6 messages Jul 5, 2024 – Jul 25, 2024

lore
Share & tools
lore
#0 lore raw

v2 cover hw/cxl: Misc minor improvements

v2: Switch patch 1 from a patch that removes the cfmws_list element
of CXLState to a fix that makes use of it to fix a crash. Now based on master as precursors merged.

Before pushing on to more significant features a few unrelated patches doing tidying up + one to avoid people setting the memory backend both for a CXL type 3 device and as normal RAM in an attempt to get SRAT to cover it correctly. We've had multiple 'bug' reports from this and if nothing else I'd like to stop getting those!

Fan Ni (1):
hw/cxl/cxl-mailbox-utils: remove unneeded mailbox output payload space
zeroing

Jonathan Cameron (1):
hw/cxl: Check for multiple mappings of memory backends.

Zhao Liu (1):
hw/cxl/cxl-host: Fix segmentation fault when getting cxl-fmw property

 hw/cxl/cxl-host.c          |  3 ++-
 hw/cxl/cxl-mailbox-utils.c |  7 -------
 hw/mem/cxl_type3.c         | 15 +++++++++++++++
 3 files changed, 17 insertions(+), 8 deletions(-)
Signature · 1 line
2.43.0
#1 lore raw

PATCH v2 1/3 hw/cxl/cxl-host: Fix segmentation fault when getting cxl-fmw property

From: Zhao Liu <[email protected]>

QEMU crashes (Segmentation fault) when getting cxl-fmw property via qmp:

(QEMU) qom-get path=machine property=cxl-fmw

This issue is caused by accessing wrong callback (opaque) type in machine_get_cfmw().

cxl_machine_init() sets the callback as CXLState * type but machine_get_cfmw() treats the callback as CXLFixedMemoryWindowOptionsList **.

Fix this error by casting opaque to CXLState * type in machine_get_cfmw().

Fixes 03b39fcf64bc ("hw/cxl: Make the CXL fixed memory window setup a machine parameter.")
---
 hw/cxl/cxl-host.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
hw/cxl/cxl-host.c +2 -1
index c5f5fcfd64..e9f2543c43 100644
@@ -315,7 +315,8 @@ static void machine_set_cxl(Object *obj, Visitor *v, const char *name,
 static void machine_get_cfmw(Object *obj, Visitor *v, const char *name,
                              void *opaque, Error **errp)
 {
-    CXLFixedMemoryWindowOptionsList **list = opaque;
+    CXLState *state = opaque;
+    CXLFixedMemoryWindowOptionsList **list = &state->cfmw_list;
 
     visit_type_CXLFixedMemoryWindowOptionsList(v, name, list, errp);
 }
Signature · 1 line
2.43.0
#2 lore raw

Re: [PATCH v2 1/3] hw/cxl/cxl-host: Fix segmentation fault when getting cxl-fmw property

Quoted 17 lines · Jonathan Cameron via

From: Zhao Liu <[email protected]>

QEMU crashes (Segmentation fault) when getting cxl-fmw property via qmp:

(QEMU) qom-get path=machine property=cxl-fmw

This issue is caused by accessing wrong callback (opaque) type in machine_get_cfmw().

cxl_machine_init() sets the callback as CXLState * type but machine_get_cfmw() treats the callback as CXLFixedMemoryWindowOptionsList **.

Fix this error by casting opaque to CXLState * type in machine_get_cfmw().

Quoted 6 lines
Fixes 03b39fcf64bc ("hw/cxl: Make the CXL fixed memory window setup a machine parameter.")

Is this a stable@ material?

Thanks,

/mjt

#3 lore raw

Re: [PATCH v2 1/3] hw/cxl/cxl-host: Fix segmentation fault when getting cxl-fmw property

Quoted 18 lines · On Wed, 24 Jul 2024 07:53:48 +0300 Michael Tokarev <[email protected]> wrote:
Quoted 17 lines · Jonathan Cameron via

From: Zhao Liu <[email protected]>

QEMU crashes (Segmentation fault) when getting cxl-fmw property via qmp:

(QEMU) qom-get path=machine property=cxl-fmw

This issue is caused by accessing wrong callback (opaque) type in machine_get_cfmw().

cxl_machine_init() sets the callback as CXLState * type but machine_get_cfmw() treats the callback as CXLFixedMemoryWindowOptionsList **.

Fix this error by casting opaque to CXLState * type in machine_get_cfmw().

Quoted 8 lines
Quoted 6 lines
Fixes 03b39fcf64bc ("hw/cxl: Make the CXL fixed memory window setup a machine parameter.")

Is this a stable@ material?

Yes please

Thanks,

Jonathan

Quoted 5 lines

Thanks,

/mjt

#4 lore raw

PATCH v2 2/3 hw/cxl/cxl-mailbox-utils: remove unneeded mailbox output payload space zeroing

In reply to Jonathan Cameron

From: Fan Ni <[email protected]>

The whole mailbox output payload space is already zeroed after copying out the input payload, which happens before processing the specific mailbox command:
https://elixir.bootlin.com/qemu/v8.2.1/source/hw/cxl/cxl-device-utils.c#L204

---
 hw/cxl/cxl-mailbox-utils.c | 7 -------
 1 file changed, 7 deletions(-)
hw/cxl/cxl-mailbox-utils.c +0 -7
index 74eeb6fde7..facec42dc8 100644
@@ -235,7 +235,6 @@ static CXLRetCode cmd_events_get_records(const struct cxl_cmd *cmd,
     log_type = payload_in[0];
 
     pl = (CXLGetEventPayload *)payload_out;
-    memset(pl, 0, sizeof(*pl));
 
     max_recs = (cxlds->payload_size - CXL_EVENT_PAYLOAD_HDR_SIZE) /
                 CXL_EVENT_RECORD_SIZE;
@@ -273,7 +272,6 @@ static CXLRetCode cmd_events_get_interrupt_policy(const struct cxl_cmd *cmd,
     CXLEventLog *log;
 
     policy = (CXLEventInterruptPolicy *)payload_out;
-    memset(policy, 0, sizeof(*policy));
 
     log = &cxlds->event_logs[CXL_EVENT_TYPE_INFO];
     if (log->irq_enabled) {
@@ -372,7 +370,6 @@ static CXLRetCode cmd_infostat_identify(const struct cxl_cmd *cmd,
     QEMU_BUILD_BUG_ON(sizeof(*is_identify) != 18);
 
     is_identify = (void *)payload_out;
-    memset(is_identify, 0, sizeof(*is_identify));
     is_identify->pcie_vid = class->vendor_id;
     is_identify->pcie_did = class->device_id;
     if (object_dynamic_cast(OBJECT(cci->d), TYPE_CXL_USP)) {
@@ -606,7 +603,6 @@ static CXLRetCode cmd_infostat_bg_op_sts(const struct cxl_cmd *cmd,
     QEMU_BUILD_BUG_ON(sizeof(*bg_op_status) != 8);
 
     bg_op_status = (void *)payload_out;
-    memset(bg_op_status, 0, sizeof(*bg_op_status));
     bg_op_status->status = cci->bg.complete_pct << 1;
     if (cci->bg.runtime > 0) {
         bg_op_status->status |= 1U << 0;
@@ -647,7 +643,6 @@ static CXLRetCode cmd_firmware_update_get_info(const struct cxl_cmd *cmd,
     }
 
     fw_info = (void *)payload_out;
-    memset(fw_info, 0, sizeof(*fw_info));
 
     fw_info->slots_supported = 2;
     fw_info->slot_info = BIT(0) | BIT(3);
@@ -805,7 +800,6 @@ static CXLRetCode cmd_identify_memory_device(const struct cxl_cmd *cmd,
     }
 
     id = (void *)payload_out;
-    memset(id, 0, sizeof(*id));
 
     snprintf(id->fw_revision, 0x10, "BWFW VERSION %02d", 0);
 
@@ -1095,7 +1089,6 @@ static CXLRetCode cmd_media_get_poison_list(const struct cxl_cmd *cmd,
     out_pl_len = sizeof(*out) + record_count * sizeof(out->records[0]);
     assert(out_pl_len <= CXL_MAILBOX_MAX_PAYLOAD_SIZE);
 
-    memset(out, 0, out_pl_len);
     QLIST_FOREACH(ent, poison_list, node) {
         uint64_t start, stop;
 
Signature · 1 line
2.43.0
#5 lore raw

PATCH v2 3/3 hw/cxl: Check for multiple mappings of memory backends.

In reply to Jonathan Cameron

Similar protection to that provided for -numa memdev=x
to make sure that memory used to back a type3 device is not also mapped as normal RAM, or for multiple type3 devices.

This is an easy footgun to remove and seems multiple people have run into it.

---
 hw/mem/cxl_type3.c | 15 +++++++++++++++
 1 file changed, 15 insertions(+)
hw/mem/cxl_type3.c +15 -0
index 35ac59883a..e7fbbb4d51 100644
@@ -737,6 +737,11 @@ static bool cxl_setup_memory(CXLType3Dev *ct3d, Error **errp)
             error_setg(errp, "volatile memdev must have backing device");
             return false;
         }
+        if (host_memory_backend_is_mapped(ct3d->hostvmem)) {
+            error_setg(errp, "memory backend %s can't be used multiple times.",
+               object_get_canonical_path_component(OBJECT(ct3d->hostvmem)));
+            return false;
+        }
         memory_region_set_nonvolatile(vmr, false);
         memory_region_set_enabled(vmr, true);
         host_memory_backend_set_mapped(ct3d->hostvmem, true);
@@ -760,6 +765,11 @@ static bool cxl_setup_memory(CXLType3Dev *ct3d, Error **errp)
             error_setg(errp, "persistent memdev must have backing device");
             return false;
         }
+        if (host_memory_backend_is_mapped(ct3d->hostpmem)) {
+            error_setg(errp, "memory backend %s can't be used multiple times.",
+               object_get_canonical_path_component(OBJECT(ct3d->hostpmem)));
+            return false;
+        }
         memory_region_set_nonvolatile(pmr, true);
         memory_region_set_enabled(pmr, true);
         host_memory_backend_set_mapped(ct3d->hostpmem, true);
@@ -790,6 +800,11 @@ static bool cxl_setup_memory(CXLType3Dev *ct3d, Error **errp)
             return false;
         }
 
+        if (host_memory_backend_is_mapped(ct3d->dc.host_dc)) {
+            error_setg(errp, "memory backend %s can't be used multiple times.",
+               object_get_canonical_path_component(OBJECT(ct3d->dc.host_dc)));
+            return false;
+        }
         /*
          * Set DC regions as volatile for now, non-volatile support can
          * be added in the future if needed.
Signature · 1 line
2.43.0