5
5
[width=100%]
6
6
[%header, cols="5,25"]
7
7
|===
8
- | ID# ^| Requirement
8
+ | ID# ^| Rule
9
9
| CTI_010 | The `time` CSR MUST increment at a constant frequency and the count
10
10
MUST be in units of 1 ns. The frequency at which the CSR provides
11
11
an updated time value MUST be at least 100 MHz.
15
15
| CTI_020 | The `time` counter MUST appear to be always on and MUST appear to
16
16
not lose its count across hart low power idle states, including
17
17
when the hart is powered off.
18
- 2+a| _This requirement does not apply to system power states such as G3 (power
18
+ 2+a| _This rule does not apply to system power states such as G3 (power
19
19
off), S3 (Suspend to RAM), or S4 (Hibernate)._ +
20
20
+
21
21
_Losing `time` count across hart low power idle states may lead to the
@@ -37,7 +37,7 @@ deliver external interrupts to the RISC-V application processor harts.
37
37
[width=100%]
38
38
[%header, cols="5,25"]
39
39
|===
40
- | ID# ^| Requirement
40
+ | ID# ^| Rule
41
41
| IIC_010 | The RISC-V Advanced Interrupt Architecture cite:[AIA] MUST be
42
42
supported.
43
43
@@ -107,15 +107,15 @@ deliver external interrupts to the RISC-V application processor harts.
107
107
[width=100%]
108
108
[%header, cols="5,25"]
109
109
|===
110
- | ID# ^| Requirement
110
+ | ID# ^| Rule
111
111
| IOM_010 | All IOMMUs in the SoC MUST support the RISC-V IOMMU specification
112
112
cite:[IOMMU].
113
113
114
114
| IOM_020 a| All DMA capable peripherals (RCiEP and non-PCIe devices) and all
115
115
PCIe root ports accessible by software on the RISC-V application
116
116
processor harts MUST be governed by an IOMMU. +
117
117
+
118
- Initiators, such as the following, are exempt from this requirement :
118
+ Initiators, such as the following, are exempt from this rule :
119
119
120
120
* Interrupt controllers, such as the APLIC.
121
121
* IOMMUs.
@@ -130,7 +130,7 @@ deliver external interrupts to the RISC-V application processor harts.
130
130
provided by the IOMMU enables usages such as passthrough of such devices to
131
131
virtual machines, shared virtual addressing, etc._ +
132
132
+
133
- _The number of IOMMUs implemented in the SoC to satisfy requirement IOM_020
133
+ _The number of IOMMUs implemented in the SoC to satisfy rule IOM_020
134
134
is `UNSPECIFIED`._
135
135
136
136
| IOM_030 | The IOMMU governing a PCIe root port MUST support at least 16-bit
@@ -396,7 +396,7 @@ hierarchy domain originating at each PCIe root port.
396
396
[width=100%]
397
397
[%header, cols="5,25"]
398
398
|===
399
- | ID# ^| Requirement
399
+ | ID# ^| Rule
400
400
| ECM_010 a| The ECAM address ranges MUST have the following physical memory
401
401
attributes (PMAs):
402
402
@@ -416,7 +416,7 @@ hierarchy domain originating at each PCIe root port.
416
416
2+| _Besides performing a write, software executing on a hart must not
417
417
require any additional actions to achieve this property._ +
418
418
+
419
- _This requirement satisfies the processor and host bridge implementation
419
+ _This rule satisfies the processor and host bridge implementation
420
420
requirement mentioned in the “Ordering Considerations for the Enhanced
421
421
Configuration Access Mechanism” implementation note of the PCIe 6.0
422
422
specification._
@@ -503,7 +503,7 @@ hierarchy domain originating at each PCIe root port.
503
503
[width=100%]
504
504
[%header, cols="5,25"]
505
505
|===
506
- | ID# ^| Requirement
506
+ | ID# ^| Rule
507
507
| MMS_010 | The SoC MUST support designating, for each hierarchy domain, one or
508
508
more ranges of system physical addresses that may be used for
509
509
mapping memory space of endpoints in that hierarchy domain using
@@ -526,8 +526,7 @@ hierarchy domain originating at each PCIe root port.
526
526
and system requirements._
527
527
528
528
| MMS_030 a| The system physical address ranges designated for mapping endpoint
529
- memory spaces have the following physical memory attribute (PMAs)
530
- requirements:
529
+ memory spaces have the following physical memory attribute (PMAs):
531
530
532
531
* MUST be not cacheable, non-idempotent, coherent, strongly-ordered
533
532
(I/O ordering) I/O region.
@@ -628,7 +627,7 @@ devices, and SR-IOV capable devices.
628
627
[width=100%]
629
628
[%header, cols="5,25"]
630
629
|===
631
- | ID# ^| Requirement
630
+ | ID# ^| Rule
632
631
| ACS_010 a| PCIe root ports and SoC integrated downstream switch ports MUST
633
632
support the following PCIe access control services (ACS) controls:
634
633
@@ -673,7 +672,7 @@ space of an endpoint or RCiEP.
673
672
[width=100%]
674
673
[%header, cols="5,25"]
675
674
|===
676
- | ID# ^| Requirement
675
+ | ID# ^| Rule
677
676
| ADR_010 | The host bridge MUST request IOMMU translations for addresses
678
677
(Translated, Untranslated, or a PCIe ATS address translation
679
678
request) used in the request by endpoints and RCiEPs.
@@ -744,7 +743,7 @@ messages or completions.
744
743
[width=100%]
745
744
[%header, cols="5,25"]
746
745
|===
747
- | ID# ^| Requirement
746
+ | ID# ^| Rule
748
747
| IDR_010 | Configuration requests from endpoints and RCiEP MUST be treated as
749
748
Unsupported Requests.
750
749
@@ -766,7 +765,7 @@ messages or completions.
766
765
[width=100%]
767
766
[%header, cols="5,25"]
768
767
|===
769
- | ID# ^| Requirement
768
+ | ID# ^| Rule
770
769
| CCS_010 | The host bridge MUST enforce PCIe memory ordering rules and SHOULD
771
770
support the relaxed ordering (RO) and ID-based ordering (IDO).
772
771
2+| _An implementation may occasionally or never permit the relaxations allowed
@@ -836,7 +835,7 @@ mechanism in PCIe.
836
835
837
836
[%header, cols="5,25"]
838
837
|===
839
- | ID# ^| Requirement
838
+ | ID# ^| Rule
840
839
| MSI_010 | Message Signaled Interrupts MUST be supported.
841
840
842
841
| MSI_020 | SoC MUST NOT support INTx virtual wire based interrupt signaling.
@@ -853,7 +852,7 @@ mechanism in PCIe.
853
852
[width=100%]
854
853
[%header, cols="5,25"]
855
854
|===
856
- | ID# ^| Requirement
855
+ | ID# ^| Rule
857
856
| PTM_010 | PCIe root ports MAY support PCIe PTM capability.
858
857
2+| _Several applications such as instrumentation, media servers, telecom
859
858
servers, etc. require high precision monitoring and tracking of time. The
@@ -883,7 +882,7 @@ mechanism in PCIe.
883
882
[width=100%]
884
883
[%header, cols="5,25"]
885
884
|===
886
- | ID# ^| Requirement
885
+ | ID# ^| Rule
887
886
| AER_010 | PCIe root ports MUST support advanced error reporting (AER)
888
887
capability for reporting errors from connected devices or the
889
888
errors detected by the root port itself.
@@ -928,7 +927,7 @@ mechanism in PCIe.
928
927
[width=100%]
929
928
[%header, cols="5,25"]
930
929
|===
931
- | ID# ^| Requirement
930
+ | ID# ^| Rule
932
931
| VSR_010 a| Vendor specific registers in the root ports, host bridge, RCiEP,
933
932
and RCRB MUST be implemented using one or more of the following
934
933
capabilities:
@@ -958,7 +957,7 @@ mechanism in PCIe.
958
957
[width=100%]
959
958
[%header, cols="5,25"]
960
959
|===
961
- | ID# ^| Requirement
960
+ | ID# ^| Rule
962
961
| SID_010 | SoC-integrated PCIe devices MUST implement all software visible
963
962
rules defined by the PCIe specification 6.0 for an EP or RCiEP as
964
963
applicable.
@@ -1026,7 +1025,7 @@ mechanism in PCIe.
1026
1025
[width=100%]
1027
1026
[%header, cols="5,25"]
1028
1027
|===
1029
- | ID# ^| Requirement
1028
+ | ID# ^| Rule
1030
1029
| RAS_010 | The level of RAS implemented by the SoC is `UNSPECIFIED`.
1031
1030
2+| _The level of RAS implemented by an SoC depends on the reliability goals
1032
1031
established for the SoC, which are commonly measured using metrics such as
@@ -1153,7 +1152,7 @@ mechanism in PCIe.
1153
1152
increment the corrected error counter only if the error differs from a
1154
1153
previously reported one. Additionally, some hardware units could
1155
1154
incorporate low-pass filters like leaky buckets, which regulate the rate at
1156
- which corrected errors are reported and counted. This requirement pertains
1155
+ which corrected errors are reported and counted. This rule pertains
1157
1156
to corrected errors tracked by the error record once the hardware component
1158
1157
determines reporting and counting based on its specific filtering rules._
1159
1158
|===
@@ -1173,7 +1172,7 @@ and more.
1173
1172
[width=100%]
1174
1173
[%header, cols="5,25"]
1175
1174
|===
1176
- | ID# ^| Requirement
1175
+ | ID# ^| Rule
1177
1176
| QOS_010 | The SoC SHOULD incorporate QoS mechanisms to mitigate unwarranted
1178
1177
performance interference that arises when multiple workloads access
1179
1178
shared resources like caches and system memory.
@@ -1280,7 +1279,7 @@ data centers and enterprises.
1280
1279
[width=100%]
1281
1280
[%header, cols="5,25"]
1282
1281
|===
1283
- | ID# ^| Requirement
1282
+ | ID# ^| Rule
1284
1283
| MNG_010 | The SoC SHOULD incorporate support for an x1 PCIe lane, preferably
1285
1284
Gen 5, but at least Gen 3, to establish a connection with the BMC.
1286
1285
2+| _This interface is commonly linked to a BMC as a PCIe endpoint, serving
@@ -1314,7 +1313,7 @@ data centers and enterprises.
1314
1313
[width=100%]
1315
1314
[%header, cols="5,25"]
1316
1315
|===
1317
- | ID# ^| Requirement
1316
+ | ID# ^| Rule
1318
1317
| SPM_010 a| Significant caches within the SoC SHOULD incorporate an HPM capable
1319
1318
of counting:
1320
1319
@@ -1358,7 +1357,7 @@ data centers and enterprises.
1358
1357
[width=100%]
1359
1358
[%header, cols="5,25"]
1360
1359
|===
1361
- | ID# ^| Requirement
1360
+ | ID# ^| Rule
1362
1361
| SEC_010 a| The Server SoC MUST implement a hardware RoT as the _primary_ root
1363
1362
of trust.
1364
1363
2+| _A root of trust (RoT) is the foundation on which all secure operations of a
0 commit comments