1. tech-unixplatformspec@lists.riscv.org
  2. Messages
   Date   
SBI Debug Trigger Extension Proposal (Draft v4)

Anup Patel
 

Hi All,

To support native debugging in S-mode and VS-mode, we need
a SBI Debug Trigger extension which complements the Sdtrig
extension.

Below is the draft proposal of the SBI Debug Trigger Extension...

Many thanks to folks (CC'ed here) who provided early feedback
and helped improve this proposal.

The proposed SBI extension is highly inspired from Linux HW
breakpoint infrastructure which is built on top of the Linux
perf subsystem.

Please review and provide your feedback.

Regards,
Anup

SBI Debug Trigger Extension (EID #0x44425452 "DBTR")
====================================================

The RISC-V Sdtrig ISA extension allows machine-mode software to
directly configure debug triggers which in-turn allows native
(or hosted) debugging without any external debugger.

The SBI debug trigger extension defines a SBI based abstraction to
provide native debugging for supervisor-mode software such that it:
1) Suitable for the rich operating systems and hypervisors running
in supervisor-mode.
2) Allows Guest (VS-mode) and Hypervisor (HS-mode) to share debug
triggers on a HART.

Some of the functions defined by this SBI extension have physical
memory as parameters for input and/or output. This physical memory
parameter MUST satisfy following requirements:
1) The base address of the physical memory MUST be 16-byte aligned
so that it is machine word aligned for RV32, RV64 and RV128.
2) The SBI implementation MUST check that the supervisor-mode
software is allowed to read and write the specified physical
memory on the calling HART.
3) The SBI implementation MUST access the specified physical
memory using the PMA attribute.
Note: If the supervisor-mode software access the same physical
memory using memory type different from PMA then a loss of
coherence or unexpected memory ordering may occur and the
invoking software should follow the rules and sequences defined
in the RISC-V Svpbmt specification to prevent the loss of
coherence and memory ordering.

Each debug trigger is assigned a logical number called `trig_idx`
by the SBI implementation. If total number of debug triggers are
`N` then `0 <= trig_idx < N`.

The configuration of each debug trigger is expressed by three
entities `trig_tdata1`, `trig_tdata2`, and `trig_tdata3` which
are same as the `tdata1`, `tdata2`, and `tdata3` CSRs defined by
the RISC-V Sdtrig specification but with the following additional
constraints:
1) The `trig_tdata1.dmode` bit should always be zero.
2) The `trig_tdata1.m` bit should always be zero.
3) The `trig_tdata1.action` bits should always be zero.

The SBI implementation must also maintain an additional software
state for each debug trigger called `trig_state` which is encoded
as follows:
reserved [XLEN-1:5] - Reserved for future use.
vs [4:4] - Saved copy of the `trig_tdata1.vs` bit.
vu [3:3] - Saved copy of the `trig_tdata1.vu` bit.
s [2:2] - Saved copy of the `trig_tdata1.s` bit.
u [1:1] - Saved copy of the `trig_tdata1.u` bit.
mapped [0:0] - When set, the trigger has been mapped to
some HW debug trigger.

Function: Get number of triggers (FID #0)
-----------------------------------------

struct sbiret sbi_debug_num_triggers(unsigned long trig_tdata1)

Get the number of per-HART debug triggers which can support the
trigger configuration specified by `trig_tdata1` parameter.

The SBI implementation must return the total number of per-HART
debug triggers when `trig_tdata1 == 0`.

Returns the number of per-HART debug triggers in `sbiret.value`
and always returns SBI_SUCCESS in `sbiret.error`.

Function: Read triggers (FID #1)
--------------------------------

struct sbiret sbi_debug_read_triggers(unsigned long trig_idx_base,
unsigned long trig_count,
unsigned long out_addr_div_by_16)

Read the trigger state and configuration for a range of debug triggers
specified by `trig_idx_base` and `trig_count` parameters on the calling
HART.

The state and configuration of triggers will be written to an output
memory of size `trig_count * 4 * (XLEN / 8)` bytes. The base physical
address of the output memory right shifted by 4 is specified by the
`out_addr_div_by_16` parameter.

For i'th debug trigger in the specified range, the trigger configuration
consists of four XLEN-bit words written in little-endian format at
`offset = i * 4 * (XLEN / 8)` of the output memory. The contents of
output trigger configuration words are as follows:
word[0] = trig_state
word[1] = trig_tdata1
word[2] = trig_tdata2
word[3] = trig_tdata3

The possible error codes returned in `sbiret.error` are below.

Errors:
SBI_SUCCESS - Configuration read successfully.
SBI_ERR_INVALID_ADDRESS - The physical memory pointed by the
`out_addr_div_by_16` parameter is not
accessible to supervisor-mode.
SBI_ERR_INVALID_PARAM - Either `trig_idx_base + trig_count` or
`trig_idx_base` is greater than the total
number of triggers.

Function: Install triggers (FID #2)
-----------------------------------

struct sbiret sbi_debug_install_triggers(unsigned long trig_count,
unsigned long in_addr_div_by_16,
unsigned long out_addr_div_by_16)

Install debug triggers based on an array of trigger configurations
specified through an input memory and return the `trig_idx` of each
installed debug trigger in an output memory.

The `trig_count` parameter represents the number of debug triggers
to be installed.

The base physical address of the input memory right shifted by 4
is specified by the `in_addr_div_by_16` parameter and the size of
input memory is `trig_count * 4 * (XLEN / 8)` bytes. The i'th input
trigger configuration is at `offset = i * 4 * (XLEN / 8)` of the
input memory consisting of four consecutive XLEN-bit words in
little-endian format. The contents of each trigger configuration
are follows:
word[0] = Reserved for future use and should be zero
word[1] = trig_tdata1
word[2] = trig_tdata2
word[3] = trig_tdata3

The base physical address of the output memory right shifted by
4 is specified by the `out_addr_div_by_16` parameter and the size
of output memory is `trig_count * (XLEN / 8)` bytes. The `trig_idx`
assigned for the i'th input trigger configuration is written at
`offset = i * (XLEN / 8)` of the output memory in little-endian
format.

For each input trigger configuration in the input memory, the SBI
implementation should:
1) Map a free `trig_idx` with an unused HW debug trigger that
matches the trigger configuration.
2) Save a copy of the `trig_tdata1.vs`, `trig_tdata1.vu`,
`trig_tdata1.s`, and `trig_tdata.u` bits in `trig_state`.
3) Update the `tdata1`, `tdata2`, and `tdata3` CSRs of the
HW debug trigger.

Additionally for each input trigger configuration chain, the
SBI implementation MUST assign contiguous `trig_idx` values
and contiguous HW debug triggers when installing debug triggers.

The last trigger configuration in the input memory MUST not have
`trig_tdata1.chain == 1` for `trig_tdata1.type = 2 or 6` to prevent
incomplete trigger configuration chain in input memory.

This function succeeds and output memory is written with `trig_idx`
values only if the SBI implementation is able to install a debug
trigger for each input trigger configuration.

The possible error codes returned in `sbiret.error` are below.

Errors:
SBI_SUCCESS - Triggers installed successfully. The
`sbiret.value` will be set to zero.
SBI_ERR_INVALID_ADDRESS - The physical memory pointed by the
`in_addr_div_by_16` or `out_addr_div_by_16`
parameter is not accessible to supervisor-mode.
The `sbiret.value` will be set to zero.
SBI_ERR_INVALID_PARAM - One of the input trigger configuration has
an invalid value. The `sbiret.value` will
be set to the array index of failing input
trigger configuration.
SBI_ERR_NOT_SUPPORTED - One of the input trigger configuration can't
be programmed due to unimplemented optional
bits in `tdata1`, `tdata2`, or `tdata3` CSRs.
The `sbiret.value` will be set to the array
index of failing input trigger configuration.
SBI_ERR_FAILED - Failed to assign `trig_idx` or HW debug
trigger for one of the input trigger
configuration. The `sbiret.value` will be
set to the array index of failing input
trigger configuration.

Function: Uninstall a set of triggers (FID #3)
----------------------------------------------

struct sbiret sbi_debug_uninstall_triggers(unsigned long trig_idx_base,
unsigned long trig_idx_mask)

Uninstall a set of debug triggers specified by the `trig_idx_base` and
`trig_idx_mask` parameters on the calling HART.

For each debug trigger in the specified set of debug triggers, the
SBI implementation should:
1) Clear the `tdata1`, `tdata2`, and `tdata3` CSRs of the mapped
HW debug trigger.
2) Clear the `trig_state` of the debug trigger.
3) Unmap the HW debug trigger from `trig_idx` and free both of
them for future trigger installations.

The possible error codes returned in `sbiret.error` are below.

Errors:
SBI_SUCCESS - Trigger uninstalled successfully.
SBI_ERR_INVALID_PARAM - One of the `trig_idx` in the specified set
of debug triggers is either:
1) Greater than the total number of triggers.
2) Represents a debug trigger that is not
mapped to any HW debug trigger.

Function: Enable a set of triggers (FID #4)
-------------------------------------------

struct sbiret sbi_debug_enable_triggers(unsigned long trig_idx_base,
unsigned long trig_idx_mask)

Enable a set of debug triggers specified by `trig_idx_base` and
`trig_idx_mask` parameters on the calling HART.

To enable a debug trigger, SBI implementation MUST restore the
`vs`, `vu`, `s`, and `u` bits of the mapped HW debug trigger
from their saved copy in `trig_state`.

The possible error codes returned in `sbiret.error` are below.

Errors:
SBI_SUCCESS - Trigger enabled successfully.
SBI_ERR_INVALID_PARAM - One of the `trig_idx` in the specified set
of debug triggers is either:
1) Greater than the total number of triggers.
2) Represents a debug trigger that is not
mapped to any HW debug trigger.

Function: Disable a set of triggers (FID #5)
--------------------------------------------

struct sbiret sbi_debug_disable_triggers(unsigned long trig_idx_base,
unsigned long trig_idx_mask)

Disable a set of debug triggers specified by `trig_idx_base` and
`trig_idx_mask` parameters on the calling HART.

To disable a debug trigger, SBI implementation MUST clear the `vs`,
`vu`, `s`, and `u` bits of the mapped HW debug trigger.

The possible error codes returned in `sbiret.error` are below.

Errors:
SBI_SUCCESS - Trigger disabled successfully.
SBI_ERR_INVALID_PARAM - One of the `trig_idx` in the specified set
of debug triggers is either:
1) Greater than the total number of triggers.
2) Represents a debug trigger that is not
mapped to any HW debug trigger.
Re: [RISC-V][tech-os-a-see] [RISC-V] [tech-unprivileged] Direction of Identifying Extensions

Aaron Durbin
 



On Mon, Jul 11, 2022 at 7:54 PM Greg Favor <gfavor@...> wrote:
On Mon, Jul 11, 2022 at 11:28 AM Aaron Durbin <adurbin@...> wrote:
On Mon, Jul 11, 2022 at 12:19 PM Greg Favor <gfavor@...> wrote:
What I'm saying (or at least my leaning) is that discovery methods should use K-V pairs as they do today (with standardized K names desired).  Whereas Profiles have a more limited goal of having extension-like names for specific K-V pairs.  The latter names are different animals than the former - not only semantically but also in being short not-so-self-descriptive acronyms (not exactly what you want for K names).

In short, keep these two areas uncoupled from each other and let each struggle down its own path without getting further delayed by each other.

We would then have 2 different namespaces representing many similar constructs. Doesn't that make things more complicated? e.g. If everything that is a non-traditional extension in the Profile has a different way to identify the construct we have 2 different ways of identification. 

But the problem is that a profile is just putting an extension-style name (e.g. Zi* or Ss*) to a specific combination of an option or parameter, and a specific value for that option or parameter.  Whereas in discovery structures one would choose at least somewhat descriptive intuitive names for options and parameters, and would use booleans, integers, enums, etc. - as appropriate - to specify values for the options and parameters.  The latter would be clean and clear and readily captured by a schema.

My point is that the information, regardless of format, needs to be conveyed in some manner to the OS. There's no getting around that aspect. We need to identify 1. all the things the OS needs and 2. how to convey that information for consumption. We're currently talking about #1 and whether a name or whatever is applied to a construct that construct still needs to be properly conveyed. The Profile proposal in its current form is providing names to some of those constructs. Whether that continues and we embrace it going forward is now up for discussion it seems, but that discussion doesn't negate the fact that we have to enumerate many things to the OS for it to make correct decisions. 


Instead using a very large number of terse extension-style names that cover the combinatorial namespace of all allowed K-V pairs, let alone needing to define a universal naming scheme to follow across the entire architecture now and into the future, seems to result in less comprehensible discovery info and will push out even farther into the future having all this worked out.  Whereas developing simple conventions for naming of options and parameters, and for specifying allowed values, and then having each TG that is developing an extension define the specific names (and allowed values) for the options and parameters of the extension, is relatively straightforward and scalable.  This is part of what tech-config was trying to enable.  (For past extensions, there of course is remedial work to be done, but the idea remains.)

I don't think the having a name for something or not reduces comprehensibility of discovery info. It all needs to be enumerated in one form or another. The question is how to get to a consistent way to describing these constructs across the ecosystem. The divide and conquer approach will work, but I'm not sure we'll get consistent names. If that's not a goal, so be it. Either way, we need to go and backfill things that don't have names/identifiers.
 

In short we need to divide-and-conquer the overall issue and not pile everything into one big ball of yarn.  And let's focus on supporting UD/DT/ACPI discovery methods.  Practically speaking, let's leave the relatively very limited naming in profiles of line item mandates, to the profiles.  Let these be two different and only superficially related problems.   (Ok, I'll step off my soapbox now.)

In order to focus on the discovery methods we need to have the set of things needed by each level the software to be enumerated and provide a way to name/identify those pieces. As you noted above, there may be different types needed, but that requirement falls out from the data gatherig step. However, we'll still need to figure out how to encode all this information anyway for DT and ACPI.
 

Greg
Re: [RISC-V] [tech-aia] [RISC-V][tech-os-a-see] [RISC-V] [tech-unixplatformspec] Review request for ACPI ECRs

Philipp Tomsich <philipp.tomsich@...>
 

Allen,

Rereading your comment, I realised that something was off (as I didn't
have a line 244) — looks like I missed the additional tabs on the
inline preview of the sheet in the mail program.
Thanks for point that out!

Philipp.

On Tue, 12 Jul 2022 at 02:09, Allen Baum <allen.baum@...> wrote:

I'm not sure what you're referring to
Line 244 list Zihintpause
Lines 180-188 list Zvlsseg, Zvamo, Zvediv, Zvqmac, Zve32x, Zve32f, Zve64x, Zve64f, Zve64d
Lines 138-147 list Zba, Zbb, Zbc, Zbe, Zbf, Zbm, Zbp, Zbr, Zbs, Zbt (many of which are not yet ratified)

On Mon, Jul 11, 2022 at 1:40 PM Philipp Tomsich <philipp.tomsich@...> wrote:

Allen,

There is no 'RVB'... and even the basic options on Zb[abcs] open up 16
possibilities.
Similarly 'V' is a not a single option, but multiple Zv* options.
Finally, Zihintpause is missing.

Cheers,
Philipp.

On Mon, 11 Jul 2022 at 22:16, Allen Baum <allen.baum@...> wrote:

FYI: here is a rough initial draft of the architectural options that I've been able to dig up.
I know that this is very, very likely to be incomplete.

This is written from the perspective of how to make our reference model model the behavior of a device under test.
That's a very different use case than what is needed for ACPI, so I have no clue which of these options require a
parameter to be supplied (as opposed to defining the parameter) and which are don't cares.

Most of this was derived from the Imperas simulator options.

Sail will eventually have CSR definitions that will be supplied from a YAML formatted riscv-config file that describes the properties of a CSR
(e.g. which are implemented, which bits are RW, what the RO values are, what the ranges of legal values are for RW fields, and what the illegal->legal mapping function is for WARL fields). Entries marked "use CSR definitions" means that don't need to be given names, but can be derived from that YAML description (though the profiles names are restrictions on top of the architecturally allowed behaviors for a particular profile)

The unified discovery spec will be encoding some of these as well, and perhaps some other things which could be derived from this.
Re: [RISC-V][tech-os-a-see] [RISC-V] [tech-unprivileged] Direction of Identifying Extensions

Greg Favor
 

On Mon, Jul 11, 2022 at 11:28 AM Aaron Durbin <adurbin@...> wrote:
On Mon, Jul 11, 2022 at 12:19 PM Greg Favor <gfavor@...> wrote:
What I'm saying (or at least my leaning) is that discovery methods should use K-V pairs as they do today (with standardized K names desired).  Whereas Profiles have a more limited goal of having extension-like names for specific K-V pairs.  The latter names are different animals than the former - not only semantically but also in being short not-so-self-descriptive acronyms (not exactly what you want for K names).

In short, keep these two areas uncoupled from each other and let each struggle down its own path without getting further delayed by each other.

We would then have 2 different namespaces representing many similar constructs. Doesn't that make things more complicated? e.g. If everything that is a non-traditional extension in the Profile has a different way to identify the construct we have 2 different ways of identification. 

But the problem is that a profile is just putting an extension-style name (e.g. Zi* or Ss*) to a specific combination of an option or parameter, and a specific value for that option or parameter.  Whereas in discovery structures one would choose at least somewhat descriptive intuitive names for options and parameters, and would use booleans, integers, enums, etc. - as appropriate - to specify values for the options and parameters.  The latter would be clean and clear and readily captured by a schema.

Instead using a very large number of terse extension-style names that cover the combinatorial namespace of all allowed K-V pairs, let alone needing to define a universal naming scheme to follow across the entire architecture now and into the future, seems to result in less comprehensible discovery info and will push out even farther into the future having all this worked out.  Whereas developing simple conventions for naming of options and parameters, and for specifying allowed values, and then having each TG that is developing an extension define the specific names (and allowed values) for the options and parameters of the extension, is relatively straightforward and scalable.  This is part of what tech-config was trying to enable.  (For past extensions, there of course is remedial work to be done, but the idea remains.)

In short we need to divide-and-conquer the overall issue and not pile everything into one big ball of yarn.  And let's focus on supporting UD/DT/ACPI discovery methods.  Practically speaking, let's leave the relatively very limited naming in profiles of line item mandates, to the profiles.  Let these be two different and only superficially related problems.   (Ok, I'll step off my soapbox now.)

Greg
Re: [RISC-V][privileged-software] [RISC-V][tech-os-a-see] [RISC-V] [tech-unprivileged] Direction of Identifying Extensions

Greg Favor
 

On Mon, Jul 11, 2022 at 2:40 PM Philipp Tomsich <philipp.tomsich@...> wrote:
Duplicating the information (i.e. having the individual extensions marked and the profile itself indicated) seems like an error-prone proposal, unless we mandate/enforce consistency checking.
Given that this does not provide a significant benefit, it seems safer to stick with the canonical form listing the base and individual extensions.

What is the expected benefit from duplicating this information?

This was a point I tried to make in another thread about tech-config and profiles.  Support for a profile implies support for all the extensions mandated by the profile (but still leaves unspecified the support for further extensions).

So the question - that should be answered by the direct software user community of tech-config discovery structures - is whether the rest of a tech-config structure must only specify supported extensions not mandated by a profile, or must it also specify (redundantly) all the supported individual extensions, or is any degree of redundancy from zero to full allowed?

For example, having full redundancy would allow someone to simply check for presence of a specific extension of interest - without having to also check profile presence bits and understand what each of them implies.  Conversely, does someone feel that any redundancy creates confusion or complications for software using tech-config structures?

The rest of us can raise questions, but the direct users of tech-config are the best ones to determine the right answer.

Greg
Re: [RISC-V] [tech-aia] [RISC-V][tech-os-a-see] [RISC-V] [tech-unixplatformspec] Review request for ACPI ECRs

Allen Baum
 

Ah, never mind. My version of the spreadsheet opens to sheet3, and that's what has the complete list. The other sheets were used to develop this one.
I will rename the sheet and put it in the front. and also move unratified extensions to another sheet

On Mon, Jul 11, 2022 at 5:09 PM Allen Baum via lists.riscv.org <allen.baum=esperantotech.com@...> wrote:
I'm not sure what you're referring to
Line 244 list Zihintpause
Lines 180-188 list Zvlsseg, Zvamo, Zvediv, Zvqmac, Zve32x, Zve32f, Zve64x, Zve64f, Zve64d
Lines 138-147 list Zba, Zbb, Zbc, Zbe, Zbf, Zbm, Zbp, Zbr, Zbs, Zbt (many of which are not yet ratified)

On Mon, Jul 11, 2022 at 1:40 PM Philipp Tomsich <philipp.tomsich@...> wrote:
Allen,

There is no 'RVB'... and even the basic options on Zb[abcs] open up 16
possibilities.
Similarly 'V' is a not a single option, but multiple Zv* options.
Finally, Zihintpause is missing.

Cheers,
Philipp.

On Mon, 11 Jul 2022 at 22:16, Allen Baum <allen.baum@...> wrote:
>
> FYI: here is a rough initial draft of the architectural options that I've been able to dig up.
> I know that this is very, very likely to be incomplete.
>
> This is written from the perspective of how to make our reference model model the behavior of a device under test.
> That's a very different use case than what is needed for ACPI,  so I have no clue which of these options require a
> parameter to be supplied (as opposed to defining the parameter) and which are don't cares.
>
> Most of this was derived from the Imperas simulator options.
>
> Sail will eventually have CSR definitions that will be supplied from a YAML formatted riscv-config file that describes the properties of a CSR
> (e.g. which are implemented, which bits are RW, what the RO values are, what the ranges of legal values are for RW fields, and what the illegal->legal mapping function is for WARL fields). Entries marked "use CSR definitions" means that don't need to be given names, but can be derived from that YAML description (though the profiles names are restrictions on top of the architecturally allowed behaviors for a particular profile)
>
> The unified discovery spec will be encoding some of these as well, and perhaps some other things which could be derived from this.

?
arch-option list.xlsx

Re: [RISC-V] [tech-aia] [RISC-V][tech-os-a-see] [RISC-V] [tech-unixplatformspec] Review request for ACPI ECRs

Allen Baum
 

They are marked in the spreadsheet as "Ignored in v1.0", for what it's worth, 
but I could make it more obvious (or better, move them to a separate sheet-- done)

On Mon, Jul 11, 2022 at 5:53 PM Greg Favor <gfavor@...> wrote:
On Mon, Jul 11, 2022, 5:09 PM Allen Baum <allen.baum@...> wrote:
I'm not sure what you're referring to
Line 244 list Zihintpause
Lines 180-188 list Zvlsseg, Zvamo, Zvediv, Zvqmac, Zve32x, Zve32f, Zve64x, Zve64f, Zve64d
Lines 138-147 list Zba, Zbb, Zbc, Zbe, Zbf, Zbm, Zbp, Zbr, Zbs, Zbt (many of which are not yet ratified)

Alan,

Given the recent troubles with people having taken extensions in the priv/unpriv spec documents as being official even though they are listed as unratified, I would suggest not listing extensions that have no active plan to be ratified nor any group actively or planning to actively pursue ratification - in particular the unratified bitmanip extensions.

Similarly, no dev effort should be spent on these possibly never to be ratified extension proposals.

Better to stick to ratified extensions, or clearly mark extensions that are working towards being ratified over the coming year.

Greg 



On Mon, Jul 11, 2022 at 1:40 PM Philipp Tomsich <philipp.tomsich@...> wrote:
Allen,

There is no 'RVB'... and even the basic options on Zb[abcs] open up 16
possibilities.
Similarly 'V' is a not a single option, but multiple Zv* options.
Finally, Zihintpause is missing.

Cheers,
Philipp.

On Mon, 11 Jul 2022 at 22:16, Allen Baum <allen.baum@...> wrote:
>
> FYI: here is a rough initial draft of the architectural options that I've been able to dig up.
> I know that this is very, very likely to be incomplete.
>
> This is written from the perspective of how to make our reference model model the behavior of a device under test.
> That's a very different use case than what is needed for ACPI,  so I have no clue which of these options require a
> parameter to be supplied (as opposed to defining the parameter) and which are don't cares.
>
> Most of this was derived from the Imperas simulator options.
>
> Sail will eventually have CSR definitions that will be supplied from a YAML formatted riscv-config file that describes the properties of a CSR
> (e.g. which are implemented, which bits are RW, what the RO values are, what the ranges of legal values are for RW fields, and what the illegal->legal mapping function is for WARL fields). Entries marked "use CSR definitions" means that don't need to be given names, but can be derived from that YAML description (though the profiles names are restrictions on top of the architecturally allowed behaviors for a particular profile)
>
> The unified discovery spec will be encoding some of these as well, and perhaps some other things which could be derived from this.
Re: [RISC-V] [tech-aia] [RISC-V][tech-os-a-see] [RISC-V] [tech-unixplatformspec] Review request for ACPI ECRs

Greg Favor
 

On Mon, Jul 11, 2022, 5:09 PM Allen Baum <allen.baum@...> wrote:
I'm not sure what you're referring to
Line 244 list Zihintpause
Lines 180-188 list Zvlsseg, Zvamo, Zvediv, Zvqmac, Zve32x, Zve32f, Zve64x, Zve64f, Zve64d
Lines 138-147 list Zba, Zbb, Zbc, Zbe, Zbf, Zbm, Zbp, Zbr, Zbs, Zbt (many of which are not yet ratified)

Alan,

Given the recent troubles with people having taken extensions in the priv/unpriv spec documents as being official even though they are listed as unratified, I would suggest not listing extensions that have no active plan to be ratified nor any group actively or planning to actively pursue ratification - in particular the unratified bitmanip extensions.

Similarly, no dev effort should be spent on these possibly never to be ratified extension proposals.

Better to stick to ratified extensions, or clearly mark extensions that are working towards being ratified over the coming year.

Greg 



On Mon, Jul 11, 2022 at 1:40 PM Philipp Tomsich <philipp.tomsich@...> wrote:
Allen,

There is no 'RVB'... and even the basic options on Zb[abcs] open up 16
possibilities.
Similarly 'V' is a not a single option, but multiple Zv* options.
Finally, Zihintpause is missing.

Cheers,
Philipp.

On Mon, 11 Jul 2022 at 22:16, Allen Baum <allen.baum@...> wrote:
>
> FYI: here is a rough initial draft of the architectural options that I've been able to dig up.
> I know that this is very, very likely to be incomplete.
>
> This is written from the perspective of how to make our reference model model the behavior of a device under test.
> That's a very different use case than what is needed for ACPI,  so I have no clue which of these options require a
> parameter to be supplied (as opposed to defining the parameter) and which are don't cares.
>
> Most of this was derived from the Imperas simulator options.
>
> Sail will eventually have CSR definitions that will be supplied from a YAML formatted riscv-config file that describes the properties of a CSR
> (e.g. which are implemented, which bits are RW, what the RO values are, what the ranges of legal values are for RW fields, and what the illegal->legal mapping function is for WARL fields). Entries marked "use CSR definitions" means that don't need to be given names, but can be derived from that YAML description (though the profiles names are restrictions on top of the architecturally allowed behaviors for a particular profile)
>
> The unified discovery spec will be encoding some of these as well, and perhaps some other things which could be derived from this.
Re: [RISC-V] [tech-aia] [RISC-V][tech-os-a-see] [RISC-V] [tech-unixplatformspec] Review request for ACPI ECRs

Greg Favor
 

Allen,

You should check with the tpm's as to where there is a complete list of all ratified extensions.  Somewhat equivalently the RVA22 S and U profiles, by requirement, list and categorize all S and U related extensions; only M related extensions aren't covered.

Cross-checking between one of these and your list would probably a good thing to do to find anything you missed.

Greg 

On Mon, Jul 11, 2022, 1:40 PM Philipp Tomsich <philipp.tomsich@...> wrote:
Allen,

There is no 'RVB'... and even the basic options on Zb[abcs] open up 16
possibilities.
Similarly 'V' is a not a single option, but multiple Zv* options.
Finally, Zihintpause is missing.

Cheers,
Philipp.

On Mon, 11 Jul 2022 at 22:16, Allen Baum <allen.baum@...> wrote:
>
> FYI: here is a rough initial draft of the architectural options that I've been able to dig up.
> I know that this is very, very likely to be incomplete.
>
> This is written from the perspective of how to make our reference model model the behavior of a device under test.
> That's a very different use case than what is needed for ACPI,  so I have no clue which of these options require a
> parameter to be supplied (as opposed to defining the parameter) and which are don't cares.
>
> Most of this was derived from the Imperas simulator options.
>
> Sail will eventually have CSR definitions that will be supplied from a YAML formatted riscv-config file that describes the properties of a CSR
> (e.g. which are implemented, which bits are RW, what the RO values are, what the ranges of legal values are for RW fields, and what the illegal->legal mapping function is for WARL fields). Entries marked "use CSR definitions" means that don't need to be given names, but can be derived from that YAML description (though the profiles names are restrictions on top of the architecturally allowed behaviors for a particular profile)
>
> The unified discovery spec will be encoding some of these as well, and perhaps some other things which could be derived from this.
Re: [RISC-V] [tech-aia] [RISC-V][tech-os-a-see] [RISC-V] [tech-unixplatformspec] Review request for ACPI ECRs

Allen Baum
 

I'm not sure what you're referring to
Line 244 list Zihintpause
Lines 180-188 list Zvlsseg, Zvamo, Zvediv, Zvqmac, Zve32x, Zve32f, Zve64x, Zve64f, Zve64d
Lines 138-147 list Zba, Zbb, Zbc, Zbe, Zbf, Zbm, Zbp, Zbr, Zbs, Zbt (many of which are not yet ratified)

On Mon, Jul 11, 2022 at 1:40 PM Philipp Tomsich <philipp.tomsich@...> wrote:
Allen,

There is no 'RVB'... and even the basic options on Zb[abcs] open up 16
possibilities.
Similarly 'V' is a not a single option, but multiple Zv* options.
Finally, Zihintpause is missing.

Cheers,
Philipp.

On Mon, 11 Jul 2022 at 22:16, Allen Baum <allen.baum@...> wrote:
>
> FYI: here is a rough initial draft of the architectural options that I've been able to dig up.
> I know that this is very, very likely to be incomplete.
>
> This is written from the perspective of how to make our reference model model the behavior of a device under test.
> That's a very different use case than what is needed for ACPI,  so I have no clue which of these options require a
> parameter to be supplied (as opposed to defining the parameter) and which are don't cares.
>
> Most of this was derived from the Imperas simulator options.
>
> Sail will eventually have CSR definitions that will be supplied from a YAML formatted riscv-config file that describes the properties of a CSR
> (e.g. which are implemented, which bits are RW, what the RO values are, what the ranges of legal values are for RW fields, and what the illegal->legal mapping function is for WARL fields). Entries marked "use CSR definitions" means that don't need to be given names, but can be derived from that YAML description (though the profiles names are restrictions on top of the architecturally allowed behaviors for a particular profile)
>
> The unified discovery spec will be encoding some of these as well, and perhaps some other things which could be derived from this.
Re: [RISC-V][privileged-software] [RISC-V][tech-os-a-see] [RISC-V] [tech-unprivileged] Direction of Identifying Extensions

Philipp Tomsich <philipp.tomsich@...>
 

Duplicating the information (i.e. having the individual extensions marked and the profile itself indicated) seems like an error-prone proposal, unless we mandate/enforce consistency checking.
Given that this does not provide a significant benefit, it seems safer to stick with the canonical form listing the base and individual extensions.

What is the expected benefit from duplicating this information?

Philipp.

On Mon, 11 Jul 2022 at 23:15, Allen Baum <allen.baum@...> wrote:
Greg and I did discuss having Unified Discovery (UD) description macros for profiles, e.g. an RV22A bit that indicates support for all of the individual extensions (and option value ranges) required to meet the profile requirements. 
This was in addition to, not replacement of, all the individual requirements, so you'd still have to set the values for each individual required extension.

The idea of have a macro "bit" (perhaps) for a union of different features, and a way to  carve out/remove (and probably replace) some specific extension or option is an interesting one, but sounds fraught with possible weird corner cases.

On Mon, Jul 11, 2022 at 11:28 AM Aaron Durbin <adurbin@...> wrote:


On Mon, Jul 11, 2022 at 12:19 PM Greg Favor <gfavor@...> wrote:
On Mon, Jul 11, 2022 at 11:02 AM Aaron Durbin <adurbin@...> wrote:
On Fri, Jul 8, 2022 at 3:25 PM Greg Favor <gfavor@...> wrote:
To add on to Alan's point.  These "extension" names were created simply to represent individual line items in the Profiles.  In many cases these represent a one or two sentence statement in a Profile about an optional behavior or about a specific parameter value.
Any effort to standardize key names and value names across discovery schemes should be kept separate from the acronymic naming in Profiles of specific line items (aka specific key-value pairs).

It's very much the same exercise where the Profile specific options are more narrowly scoped. However, I would imagine the ones created for the Profiles to be a subset of the larger KV space as you named it. If we're going down that road, it would be good to have things consistent. 

What I'm saying (or at least my leaning) is that discovery methods should use K-V pairs as they do today (with standardized K names desired).  Whereas Profiles have a more limited goal of having extension-like names for specific K-V pairs.  The latter names are different animals than the former - not only semantically but also in being short not-so-self-descriptive acronyms (not exactly what you want for K names).

In short, keep these two areas uncoupled from each other and let each struggle down its own path without getting further delayed by each other.

We would then have 2 different namespaces representing many similar constructs. Doesn't that make things more complicated? e.g. If everything that is a non-traditional extension in the Profile has a different way to identify the construct we have 2 different ways of identification. 

Somewhat related, will we have a common way to refer to a Profile that identifies itself as a Profile that can be conveyed as a whole to an OS? This discussion is very much analogous to the toolchain discussion which leads to implementations that may not entirely implement a Profile may want set operations for carve outs for non-conformance. But either way, the SW that is being provided this information has to have a way to identify and keep track of these concepts anyway because it needs to deal both with Profiles proper and the exploded components regardless.

 
 If we're going to not call them extensions, I suggest coining something we can use more accurately then. I don't have any great ideas atm.

As Alan mentioned, him and I threw around names like "profentions" (aka profile extensions), although more out of amusement than practical value.

Practically speaking they should have a reasonably self-descriptive (yet short-ish) name.  Maybe simply "mandate names"?  This also avoids any confusion with extension names and with K and V names that would be used in discovery structures.  This also covers mandate items that are one sentence behavioral descriptions (of either a behavioral allowance or disallowance).


Greg
Re: [RISC-V][tech-os-a-see] [RISC-V] [tech-unprivileged] Direction of Identifying Extensions

Allen Baum
 

Greg and I did discuss having Unified Discovery (UD) description macros for profiles, e.g. an RV22A bit that indicates support for all of the individual extensions (and option value ranges) required to meet the profile requirements. 
This was in addition to, not replacement of, all the individual requirements, so you'd still have to set the values for each individual required extension.

The idea of have a macro "bit" (perhaps) for a union of different features, and a way to  carve out/remove (and probably replace) some specific extension or option is an interesting one, but sounds fraught with possible weird corner cases.

On Mon, Jul 11, 2022 at 11:28 AM Aaron Durbin <adurbin@...> wrote:


On Mon, Jul 11, 2022 at 12:19 PM Greg Favor <gfavor@...> wrote:
On Mon, Jul 11, 2022 at 11:02 AM Aaron Durbin <adurbin@...> wrote:
On Fri, Jul 8, 2022 at 3:25 PM Greg Favor <gfavor@...> wrote:
To add on to Alan's point.  These "extension" names were created simply to represent individual line items in the Profiles.  In many cases these represent a one or two sentence statement in a Profile about an optional behavior or about a specific parameter value.
Any effort to standardize key names and value names across discovery schemes should be kept separate from the acronymic naming in Profiles of specific line items (aka specific key-value pairs).

It's very much the same exercise where the Profile specific options are more narrowly scoped. However, I would imagine the ones created for the Profiles to be a subset of the larger KV space as you named it. If we're going down that road, it would be good to have things consistent. 

What I'm saying (or at least my leaning) is that discovery methods should use K-V pairs as they do today (with standardized K names desired).  Whereas Profiles have a more limited goal of having extension-like names for specific K-V pairs.  The latter names are different animals than the former - not only semantically but also in being short not-so-self-descriptive acronyms (not exactly what you want for K names).

In short, keep these two areas uncoupled from each other and let each struggle down its own path without getting further delayed by each other.

We would then have 2 different namespaces representing many similar constructs. Doesn't that make things more complicated? e.g. If everything that is a non-traditional extension in the Profile has a different way to identify the construct we have 2 different ways of identification. 

Somewhat related, will we have a common way to refer to a Profile that identifies itself as a Profile that can be conveyed as a whole to an OS? This discussion is very much analogous to the toolchain discussion which leads to implementations that may not entirely implement a Profile may want set operations for carve outs for non-conformance. But either way, the SW that is being provided this information has to have a way to identify and keep track of these concepts anyway because it needs to deal both with Profiles proper and the exploded components regardless.

 
 If we're going to not call them extensions, I suggest coining something we can use more accurately then. I don't have any great ideas atm.

As Alan mentioned, him and I threw around names like "profentions" (aka profile extensions), although more out of amusement than practical value.

Practically speaking they should have a reasonably self-descriptive (yet short-ish) name.  Maybe simply "mandate names"?  This also avoids any confusion with extension names and with K and V names that would be used in discovery structures.  This also covers mandate items that are one sentence behavioral descriptions (of either a behavioral allowance or disallowance).


Greg
Re: [RISC-V] [tech-aia] [RISC-V][tech-os-a-see] [RISC-V] [tech-unixplatformspec] Review request for ACPI ECRs

Philipp Tomsich <philipp.tomsich@...>
 

Allen,

There is no 'RVB'... and even the basic options on Zb[abcs] open up 16
possibilities.
Similarly 'V' is a not a single option, but multiple Zv* options.
Finally, Zihintpause is missing.

Cheers,
Philipp.

On Mon, 11 Jul 2022 at 22:16, Allen Baum <allen.baum@...> wrote:

FYI: here is a rough initial draft of the architectural options that I've been able to dig up.
I know that this is very, very likely to be incomplete.

This is written from the perspective of how to make our reference model model the behavior of a device under test.
That's a very different use case than what is needed for ACPI, so I have no clue which of these options require a
parameter to be supplied (as opposed to defining the parameter) and which are don't cares.

Most of this was derived from the Imperas simulator options.

Sail will eventually have CSR definitions that will be supplied from a YAML formatted riscv-config file that describes the properties of a CSR
(e.g. which are implemented, which bits are RW, what the RO values are, what the ranges of legal values are for RW fields, and what the illegal->legal mapping function is for WARL fields). Entries marked "use CSR definitions" means that don't need to be given names, but can be derived from that YAML description (though the profiles names are restrictions on top of the architecturally allowed behaviors for a particular profile)

The unified discovery spec will be encoding some of these as well, and perhaps some other things which could be derived from this.
Re: [RISC-V] [tech-aia] [RISC-V][tech-os-a-see] [RISC-V] [tech-unixplatformspec] Review request for ACPI ECRs

Allen Baum
 

FYI: here is a rough initial draft of the architectural options that I've been able to dig up.
I know that this is very, very likely to be incomplete.

This is written from the perspective of how to make our reference model model the behavior of a device under test.
That's a very different use case than what is needed for ACPI,  so I have no clue which of these options require a 
parameter to be supplied (as opposed to defining the parameter) and which are don't cares.

Most of this was derived from the Imperas simulator options. 

Sail will eventually have CSR definitions that will be supplied from a YAML formatted riscv-config file that describes the properties of a CSR
(e.g. which are implemented, which bits are RW, what the RO values are, what the ranges of legal values are for RW fields, and what the illegal->legal mapping function is for WARL fields). Entries marked "use CSR definitions" means that don't need to be given names, but can be derived from that YAML description (though the profiles names are restrictions on top of the architecturally allowed behaviors for a particular profile)

The unified discovery spec will be encoding some of these as well, and perhaps some other things which could be derived from this.

?
arch-option list.xlsx

Re: [RISC-V][tech-os-a-see] [RISC-V] [tech-unprivileged] Direction of Identifying Extensions

Aaron Durbin
 



On Mon, Jul 11, 2022 at 12:19 PM Greg Favor <gfavor@...> wrote:
On Mon, Jul 11, 2022 at 11:02 AM Aaron Durbin <adurbin@...> wrote:
On Fri, Jul 8, 2022 at 3:25 PM Greg Favor <gfavor@...> wrote:
To add on to Alan's point.  These "extension" names were created simply to represent individual line items in the Profiles.  In many cases these represent a one or two sentence statement in a Profile about an optional behavior or about a specific parameter value.
Any effort to standardize key names and value names across discovery schemes should be kept separate from the acronymic naming in Profiles of specific line items (aka specific key-value pairs).

It's very much the same exercise where the Profile specific options are more narrowly scoped. However, I would imagine the ones created for the Profiles to be a subset of the larger KV space as you named it. If we're going down that road, it would be good to have things consistent. 

What I'm saying (or at least my leaning) is that discovery methods should use K-V pairs as they do today (with standardized K names desired).  Whereas Profiles have a more limited goal of having extension-like names for specific K-V pairs.  The latter names are different animals than the former - not only semantically but also in being short not-so-self-descriptive acronyms (not exactly what you want for K names).

In short, keep these two areas uncoupled from each other and let each struggle down its own path without getting further delayed by each other.

We would then have 2 different namespaces representing many similar constructs. Doesn't that make things more complicated? e.g. If everything that is a non-traditional extension in the Profile has a different way to identify the construct we have 2 different ways of identification. 

Somewhat related, will we have a common way to refer to a Profile that identifies itself as a Profile that can be conveyed as a whole to an OS? This discussion is very much analogous to the toolchain discussion which leads to implementations that may not entirely implement a Profile may want set operations for carve outs for non-conformance. But either way, the SW that is being provided this information has to have a way to identify and keep track of these concepts anyway because it needs to deal both with Profiles proper and the exploded components regardless.

 
 If we're going to not call them extensions, I suggest coining something we can use more accurately then. I don't have any great ideas atm.

As Alan mentioned, him and I threw around names like "profentions" (aka profile extensions), although more out of amusement than practical value.

Practically speaking they should have a reasonably self-descriptive (yet short-ish) name.  Maybe simply "mandate names"?  This also avoids any confusion with extension names and with K and V names that would be used in discovery structures.  This also covers mandate items that are one sentence behavioral descriptions (of either a behavioral allowance or disallowance).


Greg
Re: [RISC-V][tech-os-a-see] [RISC-V] [tech-unprivileged] Direction of Identifying Extensions

Greg Favor
 

On Mon, Jul 11, 2022 at 11:02 AM Aaron Durbin <adurbin@...> wrote:
On Fri, Jul 8, 2022 at 3:25 PM Greg Favor <gfavor@...> wrote:
To add on to Alan's point.  These "extension" names were created simply to represent individual line items in the Profiles.  In many cases these represent a one or two sentence statement in a Profile about an optional behavior or about a specific parameter value.
Any effort to standardize key names and value names across discovery schemes should be kept separate from the acronymic naming in Profiles of specific line items (aka specific key-value pairs).

It's very much the same exercise where the Profile specific options are more narrowly scoped. However, I would imagine the ones created for the Profiles to be a subset of the larger KV space as you named it. If we're going down that road, it would be good to have things consistent. 

What I'm saying (or at least my leaning) is that discovery methods should use K-V pairs as they do today (with standardized K names desired).  Whereas Profiles have a more limited goal of having extension-like names for specific K-V pairs.  The latter names are different animals than the former - not only semantically but also in being short not-so-self-descriptive acronyms (not exactly what you want for K names).

In short, keep these two areas uncoupled from each other and let each struggle down its own path without getting further delayed by each other.
 
 If we're going to not call them extensions, I suggest coining something we can use more accurately then. I don't have any great ideas atm.

As Alan mentioned, him and I threw around names like "profentions" (aka profile extensions), although more out of amusement than practical value.

Practically speaking they should have a reasonably self-descriptive (yet short-ish) name.  Maybe simply "mandate names"?  This also avoids any confusion with extension names and with K and V names that would be used in discovery structures.  This also covers mandate items that are one sentence behavioral descriptions (of either a behavioral allowance or disallowance).


Greg
Re: [RISC-V][tech-os-a-see] [RISC-V] [tech-unprivileged] Direction of Identifying Extensions

Aaron Durbin
 



On Fri, Jul 8, 2022 at 3:25 PM Greg Favor <gfavor@...> wrote:
To add on to Alan's point.  These "extension" names were created simply to represent individual line items in the Profiles.  In many cases these represent a one or two sentence statement in a Profile about an optional behavior or about a specific parameter value.

The combinatorics of taking every optional behavior, every parameter that can take on a set of possible values, and every WARL field and the allowed range of possibilities, and then creating an "extension" name for every individual possible "key- value" pair are crazy.  (This is why people instead use key-value representations.)

For Profiles, these names have been created on an as-needed basis for specific "key-value" pairs, and no further.  Their motivation was to simply have acronymic "extension" names for each of these line items.

In contrast, discovery schemes - like Device Tree, ACPI, and RISC-V Unified Discovery - want something distinctly different.  Namely standard "key" names for options, parameters, and WARL fields; standard names for the values of a "key"; and a schema for representing a set of key-value pairs.

Any effort to standardize key names and value names across discovery schemes should be kept separate from the acronymic naming in Profiles of specific line items (aka specific key-value pairs).

It's very much the same exercise where the Profile specific options are more narrowly scoped. However, I would imagine the ones created for the Profiles to be a subset of the larger KV space as you named it. If we're going down that road, it would be good to have things consistent. 
 

Lastly, I think everyone would agree that option / parameter / WARL field names, and their associated value names, are not "extensions".  And hence should not be mixed up with the "extension" names created and used in Profiles.  Any standardization of "key" names and "value" names should simply focus on usage in discovery schemes (i.e. DT, ACPI, UD).  This needs to support all possible key-value pairs.

As to whether all these acronymic line items or "key-value" pairs in Profiles are "extensions", there is probably contention, but I would at least say that these are not extensions in the usual sense.  And they should exist and arise solely in the context of Profiles (for the intended purpose of giving an extension-like acronymic name to each line item).

If we're going to not call them extensions, I suggest coining something we can use more accurately then. I don't have any great ideas atm.
 

In short, I would suggest that it is probably best to keep discussion of these two topics separate. 

Greg


On Fri, Jul 8, 2022, 1:30 PM Allen Baum <allen.baum@...> wrote:
If you're looking at the same thing I was looking at: the "extension names" are not extensions, in the usual sense.
They are names for the values of architectural options of extensions that already exist
 (which often,  but not always, are the legal set of values of a WARL CSR field which must be implemented). 
So these are constraints on the spec'ed options that are required to run SW on the platform.


On Fri, Jul 8, 2022 at 6:41 AM Aaron Durbin <adurbin@...> wrote:
Hi All,

First off, please redirect me where the most appropriate forum is to discuss this topic. I am casting a fairly wide net, but that's just trying to cover those who are impacted.  We can convene on a single list once it's identified.

The current Profiles (https://github.com/riscv/riscv-profiles/blob/main/profiles.adoc) proposal has coined quite a few new extension names to describe behaviors in the specifications that do not have formal names in the existing specifications. This particular topic came up during our discussion on ACPI bindings for AIA. However, userland, kernel, and firmware specifications are all impacted by this topic so settling on a well understood future direction will reap rewards across the ecosystem.

1. Is this the path we see RISC-V taking for the future? Assigning an identifier to behaviors (and related parameters) within the specifications?
2. If so, where will the official lists of extensions be maintained? Will there be a single source of extension identifiers? Or do people need to look at every potential specification to determine the identifiers?
3. There are some extensions that require parameters besides a binary presence. e.g. Zic64b in the Profile proposal indicates 64 byte cache block sizes for all cache block management, prefetch, and zero operations. That's fine for the Profile, but an implementation that may use different size(s) would need to encode the size in such an identifier. Therefore, I think that we should incorporate these notions more formally when defining identifiers for extension parameters.

I'd love to hear people's thoughts on this topic as it is very important, and I think it could be a good way in formalizing identifiers to all these concepts that can be used throughout the RISC-V ecosystem.

Thank you.

-Aaron
Re: [RISC-V] [tech-unprivileged] Direction of Identifying Extensions

Allen Baum
 

Unfortunately, I don't a have single term that would adequately convey the idea. 
Greg and I were batting around some invented words
(e.g. "poption" as a portmanteau of profile-option was my first attempt. It did get better, but that's a low bar)  
but nothing that you'd find in any dictionary.

On Mon, Jul 11, 2022 at 6:48 AM Aaron Durbin <adurbin@...> wrote:


On Fri, Jul 8, 2022 at 2:30 PM Allen Baum <allen.baum@...> wrote:
If you're looking at the same thing I was looking at: the "extension names" are not extensions, in the usual sense.
They are names for the values of architectural options of extensions that already exist
 (which often,  but not always, are the legal set of values of a WARL CSR field which must be implemented). 
So these are constraints on the spec'ed options that are required to run SW on the platform.


Correct. Do you have a better term that captures such notions? I fully agree that the previous definition of an extension is not the same in this case. FWIW, the profile spec uses both 'extensions' and 'options' within the current proposal for these newly coined names.
 

On Fri, Jul 8, 2022 at 6:41 AM Aaron Durbin <adurbin@...> wrote:
Hi All,

First off, please redirect me where the most appropriate forum is to discuss this topic. I am casting a fairly wide net, but that's just trying to cover those who are impacted.  We can convene on a single list once it's identified.

The current Profiles (https://github.com/riscv/riscv-profiles/blob/main/profiles.adoc) proposal has coined quite a few new extension names to describe behaviors in the specifications that do not have formal names in the existing specifications. This particular topic came up during our discussion on ACPI bindings for AIA. However, userland, kernel, and firmware specifications are all impacted by this topic so settling on a well understood future direction will reap rewards across the ecosystem.

1. Is this the path we see RISC-V taking for the future? Assigning an identifier to behaviors (and related parameters) within the specifications?
2. If so, where will the official lists of extensions be maintained? Will there be a single source of extension identifiers? Or do people need to look at every potential specification to determine the identifiers?
3. There are some extensions that require parameters besides a binary presence. e.g. Zic64b in the Profile proposal indicates 64 byte cache block sizes for all cache block management, prefetch, and zero operations. That's fine for the Profile, but an implementation that may use different size(s) would need to encode the size in such an identifier. Therefore, I think that we should incorporate these notions more formally when defining identifiers for extension parameters.

I'd love to hear people's thoughts on this topic as it is very important, and I think it could be a good way in formalizing identifiers to all these concepts that can be used throughout the RISC-V ecosystem.

Thank you.

-Aaron
Re: Review request for ACPI ECRs

Stephano Cetola
 

Slimming down the CC list a bit, as this is off topic for some lists.

On Fri, Jul 8, 2022 at 9:00 AM Heinrich Schuchardt <heinrich.schuchardt@...> wrote:
Does the Unified DIscovery Specification exit yet? Where can I find it?

 I'm in the process of getting status from the new Unified Discovery group:

We're hoping to have a concise draft definition for plummer's conference in sept.

Cheers,
Stephano
Re: [RISC-V] [tech-unprivileged] Direction of Identifying Extensions

Aaron Durbin
 



On Fri, Jul 8, 2022 at 2:38 PM Allen Baum <allen.baum@...> wrote:
I'm also unclear of the "identifier" you mention for other unmentioned option value (e.g. your 64b cache block size example).
Who is the consumer of that identifier? Is this a value you would expect to be encoded in the config-struct of on-line discovery?
I am building a list of all architectural options that aren't controlled by CSR fields, for what it is worth, 
as well as all the named option value ranges, regardless of whether they're controlled by a CSR value or not.
The intent of this is to be able to configure the Sail model to implement the behavior they control.

Identifier in the sense of a short name/phrase that refers to these ideas. Consumers of that would be both software and day-to-day discussion as it's a succinct identifier that refers to an idea in prose.

Take the Zic64b "extension": it's a shortcut for "Cache blocks must be 64 bytes in size, naturally aligned in the address space." However, there are 3 different cache block operations. So while a Profile name could refer to the whole set of bold options/extensions, there are cases where a system doesn't adhere to a Profile. So how does one convey the cache block size (even if it's 32 bytes for all 3 cache block operations)? I think having a way to identify those architectural options can be very powerful.

FWIW, I completely thought of the the architectural test work you are doing, but I failed to directly mention it in my original email, Allen. Sorry about that.
 

On Fri, Jul 8, 2022 at 1:30 PM Allen Baum via lists.riscv.org <allen.baum=esperantotech.com@...> wrote:
If you're looking at the same thing I was looking at: the "extension names" are not extensions, in the usual sense.
They are names for the values of architectural options of extensions that already exist
 (which often,  but not always, are the legal set of values of a WARL CSR field which must be implemented). 
So these are constraints on the spec'ed options that are required to run SW on the platform.


On Fri, Jul 8, 2022 at 6:41 AM Aaron Durbin <adurbin@...> wrote:
Hi All,

First off, please redirect me where the most appropriate forum is to discuss this topic. I am casting a fairly wide net, but that's just trying to cover those who are impacted.  We can convene on a single list once it's identified.

The current Profiles (https://github.com/riscv/riscv-profiles/blob/main/profiles.adoc) proposal has coined quite a few new extension names to describe behaviors in the specifications that do not have formal names in the existing specifications. This particular topic came up during our discussion on ACPI bindings for AIA. However, userland, kernel, and firmware specifications are all impacted by this topic so settling on a well understood future direction will reap rewards across the ecosystem.

1. Is this the path we see RISC-V taking for the future? Assigning an identifier to behaviors (and related parameters) within the specifications?
2. If so, where will the official lists of extensions be maintained? Will there be a single source of extension identifiers? Or do people need to look at every potential specification to determine the identifiers?
3. There are some extensions that require parameters besides a binary presence. e.g. Zic64b in the Profile proposal indicates 64 byte cache block sizes for all cache block management, prefetch, and zero operations. That's fine for the Profile, but an implementation that may use different size(s) would need to encode the size in such an identifier. Therefore, I think that we should incorporate these notions more formally when defining identifiers for extension parameters.

I'd love to hear people's thoughts on this topic as it is very important, and I think it could be a good way in formalizing identifiers to all these concepts that can be used throughout the RISC-V ecosystem.

Thank you.

-Aaron
41 - 60 of 1846