Include SME attributes in the name mangling of types #358

kmclaughlin-arm · 2024-11-06T11:53:13Z

This change extends the name mangling of types to include the SME streaming
and ZA interface. This will avoid naming conflicts which can currently arise such
as in the following example:

  void foo(void (*f)()) { f(); }
  void foo(void (*f)() __arm_streaming) { f(); }

Without this change, both functions 'foo' above will mangle to the same
name, despite the function pointers being different.

name: Pull request
about: Technical issues, document format problems, bugs in scripts or feature proposal.

Thank you for submitting a pull request!

If this PR is about a bugfix:

Please use the bugfix label and make sure to go through the checklist below.

If this PR is about a proposal:

We are looking forward to evaluate your proposal, and if possible to
make it part of the Arm C Language Extension (ACLE) specifications.

We would like to encourage you reading through the contribution
guidelines, in particular the section on submitting
a proposal.

Please use the proposal label.

As for any pull request, please make sure to go through the below
checklist.

Checklist: (mark with X those which apply)

If an issue reporting the bug exists, I have mentioned it in the
PR (do not bother creating the issue if all you want to do is
fixing the bug yourself).
I have added/updated the SPDX-FileCopyrightText lines on top
of any file I have edited. Format is SPDX-FileCopyrightText: Copyright {year} {entity or name} <{contact informations}>
(Please update existing copyright lines if applicable. You can
specify year ranges with hyphen , as in 2017-2019, and use
commas to separate gaps, as in 2018-2020, 2022).
I have updated the Copyright section of the sources of the
specification I have edited (this will show up in the text
rendered in the PDF and other output format supported). The
format is the same described in the previous item.
I have run the CI scripts (if applicable, as they might be
tricky to set up on non-*nix machines). The sequence can be
found in the contribution
guidelines. Don't
worry if you cannot run these scripts on your machine, your
patch will be automatically checked in the Actions of the pull
request.
I have added an item that describes the changes I have
introduced in this PR in the section Changes for next
release of the section Change Control/Document history
of the document. Create Changes for next release if it does
not exist. Notice that changes that are not modifying the
content and rendering of the specifications (both HTML and PDF)
do not need to be listed.
When modifying content and/or its rendering, I have checked the
correctness of the result in the PDF output (please refer to the
instructions on how to build the PDFs
locally).
The variable draftversion is set to true in the YAML header
of the sources of the specifications I have modified.
Please DO NOT add my GitHub profile to the list of contributors
in the README page of the project.

kmclaughlin-arm · 2024-11-06T11:59:30Z

@sdesmalen-arm
@efriedma-quic
@aemerson

sdesmalen-arm · 2024-11-14T15:20:01Z

main/acle.md

+with the arguments:
+
+``` c
+  __SME_ATTRS<normal_function_type, streaming_mode, za_state, zt0_state>;


Is modelling za_state and zt0_state with separate template arguments going to work if future architecture revisions might add new SME state? Say, for example, some new state Znew would be added then I'd expect functions that don't know about Znew to be mangled as if the template argument did not exist, but when it does know about Znew for it to get mangled with an extra 0 argument. This name would then not be compatible.

Should we instead define this with a single uint64_t bitmask where we define the upper bits as leading zeroes to mean "unknown state" or "no state"? At least that ensures that the name mangling doesn't change depending on what version of the ACLE is implemented.

This would not have worked for any new attributes which could be introduced in future. As suggested I have changed the template to use a single integer to represent all of the possible SME states, leaving the upper bits reserved for future state.

sdesmalen-arm · 2024-11-21T15:18:39Z

main/acle.md

+| 0       | 0x001     | Streaming Mode         |
+| 1       | 0x002     | Streaming-Compatible   |
+| 2       | 0x004     | ZA-In                  |
+| 3       | 0x008     | ZA-InOut               |
+| 4       | 0x010     | ZA-Out                 |
+| 5       | 0x020     | ZA-Preserved           |
+| 6       | 0x040     | ZT0-In                 |
+| 7       | 0x080     | ZT0-Out                |
+| 8       | 0x100     | ZT0-InOut              |
+| 9       | 0x200     | ZT0-Preserved          |
+| 10      | 0x400     | ZA State Agnostic      |


nit: use the ACLE names of the properties, also moved __arm_agnostic to bit 2.

Suggested change

| 0 | 0x001 | Streaming Mode |

| 1 | 0x002 | Streaming-Compatible |

| 2 | 0x004 | ZA-In |

| 3 | 0x008 | ZA-InOut |

| 4 | 0x010 | ZA-Out |

| 5 | 0x020 | ZA-Preserved |

| 6 | 0x040 | ZT0-In |

| 7 | 0x080 | ZT0-Out |

| 8 | 0x100 | ZT0-InOut |

| 9 | 0x200 | ZT0-Preserved |

| 10 | 0x400 | ZA State Agnostic |

| 0 | 0x001 | __arm_streaming |

| 1 | 0x002 | __arm_streaming_compatible |

| 2 | 0x004 | __arm_agnostic("sme_za_state") |

| 3 | 0x008 | __arm_in("za") |

| 4 | 0x010 | __arm_inout("za") |

| 5 | 0x020 | __arm_out("za") |

| 6 | 0x040 | __arm_preserved("za") |

| 7 | 0x080 | __arm_in("zt0") |

| 8 | 0x100 | __arm_out("zt0") |

| 9 | 0x200 | __arm_inout("zt0") |

| 10 | 0x400 | __arm_preserved("zt0") |

We could save some bits by encoding 'none, in, out, inout, preserved' using 3 bits, rather than 4.

sdesmalen-arm · 2024-11-21T15:29:08Z

main/acle.md

+SME attributes are mangled in the same way as a template:
+
+``` c
+  template<typename, unsigned> __SME_ATTRS;


Suggested change

template<typename, unsigned> __SME_ATTRS;

template<typename, uint64_t> __SME_ATTRS;

sdesmalen-arm · 2024-11-21T15:34:13Z

main/acle.md

+
+* normal_function_type is the function type without any SME attributes.
+
+* sme_state is an unsigned 64-bit integer representing the streaming mode, ZA state and ZT0 state of the function.


Suggested change

* sme_state is an unsigned 64-bit integer representing the streaming mode, ZA state and ZT0 state of the function.

* sme_state is an unsigned 64-bit integer representing the streaming and ZA properties of the function's interface.

sdesmalen-arm · 2024-11-21T15:41:54Z

main/acle.md

+| 9       | 0x200     | ZT0-Preserved          |
+| 10      | 0x400     | ZA State Agnostic      |
+
+The upper bits 11-63 are defined as leading zeroes, reserved for representing any future interface types.


Suggested change

The upper bits 11-63 are defined as leading zeroes, reserved for representing any future interface types.

Bits 11-63 are defined to be zero by this revision of the ACLE and are reserved for future type attributes.

sdesmalen-arm · 2024-11-21T16:28:04Z

main/acle.md

+| 7       | 0x080     | __arm_out("zt0")               |
+| 8       | 0x100     | __arm_preserved("zt0")         |
+
+Bits 9-63 are defined to be zero by this revision of the ACLE and are reserved for future type attributes.


That's not what I actually meant with encoding with 3 bits. I meant:

0 = no state 1 = __arm_in 2 = __arm_out 3 = __arm_inout 4 = __arm_preserves

which is 5 values, and so can be encoded with 3 bits, rather than 4.

Because __arm_inout, __arm_in and __arm_out for the same state are all mutually exclusive, I think we should encode these separately.

ironically the actual encoding is the same, but I prefer the way this is phrased :)

sdesmalen-arm · 2024-11-21T16:28:38Z

main/acle.md

+| 2       | 0x004     | __arm_agnostic("sme_za_state") |
+| 3       | 0x008     | __arm_in("za")                 |
+| 4       | 0x010     | __arm_out("za")                |
+| 5       | 0x020     | __arm_preserved("za")          |


Sorry I misspelled this one, it's __arm_preserves (not __arm_preserved)

sdesmalen-arm

LGTM with nit addressed. Please wait for @rsandifo-arm to confirm as well.

sdesmalen-arm · 2024-11-21T17:38:02Z

main/acle.md

+| 7       | 0x080     | __arm_out("zt0")               |
+| 8       | 0x100     | __arm_preserved("zt0")         |
+
+Bits 9-63 are defined to be zero by this revision of the ACLE and are reserved for future type attributes.


ironically the actual encoding is the same, but I prefer the way this is phrased :)

sdesmalen-arm · 2024-11-21T17:38:30Z

main/acle.md

+
+* normal_function_type is the function type without any SME attributes.
+
+* sme_state is an unsigned 64-bit integer representing the streaming and ZA properties of the function's interface.


nit: do we need to maintain 80char line-wrapping in this file?

rsandifo-arm

LGTM as well.

This change extends the name mangling of types to include the SME streaming and ZA interface. This will avoid naming conflicts which can currently arise such as in the following example: void foo(void (*f)()) { f(); } void foo(void (*f)() __arm_streaming) { f(); } Without this change, both functions 'foo' above will mangle to the same name, despite the function pointers being different.

…ll attributes. - Include __arm_agnostic("sme_za_state") in the bitmask

- Removed bits for ZA/ZT0 'inout', as this can be represented by setting both 'in' and 'out'.

- Added default ZA & ZT0 states to the sme_state table.

…ion types (#114209) Similar to arm_sve_vector_bits, the mangling of function types is implemented as a pseudo template if there are any SME attributes present, i.e. `__SME_ATTRS<normal_function_type, sme_state>` For example, the following function: `void f(svint8_t (*fn)() __arm_streaming) { fn(); }` would be mangled as: `_Z1fP11__SME_ATTRSIFu10__SVInt8_tELj1EE` See ARM-software/acle#358

kmclaughlin-arm mentioned this pull request Nov 6, 2024

Include SME attributes in the name mangling of types ARM-software/abi-aa#290

Closed

kmclaughlin-arm requested a review from rsandifo-arm November 6, 2024 11:58

sdesmalen-arm reviewed Nov 14, 2024

View reviewed changes

sdesmalen-arm reviewed Nov 21, 2024

View reviewed changes

sdesmalen-arm approved these changes Nov 21, 2024

View reviewed changes

rsandifo-arm approved these changes Nov 25, 2024

View reviewed changes

kmclaughlin-arm mentioned this pull request Nov 25, 2024

[Clang][AArch64] Include SME attributes in the name mangling of function types llvm/llvm-project#114209

Merged

kmclaughlin-arm added 6 commits November 26, 2024 09:27

- Added description to Changes for next release section

60bbbce

- Replace separate template arguments with one bitmask representing a…

2b5e58c

…ll attributes. - Include __arm_agnostic("sme_za_state") in the bitmask

- Use ACLE names of streaming & ZA properties.

b8cdee2

- Removed bits for ZA/ZT0 'inout', as this can be represented by setting both 'in' and 'out'.

- Include __arm_inout in the table & correct __arm_preserved

42b4fee

- Ensure 80 char line wrapping in the new changes.

8c888a3

- Added default ZA & ZT0 states to the sme_state table.

kmclaughlin-arm force-pushed the sme-keyword-mangling branch from a2d1e06 to 8c888a3 Compare November 26, 2024 09:29

kmclaughlin-arm merged commit f6190ce into ARM-software:main Nov 26, 2024
4 checks passed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Include SME attributes in the name mangling of types #358

Include SME attributes in the name mangling of types #358

kmclaughlin-arm commented Nov 6, 2024

kmclaughlin-arm commented Nov 6, 2024

sdesmalen-arm Nov 14, 2024

kmclaughlin-arm Nov 21, 2024

sdesmalen-arm Nov 21, 2024

sdesmalen-arm Nov 21, 2024

sdesmalen-arm Nov 21, 2024

sdesmalen-arm Nov 21, 2024

sdesmalen-arm Nov 21, 2024

sdesmalen-arm Nov 21, 2024

sdesmalen-arm Nov 21, 2024

sdesmalen-arm left a comment

sdesmalen-arm Nov 21, 2024

sdesmalen-arm Nov 21, 2024

rsandifo-arm left a comment

	template<typename, unsigned> __SME_ATTRS;
	template<typename, uint64_t> __SME_ATTRS;


		* normal_function_type is the function type without any SME attributes.

		* sme_state is an unsigned 64-bit integer representing the streaming mode, ZA state and ZT0 state of the function.

	The upper bits 11-63 are defined as leading zeroes, reserved for representing any future interface types.
	Bits 11-63 are defined to be zero by this revision of the ACLE and are reserved for future type attributes.

Include SME attributes in the name mangling of types #358

Include SME attributes in the name mangling of types #358

Conversation

kmclaughlin-arm commented Nov 6, 2024

kmclaughlin-arm commented Nov 6, 2024

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

sdesmalen-arm left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

rsandifo-arm left a comment

Choose a reason for hiding this comment