Re: Sashiko reviews for pahole: missing custom_remotes

[Arnaldo Carvalho de Melo <acme@kernel.org>]

On Mon, Jun 22, 2026 at 12:18:10PM -0700, Roman Gushchin wrote:
> Hi Arnaldo,
> 
> Sashiko is not really officially supporting a non-kernel code yet. It might work occasionally,
> but I can’t guarantee anything and results might be pathetic.

I'm using it locally with great results, using:

acme@number:~/git/sashiko$ cat Settings.toml | grep -B1 -A3 'dwarves\|pahole'
[mailing_lists]
track = ["linux-kernel", "dwarves"]

[nntp]
server = "nntp.lore.kernel.org"
--
[[git.custom_remotes]]
name = "pahole"
url = "git://git.kernel.org/pub/scm/devel/pahole/pahole.git"
check_all_branches = false
only_branches = ["next", "master"]

acme@number:~/git/sashiko$

Submitting series to sashiko with:

$ sashiko-cli submit FIRST_SHA^..LAST_SHA --repo /home/acme/git/pahole

With great results, for instance:

acme@number:~$ sashiko-cli show 132 | head -100
Patchset Details:
  ID:        132
  Subject:   dwarf_loader: Initial support for DW_TAG_variant_part
  Author:    Arnaldo Carvalho de Melo <acme@redhat.com>
  Status:    Failed
  Date:      2026-06-20 11:23:54

Patches (16):
  [1] dwarf_loader: Initial support for DW_TAG_variant_part (Reviewed) [Issues Found] {cache: 1322/2592 hits, 16.5M tokens saved}
  [2] dwarf_loader: Allow forcing the merge of CUs for solving inter CU tag references (Reviewed) [Reviewed] {cache: 1321/2567 hits, 16.5M tokens saved}
  [3] dwarf_loader: Initial support for DW_TAG_subprogram in DW_TAG_enumeration (Reviewed) [Issues Found] {cache: 1324/2542 hits, 16.5M tokens saved}
  [4] encoders: Fix diagnostic messages for unexpected tags in enumerations (Reviewed) [Reviewed] {cache: 818/1833 hits, 10.0M tokens saved}
  [5] dwarves_fprintf: Accumulate function__fprintf return value in enumeration printing (Failed) [Failed] {cache: 309/401 hits, 3.8M tokens saved}
  [6] dwarves: Use tag__delete for enumeration children (Reviewed) [Reviewed] {cache: 1325/2534 hits, 16.5M tokens saved}
  [7] btf_encoder: Fix types__match parameter comparison in BTF_KIND_FUNC_PROTO (Reviewed) [Reviewed] {cache: 508/1429 hits, 6.1M tokens saved}
  [8] encoders: Handle DW_TAG_subprogram in enumerations during BTF/CTF encoding (Reviewed) [Reviewed] {cache: 64/624 hits, 719.8k tokens saved}
  [9] dwarf_loader: Populate DW_TAG_variant children in DW_TAG_variant_part (Reviewed) [Issues Found] {cache: 1325/2570 hits, 16.5M tokens saved}
  [10] btf_encoder: Encode variant parts as union members in BTF (Reviewed) [Reviewed] {cache: 965/2027 hits, 11.9M tokens saved}
  [11] dwarf_loader: Handle DW_FORM_block in attr_numeric for Rust discriminant values (Reviewed) [Issues Found] {cache: 1325/2614 hits, 16.5M tokens saved}
  [12] dwarf_loader: Support DW_TAG_imported_unit for same-file partial units (Reviewed) [Issues Found] {cache: 1320/2550 hits, 16.5M tokens saved}
  [13] dwarf_loader: Fix cus__merging_cu failing to detect DW_FORM_ref_addr (Reviewed) [Reviewed] {cache: 274/1071 hits, 3.2M tokens saved}
  [14] tests: Add inter-CU type reference comparison test (Reviewed) [Reviewed] {cache: 280/1097 hits, 3.2M tokens saved}
  [15] tests: Guard cleanup() against empty outdir to prevent rm /* (Reviewed) [Reviewed] {cache: 1233/2356 hits, 15.5M tokens saved}
  [16] tests: Source test_lib.sh via dirname so tests run from any directory (Failed) [Failed] {cache: 322/405 hits, 4.2M tokens saved}

Cache: 14035/29212 hits (48.0%), 174.2M tokens saved, 530.8M tokens stored

Review Summary:
  Model:   gemini-3.1-pro-preview

Patch 1: dwarf_loader: Initial support for DW_TAG_variant_part
Critical: 0 · High: 0 · Medium: 0 · Low: 1

  [Low] 

commit 3d23ae0089bbe7cda5b12c12f58ad92e02812f2b
Author: Arnaldo Carvalho de Melo <acme@redhat.com>
Subject: dwarf_loader: Initial support for DW_TAG_variant_part

This commit adds initial scaffolding for parsing DW_TAG_variant_part entries
from Rust DWARF. The goal is to allow perf to process Rust workloads and test
pahole's pretty printing features.

> diff --git a/dwarf_loader.c b/dwarf_loader.c
> index 8ce34cbb8ed60aca..438ebd52893fdb8f 100644
> --- a/dwarf_loader.c
> +++ b/dwarf_loader.c
> @@ -508,6 +508,8 @@ static void tag__init(struct tag *tag, struct cu *cu, Dwarf_Die *die)
>  
>  	if (tag->tag == DW_TAG_imported_module || tag->tag == DW_TAG_imported_declaration)
>  		dwarf_tag__set_attr_type(dtag, type, die, DW_AT_import);
> +	else if (tag->tag == DW_TAG_variant_part)
> +		dwarf_tag__set_attr_type(dtag, type, die, DW_AT_discr);
>  	else
>  		dwarf_tag__set_attr_type(dtag, type, die, DW_AT_type);
>  

[ ... ]

> @@ -2085,9 +2099,19 @@ static int die__process_class(Dwarf_Die *die, struct type *class,
>  		case DW_TAG_GNU_template_template_param:
>  #endif
>  		case DW_TAG_subrange_type: // XXX: ADA stuff, its a type tho, will have other entries referencing it...
> -		case DW_TAG_variant_part: // XXX: Rust stuff
>  			tag__print_not_supported(die);
>  			continue;
> +		case DW_TAG_variant_part: {
> +			struct variant_part *vpart = variant_part__new(die, cu, conf);
> +
> +			if (vpart == NULL)
> +				return -ENOMEM;
> +
> +			// For rust it seems we have just one, but DWARF, according to Gemini, support having
> +			// more than one DW_TAG_variant_part for a given DW_TAG_structure_type, so future proof it
> +			type__add_variant_part(class, vpart);

[Severity: Low]
Are we missing a call to cu__table_add_tag() here before adding the
variant part?

Since variant_part tags are not added to the CU's tags table, they will be
skipped entirely during type recoding in cu__recode_dwarf_types(). This
leaves the DW_AT_discr attribute extracted earlier in tag__init() as a
raw DIE offset rather than a valid small_id, which makes the extracted
discriminant unusable.

> +			continue;
> +		}
>  		case DW_TAG_template_type_parameter: {

Patch 2: dwarf_loader: Allow forcing the merge of CUs for solving inter CU tag references
Critical: 0 (2) · High: 0 · Medium: 0 · Low: 0

  [Critical] 
  [Critical] 

commit 72c660e79f1db109e3466377d4d7f2ea187787f4
Author: Arnaldo Carvalho de Melo <acme@redhat.com>

dwarf_loader: Allow forcing the merge of CUs for solving inter CU tag references

thread 'main' (127686) panicked at library/std/src/io/stdio.rs:1165:9:
failed printing to stdout: Broken pipe (os error 32)
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
acme@number:~

Etc.

> We might explore adding support for more project going forward, but we’re not there yet.

Your call, but just as a data point, its being useful for me for pahole
development and I bet other pahole developers would like having it
enabled, since it seems to be useful with the current sashiko codebase.

- Arnaldo
 
> Thanks
> 
> > On Jun 19, 2026, at 12:30 PM, Arnaldo Carvalho de Melo <acme@kernel.org> wrote:
> > 
> > Hi Roman,
> > 
> >    The dwarves@vger.kernel.org mailing list is being tracked:
> > 
> > https://sashiko.dev/#/?list=org.kernel.vger.dwarves
> > 
> >    But all the patches for pahole are failing to apply, I think we
> > need to have the second hunk applied, the first should be there as it is
> > being tracked, can you please take a look and fix this?
> > 
> > Thanks,
> > 
> > - Arnaldo
> > 
> > diff --git a/Settings.toml b/Settings.toml
> > index 61978b51f89a0d86..666b4e8f1b7a76f3 100644
> > --- a/Settings.toml
> > +++ b/Settings.toml
> > @@ -5,7 +5,7 @@ url = "sashiko.db"
> > token = ""
> > 
> > [mailing_lists]
> > -track = ["linux-kernel"]
> > +track = ["linux-kernel", "dwarves"]
> > 
> > [nntp]
> > server = "nntp.lore.kernel.org"
> > @@ -104,6 +104,12 @@ port = 8080
> > repository_path = "third_party/linux"
> > 
> > # Track custom remotes and attempt to apply patches on their branches
> > +[[git.custom_remotes]]
> > +name = "pahole"
> > +url = "git://git.kernel.org/pub/scm/devel/pahole/pahole.git"
> > +check_all_branches = false
> > +only_branches = ["next", "master"]
> > +
> > # [[git.custom_remotes]]
> > # name = "remote-name"
> > # url = "git://my-custom.com/remote"