public inbox for dwarves@vger.kernel.org
 help / color / mirror / Atom feed
From: Ihor Solodrai <ihor.solodrai@linux.dev>
To: "Alexis Lothoré (eBPF Foundation)" <alexis.lothore@bootlin.com>,
	dwarves@vger.kernel.org
Cc: Alan Maguire <alan.maguire@oracle.com>,
	Arnaldo Carvalho de Melo <acme@redhat.com>,
	ebpf@linuxfoundation.org,
	Thomas Petazzoni <thomas.petazzoni@bootlin.com>,
	Bastien Curutchet <bastien.curutchet@bootlin.com>
Subject: Re: [PATCH] pahole: do not return an error when printing only a specific class
Date: Tue, 22 Jul 2025 16:42:03 -0700	[thread overview]
Message-ID: <e1afba7b-93f9-4ed3-9cd5-ef4ff4bebb70@linux.dev> (raw)
In-Reply-To: <20250716-lsk__abort-v1-1-586f31a9d97d@bootlin.com>

On 7/16/25 2:04 AM, Alexis Lothoré (eBPF Foundation) wrote:
> When trying to print a specific class layout with recent pahole
> versions, we get an error right after the expected class print:
> 
>    $ ./build/pahole -C test_bin_struct_packed -F dwarf tests/bin/test_bin
>    struct test_bin_struct_packed {
>            char                       a;                    /*     0     1 */
>            short int                  b;                    /*     1     2 */
>            int                        c;                    /*     3     4 */
>            long long unsigned int     d;                    /*     7     8 */
> 
>            /* size: 15, cachelines: 1, members: 4 */
>            /* last cacheline: 15 bytes */
>    } __attribute__((__packed__));
> 
>    pahole: /home/alexis/src/pahole/tests/bin/test_bin: Invalid argument
> 
> This error is due to pahole_stealer returning LSK__STOP_LOADING when
> dealing with a single class and when it is done. This LSK__STOP_LOADING
> is then propagated to dwarf_loader__worker_thread and interpreted as an
> error (and turned into a DWARF_CB_ABORT). The main issue is that this
> LSK__STOP_LOADING value is sometimes generated by actual issues during
> pahole stealing, sometimes by "nominal" execution.
> 
> Add a new LSK__ABORT status to distinguish between those nominal and
> faulty cases. Return an error only when pahole_stealer returns
> LSK__ABORT.
> 
> Signed-off-by: Alexis Lothoré (eBPF Foundation) <alexis.lothore@bootlin.com>
> ---
> See [1] for original discussions about the issue
> 
> [1] https://lore.kernel.org/dwarves/DB5VWHU0N27I.3ETC4G47KB9Q@bootlin.com/

Hi Alexis, thank you for working on this.

The change looks good to me, just one comment.

> ---
>   dwarf_loader.c |  6 ++++--
>   dwarves.h      |  1 +
>   pahole.c       | 21 +++++++++++----------
>   3 files changed, 16 insertions(+), 12 deletions(-)
> 
> diff --git a/dwarf_loader.c b/dwarf_loader.c
> index ef00cdf308166732fc0938f2eecd56d314d3354f..4e1e19c9d828e111d78012feda4c8baa0d716378 100644
> --- a/dwarf_loader.c
> +++ b/dwarf_loader.c
> @@ -3277,6 +3277,8 @@ static int cus__steal_now(struct cus *cus, struct cu *cu, struct conf_load *conf
>   		break;
>   	case LSK__KEEPIT:
>   		break;
> +	case LSK__ABORT:
> +		break;
>   	}
>   	return lsk;
>   }
> @@ -3679,7 +3681,7 @@ static void *dwarf_loader__worker_thread(void *arg)
>   			break;
>   
>   		case JOB_STEAL:
> -			if (cus__steal_now(dcus->cus, job->cu, dcus->conf) == LSK__STOP_LOADING)
> +			if (cus__steal_now(dcus->cus, job->cu, dcus->conf) == LSK__ABORT)
>   				goto out_abort;
>   			cus_queue__inc_next_cu_id();
>   			/* re-enqueue JOB_DECODE so that next CU is decoded from DWARF */
> @@ -3872,7 +3874,7 @@ static int cus__merge_and_process_cu(struct cus *cus, struct conf_load *conf,
>   		goto out_abort;
>   
>   	cu__finalize(cu, cus, conf);
> -	if (cus__steal_now(cus, cu, conf) == LSK__STOP_LOADING)
> +	if (cus__steal_now(cus, cu, conf) == LSK__ABORT)
>   		goto out_abort;

I think we should handle both LSK__STOP_LOADING and LSK__ABORT here,
otherwise the worker will keep loading DWARF (which is most of the
pahole's work) and passing cu-s to the stealer function.

I guess in many cases it'll complete without errors, but it's just
unnecessary work.

Maybe add a check for LSK__STOP_LOADING here with goto_ok?

>   
>   	return 0;
> diff --git a/dwarves.h b/dwarves.h
> index 883852f4b60a36d73bce4568cbacd8ef3cff97ea..5df1cdad9be36212bd6548e86180860daffa3f63 100644
> --- a/dwarves.h
> +++ b/dwarves.h
> @@ -43,6 +43,7 @@ enum load_steal_kind {
>   	LSK__KEEPIT,
>   	LSK__DELETE,
>   	LSK__STOP_LOADING,
> +	LSK__ABORT,
>   };
>   
>   /*
> diff --git a/pahole.c b/pahole.c
> index 333e71ab65924d2c64771be7d40f768f02e9a0f0..66448fc430c19e5137a8467026d35d2ba3125e4c 100644
> --- a/pahole.c
> +++ b/pahole.c
> @@ -3165,13 +3165,13 @@ static enum load_steal_kind pahole_stealer__btf_encode(struct cu *cu, struct con
>   
>   	if (!btf_encoder) {
>   		fprintf(stderr, "Error creating BTF encoder.\n");
> -		return LSK__STOP_LOADING;
> +		return LSK__ABORT;
>   	}
>   
>   	err = btf_encoder__encode_cu(btf_encoder, cu, conf_load);
>   	if (err < 0) {
>   		fprintf(stderr, "Error while encoding BTF.\n");
> -		return LSK__STOP_LOADING;
> +		return LSK__ABORT;
>   	}
>   
>   	return LSK__DELETE;
> @@ -3230,7 +3230,7 @@ static enum load_steal_kind pahole_stealer(struct cu *cu, struct conf_load *conf
>   		 *
>   		 * FIXME: Disabled, should use Oracle's libctf
>   		 */
> -		goto dump_and_stop;
> +		goto dump_and_abort;
>   	}
>   #endif
>   	if (class_name == NULL) {
> @@ -3281,7 +3281,7 @@ static enum load_steal_kind pahole_stealer(struct cu *cu, struct conf_load *conf
>   
>   		if (prototype->nr_args != 0 && !tag__is_struct(class)) {
>   			fprintf(stderr, "pahole: attributes are only supported with 'class' and 'struct' types\n");
> -			goto dump_and_stop;
> +			goto dump_and_abort;
>   		}
>   
>   		struct type *type = tag__type(class);
> @@ -3291,7 +3291,7 @@ static enum load_steal_kind pahole_stealer(struct cu *cu, struct conf_load *conf
>   			if (type->sizeof_member == NULL) {
>   				fprintf(stderr, "pahole: the sizeof member '%s' wasn't found in the '%s' type\n",
>   					prototype->size, prototype->name);
> -				goto dump_and_stop;
> +				goto dump_and_abort;
>   			}
>   		}
>   
> @@ -3300,7 +3300,7 @@ static enum load_steal_kind pahole_stealer(struct cu *cu, struct conf_load *conf
>   			if (type->type_member == NULL) {
>   				fprintf(stderr, "pahole: the type member '%s' wasn't found in the '%s' type\n",
>   					prototype->type, prototype->name);
> -				goto dump_and_stop;
> +				goto dump_and_abort;
>   			}
>   		}
>   
> @@ -3315,7 +3315,7 @@ static enum load_steal_kind pahole_stealer(struct cu *cu, struct conf_load *conf
>   			if (type->filter == NULL) {
>   				fprintf(stderr, "pahole: invalid filter '%s' for '%s'\n",
>   					prototype->filter, prototype->name);
> -				goto dump_and_stop;
> +				goto dump_and_abort;
>   			}
>   		}
>   
> @@ -3386,15 +3386,16 @@ static enum load_steal_kind pahole_stealer(struct cu *cu, struct conf_load *conf
>   	/*
>   	 * If we found all the entries in --class_name, stop
>   	 */
> -	if (list_empty(&class_names)) {
> -dump_and_stop:
> +	if (list_empty(&class_names))
>   		ret = LSK__STOP_LOADING;
> -	}
>   dump_it:
>   	if (first_obj_only)
>   		ret = LSK__STOP_LOADING;
>   filter_it:
>   	return ret;
> +dump_and_abort:
> +	ret = LSK__ABORT;
> +	return ret;
>   }
>   
>   static int prototypes__add(struct list_head *prototypes, const char *entry)
> 
> ---
> base-commit: 9845339bd09f9889dd741ed94f71eb92008dcdfc
> change-id: 20250716-lsk__abort-2b517860af28
> 
> Best regards,


  reply	other threads:[~2025-07-22 23:42 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-07-16  9:04 [PATCH] pahole: do not return an error when printing only a specific class Alexis Lothoré (eBPF Foundation)
2025-07-22 23:42 ` Ihor Solodrai [this message]
2025-07-23  8:39   ` Alexis Lothoré
2025-07-23 10:03     ` Alan Maguire
2025-07-23 15:37       ` Ihor Solodrai

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=e1afba7b-93f9-4ed3-9cd5-ef4ff4bebb70@linux.dev \
    --to=ihor.solodrai@linux.dev \
    --cc=acme@redhat.com \
    --cc=alan.maguire@oracle.com \
    --cc=alexis.lothore@bootlin.com \
    --cc=bastien.curutchet@bootlin.com \
    --cc=dwarves@vger.kernel.org \
    --cc=ebpf@linuxfoundation.org \
    --cc=thomas.petazzoni@bootlin.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox