Re: Encountered error while encoding BTF due to Unsupported DW_TAG_unspecified_type(0x3b)

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

Em Mon, Oct 10, 2022 at 05:08:26PM -0300, Arnaldo Carvalho de Melo escreveu:
> Em Mon, Oct 10, 2022 at 09:06:46AM -0300, Arnaldo Carvalho de Melo escreveu:
> > Em Fri, Oct 07, 2022 at 05:25:21PM -0700, Yonghong Song escreveu:
> > > For function entry_ibpb, actually the kernel has a prorotype,
 
> > > include/asm/nospec-branch.h:extern void entry_ibpb(void);
 
> > > But unfortunately since the function is defined in asm code.
> > > The actual type information is not encoded in dwarf so pahole
> > > does not have enough info from vmlinux dwarf to get func
> > > types. The compiler does not generate func types based on
> > > declarations.
 
> > Right
> 
> > > This may prevent people from tracing functions defined in
> > > asm code but this should be extremely rare.
> > > So I agree that BTF type id 0 is probably the best choice.
> > 
> > Thanks for your comments, that is the way I'll do it.
> 
> Done, some prep patches then the patch below.
> 
> If possible, please ack :-)

Its all now on my 'next' branch at:

git://git.kernel.org/pub/scm/devel/pahole/pahole.git
https://github.com/acmel/dwarves.git

https://git.kernel.org/pub/scm/devel/pahole/pahole.git/log/?h=next
 
> - Arnaldo
> 
> commit 3abc72d9d56ec0cbfffc1794d2bf5d527d1e88ba
> Author: Arnaldo Carvalho de Melo <acme@redhat.com>
> Date:   Mon Oct 10 11:20:07 2022 -0300
> 
>     btf_encoder: Encode DW_TAG_unspecified_type returning routines as void
>     
>     Since we don´t have how to encode this info in BTF, and from what we
>     saw, at least in this case:
>     
>     Built binutils from git://sourceware.org/git/binutils-gdb.git, then used
>     gcc's -B option to point to the directory with the new as, that is built
>     as as-new, so make a symlink, ending up with:
>     
>       15e20ce2324a:~/git/linux # readelf -wi ./arch/x86/entry/entry.o
>       Contents of the .debug_info section:
>     
>         Compilation Unit @ offset 0:
>          Length:        0x35 (32-bit)
>          Version:       5
>          Unit Type:     DW_UT_compile (1)
>          Abbrev Offset: 0
>          Pointer Size:  8
>        <0><c>: Abbrev Number: 1 (DW_TAG_compile_unit)
>           <d>   DW_AT_stmt_list   : 0
>           <11>   DW_AT_low_pc      : 0
>           <19>   DW_AT_high_pc     : 19
>           <1a>   DW_AT_name        : (indirect string, offset: 0): arch/x86/entry/entry.S
>           <1e>   DW_AT_comp_dir    : (indirect string, offset: 0x17): /root/git/linux
>           <22>   DW_AT_producer    : (indirect string, offset: 0x27): GNU AS 2.39.50
>           <26>   DW_AT_language    : 32769  (MIPS assembler)
>        <1><28>: Abbrev Number: 2 (DW_TAG_subprogram)
>           <29>   DW_AT_name        : (indirect string, offset: 0x36): entry_ibpb
>           <2d>   DW_AT_external    : 1
>           <2d>   DW_AT_type        : <0x37>
>           <2e>   DW_AT_low_pc      : 0
>           <36>   DW_AT_high_pc     : 19
>        <1><37>: Abbrev Number: 3 (DW_TAG_unspecified_type)
>        <1><38>: Abbrev Number: 0
>     
>     So we have that asm label encoded by GNU AS 2.39.50 as a
>     DW_TAG_subprogram that has as its DW_AT_type the DW_TAG_unspecified_type
>     0x37 that we convert to 0 (void):
>     
>       15e20ce2324a:~/git/linux # pahole -J ./arch/x86/entry/entry.o
>       15e20ce2324a:~/git/linux # pahole -JV ./arch/x86/entry/entry.o
>       btf_encoder__new: 'entry.o' doesn't have '.data..percpu' section
>       Found 0 per-CPU variables!
>       Found 1 functions!
>       File entry.o:
>       [1] FUNC_PROTO (anon) return=0 args=(void)
>       [2] FUNC entry_ibpb type_id=1
>       15e20ce2324a:~/git/linux # pfunct -F btf ./arch/x86/entry/entry.o
>       entry_ibpb
>       15e20ce2324a:~/git/linux # pfunct --proto -F btf ./arch/x86/entry/entry.o
>       void  entry_ibpb(void);
>       15e20ce2324a:~/git/linux #
>     
>       15e20ce2324a:~/git/linux # tools/bpf/bpftool/bpftool btf dump file ./arch/x86/entry/entry.o format raw
>       [1] FUNC_PROTO '(anon)' ret_type_id=0 vlen=0
>       [2] FUNC 'entry_ibpb' type_id=1 linkage=static
>       15e20ce2324a:~/git/linux #
>     
>     I think this is what can be done to avoid having to skip ASM DWARF when
>     gets widely used, i.e. binutils gets updated.
>     
>     Cc: Andrii Nakryiko <andrii.nakryiko@gmail.com>,
>     Cc: Martin Liška <mliska@suse.cz>
>     Cc: Yonghong Song <yhs@meta.com>
>     Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
> 
> diff --git a/btf_encoder.c b/btf_encoder.c
> index fb2ca77e2e9bf144..a5fa04a84ee246ee 100644
> --- a/btf_encoder.c
> +++ b/btf_encoder.c
> @@ -593,6 +593,19 @@ static int32_t btf_encoder__add_func_param(struct btf_encoder *encoder, const ch
>  	}
>  }
>  
> +static int32_t btf_encoder__tag_type(struct btf_encoder *encoder, uint32_t type_id_off, uint32_t tag_type)
> +{
> +	if (tag_type == 0)
> +		return 0;
> +
> +	if (encoder->cu->unspecified_type.tag && tag_type == encoder->cu->unspecified_type.type) {
> +		// No provision for encoding this, turn it into void.
> +		return 0;
> +	}
> +
> +	return type_id_off + tag_type;
> +}
> +
>  static int32_t btf_encoder__add_func_proto(struct btf_encoder *encoder, struct ftype *ftype, uint32_t type_id_off)
>  {
>  	struct btf *btf = encoder->btf;
> @@ -603,7 +616,7 @@ static int32_t btf_encoder__add_func_proto(struct btf_encoder *encoder, struct f
>  
>  	/* add btf_type for func_proto */
>  	nr_params = ftype->nr_parms + (ftype->unspec_parms ? 1 : 0);
> -	type_id = ftype->tag.type == 0 ? 0 : type_id_off + ftype->tag.type;
> +	type_id = btf_encoder__tag_type(encoder, type_id_off, ftype->tag.type);
>  
>  	id = btf__add_func_proto(btf, type_id);
>  	if (id > 0) {
> @@ -966,6 +979,15 @@ static int btf_encoder__encode_tag(struct btf_encoder *encoder, struct tag *tag,
>  		return btf_encoder__add_enum_type(encoder, tag, conf_load);
>  	case DW_TAG_subroutine_type:
>  		return btf_encoder__add_func_proto(encoder, tag__ftype(tag), type_id_off);
> +        case DW_TAG_unspecified_type:
> +		/* Just don't encode this for now, converting anything with this type to void (0) instead.
> +		 *
> +		 * If we end up needing to encode this, one possible hack is to do as follows, as "const void".
> +		 *
> +		 * Returning zero means we skipped encoding a DWARF type.
> +		 */
> +               // btf_encoder__add_ref_type(encoder, BTF_KIND_CONST, 0, NULL, false);
> +               return 0;
>  	default:
>  		fprintf(stderr, "Unsupported DW_TAG_%s(0x%x): type: 0x%x\n",
>  			dwarf_tag_name(tag->tag), tag->tag, ref_type_id);
> @@ -1487,7 +1509,7 @@ int btf_encoder__encode_cu(struct btf_encoder *encoder, struct cu *cu, struct co
>  {
>  	uint32_t type_id_off = btf__type_cnt(encoder->btf) - 1;
>  	struct llvm_annotation *annot;
> -	int btf_type_id, tag_type_id;
> +	int btf_type_id, tag_type_id, skipped_types = 0;
>  	uint32_t core_id;
>  	struct function *fn;
>  	struct tag *pos;
> @@ -1510,8 +1532,13 @@ int btf_encoder__encode_cu(struct btf_encoder *encoder, struct cu *cu, struct co
>  	cu__for_each_type(cu, core_id, pos) {
>  		btf_type_id = btf_encoder__encode_tag(encoder, pos, type_id_off, conf_load);
>  
> +		if (btf_type_id == 0) {
> +			++skipped_types;
> +			continue;
> +		}
> +
>  		if (btf_type_id < 0 ||
> -		    tag__check_id_drift(pos, core_id, btf_type_id, type_id_off)) {
> +		    tag__check_id_drift(pos, core_id, btf_type_id + skipped_types, type_id_off)) {
>  			err = -1;
>  			goto out;
>  		}

-- 

- Arnaldo