Re: Build failures: unresolved symbol vfs_getattr

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Wed, Oct 21, 2020 at 12:42 PM Jiri Olsa <jolsa@xxxxxxxxxx> wrote:
>
> On Fri, Oct 16, 2020 at 11:38:37PM +0200, Jiri Olsa wrote:
> > On Wed, Sep 16, 2020 at 11:06:27AM +0200, Jiri Olsa wrote:
> > > On Tue, Sep 15, 2020 at 02:17:46PM +0200, Jiri Olsa wrote:
> > >
> > > SNIP
> > >
> > > >    <2><140d7aa>: Abbrev Number: 3 (DW_TAG_formal_parameter)
> > > >       <140d7ab>   DW_AT_type        : <0x140cfb6>
> > > >    <2><140d7af>: Abbrev Number: 3 (DW_TAG_formal_parameter)
> > > >       <140d7b0>   DW_AT_type        : <0x1406176>
> > > >    <2><140d7b4>: Abbrev Number: 3 (DW_TAG_formal_parameter)
> > > >       <140d7b5>   DW_AT_type        : <0x14060c9>
> > > >    <2><140d7b9>: Abbrev Number: 0
> > > >
> > > > the latter is just declaration.. but it's missing the
> > > >     <365d69d>   DW_AT_declaration : 1
> > > >
> > > > so it goes through pahole's function processing:
> > > >
> > > >   cu__encode_btf:
> > > >   ...
> > > >         cu__for_each_function(cu, core_id, fn) {
> > > >                 int btf_fnproto_id, btf_fn_id;
> > > >
> > > >                 if (fn->declaration || !fn->external)
> > > >                         continue;
> > > >   ...
> > > >
> > > >
> > > > CC-ing Frank.. any idea why is the DW_AT_declaration : 1 missing?
> > >
> > > looks like gcc issue:
> > >   https://gcc.gnu.org/bugzilla/show_bug.cgi?id=97060
> > >
> > > let's see ;-)
> >
> > so this gcc bug did not disappear and the fix might be delayed,
> > as I was told it's real complex and difficult to fix
> >
> > and it's no longer just rawhide issue, because I just started to
> > see it in Fedora 32 after updating to gcc (GCC) 10.2.1 20201005
> > (Red Hat 10.2.1-5)
> >
> > I'm checking on pahole's workaround, but so far I can't see dwarf
> > based solution for that.. any thoughts/ideas? ;-)
>
> hi,
> FYI there's still no solution yet, so far the progress is:
>
> the proposed workaround was to use the negation -> we don't have
> DW_AT_declaration tag, so let's find out instead which DW_TAG_subprogram
> tags have attached code and skip them if they don't have any:
>   https://gcc.gnu.org/bugzilla/show_bug.cgi?id=97060#c10
>
> the attached patch is doing that, but the resulting BTF is missing
> several functions due to another bug in dwarf:
>   https://bugzilla.redhat.com/show_bug.cgi?id=1890107

It seems fine if there are only few functions (especially if those are
unlikely to be traced). Do you have an estimate of how many functions
have this second DWARF bug?

>
>
> the only other workaround I can think of is to check each DW_TAG_subprogram
> tag name with vmlinux symbol to ensure it's actually present,
> I'll check on it, because as Mark suggested it might be good
> for future not to completely rely on dwarf being correct, even
> if that gcc bug gets eventually fixed

This might be a good thing to do anyways. Currently BTF contains only
global functions, but a lot of static functions that didn't end up
inlined are available for attachment, but because BTF info is not
there, we can't use fentry/fexit on them. Checking against ELF symbols
would match kallsyms, right? So we would be able to drop fn->external
requirement and have all the attachable functions available.

Have you tried this? I'm curious how good the data is and how much
bigger BTF size is with all the functions included?

>
> jirka
>
>
> ---
> diff --git a/btf_encoder.c b/btf_encoder.c
> index e90150784282..51a370d580b7 100644
> --- a/btf_encoder.c
> +++ b/btf_encoder.c
> @@ -302,8 +302,9 @@ int cu__encode_btf(struct cu *cu, int verbose, bool force,
>
>         cu__for_each_function(cu, core_id, fn) {
>                 int btf_fnproto_id, btf_fn_id;
> +               bool has_pc = !!function__addr(fn) || fn->ranges;
>
> -               if (fn->declaration || !fn->external)
> +               if (!has_pc || !fn->external)
>                         continue;
>
>                 btf_fnproto_id = btf_elf__add_func_proto(btfe, &fn->proto, type_id_off);
> diff --git a/dwarf_loader.c b/dwarf_loader.c
> index d3586aa5b0dd..4763b9118475 100644
> --- a/dwarf_loader.c
> +++ b/dwarf_loader.c
> @@ -953,6 +953,7 @@ static struct function *function__new(Dwarf_Die *die, struct cu *cu)
>                 func->declaration     = dwarf_hasattr(die, DW_AT_declaration);
>                 func->external        = dwarf_hasattr(die, DW_AT_external);
>                 func->abstract_origin = dwarf_hasattr(die, DW_AT_abstract_origin);
> +               func->ranges          = dwarf_hasattr(die, DW_AT_ranges);
>                 dwarf_tag__set_spec(func->proto.tag.priv,
>                                     attr_type(die, DW_AT_specification));
>                 func->accessibility   = attr_numeric(die, DW_AT_accessibility);
> @@ -2023,8 +2024,10 @@ static int tag__recode_dwarf_type(struct tag *tag, struct cu *cu)
>                         dtype = dwarf_cu__find_tag_by_ref(cu->priv, &dtag->abstract_origin);
>                         if (dtype == NULL)
>                                 dtype = dwarf_cu__find_tag_by_ref(cu->priv, &specification);
> -                       if (dtype != NULL)
> +                       if (dtype != NULL) {
>                                 fn->name = tag__function(dtype->tag)->name;
> +                               fn->external = tag__function(dtype->tag)->external;
> +                       }
>                         else {
>                                 fprintf(stderr,
>                                         "%s: couldn't find name for "
> diff --git a/dwarves.h b/dwarves.h
> index 7c4254eded1f..3204f69abfe5 100644
> --- a/dwarves.h
> +++ b/dwarves.h
> @@ -813,6 +813,7 @@ struct function {
>         uint8_t          virtuality:2; /* DW_VIRTUALITY_{none,virtual,pure_virtual} */
>         uint8_t          declaration:1;
>         uint8_t          btf:1;
> +       uint8_t          ranges:1;
>         int32_t          vtable_entry;
>         struct list_head vtable_node;
>         /* fields used by tools */
>



[Index of Archives]     [Linux Samsung SoC]     [Linux Rockchip SoC]     [Linux Actions SoC]     [Linux for Synopsys ARC Processors]     [Linux NFS]     [Linux NILFS]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]


  Powered by Linux