public inbox for libc-alpha@sourceware.org
 help / color / mirror / Atom feed
From: Adhemerval Zanella <adhemerval.zanella@linaro.org>
To: libc-alpha@sourceware.org, "H.J. Lu" <hjl.tools@gmail.com>,
	Carlos O'Donell <carlos@redhat.com>
Subject: Re: [PATCH] ifuncmain6pie: Remove the circular IFUNC dependency [BZ #20019]
Date: Wed, 13 Jan 2021 16:43:36 -0300	[thread overview]
Message-ID: <5a43912e-d46e-6bb9-61f3-aca60a31efaf@linaro.org> (raw)
In-Reply-To: <CAMe9rOryuOAHN=7yk8Ym=qEqseznYwN+ZSOnMep5eS8299TpYA@mail.gmail.com>



On 04/01/2021 19:38, H.J. Lu via Libc-alpha wrote:
> On Mon, Jan 4, 2021 at 1:20 PM Carlos O'Donell <carlos@redhat.com> wrote:
>>
> ...
>>>
>>> [hjl@gnu-cfl-2 build-x86_64-linux]$ ./elf/ifuncmain6pie --direct
>>> ./elf/ifuncmain6pie: IFUNC symbol 'foo' referenced in
>>> '/export/build/gnu/tools-build/glibc/build-x86_64-linux/elf/ifuncmod6.so'
>>> is defined in the executable and creates an unsatisfiable circular
>>> dependency.
>>> [hjl@gnu-cfl-2 build-x86_64-linux]$
>>>
>>> The message is correct.  Should we update the testcase to avoid it?
>>
>> Yes, but it is still possible to support this with lazy binding?
>>
>> Should ifuncmain6pie be explicitly compiled with -Wl,-z,lazy to
>> bypass selection from the toolchain?
> 
> The problem is non-JUMP_SLOT relocations.  Here is a patch to
> remove them.   OK for master?
> 
> Thanks.
> 

I am getting a failure for elf/ifuncmain6pie for a couple of days:

$ ./testrun.sh elf/ifuncmain6pie
elf/ifuncmain6pie: IFUNC symbol 'foo' referenced in '/home/azanella/Projects/glibc/build/x86_64-linux-gnu/elf/ifuncmod6.so' is defined in the executable and creates an unsatisfiable circular dependency.

The patch looks ok for 2.33.

However, I think it should be been added along with 6ea5b57afa5
fix in first place. Why hasn't it shown in your make check? Does the
failure depend on a binutils version?


> diff --git a/elf/Makefile b/elf/Makefile
> index 543800f4be..c41d11693b 100644
> --- a/elf/Makefile
> +++ b/elf/Makefile
> @@ -1403,6 +1403,8 @@ CFLAGS-ifuncmain7pie.c += $(pie-ccflag)
>  CFLAGS-ifuncmain9pie.c += $(pie-ccflag)
>  CFLAGS-tst-ifunc-textrel.c += $(pic-ccflag)
>  
> +LDFLAGS-ifuncmain6pie = -Wl,-z,lazy
> +
>  $(objpfx)ifuncmain1pie: $(objpfx)ifuncmod1.so
>  $(objpfx)ifuncmain1staticpie: $(objpfx)ifuncdep1pic.o
>  $(objpfx)ifuncmain1vispie: $(objpfx)ifuncmod1.so
> diff --git a/elf/ifuncmain6pie.c b/elf/ifuncmain6pie.c
> index 04faeb86ef..4a01906836 100644
> --- a/elf/ifuncmain6pie.c
> +++ b/elf/ifuncmain6pie.c
> @@ -9,7 +9,6 @@
>  #include "ifunc-sel.h"
>  
>  typedef int (*foo_p) (void);
> -extern foo_p foo_ptr;
>  
>  static int
>  one (void)
> @@ -28,20 +27,17 @@ foo_ifunc (void)
>  }
>  
>  extern int foo (void);
> -extern foo_p get_foo (void);
> +extern int call_foo (void);
>  extern foo_p get_foo_p (void);
>  
> -foo_p my_foo_ptr = foo;
> +foo_p foo_ptr = foo;
>  
>  int
>  main (void)
>  {
>    foo_p p;
>  
> -  p = get_foo ();
> -  if (p != foo)
> -    abort ();
> -  if ((*p) () != -30)
> +  if (call_foo () != -30)
>      abort ();
>  
>    p = get_foo_p ();
> @@ -52,12 +48,8 @@ main (void)
>  
>    if (foo_ptr != foo)
>      abort ();
> -  if (my_foo_ptr != foo)
> -    abort ();
>    if ((*foo_ptr) () != -30)
>      abort ();
> -  if ((*my_foo_ptr) () != -30)
> -    abort ();
>    if (foo () != -30)
>      abort ();
>  
> diff --git a/elf/ifuncmod6.c b/elf/ifuncmod6.c
> index 2e16c1d06d..2f6d0715e6 100644
> --- a/elf/ifuncmod6.c
> +++ b/elf/ifuncmod6.c
> @@ -4,7 +4,7 @@ extern int foo (void);
>  
>  typedef int (*foo_p) (void);
>  
> -foo_p foo_ptr = foo;
> +extern foo_p foo_ptr;
>  
>  foo_p
>  get_foo_p (void)
> @@ -12,8 +12,8 @@ get_foo_p (void)
>    return foo_ptr;
>  }
>  
> -foo_p
> -get_foo (void)
> +int
> +call_foo (void)
>  {
> -  return foo;
> +  return foo ();
>  }
> -- 
> 2.29.2

  reply	other threads:[~2021-01-13 19:43 UTC|newest]

Thread overview: 14+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-12-28 14:11 [PATCH] x86: Check IFUNC definition in unrelocated executable " H.J. Lu
2021-01-04 18:47 ` Carlos O'Donell
2021-01-04 19:34   ` V2 " H.J. Lu
2021-01-04 19:50     ` Carlos O'Donell
2021-01-04 19:59       ` H.J. Lu
2021-01-04 22:57         ` H.J. Lu
2021-01-05 13:03           ` Carlos O'Donell
2021-01-05 15:14             ` H.J. Lu
2021-01-04 20:44       ` H.J. Lu
2021-01-04 21:20         ` Carlos O'Donell
2021-01-04 22:38           ` [PATCH] ifuncmain6pie: Remove the circular IFUNC dependency " H.J. Lu
2021-01-13 19:43             ` Adhemerval Zanella [this message]
2021-01-13 19:48               ` H.J. Lu
2021-01-14 13:10                 ` Adhemerval Zanella

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=5a43912e-d46e-6bb9-61f3-aca60a31efaf@linaro.org \
    --to=adhemerval.zanella@linaro.org \
    --cc=carlos@redhat.com \
    --cc=hjl.tools@gmail.com \
    --cc=libc-alpha@sourceware.org \
    /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;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).