public inbox for libc-alpha@sourceware.org
 help / color / mirror / Atom feed
From: "H.J. Lu" <hjl.tools@gmail.com>
To: Adhemerval Zanella <adhemerval.zanella@linaro.org>
Cc: GNU C Library <libc-alpha@sourceware.org>,
	Florian Weimer <fweimer@redhat.com>
Subject: Re: [PATCH v7 2/4] elf: Add a test for PT_LOAD segments with mixed p_align [BZ #28676]
Date: Tue, 18 Jan 2022 13:53:58 -0800	[thread overview]
Message-ID: <CAMe9rOru9Tt+Movx-mRakJ2--GxjRjD0MLJTooG_Qxs6--HzNA@mail.gmail.com> (raw)
In-Reply-To: <5d0957aa-5128-601e-fece-837cdc6e6a20@linaro.org>

On Tue, Jan 18, 2022 at 8:26 AM Adhemerval Zanella
<adhemerval.zanella@linaro.org> wrote:
>
>
>
> On 03/01/2022 20:04, H.J. Lu wrote:
> > Add tst-p_alignmod1-edit to edit the copy of tst-alignmod3.so to reduce
> > p_align of the first PT_LOAD segment by half and verify that the shared
> > library is mapped with the maximum p_align of all PT_LOAD segments.
> > ---
> >  elf/Makefile               |  18 ++++++
> >  elf/tst-elf-edit.h         | 126 +++++++++++++++++++++++++++++++++++++
> >  elf/tst-p_align1.c         |  27 ++++++++
> >  elf/tst-p_alignmod1-edit.c |  34 ++++++++++
> >  4 files changed, 205 insertions(+)
> >  create mode 100644 elf/tst-elf-edit.h
> >  create mode 100644 elf/tst-p_align1.c
> >  create mode 100644 elf/tst-p_alignmod1-edit.c
> >
> > diff --git a/elf/Makefile b/elf/Makefile
> > index 883578ee36..9c831faac1 100644
> > --- a/elf/Makefile
> > +++ b/elf/Makefile
> > @@ -403,6 +403,13 @@ modules-names = testobj1 testobj2 testobj3 testobj4 testobj5 testobj6 \
> >               tst-dl_find_object-mod8 \
> >               tst-dl_find_object-mod9 \
> >
> > +ifeq (yes,$(build-shared))
> > +tests += \
> > +  tst-p_alignmod1-edit \
> > +  tst-p_align1 \
> > +
> > +endif
> > +
> >  # Most modules build with _ISOMAC defined, but those filtered out
> >  # depend on internal headers.
> >  modules-names-tests = $(filter-out ifuncmod% tst-tlsmod%,\
> > @@ -2079,3 +2086,14 @@ CFLAGS-tst-dl_find_object-mod6.c += -funwind-tables
> >  CFLAGS-tst-dl_find_object-mod7.c += -funwind-tables
> >  CFLAGS-tst-dl_find_object-mod8.c += -funwind-tables
> >  CFLAGS-tst-dl_find_object-mod9.c += -funwind-tables
> > +
> > +$(objpfx)tst-p_align1: $(objpfx)tst-p_alignmod1.so
> > +
> > +# Make a copy of tst-alignmod3.so and lower p_align of the first PT_LOAD
> > +# segment.
> > +$(objpfx)tst-p_alignmod1.so: $(objpfx)tst-p_alignmod1-edit \
> > +                          $(objpfx)tst-alignmod3.so
>
> tst-alignmod3.so is only provided in the last patch, I think it would be better
> to move to this patch as tst-alignmod-base, not  and it instead by copying and changing
> on each test.

It is tst-alignmod3.so, not tst-p_alignmod3.so.  In any case, I added
tst-p_alignmod-base.so.

> > +     rm -f $@
> > +     cp $(objpfx)tst-alignmod3.so $@
> > +     $(test-wrapper-env) $(run-program-env) $(rtld-prefix) \
> > +       $(objpfx)tst-p_alignmod1-edit $@
> > diff --git a/elf/tst-elf-edit.h b/elf/tst-elf-edit.h
> > new file mode 100644
> > index 0000000000..ce98cf012c
> > --- /dev/null
> > +++ b/elf/tst-elf-edit.h
> > @@ -0,0 +1,126 @@
> > +/* Update p_align of the first PT_LOAD segment.
> > +   Copyright (C) 2021 Free Software Foundation, Inc.
> > +   This file is part of the GNU C Library.
> > +
> > +   The GNU C Library is free software; you can redistribute it and/or
> > +   modify it under the terms of the GNU Lesser General Public
> > +   License as published by the Free Software Foundation; either
> > +   version 2.1 of the License, or (at your option) any later version.
> > +
> > +   The GNU C Library is distributed in the hope that it will be useful,
> > +   but WITHOUT ANY WARRANTY; without even the implied warranty of
> > +   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
> > +   Lesser General Public License for more details.
> > +
> > +   You should have received a copy of the GNU Lesser General Public
> > +   License along with the GNU C Library; if not, see
> > +   <https://www.gnu.org/licenses/>.  */
> > +
> > +#include <stdio.h>
> > +#include <stdbool.h>
> > +#include <unistd.h>
> > +#include <fcntl.h>
> > +#include <link.h>
> > +#include <error.h>
> > +#include <errno.h>
> > +#include <sys/stat.h>
> > +#include <sys/mman.h>
> > +
> > +const char *file_name;
> > +
> > +static size_t update_p_align (size_t);
> > +
> > +int
> > +main (int argc, char ** argv)
> > +{
> > +  if (argc != 2)
> > +    {
> > +      printf ("Usage: %s: file\n", argv[0]);
> > +      return 0;
> > +    }
> > +
> > +  file_name = argv[1];
> > +  struct stat statbuf;
> > +  int errno_saved;
> > +
> > +  if (stat (file_name, &statbuf) < 0)
> > +    error (1, errno, "%s: not exist", file_name);
> > +
> > +  ElfW(Ehdr) *ehdr;
> > +
> > +  if (statbuf.st_size < sizeof (*ehdr))
> > +    error (1, 0, "%s: too small", file_name);
> > +
> > +  int fd = open (file_name, O_RDWR);
> > +  if (fd < 0)
> > +    error (1, errno, "%s: can't open", file_name);
> > +
> > +  /* Map in the whole file.  */
> > +  void *base = mmap (NULL, statbuf.st_size, PROT_READ | PROT_WRITE,
> > +                  MAP_SHARED, fd, 0);
> > +  if (base == MAP_FAILED)
> > +    {
> > +      errno_saved = errno;
> > +      close (fd);
> > +      error (1, errno_saved, "%s: failed to map", file_name);
> > +    }
> > +
> > +  ehdr = (ElfW(Ehdr) *) base;
> > +  if (ehdr->e_ident[EI_MAG0] != ELFMAG0
> > +      || ehdr->e_ident[EI_MAG1] != ELFMAG1
> > +      || ehdr->e_ident[EI_MAG2] != ELFMAG2
> > +      || ehdr->e_ident[EI_MAG3] != ELFMAG3)
> > +    {
> > +      close (fd);
> > +      error (1, 0, "%s: bad ELF header", file_name);
> > +    }
> > +
> > +  if (ehdr->e_type != ET_DYN)
> > +    {
> > +      close (fd);
> > +      error (1, 0, "%s: not shared library", file_name);
> > +    }
> > +
> > +  bool unsupported_class = true;
> > +  switch (ehdr->e_ident[EI_CLASS])
> > +    {
> > +    default:
> > +      break;
> > +
> > +    case ELFCLASS32:
> > +      unsupported_class = __ELF_NATIVE_CLASS != 32;
> > +      break;
> > +
> > +    case ELFCLASS64:
> > +      unsupported_class = __ELF_NATIVE_CLASS != 64;
> > +      break;
> > +    }
> > +
> > +  if (unsupported_class)
> > +    {
> > +      close (fd);
> > +      error (1, 0, "%s: unsupported ELF class: %d",
> > +          file_name, ehdr->e_ident[EI_CLASS]);
> > +    }
> > +
> > +  size_t phdr_size = sizeof (ElfW(Phdr)) * ehdr->e_phentsize;
> > +  if (statbuf.st_size < (ehdr->e_phoff + phdr_size))
> > +    {
> > +      close (fd);
> > +      error (1, 0, "%s: too small", file_name);
> > +    }
> > +
> > +  ElfW(Phdr) *phdr = (ElfW(Phdr) *) (base + ehdr->e_phoff);
> > +  for (int i = 0; i < ehdr->e_phnum; i++, phdr++)
> > +    if (phdr->p_type == PT_LOAD)
> > +      {
> > +     /* Update p_align of the first PT_LOAD segment.  */
> > +     phdr->p_align = update_p_align (phdr->p_align);
> > +     break;
> > +      }
> > +
> > +  munmap (base, statbuf.st_size);
> > +  close (fd);
> > +
> > +  return 0;
> > +}
>
> Look ok, although I think it would be simpler to build it once and use an
> argument of how to change the alignment instead of build once binary
> per test.  I am not if it is worth to use libsupport here.

This can be more flexible than an argument.  I prefer to keep
it ASIS.

> > diff --git a/elf/tst-p_align1.c b/elf/tst-p_align1.c
> > new file mode 100644
> > index 0000000000..cab9793220
> > --- /dev/null
> > +++ b/elf/tst-p_align1.c
> > @@ -0,0 +1,27 @@
> > +/* Check different alignments of PT_LOAD segments in a shared library.
> > +   Copyright (C) 2021 Free Software Foundation, Inc.
> > +   This file is part of the GNU C Library.
> > +
> > +   The GNU C Library is free software; you can redistribute it and/or
> > +   modify it under the terms of the GNU Lesser General Public
> > +   License as published by the Free Software Foundation; either
> > +   version 2.1 of the License, or (at your option) any later version.
> > +
> > +   The GNU C Library is distributed in the hope that it will be useful,
> > +   but WITHOUT ANY WARRANTY; without even the implied warranty of
> > +   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
> > +   Lesser General Public License for more details.
> > +
> > +   You should have received a copy of the GNU Lesser General Public
> > +   License along with the GNU C Library; if not, see
> > +   <https://www.gnu.org/licenses/>.  */
> > +
> > +extern int do_load_test (void);
>
> The do_load_test is defined as returning void on elf/tst-alignmod3.c,
> but it redefined here as returning 'int'.  This is causing a failure
> on aarch64.
>
> Maybe it would better to move it to a proper header.

Done.

> > +
> > +static int
> > +do_test (void)
> > +{
> > +  return do_load_test ();
> > +}
> > +
> > +#include <support/test-driver.c>
>
> And I think 'is_aligned' from sysdeps/generic/tst-stack-align.h
> seems to have the inverted semantic that makes things confusing
> (elf/tst-alignmod3.c check if it is equal to 0 to actually check
> if pointer is aligned).
>
> Maybe it would be better to return bool and 'true' if the pointer
> is indeed aligned (similar to what IS_ALIGNED/PTR_IS_ALIGNED
> internal macros do).

Fixed.

>    bool
>    __attribute__ ((weak, noclone, noinline))
>    is_aligned (void *p, int align)
>    {
>       return (((uintptr_t) p) & (align - 1)) == 0;
>    }
>
> > diff --git a/elf/tst-p_alignmod1-edit.c b/elf/tst-p_alignmod1-edit.c
> > new file mode 100644
> > index 0000000000..2aa0e7ccdd
> > --- /dev/null
> > +++ b/elf/tst-p_alignmod1-edit.c
> > @@ -0,0 +1,34 @@
> > +/* Reduce p_align of the first PT_LOAD segment by half.
> > +   Copyright (C) 2021 Free Software Foundation, Inc.
> > +   This file is part of the GNU C Library.
> > +
> > +   The GNU C Library is free software; you can redistribute it and/or
> > +   modify it under the terms of the GNU Lesser General Public
> > +   License as published by the Free Software Foundation; either
> > +   version 2.1 of the License, or (at your option) any later version.
> > +
> > +   The GNU C Library is distributed in the hope that it will be useful,
> > +   but WITHOUT ANY WARRANTY; without even the implied warranty of
> > +   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
> > +   Lesser General Public License for more details.
> > +
> > +   You should have received a copy of the GNU Lesser General Public
> > +   License along with the GNU C Library; if not, see
> > +   <https://www.gnu.org/licenses/>.  */
> > +
> > +#include "tst-elf-edit.h"
> > +
> > +/* Reduce p_align by half.  */
> > +
> > +static size_t
> > +update_p_align (size_t p_align)
> > +{
> > +  size_t pagesize = sysconf (_SC_PAGESIZE);
> > +  size_t new_p_align = p_align >> 1;
> > +
> > +  if (new_p_align < pagesize)
> > +    error (1, 0, "%s: new p_align (0x%zx) < page size (0x%zx)",
> > +        file_name, new_p_align, pagesize);
> > +
> > +  return new_p_align;
> > +}

Thanks.

-- 
H.J.

  reply	other threads:[~2022-01-18 21:54 UTC|newest]

Thread overview: 14+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-01-03 23:04 [PATCH v7 0/4] Properly align all PT_LOAD segments with tests H.J. Lu
2022-01-03 23:04 ` [PATCH v7 1/4] elf: Properly align all PT_LOAD segments [BZ #28676] H.J. Lu
2022-01-18 17:48   ` Adhemerval Zanella
2022-01-18 21:46     ` H.J. Lu
2022-01-03 23:04 ` [PATCH v7 2/4] elf: Add a test for PT_LOAD segments with mixed p_align " H.J. Lu
2022-01-18 16:25   ` Adhemerval Zanella
2022-01-18 21:53     ` H.J. Lu [this message]
2022-01-03 23:04 ` [PATCH v7 3/4] elf: Add a test for PT_LOAD segments with p_align == 1 [BZ #28688] H.J. Lu
2022-01-18 16:32   ` Adhemerval Zanella
2022-01-18 16:36     ` Adhemerval Zanella
2022-01-18 21:55     ` H.J. Lu
2022-01-03 23:04 ` [PATCH v7 4/4] elf: Add a test for PT_LOAD segments with invalid p_align " H.J. Lu
2022-01-18 18:05   ` Adhemerval Zanella
2022-01-18 21:50     ` H.J. Lu

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=CAMe9rOru9Tt+Movx-mRakJ2--GxjRjD0MLJTooG_Qxs6--HzNA@mail.gmail.com \
    --to=hjl.tools@gmail.com \
    --cc=adhemerval.zanella@linaro.org \
    --cc=fweimer@redhat.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).