public inbox for libc-alpha@sourceware.org
 help / color / mirror / Atom feed
From: Adhemerval Zanella <adhemerval.zanella@linaro.org>
To: "H.J. Lu" <hjl.tools@gmail.com>, libc-alpha@sourceware.org
Cc: 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 13:25:59 -0300	[thread overview]
Message-ID: <5d0957aa-5128-601e-fece-837cdc6e6a20@linaro.org> (raw)
In-Reply-To: <20220103230433.1907240-3-hjl.tools@gmail.com>



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 and it instead by copying and changing
on each test.

> +	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.

> 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.

> +
> +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).

   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;
> +}

  reply	other threads:[~2022-01-18 16:26 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 [this message]
2022-01-18 21:53     ` H.J. Lu
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=5d0957aa-5128-601e-fece-837cdc6e6a20@linaro.org \
    --to=adhemerval.zanella@linaro.org \
    --cc=fweimer@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).