public inbox for libc-alpha@sourceware.org
 help / color / mirror / Atom feed
From: Joe Simmons-Talbott <josimmon@redhat.com>
To: Adhemerval Zanella Netto <adhemerval.zanella@linaro.org>
Cc: libc-alpha@sourceware.org, Andreas Schwab <schwab@suse.de>,
	 "Carlos O'Donell" <carlos@redhat.com>
Subject: Re: [PATCH v3] elf/rtld: Count skipped environment variables for enable_secure
Date: Fri, 12 Apr 2024 11:23:48 -0400	[thread overview]
Message-ID: <CAAQBMsOxdKtOfDxgv+kkcaaghb0A8n0XwG=h=ih3=7w-4+o2hw@mail.gmail.com> (raw)
In-Reply-To: <2c2922df-4900-4efb-b317-6d6760c5a61c@linaro.org>

On Fri, Apr 12, 2024 at 10:14 AM Adhemerval Zanella Netto
<adhemerval.zanella@linaro.org> wrote:
>
>
>
> On 11/04/24 13:53, Joe Simmons-Talbott wrote:
> > When using the glibc.rtld.enable_secure tunable we need to keep track of
> > the count of environment variables we skip due to __libc_enable_secure
> > being set and adjust the auxv section of the stack.  This fixes an
> > assertion when running ld.so directly with glibc.rtld.enable_secure set.
> > Add a testcase that ensures the assert is not hit.
> >
> > elf/rtld.c:1324   assert (auxv == sp + 1);
> > ---
> > Changes to v2:
> > * Add testcase that triggers the assertion before this patch and does
> >   not after the patch.
> >
> > Changes to v1:
> > * Add comment explaining how and why skip_env is adjusted.
> >
> >  elf/Makefile                         |  3 +++
> >  elf/rtld.c                           | 31 +++++++++++++++++++-------
> >  elf/tst-tunables-enable_secure-env.c | 33 ++++++++++++++++++++++++++++
> >  3 files changed, 59 insertions(+), 8 deletions(-)
> >  create mode 100644 elf/tst-tunables-enable_secure-env.c
> >
> > diff --git a/elf/Makefile b/elf/Makefile
> > index 6dad11bcfb..5c35db1799 100644
> > --- a/elf/Makefile
> > +++ b/elf/Makefile
> > @@ -315,6 +315,7 @@ tests := \
> >    tst-leaks1 \
> >    tst-stringtable \
> >    tst-tls9 \
> > +  tst-tunables-enable_secure-env \
> >    # tests
> >
> >  tests-internal := \
> > @@ -336,6 +337,8 @@ tst-tls9-static-ENV = $(static-dlopen-environment)
> >  tst-single_threaded-static-dlopen-ENV = $(static-dlopen-environment)
> >  tst-rootdir-ENV = LD_LIBRARY_PATH=/
> >
> > +tst-tunables-enable_secure-env-ENV = GLIBC_TUNABLES=glibc.rtld.enable_secure=1
> > +
>
> I think since the idea is to check for running ld.so directly, you will need to
> explicit issue the loader:
>
> diff --git a/elf/Makefile b/elf/Makefile
> index 5c35db1799..1598fbd6d7 100644
> --- a/elf/Makefile
> +++ b/elf/Makefile
> @@ -337,8 +337,6 @@ tst-tls9-static-ENV = $(static-dlopen-environment)
>  tst-single_threaded-static-dlopen-ENV = $(static-dlopen-environment)
>  tst-rootdir-ENV = LD_LIBRARY_PATH=/
>
> -tst-tunables-enable_secure-env-ENV = GLIBC_TUNABLES=glibc.rtld.enable_secure=1
> -
>  tests += \
>    argv0test \
>    constload1 \
> @@ -1192,6 +1190,7 @@ tests-special += \
>    $(objpfx)tst-trace3.out \
>    $(objpfx)tst-trace4.out \
>    $(objpfx)tst-trace5.out \
> +  $(objpfx)tst-tunables-enable_secure-env.out \
>    $(objpfx)tst-unused-dep-cmp.out \
>    $(objpfx)tst-unused-dep.out \
>    # tests-special
> @@ -2219,6 +2218,13 @@ $(objpfx)tst-unused-dep-cmp.out: $(objpfx)tst-unused-dep.out
>         cmp $< /dev/null > $@; \
>         $(evaluate-test)
>
> +$(objpfx)tst-tunables-enable_secure-env.out: $(objpfx)tst-tunables-enable_secure-env
> +       $(test-wrapper-env) \
> +       GLIBC_TUNABLES=glibc.rtld.enable_secure=1 \
> +       $(rtld-prefix) \
> +         $< > $@; \
> +       $(evaluate-test)
> +
>  $(objpfx)tst-audit11.out: $(objpfx)tst-auditmod11.so $(objpfx)tst-audit11mod1.so
>  tst-audit11-ENV = LD_AUDIT=$(objpfx)tst-auditmod11.so
>  $(objpfx)tst-audit11mod1.so: $(objpfx)tst-audit11mod2.so
>
> Otherwise this won't check the bug with --enable-hardcoded-path-in-tests.
>
> The rest looks ok.

Thanks for the review.  I've posted a v4 [1] with this change.

Thanks,
Joe

[1] https://sourceware.org/pipermail/libc-alpha/2024-April/156040.html

>
> >  tests += \
> >    argv0test \
> >    constload1 \
> > diff --git a/elf/rtld.c b/elf/rtld.c
> > index 18d73f19c6..d116a436f5 100644
> > --- a/elf/rtld.c
> > +++ b/elf/rtld.c
> > @@ -155,7 +155,7 @@ static void dl_main_state_init (struct dl_main_state *state);
> >     Since all of them start with `LD_' we are a bit smarter while finding
> >     all the entries.  */
> >  extern char **_environ attribute_hidden;
> > -static void process_envvars (struct dl_main_state *state);
> > +static int process_envvars (struct dl_main_state *state);
> >
> >  int _dl_argc attribute_relro attribute_hidden;
> >  char **_dl_argv attribute_relro = NULL;
> > @@ -1289,7 +1289,7 @@ rtld_setup_main_map (struct link_map *main_map)
> >     _dl_argv and _dl_argc accordingly.  Those arguments are removed from
> >     argv here.  */
> >  static void
> > -_dl_start_args_adjust (int skip_args)
> > +_dl_start_args_adjust (int skip_args, int skip_env)
> >  {
> >    void **sp = (void **) (_dl_argv - skip_args - 1);
> >    void **p = sp + skip_args;
> > @@ -1321,7 +1321,7 @@ _dl_start_args_adjust (int skip_args)
> >    while (*p != NULL);
> >
> >  #ifdef HAVE_AUX_VECTOR
> > -  void **auxv = (void **) GLRO(dl_auxv) - skip_args;
> > +  void **auxv = (void **) GLRO(dl_auxv) - skip_args - skip_env;
> >    GLRO(dl_auxv) = (ElfW(auxv_t) *) auxv; /* Aliasing violation.  */
> >    assert (auxv == sp + 1);
> >
> > @@ -1352,6 +1352,7 @@ dl_main (const ElfW(Phdr) *phdr,
> >    unsigned int i;
> >    bool rtld_is_main = false;
> >    void *tcbp = NULL;
> > +  int skip_env = 0;
> >
> >    struct dl_main_state state;
> >    dl_main_state_init (&state);
> > @@ -1365,7 +1366,7 @@ dl_main (const ElfW(Phdr) *phdr,
> >  #endif
> >
> >    /* Process the environment variable which control the behaviour.  */
> > -  process_envvars (&state);
> > +  skip_env = process_envvars (&state);
> >
> >  #ifndef HAVE_INLINED_SYSCALLS
> >    /* Set up a flag which tells we are just starting.  */
> > @@ -1630,7 +1631,7 @@ dl_main (const ElfW(Phdr) *phdr,
> >          _dl_argv[0] = argv0;
> >
> >        /* Adjust arguments for the application entry point.  */
> > -      _dl_start_args_adjust (_dl_argv - orig_argv);
> > +      _dl_start_args_adjust (_dl_argv - orig_argv, skip_env);
> >      }
> >    else
> >      {
> > @@ -2534,11 +2535,12 @@ a filename can be specified using the LD_DEBUG_OUTPUT environment variable.\n");
> >      }
> >  }
> >
> > -static void
> > +static int
> >  process_envvars_secure (struct dl_main_state *state)
> >  {
> >    char **runp = _environ;
> >    char *envline;
> > +  int skip_env = 0;
> >
> >    while ((envline = _dl_next_ld_env_entry (&runp)) != NULL)
> >      {
> > @@ -2580,6 +2582,14 @@ process_envvars_secure (struct dl_main_state *state)
> >    const char *nextp = UNSECURE_ENVVARS;
> >    do
> >      {
> > +      /* Keep track of the number of environment variables that were set in
> > +         the environment and are unset below.  Use getenv() which returns
> > +      non-NULL if the variable is set in the environment.  This count is
> > +      needed if we need to adjust the location of the AUX vector on the
> > +      stack when running ld.so directly. */
> > +      if (getenv (nextp) != NULL)
> > +        skip_env++;
> > +
> >        unsetenv (nextp);
> >        nextp = strchr (nextp, '\0') + 1;
> >      }
> > @@ -2592,6 +2602,8 @@ process_envvars_secure (struct dl_main_state *state)
> >        || state->mode != rtld_mode_normal
> >        || state->version_info)
> >      _exit (5);
> > +
> > +  return skip_env;
> >  }
> >
> >  static void
> > @@ -2745,13 +2757,16 @@ process_envvars_default (struct dl_main_state *state)
> >      }
> >  }
> >
> > -static void
> > +static int
> >  process_envvars (struct dl_main_state *state)
> >  {
> > +  int skip_env = 0;
> >    if (__glibc_unlikely (__libc_enable_secure))
> > -    process_envvars_secure (state);
> > +    skip_env += process_envvars_secure (state);
> >    else
> >      process_envvars_default (state);
> > +
> > +  return skip_env;
> >  }
> >
> >  #if HP_TIMING_INLINE
> > diff --git a/elf/tst-tunables-enable_secure-env.c b/elf/tst-tunables-enable_secure-env.c
> > new file mode 100644
> > index 0000000000..24e846f299
> > --- /dev/null
> > +++ b/elf/tst-tunables-enable_secure-env.c
> > @@ -0,0 +1,33 @@
> > +/* Check enable_secure tunable handles removed ENV variables without
> > +   assertions.
> > +   Copyright (C) 2024 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 <support/capture_subprocess.h>
> > +#include <support/check.h>
> > +
> > +static int
> > +do_test (int argc, char *argv[])
> > +{
> > +  /* Ensure that no assertions are hit when a dynamically linked application
> > +     runs.  This test requires that GLIBC_TUNABLES=glibc.rtld.enable_secure=1
> > +     is set. */
> > +  return 0;
> > +}
> > +
> > +#define TEST_FUNCTION_ARGV do_test
> > +#include <support/test-driver.c>
>


-- 
Joe Simmons-Talbott


      reply	other threads:[~2024-04-12 15:24 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-04-11 16:53 Joe Simmons-Talbott
2024-04-12 14:13 ` Adhemerval Zanella Netto
2024-04-12 15:23   ` Joe Simmons-Talbott [this message]

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='CAAQBMsOxdKtOfDxgv+kkcaaghb0A8n0XwG=h=ih3=7w-4+o2hw@mail.gmail.com' \
    --to=josimmon@redhat.com \
    --cc=adhemerval.zanella@linaro.org \
    --cc=carlos@redhat.com \
    --cc=libc-alpha@sourceware.org \
    --cc=schwab@suse.de \
    /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).