From: Adhemerval Zanella Netto <adhemerval.zanella@linaro.org>
To: Joe Simmons-Talbott <josimmon@redhat.com>, libc-alpha@sourceware.org
Cc: Andreas Schwab <schwab@suse.de>, Carlos O'Donell <carlos@redhat.com>
Subject: Re: [PATCH v4] elf/rtld: Count skipped environment variables for enable_secure
Date: Tue, 16 Apr 2024 13:22:14 -0300 [thread overview]
Message-ID: <c1b4a33d-cb8d-4688-a0e9-285a9257aae4@linaro.org> (raw)
In-Reply-To: <20240412152126.663092-1-josimmon@redhat.com>
On 12/04/24 12:21, 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);
LGTM, thanks.
Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
> ---
> Changes to v3:
> * Run the testcase with ld.so explicitly.
>
> 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 | 10 +++++++++
> elf/rtld.c | 31 +++++++++++++++++++-------
> elf/tst-tunables-enable_secure-env.c | 33 ++++++++++++++++++++++++++++
> 3 files changed, 66 insertions(+), 8 deletions(-)
> create mode 100644 elf/tst-tunables-enable_secure-env.c
>
> diff --git a/elf/Makefile b/elf/Makefile
> index 6dad11bcfb..ea8a078f3e 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 := \
> @@ -1189,6 +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
> @@ -2216,6 +2218,14 @@ $(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
> 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");
> }
> }
> \f
> -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>
prev parent reply other threads:[~2024-04-16 16:22 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-04-12 15:21 Joe Simmons-Talbott
2024-04-16 16:22 ` Adhemerval Zanella Netto [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=c1b4a33d-cb8d-4688-a0e9-285a9257aae4@linaro.org \
--to=adhemerval.zanella@linaro.org \
--cc=carlos@redhat.com \
--cc=josimmon@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).