public inbox for libc-alpha@sourceware.org
 help / color / mirror / Atom feed
From: Florian Weimer <fw@deneb.enyo.de>
To: Szabolcs Nagy <szabolcs.nagy@arm.com>
Cc: libc-alpha@sourceware.org,  nd@arm.com
Subject: Re: [PATCH] libio: Disable vtable validation in case of interposition [BZ #23313]
Date: Wed, 20 Jun 2018 07:58:00 -0000	[thread overview]
Message-ID: <87woutzyvb.fsf@mid.deneb.enyo.de> (raw)
In-Reply-To: <87h8ly3805.fsf@mid.deneb.enyo.de> (Florian Weimer's message of "Tue, 19 Jun 2018 21:12:58 +0200")

* Florian Weimer:

> * Florian Weimer:
>
>> * Szabolcs Nagy:
>>
>>> the check looks ok to me
>>> i think hidden symbol alias would work too
>>>
>>> return &_IO_2_1_stdin == &_IO_2_1_stdin_internal_alias || ...;
>>>
>>> and it may be a bit nicer than checking the section.. i'm not sure
>>
>> I had hoped that GCC would combine the comparisons in some way, but it
>> doesn't do that either way.
>>
>> The hidden alias approach also has the advantage that it does not
>> touch the _IO_MTSAFE_IO code.
>>
>>
>> 2018-06-19  Florian Weimer  <fweimer@redhat.com>
>>
>> 	[BZ #23313]
>> 	* libio/libioP.h [SHARED] (_IO_2_1_stdin_hidden)
>> 	(_IO_2_1_stdout_hidden, _IO_2_1_stderr_hidden): Declare.
>> 	* libio/stdfiles.c [SHARED] (_IO_2_1_stdin_hidden)
>> 	(_IO_2_1_stdout_hidden, _IO_2_1_stderr_hidden): Define aliases.
>> 	* libio/vtables.c (stdfiles_interposed): New function.
>> 	(_IO_vtable_check): Call it.
>
> This seems to be overly conservative.  I think we do not actually have
> to care about interposition.  We should read the vtables in the
> libc.so startup code and check if they match the values in stdfiles.c.
> This way, vtable validation is not disabled if there is merely a copy
> relocation.

I suppose this is the best way to implement this.  No regressions in
the test suite in x86-64, the vtables test I posted still passes, and
yadex still starts, too.

Since this code only runs once, it should be okay from a hardening
standpoint.

libio: Detect foreign vtables in interposed standard streams [BZ #23313]

2018-06-20  Florian Weimer  <fweimer@redhat.com>

	[BZ #23313]
	* libio/vtables.c (check_stdfiles_vtables): New ELF constructor.

diff --git a/libio/vtables.c b/libio/vtables.c
index 9fd4ccf642..5b1b581984 100644
--- a/libio/vtables.c
+++ b/libio/vtables.c
@@ -71,3 +71,17 @@ _IO_vtable_check (void)
 
   __libc_fatal ("Fatal error: glibc detected an invalid stdio handle\n");
 }
+
+/* Some variants of libstdc++ interpose _IO_2_1_stdin_ etc. and
+   install their own vtables directly, without calling _IO_init or
+   other functions.  Detect this by looking at the vtables values
+   during startup, and disable vtable validation in this case.  */
+__attribute__ ((constructor))
+static void
+check_stdfiles_vtables (void)
+{
+  if (_IO_2_1_stdin_.vtable != &_IO_file_jumps
+      || _IO_2_1_stdout_.vtable != &_IO_file_jumps
+      || _IO_2_1_stderr_.vtable != &_IO_file_jumps)
+    IO_set_accept_foreign_vtables (&_IO_vtable_check);
+}

  reply	other threads:[~2018-06-20  7:58 UTC|newest]

Thread overview: 9+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-06-19  9:56 Florian Weimer
2018-06-19 10:38 ` Szabolcs Nagy
2018-06-19 11:27   ` Florian Weimer
2018-06-19 19:13     ` Florian Weimer
2018-06-20  7:58       ` Florian Weimer [this message]
2018-06-21 21:20         ` Florian Weimer
2018-06-22 10:16           ` Szabolcs Nagy
2018-06-22 10:51             ` Florian Weimer
2018-06-26  9:33               ` Florian Weimer

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=87woutzyvb.fsf@mid.deneb.enyo.de \
    --to=fw@deneb.enyo.de \
    --cc=libc-alpha@sourceware.org \
    --cc=nd@arm.com \
    --cc=szabolcs.nagy@arm.com \
    /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).