public inbox for libc-alpha@sourceware.org
 help / color / mirror / Atom feed
From: Adhemerval Zanella <adhemerval.zanella@linaro.org>
To: libc-alpha@sourceware.org
Subject: Re: [PATCH 2/6] New internal function __access_noerrno
Date: Tue, 08 Nov 2016 17:10:00 -0000	[thread overview]
Message-ID: <317e63c8-1681-84b1-e278-c6ecae71e657@linaro.org> (raw)
In-Reply-To: <1477320168-23397-3-git-send-email-siddhesh@sourceware.org>



On 24/10/2016 12:42, Siddhesh Poyarekar wrote:
> Implement an internal version of __access called __access_noerrno that
> avoids setting errno.  This is useful to check accessibility of files
> very early on in process startup i.e. before TLS setup.  This allows
> tunables to replace MALLOC_CHECK_ safely (i.e. check existence of
> /etc/suid-debug to enable/disable MALLOC_CHECK) and at the same time
> initialize very early so that it can override IFUNCs.
> 
> 	* include/unistd.h [IS_IN (rtld) || !defined SHARED]: Declare
> 	__access_noerrno.
> 	* io/Makefile (routines): Add access_noerrno.
> 	* io/access.c (__ACCESS)[!__ACCESS]: Define as __access.
> 	(__access): Rename to __ACCESS.
> 	[!NOERRNO]: Retain default __access logic.
> 	* io/access_noerrno.c: New file.
> 	* sysdeps/mach/hurd/access.c (__ACCESS)[!__ACCESS]: Define as
> 	__access.
> 	(__HURD_FAIL): New macro.
> 	(__access): Rename to __ACCESS.  Use __HURD_FAIL instead of
> 	__hurd_fail.
> 	[!NOERRNO]: Set weak alias to access.
> 	* sysdeps/nacl/access.c (__ACCESS)[!__ACCESS]: Define as
> 	__access.
> 	(DO_NACL_CALL): New macro.
> 	(__access): Rename to __ACCESS.  Use DO_NACL_CALL instead of
> 	NACL_CALL.
> 	[!NOERRNO]: Set weak alias to access.
> 	* sysdeps/nacl/nacl-interfaces.h (NACL_CALL_NOERRNO): New
> 	macro.
> 	* sysdeps/unix/access_noerrno.c: New file.
> 	* sysdeps/unix/sysv/linux/generic/access.c: Include sysdep.h.
> 	(__ACCESS)[!__ACCESS]: Define as __access.
> 	(__access): Rename to __ACCESS.
> 	[NOERRNO]: Call faccessat syscall without setting errno.
> ---
>  include/unistd.h                         |  6 +++++
>  io/Makefile                              |  1 +
>  io/access.c                              | 10 ++++++++-
>  io/access_noerrno.c                      | 21 ++++++++++++++++++
>  sysdeps/mach/hurd/access.c               | 20 +++++++++++++----
>  sysdeps/nacl/access.c                    | 16 ++++++++++++--
>  sysdeps/nacl/nacl-interfaces.h           |  4 ++++
>  sysdeps/unix/access_noerrno.c            | 38 ++++++++++++++++++++++++++++++++
>  sysdeps/unix/sysv/linux/generic/access.c | 19 +++++++++++++++-
>  9 files changed, 127 insertions(+), 8 deletions(-)
>  create mode 100644 io/access_noerrno.c
>  create mode 100644 sysdeps/unix/access_noerrno.c
> 
> diff --git a/include/unistd.h b/include/unistd.h
> index d2802b2..6144f41 100644
> --- a/include/unistd.h
> +++ b/include/unistd.h
> @@ -181,6 +181,12 @@ extern int __getlogin_r_loginuid (char *name, size_t namesize)
>  #   include <dl-unistd.h>
>  #  endif
>  
> +#  if IS_IN (rtld) || !defined SHARED
> +/* __access variant that does not set errno.  Used in very early initialization
> +   code in libc.a and ld.so.  */

Is this comment correct? Checking the patch I am seeing it builds
only for libc and there is no resulting __access_noerrno on ld.so.

> diff --git a/sysdeps/unix/access_noerrno.c b/sysdeps/unix/access_noerrno.c
> new file mode 100644
> index 0000000..1ff90a2
> --- /dev/null
> +++ b/sysdeps/unix/access_noerrno.c
> @@ -0,0 +1,38 @@
> +/* Test for access to a file but do not set errno on error.
> +   Copyright (C) 2016 Free Software Foundation, Inc.
> +   This file is part of the GNU C Library.
> +   Contributed by Chris Metcalf <cmetcalf@tilera.com>, 2011.
> +
> +   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
> +   <http://www.gnu.org/licenses/>.  */
> +
> +#include <errno.h>
> +#include <stddef.h>
> +#include <unistd.h>
> +#include <fcntl.h>
> +#include <sysdep-cancel.h>
> +#include <sysdep.h>
> +
> +/* Test for access to FILE.  */
> +int
> +__access_noerrno (const char *file, int type)
> +{
> +  INTERNAL_SYSCALL_DECL (err);
> +  int res;
> +  res = INTERNAL_SYSCALL (access, err, 2, file, type);
> +  if (INTERNAL_SYSCALL_ERROR_P (res, err))
> +    return INTERNAL_SYSCALL_ERRNO (res, err);
> +  else
> +    return 0;
> +}

I think it would be simpler to just 1. consolidation Linux access
implementation and 2. add the '_noerrno' on same file.

The 1. would be simpler to just:

   1.1. Remove access from sysdeps/unix/syscalls.list
   1.2. Move sysdeps/unix/sysv/linux/generic/access.c to
	sysdeps/unix/sysv/linux/access.c
   1.3. And implement access checking for __NR_access and
	using __NR_facessat if is not defined.  Something
	like:

[...]
/* Test for access to FILE.  */
int
__access (const char *file, int type)
{
#if __NR_access
  return INLINE_SYSCALL_CALL (access, file, type);
#else
  return INLINE_SYSCALL_CALL (faccessat, AT_FDCWD, file, type);
#endif
}
[...]

Then __access_noerro could be just implemented on same file.
I think it has the advantage of not splitting the access
on multiple files.

I think for hurd and nacl it would also result in simplify
code.  What do you think?


> diff --git a/sysdeps/unix/sysv/linux/generic/access.c b/sysdeps/unix/sysv/linux/generic/access.c
> index 586aa93..5bafc06 100644
> --- a/sysdeps/unix/sysv/linux/generic/access.c
> +++ b/sysdeps/unix/sysv/linux/generic/access.c
> @@ -21,11 +21,28 @@
>  #include <unistd.h>
>  #include <fcntl.h>
>  #include <sysdep-cancel.h>
> +#include <sysdep.h>
> +
> +#ifndef __ACCESS
> +# define __ACCESS __access
> +#endif
>  
>  /* Test for access to FILE.  */
>  int
> -__access (const char *file, int type)
> +__ACCESS (const char *file, int type)
>  {
> +#ifndef NOERRNO
>    return INLINE_SYSCALL (faccessat, 3, AT_FDCWD, file, type);
> +#else
> +  INTERNAL_SYSCALL_DECL (err);
> +  int res;
> +  res = INTERNAL_SYSCALL (faccessat, err, 3, AT_FDCWD, file, type);
> +  if (INTERNAL_SYSCALL_ERROR_P (res, err))
> +    return INTERNAL_SYSCALL_ERRNO (res, err);
> +  else
> +    return 0;
> +#endif
>  }
> +#ifndef NOERRNO
>  weak_alias (__access, access)
> +#endif
> 

  reply	other threads:[~2016-11-08 17:10 UTC|newest]

Thread overview: 21+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-10-24 14:43 [PATCH v5 0/6] glibc tunables Siddhesh Poyarekar
2016-10-24 14:43 ` [PATCH 5/6] Enhance --enable-tunables to select tunables frontend at build time Siddhesh Poyarekar
2016-10-24 14:43 ` [PATCH 6/6] User manual documentation for tunables Siddhesh Poyarekar
2016-10-24 17:22   ` Joseph Myers
2016-10-26  9:53     ` Siddhesh Poyarekar
2016-10-24 14:43 ` [PATCH 2/6] New internal function __access_noerrno Siddhesh Poyarekar
2016-11-08 17:10   ` Adhemerval Zanella [this message]
2016-11-08 19:00     ` Siddhesh Poyarekar
2016-11-10  5:47       ` Siddhesh Poyarekar
2016-11-10 12:15         ` Adhemerval Zanella
2016-11-10 13:05           ` Siddhesh Poyarekar
2016-11-10 16:07             ` Adhemerval Zanella
2016-10-24 14:43 ` [PATCH 4/6] Initialize tunable list with the GLIBC_TUNABLES environment variable Siddhesh Poyarekar
2016-10-24 15:01   ` Andreas Schwab
2016-10-26  9:48     ` Siddhesh Poyarekar
2016-10-24 14:43 ` [PATCH 3/6] Add framework for tunables Siddhesh Poyarekar
2016-10-24 14:43 ` [PATCH 1/6] Static inline functions for mallopt helpers Siddhesh Poyarekar
2016-10-26 17:51   ` DJ Delorie
2016-10-27  3:12     ` Siddhesh Poyarekar
2016-11-03 10:26 ` [PING][PATCH v5 0/6] glibc tunables Siddhesh Poyarekar
2016-11-08  5:37   ` [PING 2][PATCH " Siddhesh Poyarekar

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=317e63c8-1681-84b1-e278-c6ecae71e657@linaro.org \
    --to=adhemerval.zanella@linaro.org \
    --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).