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 2/5] resolv: Use <file_change_detection.h> in __resolv_conf_get_current
Date: Thu, 13 Feb 2020 20:53:00 -0000	[thread overview]
Message-ID: <6bb92349-0f68-dc7b-0b57-1f68a028b55a@linaro.org> (raw)
In-Reply-To: <c657cfe9aedbc48a368f11442f453190ed1cf1e9.1579631655.git.fweimer@redhat.com>



On 21/01/2020 15:41, Florian Weimer wrote:
> Only minor functional changes (i.e., regarding the handling of
> directories, which are now treated as empty files).

LGTM, thanks.

Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>

> ---
>  resolv/resolv_conf.c | 43 ++++++++-----------------------------------
>  1 file changed, 8 insertions(+), 35 deletions(-)
> 
> diff --git a/resolv/resolv_conf.c b/resolv/resolv_conf.c
> index 08c50ef19e..d954ba9a5a 100644
> --- a/resolv/resolv_conf.c
> +++ b/resolv/resolv_conf.c
> @@ -24,6 +24,7 @@
>  #include <resolv-internal.h>
>  #include <sys/stat.h>
>  #include <libc-symbols.h>
> +#include <file_change_detection.h>
>  
>  /* _res._u._ext.__glibc_extension_index is used as an index into a
>     struct resolv_conf_array object.  The intent of this construction
> @@ -68,12 +69,8 @@ struct resolv_conf_global
>    /* Cached current configuration object for /etc/resolv.conf.  */
>    struct resolv_conf *conf_current;
>  
> -  /* These properties of /etc/resolv.conf are used to check if the
> -     configuration needs reloading.  */
> -  struct timespec conf_mtime;
> -  struct timespec conf_ctime;
> -  off64_t conf_size;
> -  ino64_t conf_ino;
> +  /* File system identification for /etc/resolv.conf.  */
> +  struct file_change_detection file_resolve_conf;
>  };
>  
>  /* Lazily allocated storage for struct resolv_conf_global.  */

Ok.

> @@ -123,37 +120,16 @@ conf_decrement (struct resolv_conf *conf)
>  struct resolv_conf *
>  __resolv_conf_get_current (void)
>  {
> -  struct stat64 st;
> -  if (stat64 (_PATH_RESCONF, &st) != 0)
> -    {
> -    switch (errno)
> -      {
> -      case EACCES:
> -      case EISDIR:
> -      case ELOOP:
> -      case ENOENT:
> -      case ENOTDIR:
> -      case EPERM:
> -        /* Ignore errors due to file system contents.  */
> -        memset (&st, 0, sizeof (st));
> -        break;
> -      default:
> -        /* Other errors are fatal.  */
> -        return NULL;
> -      }
> -    }
> +  struct file_change_detection initial;
> +  if (!file_change_detection_for_path (&initial, _PATH_RESCONF))
> +    return NULL;
>  

Ok.

>    struct resolv_conf_global *global_copy = get_locked_global ();
>    if (global_copy == NULL)
>      return NULL;
>    struct resolv_conf *conf;
>    if (global_copy->conf_current != NULL
> -      && (global_copy->conf_mtime.tv_sec == st.st_mtim.tv_sec
> -          && global_copy->conf_mtime.tv_nsec == st.st_mtim.tv_nsec
> -          && global_copy->conf_ctime.tv_sec == st.st_ctim.tv_sec
> -          && global_copy->conf_ctime.tv_nsec == st.st_ctim.tv_nsec
> -          && global_copy->conf_ino == st.st_ino
> -          && global_copy->conf_size == st.st_size))
> +      && file_is_unchanged (&initial, &global_copy->file_resolve_conf))
>      /* We can reuse the cached configuration object.  */
>      conf = global_copy->conf_current;
>    else
> @@ -171,10 +147,7 @@ __resolv_conf_get_current (void)
>               read could be a newer version of the file, but this does
>               not matter because this will lead to an extraneous reload
>               later.  */
> -          global_copy->conf_mtime = st.st_mtim;
> -          global_copy->conf_ctime = st.st_ctim;
> -          global_copy->conf_ino = st.st_ino;
> -          global_copy->conf_size = st.st_size;
> +          global_copy->file_resolve_conf = initial;
>          }
>      }
>  
> 

Ok.

  reply	other threads:[~2020-02-13 20:53 UTC|newest]

Thread overview: 15+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-01-21 18:41 [PATCH 0/5] Race condition in /etc/resolv.conf reloading (bug 25420) Florian Weimer
2020-01-21 18:41 ` [PATCH 1/5] Add internal <file_change_detection.h> header file Florian Weimer
2020-02-10 19:47   ` Adhemerval Zanella
2020-02-10 19:58     ` Florian Weimer
2020-02-10 20:57       ` Adhemerval Zanella
2020-01-21 18:42 ` [PATCH 4/5] resolv: Enhance __resolv_conf_load to capture file change data Florian Weimer
2020-02-13 21:33   ` Adhemerval Zanella
2020-01-21 18:42 ` [PATCH 3/5] resolv: Fix file handle leak in __resolv_conf_load [BZ #25429] Florian Weimer
2020-02-13 21:01   ` Adhemerval Zanella
2020-02-13 21:08     ` Florian Weimer
2020-02-13 21:30       ` Adhemerval Zanella
2020-01-21 18:42 ` [PATCH 2/5] resolv: Use <file_change_detection.h> in __resolv_conf_get_current Florian Weimer
2020-02-13 20:53   ` Adhemerval Zanella [this message]
2020-01-21 21:28 ` [PATCH 5/5] resolv: Fix ABA race in /etc/resolv.conf change detection [BZ #25420] Florian Weimer
2020-02-13 21:59   ` Adhemerval Zanella

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=6bb92349-0f68-dc7b-0b57-1f68a028b55a@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).