Re: [PATCH] Use new oom_score_adj without a new compile-time constant

From: Dan McGee <dan(at)archlinux(dot)org>
To: pgsql-hackers(at)postgresql(dot)org
Subject: Re: [PATCH] Use new oom_score_adj without a new compile-time constant
Date: 2011-09-19 20:36:50
Message-ID: CAEik5nPuNBHZgOGX3-baU=WKD353Hrw5qnWrw8dr6iViQ6553A@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On Mon, Sep 19, 2011 at 3:11 PM, Dan McGee <dan(at)archlinux(dot)org> wrote:
> This is one way to prevent the kernel warning message without having to
> introduce a new constant. Scale the old oom_adj-style value the same way
> the kernel internally does it and use that instead if oom_score_adj is
> available for writing.
>
> Signed-off-by: Dan McGee <dan(at)archlinux(dot)org>
> ---
>
> This addresses some of the concerns raised on the ML and will at least keep
> those of us running modern kernels happy.
>
> Alternatively one could switch the symbol used to be the new style and have the
> old one computed; however this is a pain for legacy vs. current versions.
>
>  src/backend/postmaster/fork_process.c |   22 +++++++++++++++++++++-
>  1 files changed, 21 insertions(+), 1 deletions(-)
>
> diff --git a/src/backend/postmaster/fork_process.c b/src/backend/postmaster/fork_process.c
> index b2fe9a1..3cded54 100644
> --- a/src/backend/postmaster/fork_process.c
> +++ b/src/backend/postmaster/fork_process.c
> @@ -81,16 +81,36 @@ fork_process(void)
>                         * Use open() not stdio, to ensure we control the open flags. Some
>                         * Linux security environments reject anything but O_WRONLY.
>                         */
> -                       int                     fd = open("/proc/self/oom_adj", O_WRONLY, 0);
> +                       int                     fd = open("/proc/self/oom_score_adj", O_WRONLY, 0);
>
>                        /* We ignore all errors */
>                        if (fd >= 0)
>                        {
>                                char            buf[16];
> +                               int             oom_score_adj;
>
> +                               /*
> +                                * The compile-time value is the old style oom_adj;
> +                                * scale it the same way the kernel does to
> +                                * convert to the new style oom_score_adj. This
> +                                * should become a constant at compile time.
> +                                * Valid values range from -17 (never kill) to
> +                                * 15; no attempt of validation is done.
> +                                */
> +                               oom_score_adj = LINUX_OOM_ADJ * 1000 / 17;
>                                snprintf(buf, sizeof(buf), "%d\n", LINUX_OOM_ADJ);
Of course it would help to actually use the computed value here; this should be:
snprintf(buf, sizeof(buf), "%d\n",
oom_score_adj);

>                                (void) write(fd, buf, strlen(buf));
>                                close(fd);
> +                       } else if (errno == EEXIST) {
> +                               int             fd = open("/proc/self/oom_adj", O_WRONLY, 0);
> +                               if (fd >= 0)
> +                               {
> +                                       char    buf[16];
> +
> +                                       snprintf(buf, sizeof(buf), "%d\n", LINUX_OOM_ADJ);
> +                                       (void) write(fd, buf, strlen(buf));
> +                                       close(fd);
> +                               }
>                        }
>                }
>  #endif   /* LINUX_OOM_ADJ */
> --
> 1.7.6.1
>
>

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Greg Smith 2011-09-19 20:44:17 Re: Is there really no interest in SQL Standard?
Previous Message Dan McGee 2011-09-19 20:11:18 [PATCH] Use new oom_score_adj without a new compile-time constant