Received-SPF: pass (google.com: domain of linux-nfs-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Subject: Re: [PATCH 2/5] svc: Batch allocations of pollfds
To:     Doug Nazar <nazard@nazar.ca>, libtirpc-devel@lists.sourceforge.net
Cc:     linux-nfs@vger.kernel.org
References: <20200722053445.27987-1-nazard@nazar.ca>
 <20200722053445.27987-3-nazard@nazar.ca>
From:   Steve Dickson <SteveD@RedHat.com>
Message-ID: <fc8ebe17-7e16-6be5-2323-6111a393510e@RedHat.com>
Date:   Wed, 29 Jul 2020 10:20:16 -0400
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101
 Thunderbird/68.5.0
MIME-Version: 1.0
In-Reply-To: <20200722053445.27987-3-nazard@nazar.ca>
Content-Type: text/plain; charset=utf-8
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Sender: linux-nfs-owner@vger.kernel.org
Precedence: bulk


On 7/22/20 1:34 AM, Doug Nazar wrote:
> Signed-off-by: Doug Nazar <nazard@nazar.ca>
> ---
>  src/svc.c | 16 +++++++++-------
>  1 file changed, 9 insertions(+), 7 deletions(-)
> 
> diff --git a/src/svc.c b/src/svc.c
> index 6db164b..57f7ba3 100644
> --- a/src/svc.c
> +++ b/src/svc.c
> @@ -54,6 +54,7 @@
>  #include "rpc_com.h"
>  
>  #define	RQCRED_SIZE	400	/* this size is excessive */
> +#define	SVC_POLLFD_INCREMENT	16
>  
>  #define max(a, b) (a > b ? a : b)
>  
> @@ -107,6 +108,7 @@ xprt_register (xprt)
>    if (sock < _rpc_dtablesize())
>      {
>        int i;
> +      size_t size;
>        struct pollfd *new_svc_pollfd;
>  
>        __svc_xports[sock] = xprt;
> @@ -126,17 +128,17 @@ xprt_register (xprt)
>              goto unlock;
>            }
>  
> -      new_svc_pollfd = (struct pollfd *) realloc (svc_pollfd,
> -                                                  sizeof (struct pollfd)
> -                                                  * (svc_max_pollfd + 1));
> +      size = sizeof (struct pollfd) * (svc_max_pollfd + SVC_POLLFD_INCREMENT);
> +      new_svc_pollfd = (struct pollfd *) realloc (svc_pollfd, size);
>        if (new_svc_pollfd == NULL) /* Out of memory */
>          goto unlock;
>        svc_pollfd = new_svc_pollfd;
> -      ++svc_max_pollfd;
> +      svc_max_pollfd += SVC_POLLFD_INCREMENT;
>  
> -      svc_pollfd[svc_max_pollfd - 1].fd = sock;
> -      svc_pollfd[svc_max_pollfd - 1].events = (POLLIN | POLLPRI |
> -                                               POLLRDNORM | POLLRDBAND);
> +      svc_pollfd[i].fd = sock;
> +      svc_pollfd[i].events = (POLLIN | POLLPRI | POLLRDNORM | POLLRDBAND);
> +      for (++i; i < svc_max_pollfd; ++i)
> +        svc_pollfd[i].fd = -1;
>      }
>  unlock:
>    rwlock_unlock (&svc_fd_lock);
> 
Just curious as why batch allocations are need? What problem does it solve?

steved.