Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751726Ab0FCQGY (ORCPT ); Thu, 3 Jun 2010 12:06:24 -0400 Received: from fxip-0047f.externet.hu ([88.209.222.127]:36780 "EHLO pomaz-ex.szeredi.hu" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751250Ab0FCQGX (ORCPT ); Thu, 3 Jun 2010 12:06:23 -0400 To: Jens Axboe CC: miklos@szeredi.hu, mtk.manpages@gmail.com, linux-kernel@vger.kernel.org In-reply-to: <4C0793F6.4070107@fusionio.com> (message from Jens Axboe on Thu, 03 Jun 2010 13:37:26 +0200) Subject: Re: [patch] pipe: add support for shrinking and growing pipes References: <20100524070552.GR23411@kernel.dk> <20100524173551.GU23411@kernel.dk> <20100524175649.GV23411@kernel.dk> <20100601074534.GL1660@kernel.dk> <20100603061039.GD3564@kernel.dk> <20100603070126.GJ3564@kernel.dk> <4C078610.6020901@fusionio.com> <4C078E2B.7090107@fusionio.com> <4C0793F6.4070107@fusionio.com> Message-Id: From: Miklos Szeredi Date: Thu, 03 Jun 2010 18:06:19 +0200 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 4578 Lines: 150 On Thu, 03 Jun 2010, Jens Axboe wrote: > On 2010-06-03 13:32, Miklos Szeredi wrote: > > On Thu, 03 Jun 2010, Jens Axboe wrote: > >> I looked at that when I did the original pipe-max-pages, but it seems > >> we're not consistent in this regard. Anyway, here's what I came up with. > >> Not tested yet, and I hope thunderbird doesn't fsck up the patch. > > > > Unfortunately it did (line wrap damage). > > I suspected it might. This better? > > diff --git a/fs/pipe.c b/fs/pipe.c > index f98fae3..f7f62f3 100644 > --- a/fs/pipe.c > +++ b/fs/pipe.c > @@ -26,9 +26,14 @@ > > /* > * The max size that a non-root user is allowed to grow the pipe. Can > - * be set by root in /proc/sys/fs/pipe-max-pages > + * be set by root in /proc/sys/fs/pipe_max_size > */ > -unsigned int pipe_max_pages = PIPE_DEF_BUFFERS * 16; > +unsigned int pipe_max_size = 1048576; > + > +/* > + * Minimum pipe size, as required by POSIX > + */ > +unsigned int pipe_min_size = PAGE_SIZE; > > /* > * We use a start+len construction, which provides full use of the > @@ -1156,6 +1161,35 @@ static long pipe_set_size(struct pipe_inode_info *pipe, unsigned long nr_pages) > return nr_pages * PAGE_SIZE; > } > > +/* > + * Currently we rely on the pipe array holding a power-of-2 number > + * of pages. > + */ > +static inline unsigned int round_pipe_size(unsigned int size) > +{ > + unsigned long nr_pages; > + > + nr_pages = (size + PAGE_SIZE - 1) >> PAGE_SHIFT; > + return roundup_pow_of_two(nr_pages) << PAGE_SHIFT; > +} > + > +/* > + * This should work even if CONFIG_PROC_FS isn't set, as proc_dointvec_minmax > + * will return an error. > + */ > +int pipe_proc_fn(struct ctl_table *table, int write, void __user *buf, > + size_t *lenp, loff_t *ppos) > +{ > + int ret; > + > + ret = proc_dointvec_minmax(table, write, buf, lenp, ppos); > + if (ret < 0 || !write) > + return ret; > + > + pipe_max_size = round_pipe_size(pipe_max_size); > + return ret; > +} > + > long pipe_fcntl(struct file *file, unsigned int cmd, unsigned long arg) > { > struct pipe_inode_info *pipe; > @@ -1169,23 +1203,19 @@ long pipe_fcntl(struct file *file, unsigned int cmd, unsigned long arg) > > switch (cmd) { > case F_SETPIPE_SZ: { > - unsigned long nr_pages; > + unsigned int size, nr_pages; > > - /* > - * Currently the array must be a power-of-2 size, so adjust > - * upwards if needed. > - */ > - nr_pages = (arg + PAGE_SIZE - 1) >> PAGE_SHIFT; > - nr_pages = roundup_pow_of_two(nr_pages); > + size = round_pipe_size(arg); > + nr_pages = size >> PAGE_SHIFT; > > - if (!capable(CAP_SYS_RESOURCE) && nr_pages > pipe_max_pages) { > + if (!capable(CAP_SYS_RESOURCE) && size > pipe_max_size) { > ret = -EPERM; > goto out; > - } else if (nr_pages < 1) { > + } else if (nr_pages < PAGE_SIZE) { This should either be "nr_pages < 1" or "size < PAGE_SIZE". Miklos > ret = -EINVAL; > goto out; > } > - ret = pipe_set_size(pipe, arg); > + ret = pipe_set_size(pipe, nr_pages); > break; > } > case F_GETPIPE_SZ: > diff --git a/include/linux/pipe_fs_i.h b/include/linux/pipe_fs_i.h > index 16de393..4457969 100644 > --- a/include/linux/pipe_fs_i.h > +++ b/include/linux/pipe_fs_i.h > @@ -139,7 +139,9 @@ void pipe_lock(struct pipe_inode_info *); > void pipe_unlock(struct pipe_inode_info *); > void pipe_double_lock(struct pipe_inode_info *, struct pipe_inode_info *); > > -extern unsigned int pipe_max_pages; > +extern unsigned int pipe_max_size, pipe_min_size; > +int pipe_proc_fn(struct ctl_table *, int, void __user *, size_t *, loff_t *); > + > > /* Drop the inode semaphore and wait for a pipe event, atomically */ > void pipe_wait(struct pipe_inode_info *pipe); > diff --git a/kernel/sysctl.c b/kernel/sysctl.c > index 997080f..2f9b3a6 100644 > --- a/kernel/sysctl.c > +++ b/kernel/sysctl.c > @@ -1471,12 +1471,12 @@ static struct ctl_table fs_table[] = { > }, > #endif > { > - .procname = "pipe-max-pages", > - .data = &pipe_max_pages, > + .procname = "pipe_max_size", > + .data = &pipe_max_size, > .maxlen = sizeof(int), > .mode = 0644, > - .proc_handler = &proc_dointvec_minmax, > - .extra1 = &two, > + .proc_handler = &pipe_proc_fn, > + .extra1 = &pipe_min_size, > }, > /* > * NOTE: do not add new entries to this table unless you have read > > -- > Jens Axboe > > -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/