* Re: [patch 1/4][resend] fuse-procfs: proxy proc files
[not found] ` <20090904165149.491603361-+lpkiSBCrgqwGcPS5fe+fxcU4LlsiAD9x8DZ0lRKBDr1ENwx4SLHqw@public.gmane.org>
@ 2009-09-11 2:57 ` Serge E. Hallyn
2009-09-14 20:38 ` Krzysztof Taraszka
1 sibling, 0 replies; 5+ messages in thread
From: Serge E. Hallyn @ 2009-09-11 2:57 UTC (permalink / raw)
To: Daniel Lezcano
Cc: containers-cunTk1MwBs9QetFLy7KEm3xJsTq8ys+cHZ5vskTnxNA,
kt-S89nZTSLPHGGdvJs77BJ7Q,
lxc-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Quoting Daniel Lezcano (daniel.lezcano-GANU6spQydw@public.gmane.org):
> This patch makes possible to mount the fuse-procfs on top of /proc
> and display the content of /proc via fuse.
>
> Signed-off-by: Daniel Lezcano <daniel.lezcano-GANU6spQydw@public.gmane.org>
Thanks, Daniel. This is neat. How were you thinking of
configuring this? A config file read at mount?
Actually one question I have is:
> +static int procfs_read(const char *path, char *buf, size_t size,
> + off_t offset, struct fuse_file_info *fi)
> +{
> + int ret;
> + struct procfs_file *pfile = (typeof(pfile))fi->fh;;
> +
> + switch (pfile->type) {
> +
> + case PROCFS_PROXY:
> + ret = read(pfile->file.proxy.fd, buf, size);
Are you supposed to do something with the offset? I don't see
any seek method in fuse_operations, does fuse keep the fp
internally and send it along as offset to read?
> + if (ret < 0)
> + return -errno;
> + break;
> + }
> +
> + return ret;
> +}
-serge
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [patch 1/4][resend] fuse-procfs: proxy proc files
[not found] ` <20090904165149.491603361-+lpkiSBCrgqwGcPS5fe+fxcU4LlsiAD9x8DZ0lRKBDr1ENwx4SLHqw@public.gmane.org>
2009-09-11 2:57 ` [patch 1/4][resend] fuse-procfs: proxy proc files Serge E. Hallyn
@ 2009-09-14 20:38 ` Krzysztof Taraszka
[not found] ` <ac1c4bf20909141338p3e13c1d6sc71899b7859bd963-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
1 sibling, 1 reply; 5+ messages in thread
From: Krzysztof Taraszka @ 2009-09-14 20:38 UTC (permalink / raw)
To: Daniel Lezcano
Cc: containers-cunTk1MwBs9QetFLy7KEm3xJsTq8ys+cHZ5vskTnxNA,
lxc-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Hi Daniel,
right now I do not have any new tasks so I can take a look your patches
and... they won't to apply (patches from 2 to 4).
Can you send me your current procfs.c single file with patches added?
Thanks
--
K
2009/9/4 Daniel Lezcano <daniel.lezcano-GANU6spQydw@public.gmane.org>
> This patch makes possible to mount the fuse-procfs on top of /proc
> and display the content of /proc via fuse.
>
> Signed-off-by: Daniel Lezcano <daniel.lezcano-GANU6spQydw@public.gmane.org>
> ---
> procfs.c | 246
> +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
> 1 file changed, 246 insertions(+)
>
> Index: lxcfs/procfs.c
> ===================================================================
> --- /dev/null
> +++ lxcfs/procfs.c
> @@ -0,0 +1,246 @@
> +
> +#include <fuse.h>
> +
> +#include <stdio.h>
> +#include <stdlib.h>
> +#include <string.h>
> +#include <libgen.h>
> +#include <errno.h>
> +#include <dirent.h>
> +#include <mntent.h>
> +#include <limits.h>
> +#include <fcntl.h>
> +#include <sys/types.h>
> +#include <sys/stat.h>
> +#include <sys/param.h>
> +
> +enum {
> + PROCFS_PROXY,
> +};
> +
> +struct procfs_info {
> + DIR *procdir;
> + DIR *subdir;
> +};
> +
> +struct proxy_file {
> + int fd;
> +};
> +
> +struct procfs_file {
> + int type;
> + union {
> + struct proxy_file proxy;
> + } file;
> +};
> +
> +static int procfs_readlink(const char *path, char *buf, size_t bufsiz)
> +{
> + struct fuse_context *context = fuse_get_context();
> + struct procfs_info *fsinfo = context->private_data;
> + char *bname = strchr(path, '/');
> + int ret;
> +
> + if (strcmp(path, "/"))
> + bname += 1;
> +
> + ret = readlinkat(dirfd(fsinfo->procdir), bname, buf, bufsiz);
> + if (ret < 0)
> + return -errno;
> +
> + buf[ret] = '\0';
> +
> + return 0;
> +}
> +
> +static int procfs_getattr(const char *path, struct stat *stbuf)
> +{
> + struct fuse_context *context = fuse_get_context();
> + struct procfs_info *fsinfo = context->private_data;
> + char *bname = strchr(path, '/');
> +
> + if (strcmp(path, "/"))
> + bname += 1;
> +
> + if (fstatat(dirfd(fsinfo->procdir), bname, stbuf,
> AT_SYMLINK_NOFOLLOW))
> + return -errno;
> +
> + return 0;
> +}
> +
> +static int procfs_open(const char *path, struct fuse_file_info *fi)
> +{
> + struct fuse_context *context = fuse_get_context();
> + struct procfs_info *fsinfo = context->private_data;
> + char *bname = strchr(path, '/');
> + struct procfs_file *pfile;
> +
> + if (strcmp(path, "/"))
> + bname += 1;
> +
> + pfile = malloc(sizeof(*pfile));
> + if (!pfile)
> + return -ENOMEM;
> +
> + pfile->type = PROCFS_PROXY;
> + pfile->file.proxy.fd = openat(dirfd(fsinfo->procdir), bname,
> fi->flags);
> + if (pfile->file.proxy.fd < 0) {
> + free(pfile);
> + return -errno;
> + }
> +
> + fi->fh = (typeof(fi->fh))pfile;
> +
> + return 0;
> +}
> +
> +static int procfs_read(const char *path, char *buf, size_t size,
> + off_t offset, struct fuse_file_info *fi)
> +{
> + int ret;
> + struct procfs_file *pfile = (typeof(pfile))fi->fh;;
> +
> + switch (pfile->type) {
> +
> + case PROCFS_PROXY:
> + ret = read(pfile->file.proxy.fd, buf, size);
> + if (ret < 0)
> + return -errno;
> + break;
> + }
> +
> + return ret;
> +}
> +
> +static int procfs_write(const char *path, const char *buf, size_t size,
> + off_t offset, struct fuse_file_info *fi)
> +{
> + int ret;
> + struct procfs_file *pfile = (typeof(pfile))fi->fh;;
> +
> + switch (pfile->type) {
> +
> + case PROCFS_PROXY:
> + ret = write(pfile->file.proxy.fd, buf, size);
> + if (ret < 0)
> + return -errno;
> + break;
> + }
> +
> + return ret;
> +}
> +
> +static int procfs_release(const char *path, struct fuse_file_info *fi)
> +{
> + struct procfs_file *pfile = (typeof(pfile))fi->fh;;
> +
> + switch (pfile->type) {
> +
> + case PROCFS_PROXY:
> + if (close(pfile->file.proxy.fd))
> + return -errno;
> + break;
> + }
> +
> + return 0;
> +}
> +
> +static int procfs_opendir(const char *path, struct fuse_file_info *fi)
> +{
> + struct fuse_context *context = fuse_get_context();
> + struct procfs_info *fsinfo = context->private_data;
> + char *bname = strchr(path, '/');
> + int fd;
> +
> + if (!strcmp(path, "/")) {
> + fsinfo->subdir = fsinfo->procdir;
> + return 0;
> + }
> +
> + fd = openat(dirfd(fsinfo->procdir), bname + 1, O_DIRECTORY);
> + if (fd < 0)
> + return -errno;
> +
> + fsinfo->subdir = fdopendir(fd);
> +
> + return 0;
> +}
> +
> +static int procfs_readdir(const char *path, void *buf, fuse_fill_dir_t
> filler,
> + off_t offset, struct fuse_file_info *fi)
> +{
> + struct fuse_context *context = fuse_get_context();
> + struct procfs_info *fsinfo = context->private_data;
> + struct dirent dirent, *direntp;
> +
> + if (!offset)
> + rewinddir(fsinfo->subdir);
> +
> + while (!readdir_r(fsinfo->subdir, &dirent, &direntp) && direntp)
> + if (filler(buf, direntp->d_name, NULL, offset++))
> + break;
> +
> + return 0;
> +}
> +
> +static int procfs_releasedir(const char *path, struct fuse_file_info *fi)
> +{
> + struct fuse_context *context = fuse_get_context();
> + struct procfs_info *fsinfo = context->private_data;
> + DIR *subdir = fsinfo->subdir;
> +
> + fsinfo->subdir = NULL;
> + if (subdir && subdir != fsinfo->procdir)
> + closedir(subdir);
> +
> + return 0;
> +}
> +
> +static void *procfs_init(struct fuse_conn_info *conn)
> +{
> + struct procfs_info *fsinfo;
> +
> + fsinfo = malloc(sizeof(*fsinfo));
> + if (!fsinfo)
> + return NULL;
> +
> + fsinfo->procdir = opendir("/proc");
> + if (!fsinfo->procdir)
> + goto out_free_fsinfo;
> +
> + fsinfo->subdir = NULL;
> +out:
> + return fsinfo;
> +
> +out_free_fsinfo:
> + free(fsinfo);
> + fsinfo = NULL;
> + goto out;
> +}
> +
> +static void procfs_destroy(void *private_data)
> +{
> + struct procfs_info *fsinfo = private_data;
> +
> + closedir(fsinfo->procdir);
> + free(fsinfo);
> +}
> +
> +static struct fuse_operations procfs_ops = {
> + .readlink = procfs_readlink,
> + .getattr = procfs_getattr,
> + .open = procfs_open,
> + .read = procfs_read,
> + .write = procfs_write,
> + .opendir = procfs_opendir,
> + .releasedir = procfs_releasedir,
> + .readdir = procfs_readdir,
> + .release = procfs_release,
> + .init = procfs_init,
> + .destroy = procfs_destroy,
> +};
> +
> +int main(int argc, char *argv[])
> +{
> + return fuse_main(argc, argv, &procfs_ops, NULL);
> +}
>
>
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [patch 1/4][resend] fuse-procfs: proxy proc files
[not found] ` <ac1c4bf20909141338p3e13c1d6sc71899b7859bd963-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
@ 2009-09-14 21:13 ` Daniel Lezcano
[not found] ` <4AAEB1FA.3070102-GANU6spQydw@public.gmane.org>
0 siblings, 1 reply; 5+ messages in thread
From: Daniel Lezcano @ 2009-09-14 21:13 UTC (permalink / raw)
To: kt-S89nZTSLPHGGdvJs77BJ7Q
Cc: containers-cunTk1MwBs9QetFLy7KEm3xJsTq8ys+cHZ5vskTnxNA,
lxc-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
[-- Attachment #1: Type: text/plain, Size: 384 bytes --]
Krzysztof Taraszka wrote:
> Hi Daniel,
>
> right now I do not have any new tasks so I can take a look your patches
> and... they won't to apply (patches from 2 to 4).
> Can you send me your current procfs.c single file with patches added?
>
> Thanks
>
Sure, in attachment the procfs.tar.gz which is a directory containing
the patches.
Just untar and quilt push -a.
-- Daniel
[-- Attachment #2: procfs.tar.gz --]
[-- Type: application/x-gzip, Size: 6688 bytes --]
[-- Attachment #3: Type: text/plain, Size: 206 bytes --]
_______________________________________________
Containers mailing list
Containers-cunTk1MwBs9QetFLy7KEm3xJsTq8ys+cHZ5vskTnxNA@public.gmane.org
https://lists.linux-foundation.org/mailman/listinfo/containers
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [patch 1/4][resend] fuse-procfs: proxy proc files
[not found] ` <4AAEB1FA.3070102-GANU6spQydw@public.gmane.org>
@ 2009-09-28 9:33 ` Krzysztof Taraszka
[not found] ` <ac1c4bf20909280233r5cd94343m30ac00c6ef9e1d10-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
0 siblings, 1 reply; 5+ messages in thread
From: Krzysztof Taraszka @ 2009-09-28 9:33 UTC (permalink / raw)
To: Daniel Lezcano
Cc: containers-cunTk1MwBs9QetFLy7KEm3xJsTq8ys+cHZ5vskTnxNA,
lxc-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Thank you Daniel.
I have a little problem with fuse and containers. Do I have to add the fuse
device to the container?
When I tried to mount procfs I received:
container2:/# ./procfs -odirect_io /proc/meminfo
fuse: device not found, try 'modprobe fuse' first
The fuse module was loaded on host. What I did wrong? I am asking because I
am working around user space scripts for managing lxc (written in bash,
init.d script, few lxc-create-* scripts based on distro, etc) and I would
like to add this development future to this scripts.
Right now I have few nice and good looking userspace scripts.
Would be great If I may do the deb and rpm package with lxc user scripts and
lxc-tools or add my scripts to lxc-tools tree.
More about that in this week in the different thread. Hope you will give me
the direction guys.
2009/9/14 Daniel Lezcano <daniel.lezcano-GANU6spQydw@public.gmane.org>
> Krzysztof Taraszka wrote:
>
>> Hi Daniel,
>>
>> right now I do not have any new tasks so I can take a look your patches
>> and... they won't to apply (patches from 2 to 4).
>> Can you send me your current procfs.c single file with patches added?
>>
>> Thanks
>>
>>
> Sure, in attachment the procfs.tar.gz which is a directory containing the
> patches.
> Just untar and quilt push -a.
>
> -- Daniel
>
>
>
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [patch 1/4][resend] fuse-procfs: proxy proc files
[not found] ` <ac1c4bf20909280233r5cd94343m30ac00c6ef9e1d10-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
@ 2009-09-28 15:47 ` Daniel Lezcano
0 siblings, 0 replies; 5+ messages in thread
From: Daniel Lezcano @ 2009-09-28 15:47 UTC (permalink / raw)
To: kt-S89nZTSLPHGGdvJs77BJ7Q
Cc: containers-cunTk1MwBs9QetFLy7KEm3xJsTq8ys+cHZ5vskTnxNA,
lxc-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Krzysztof Taraszka wrote:
> Thank you Daniel.
> I have a little problem with fuse and containers. Do I have to add the fuse
> device to the container?
> When I tried to mount procfs I received:
>
> container2:/# ./procfs -odirect_io /proc/meminfo
> fuse: device not found, try 'modprobe fuse' first
>
> The fuse module was loaded on host. What I did wrong? I am asking because I
> am working around user space scripts for managing lxc (written in bash,
> init.d script, few lxc-create-* scripts based on distro, etc) and I would
> like to add this development future to this scripts.
> Right now I have few nice and good looking userspace scripts.
> Would be great If I may do the deb and rpm package with lxc user scripts and
> lxc-tools or add my scripts to lxc-tools tree.
>
Let's see :)
> More about that in this week in the different thread. Hope you will give me
> the direction guys.
>
I think the command line is not right. Fuse uses at the /proc directory
when initializating, so that have to be done in two steps:
Can you try:
lxc-execute -n foo /bin/bash
(from the shell in the container):
procfs -o direct_io /tmp/<tmpdir>
mount --bind /tmp/<tmpdir> /proc
echo 268435456 > /cgroup/foo/memory.memsw.limit_in_bytes
echo 268435456 > /cgroup/foo/memory.limit_in_bytes
* before exiting the container:
umount /proc
fusermount -u /tmp/<tmpfile>
If you do 'ls /proc', you should see everything expect the /proc/sys
directory.
If you do 'cat /proc/meminfo', you should see:
MemTotal: 262144 kB
MemFree: 250684 kB
SwapTotal: 262144 kB
SwapFree: 0 kB
^ permalink raw reply [flat|nested] 5+ messages in thread
end of thread, other threads:[~2009-09-28 15:47 UTC | newest]
Thread overview: 5+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
[not found] <20090904162556.407556958@mai-009101017029.toulouse-stg.fr.ibm.com>
[not found] ` <20090904165149.491603361@mai-009101017029.toulouse-stg.fr.ibm.com>
[not found] ` <20090904165149.491603361-+lpkiSBCrgqwGcPS5fe+fxcU4LlsiAD9x8DZ0lRKBDr1ENwx4SLHqw@public.gmane.org>
2009-09-11 2:57 ` [patch 1/4][resend] fuse-procfs: proxy proc files Serge E. Hallyn
2009-09-14 20:38 ` Krzysztof Taraszka
[not found] ` <ac1c4bf20909141338p3e13c1d6sc71899b7859bd963-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
2009-09-14 21:13 ` Daniel Lezcano
[not found] ` <4AAEB1FA.3070102-GANU6spQydw@public.gmane.org>
2009-09-28 9:33 ` Krzysztof Taraszka
[not found] ` <ac1c4bf20909280233r5cd94343m30ac00c6ef9e1d10-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
2009-09-28 15:47 ` Daniel Lezcano
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.