lkml.org 
[lkml]   [2010]   [May]   [1]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
From
Subject[PATCH v21 079/100] c/r: checkpoint/restart epoll sets
Date
From: Matt Helsley <matthltc@us.ibm.com>

Save/restore epoll items during checkpoint/restart respectively.

Output the epoll header and items separately. Chunk the output much
like the pid array gets chunked. This ensures that even sub-order 0
allocations will enable checkpoint of large epoll sets. A subsequent
patch will do something similar for the restore path.

On restart, we grab a piece of memory suitable to store a "chunk" of
items for input. Read the input one chunk at a time and add epoll
items for each item in the chunk.

Changelog [v21]:
- Put file_ops->checkpoint under CONFIG_CHECKPOINT
Changelog [v19]:
- [Oren Laadan] Fix broken compilation for no-c/r architectures
Changelog [v19-rc1]:
- [Oren Laadan] Return -EBUSY (not BUG_ON) if fd is gone on restart
- [Oren Laadan] Fix the chunk size instead of auto-tune

Changelog v5:
Fix potential recursion during collect.
Replace call to ckpt_obj_collect() with ckpt_collect_file().
[Oren]
Fix checkpoint leak detection when there are more items than
expected.
Cleanup/simplify error write paths. (will complicate in a later
patch) [Oren]
Remove files_deferq bits. [Oren]
Remove extra newline. [Oren]
Remove aggregate check on number of watches added. [Oren]
This is OK since these will be done individually anyway.
Remove check for negative objrefs during restart. [Oren]
Fixup comment regarding race that indicates checkpoint leaks.
[Oren]
s/ckpt_read_obj/ckpt_read_buf_type/ [Oren]
Patch for lots of epoll items follows.
Moved sys_close(epfd) right under fget(). [Oren]
Use CKPT_HDR_BUFFER rather than custome ckpt_read/write_*
This makes it more similar to the pid array code. [Oren]
It also simplifies the error recovery paths.
Tested polling a pipe and 50,000 UNIX sockets.

Changelog v4: ckpt-v18
Use files_deferq as submitted by Dan Smith
Cleanup to only report >= 1 items when debugging.

Changelog v3: [unposted]
Removed most of the TODOs -- the remainder will be removed by
subsequent patches.
Fixed missing ep_file_collect() [Serge]
Rather than include checkpoint_hdr.h declare (but do not define)
the two structs needed in eventpoll.h [Oren]
Complain with ckpt_write_err() when we detect checkpoint obj
leaks. [Oren]
Remove redundant is_epoll_file() check in collect. [Oren]
Move epfile_objref lookup to simplify error handling. [Oren]
Simplify error handling with early return in
ep_eventpoll_checkpoint(). [Oren]
Cleaned up a comment. [Oren]
Shorten CKPT_HDR_FILE_EPOLL_ITEMS (-FILE) [Oren]
Renumbered to indicate that it follows the file table.
Renamed the epoll struct in checkpoint_hdr.h [Oren]
Also renamed substruct.
Fixup return of empty ep_file_restore(). [Oren]
Changed some error returns. [Oren]
Changed some tests to BUG_ON(). [Oren]
Factored out watch insert with epoll_ctl() into do_epoll_ctl().
[Cedric, Oren]

Cc: Davide Libenzi <davidel@xmailserver.org>
Cc: linux-fsdevel@vger.kernel.org
Signed-off-by: Matt Helsley <matthltc@us.ibm.com>
Acked-by: Oren Laadan <orenl@cs.columbia.edu>
Acked-by: Serge Hallyn <serue@us.ibm.com>
---
fs/checkpoint.c | 7 +
fs/eventpoll.c | 334 ++++++++++++++++++++++++++++++++++++----
include/linux/checkpoint_hdr.h | 18 ++
include/linux/eventpoll.h | 17 ++-
4 files changed, 347 insertions(+), 29 deletions(-)
diff --git a/fs/checkpoint.c b/fs/checkpoint.c
index 9b29a26..3bfa692 100644
--- a/fs/checkpoint.c
+++ b/fs/checkpoint.c
@@ -21,6 +21,7 @@
#include <linux/syscalls.h>
#include <linux/deferqueue.h>
#include <linux/checkpoint.h>
+#include <linux/eventpoll.h>
#include <net/sock.h>

/**************************************************************************
@@ -632,6 +633,12 @@ static struct restore_file_ops restore_file_ops[] = {
.file_type = CKPT_FILE_TTY,
.restore = tty_file_restore,
},
+ /* epoll */
+ {
+ .file_name = "EPOLL",
+ .file_type = CKPT_FILE_EPOLL,
+ .restore = ep_file_restore,
+ },
};

static void *restore_file(struct ckpt_ctx *ctx)
diff --git a/fs/eventpoll.c b/fs/eventpoll.c
index bd056a5..99920d2 100644
--- a/fs/eventpoll.c
+++ b/fs/eventpoll.c
@@ -39,6 +39,9 @@
#include <asm/mman.h>
#include <asm/atomic.h>

+#include <linux/checkpoint.h>
+#include <linux/deferqueue.h>
+
/*
* LOCKING:
* There are three level of locking required by epoll :
@@ -671,10 +674,19 @@ static unsigned int ep_eventpoll_poll(struct file *file, poll_table *wait)
return pollflags != -1 ? pollflags : 0;
}

+#ifdef CONFIG_CHECKPOINT
+static int ep_eventpoll_checkpoint(struct ckpt_ctx *ctx, struct file *file);
+static int ep_file_collect(struct ckpt_ctx *ctx, struct file *file);
+#endif
+
/* File callbacks that implement the eventpoll file behaviour */
static const struct file_operations eventpoll_fops = {
.release = ep_eventpoll_release,
- .poll = ep_eventpoll_poll
+ .poll = ep_eventpoll_poll,
+#ifdef CONFIG_CHECKPOINT
+ .checkpoint = ep_eventpoll_checkpoint,
+ .collect = ep_file_collect,
+#endif
};

/* Fast test to see if the file is an evenpoll file */
@@ -1226,35 +1238,18 @@ SYSCALL_DEFINE1(epoll_create, int, size)
* the eventpoll file that enables the insertion/removal/change of
* file descriptors inside the interest set.
*/
-SYSCALL_DEFINE4(epoll_ctl, int, epfd, int, op, int, fd,
- struct epoll_event __user *, event)
+int do_epoll_ctl(int op, int fd,
+ struct file *file, struct file *tfile,
+ struct epoll_event *epds)
{
int error;
- struct file *file, *tfile;
struct eventpoll *ep;
struct epitem *epi;
- struct epoll_event epds;
-
- error = -EFAULT;
- if (ep_op_has_event(op) &&
- copy_from_user(&epds, event, sizeof(struct epoll_event)))
- goto error_return;
-
- /* Get the "struct file *" for the eventpoll file */
- error = -EBADF;
- file = fget(epfd);
- if (!file)
- goto error_return;
-
- /* Get the "struct file *" for the target file */
- tfile = fget(fd);
- if (!tfile)
- goto error_fput;

/* The target file descriptor must support poll */
error = -EPERM;
if (!tfile->f_op || !tfile->f_op->poll)
- goto error_tgt_fput;
+ return error;

/*
* We have to check that the file structure underneath the file descriptor
@@ -1263,7 +1258,7 @@ SYSCALL_DEFINE4(epoll_ctl, int, epfd, int, op, int, fd,
*/
error = -EINVAL;
if (file == tfile || !is_file_epoll(file))
- goto error_tgt_fput;
+ return error;

/*
* At this point it is safe to assume that the "private_data" contains
@@ -1284,8 +1279,8 @@ SYSCALL_DEFINE4(epoll_ctl, int, epfd, int, op, int, fd,
switch (op) {
case EPOLL_CTL_ADD:
if (!epi) {
- epds.events |= POLLERR | POLLHUP;
- error = ep_insert(ep, &epds, tfile, fd);
+ epds->events |= POLLERR | POLLHUP;
+ error = ep_insert(ep, epds, tfile, fd);
} else
error = -EEXIST;
break;
@@ -1297,15 +1292,46 @@ SYSCALL_DEFINE4(epoll_ctl, int, epfd, int, op, int, fd,
break;
case EPOLL_CTL_MOD:
if (epi) {
- epds.events |= POLLERR | POLLHUP;
- error = ep_modify(ep, epi, &epds);
+ epds->events |= POLLERR | POLLHUP;
+ error = ep_modify(ep, epi, epds);
} else
error = -ENOENT;
break;
}
mutex_unlock(&ep->mtx);

-error_tgt_fput:
+ return error;
+}
+
+/*
+ * The following function implements the controller interface for
+ * the eventpoll file that enables the insertion/removal/change of
+ * file descriptors inside the interest set.
+ */
+SYSCALL_DEFINE4(epoll_ctl, int, epfd, int, op, int, fd,
+ struct epoll_event __user *, event)
+{
+ int error;
+ struct file *file, *tfile;
+ struct epoll_event epds;
+
+ error = -EFAULT;
+ if (ep_op_has_event(op) &&
+ copy_from_user(&epds, event, sizeof(struct epoll_event)))
+ goto error_return;
+
+ /* Get the "struct file *" for the eventpoll file */
+ error = -EBADF;
+ file = fget(epfd);
+ if (!file)
+ goto error_return;
+
+ /* Get the "struct file *" for the target file */
+ tfile = fget(fd);
+ if (!tfile)
+ goto error_fput;
+
+ error = do_epoll_ctl(op, fd, file, tfile, &epds);
fput(tfile);
error_fput:
fput(file);
@@ -1413,6 +1439,258 @@ SYSCALL_DEFINE6(epoll_pwait, int, epfd, struct epoll_event __user *, events,

#endif /* HAVE_SET_RESTORE_SIGMASK */

+#ifdef CONFIG_CHECKPOINT
+static int ep_file_collect(struct ckpt_ctx *ctx, struct file *file)
+{
+ struct rb_node *rbp;
+ struct eventpoll *ep;
+ int ret = 0;
+
+ ep = file->private_data;
+ mutex_lock(&ep->mtx);
+ for (rbp = rb_first(&ep->rbr); rbp; rbp = rb_next(rbp)) {
+ struct epitem *epi;
+
+ epi = rb_entry(rbp, struct epitem, rbn);
+ if (is_file_epoll(epi->ffd.file))
+ continue; /* Don't recurse */
+ ret = ckpt_collect_file(ctx, epi->ffd.file);
+ if (ret < 0)
+ break;
+ }
+ mutex_unlock(&ep->mtx);
+ return ret;
+}
+
+struct epoll_deferq_entry {
+ struct ckpt_ctx *ctx;
+ struct file *epfile;
+};
+
+#define CKPT_EPOLL_CHUNK (8096 / (int) sizeof(struct ckpt_eventpoll_item))
+
+static int ep_items_checkpoint(void *data)
+{
+ struct epoll_deferq_entry *dq_entry = data;
+ struct ckpt_ctx *ctx;
+ struct ckpt_hdr_eventpoll_items *h;
+ struct ckpt_eventpoll_item *items;
+ struct rb_node *rbp;
+ struct eventpoll *ep;
+ __s32 epfile_objref;
+ int num_items = 0, ret;
+
+ ctx = dq_entry->ctx;
+
+ epfile_objref = ckpt_obj_lookup(ctx, dq_entry->epfile, CKPT_OBJ_FILE);
+ BUG_ON(epfile_objref <= 0);
+
+ ep = dq_entry->epfile->private_data;
+ mutex_lock(&ep->mtx);
+ for (rbp = rb_first(&ep->rbr); rbp; rbp = rb_next(rbp))
+ num_items++;
+ mutex_unlock(&ep->mtx);
+
+ h = ckpt_hdr_get_type(ctx, sizeof(*h), CKPT_HDR_EPOLL_ITEMS);
+ if (!h)
+ return -ENOMEM;
+ h->num_items = num_items;
+ h->epfile_objref = epfile_objref;
+ ret = ckpt_write_obj(ctx, &h->h);
+ ckpt_hdr_put(ctx, h);
+ if (ret || !num_items)
+ return ret;
+
+ ret = ckpt_write_obj_type(ctx, NULL, sizeof(*items)*num_items,
+ CKPT_HDR_BUFFER);
+ if (ret < 0)
+ return ret;
+
+ items = kzalloc(sizeof(*items) * CKPT_EPOLL_CHUNK, GFP_KERNEL);
+ if (!items)
+ return -ENOMEM;
+
+ /*
+ * Walk the rbtree copying items into the chunk of memory and then
+ * writing them to the checkpoint image
+ */
+ ret = 0;
+ mutex_lock(&ep->mtx);
+ rbp = rb_first(&ep->rbr);
+ while ((num_items > 0) && rbp) {
+ int n = min(num_items, CKPT_EPOLL_CHUNK);
+ int j;
+
+ for (j = 0; rbp && j < n; j++, rbp = rb_next(rbp)) {
+ struct epitem *epi;
+ int objref;
+
+ epi = rb_entry(rbp, struct epitem, rbn);
+ items[j].fd = epi->ffd.fd;
+ items[j].events = epi->event.events;
+ items[j].data = epi->event.data;
+ objref = ckpt_obj_lookup(ctx, epi->ffd.file,
+ CKPT_OBJ_FILE);
+ if (objref <= 0)
+ goto unlock;
+ items[j].file_objref = objref;
+ }
+ ret = ckpt_kwrite(ctx, items, n*sizeof(*items));
+ if (ret < 0)
+ break;
+ num_items -= n;
+ }
+unlock:
+ mutex_unlock(&ep->mtx);
+ kfree(items);
+ if (num_items != 0 || (num_items == 0 && rbp))
+ ret = -EBUSY; /* extra item(s) -- checkpoint obj leak */
+ if (ret)
+ ckpt_err(ctx, ret, "Checkpointing epoll items.\n");
+ return ret;
+}
+
+static int ep_eventpoll_checkpoint(struct ckpt_ctx *ctx, struct file *file)
+{
+ struct ckpt_hdr_file *h;
+ struct epoll_deferq_entry dq_entry;
+ int ret = -ENOMEM;
+
+ h = ckpt_hdr_get_type(ctx, sizeof(*h), CKPT_HDR_FILE);
+ if (!h)
+ return -ENOMEM;
+ h->f_type = CKPT_FILE_EPOLL;
+ ret = checkpoint_file_common(ctx, file, h);
+ if (ret < 0)
+ goto out;
+ ret = ckpt_write_obj(ctx, &h->h);
+ if (ret < 0)
+ goto out;
+
+ /*
+ * Defer saving the epoll items until all of the ffd.file pointers
+ * have an objref; after the file table has been checkpointed.
+ */
+ dq_entry.ctx = ctx;
+ dq_entry.epfile = file;
+ ret = deferqueue_add(ctx->files_deferq, &dq_entry,
+ sizeof(dq_entry), ep_items_checkpoint, NULL);
+out:
+ ckpt_hdr_put(ctx, h);
+ return ret;
+}
+
+static int ep_items_restore(void *data)
+{
+ struct ckpt_ctx *ctx = deferqueue_data_ptr(data);
+ struct ckpt_hdr_eventpoll_items *h;
+ struct ckpt_eventpoll_item *items = NULL;
+ struct eventpoll *ep;
+ struct file *epfile = NULL;
+ int ret, num_items;
+
+ h = ckpt_read_obj_type(ctx, sizeof(*h), CKPT_HDR_EPOLL_ITEMS);
+ if (IS_ERR(h))
+ return PTR_ERR(h);
+ num_items = h->num_items;
+ epfile = ckpt_obj_fetch(ctx, h->epfile_objref, CKPT_OBJ_FILE);
+ ckpt_hdr_put(ctx, h);
+
+ /* Make sure userspace didn't give us a ref to a non-epoll file. */
+ if (IS_ERR(epfile))
+ return PTR_ERR(epfile);
+ if (!is_file_epoll(epfile))
+ return -EINVAL;
+ if (!num_items)
+ return 0;
+
+ ret = _ckpt_read_obj_type(ctx, NULL, 0, CKPT_HDR_BUFFER);
+ if (ret < 0)
+ return ret;
+ /* Make sure the items match the size we expect */
+ if (num_items != (ret / sizeof(*items)))
+ return -EINVAL;
+
+ items = kzalloc(sizeof(*items) * CKPT_EPOLL_CHUNK, GFP_KERNEL);
+ if (!items)
+ return -ENOMEM;
+
+ ep = epfile->private_data;
+
+ while (num_items > 0) {
+ int n = min(num_items, CKPT_EPOLL_CHUNK);
+ int j;
+
+ ret = ckpt_kread(ctx, items, n*sizeof(*items));
+ if (ret < 0)
+ break;
+
+ /* Restore the epoll items/watches */
+ for (j = 0; !ret && j < n; j++) {
+ struct epoll_event epev;
+ struct file *tfile;
+
+ tfile = ckpt_obj_fetch(ctx, items[j].file_objref,
+ CKPT_OBJ_FILE);
+ if (IS_ERR(tfile)) {
+ ret = PTR_ERR(tfile);
+ goto out;
+ }
+ epev.events = items[j].events;
+ epev.data = items[j].data;
+ ret = do_epoll_ctl(EPOLL_CTL_ADD, items[j].fd,
+ epfile, tfile, &epev);
+ }
+ num_items -= n;
+ }
+out:
+ kfree(items);
+ return ret;
+}
+
+struct file *ep_file_restore(struct ckpt_ctx *ctx,
+ struct ckpt_hdr_file *h)
+{
+ struct file *epfile;
+ int epfd, ret;
+
+ if (h->h.type != CKPT_HDR_FILE ||
+ h->h.len != sizeof(*h) ||
+ h->f_type != CKPT_FILE_EPOLL)
+ return ERR_PTR(-EINVAL);
+
+ epfd = sys_epoll_create1(h->f_flags & EPOLL_CLOEXEC);
+ if (epfd < 0)
+ return ERR_PTR(epfd);
+ epfile = fget(epfd);
+ sys_close(epfd); /* harmless even if an error occured */
+ if (!epfile) /* can happen with a malicious user */
+ return ERR_PTR(-EBUSY);
+
+ /*
+ * Needed before we can properly restore the watches and enforce the
+ * limit on watch numbers.
+ */
+ ret = restore_file_common(ctx, epfile, h);
+ if (ret < 0)
+ goto fput_out;
+
+ /*
+ * Defer restoring the epoll items until the file table is
+ * fully restored. Ensures that valid file objrefs will resolve.
+ */
+ ret = deferqueue_add_ptr(ctx->files_deferq, ctx,
+ ep_items_restore, NULL);
+ if (ret < 0) {
+fput_out:
+ fput(epfile);
+ epfile = ERR_PTR(ret);
+ }
+ return epfile;
+}
+
+#endif /* CONFIG_CHECKPOINT */
+
static int __init eventpoll_init(void)
{
struct sysinfo si;
diff --git a/include/linux/checkpoint_hdr.h b/include/linux/checkpoint_hdr.h
index 79e8e2d..21540d7 100644
--- a/include/linux/checkpoint_hdr.h
+++ b/include/linux/checkpoint_hdr.h
@@ -127,6 +127,8 @@ enum {
#define CKPT_HDR_TTY CKPT_HDR_TTY
CKPT_HDR_TTY_LDISC,
#define CKPT_HDR_TTY_LDISC CKPT_HDR_TTY_LDISC
+ CKPT_HDR_EPOLL_ITEMS, /* must be after file-table */
+#define CKPT_HDR_EPOLL_ITEMS CKPT_HDR_EPOLL_ITEMS

CKPT_HDR_MM = 401,
#define CKPT_HDR_MM CKPT_HDR_MM
@@ -485,6 +487,8 @@ enum file_type {
#define CKPT_FILE_SOCKET CKPT_FILE_SOCKET
CKPT_FILE_TTY,
#define CKPT_FILE_TTY CKPT_FILE_TTY
+ CKPT_FILE_EPOLL,
+#define CKPT_FILE_EPOLL CKPT_FILE_EPOLL
CKPT_FILE_MAX
#define CKPT_FILE_MAX CKPT_FILE_MAX
};
@@ -701,6 +705,20 @@ struct ckpt_hdr_file_socket {
__s32 sock_objref;
} __attribute__((aligned(8)));

+struct ckpt_hdr_eventpoll_items {
+ struct ckpt_hdr h;
+ __s32 epfile_objref;
+ __u32 num_items;
+} __attribute__((aligned(8)));
+
+/* Contained in a CKPT_HDR_BUFFER following the ckpt_hdr_eventpoll_items */
+struct ckpt_eventpoll_item {
+ __u64 data;
+ __u32 fd;
+ __s32 file_objref;
+ __u32 events;
+} __attribute__((aligned(8)));
+
/* memory layout */
struct ckpt_hdr_mm {
struct ckpt_hdr h;
diff --git a/include/linux/eventpoll.h b/include/linux/eventpoll.h
index f6856a5..52282ae 100644
--- a/include/linux/eventpoll.h
+++ b/include/linux/eventpoll.h
@@ -56,6 +56,9 @@ struct file;


#ifdef CONFIG_EPOLL
+struct ckpt_ctx;
+struct ckpt_hdr_file;
+

/* Used to initialize the epoll bits inside the "struct file" */
static inline void eventpoll_init_file(struct file *file)
@@ -95,11 +98,23 @@ static inline void eventpoll_release(struct file *file)
eventpoll_release_file(file);
}

-#else

+#ifdef CONFIG_CHECKPOINT
+extern struct file *ep_file_restore(struct ckpt_ctx *ctx,
+ struct ckpt_hdr_file *h);
+#endif
+#else
+/* !defined(CONFIG_EPOLL) */
static inline void eventpoll_init_file(struct file *file) {}
static inline void eventpoll_release(struct file *file) {}

+#ifdef CONFIG_CHECKPOINT
+static inline struct file *ep_file_restore(struct ckpt_ctx *ctx,
+ struct ckpt_hdr_file *ptr)
+{
+ return ERR_PTR(-ENOSYS);
+}
+#endif
#endif

#endif /* #ifdef __KERNEL__ */
--
1.6.3.3


\
 
 \ /
  Last update: 2010-05-01 16:41    [from the cache]
©2003-2011 Jasper Spaans