ckpt-v19-rc2
Subsystems (2)
I put together ckpt-v19-rc2 (kernel and user) that includes:
* eclone
* INET sockets
* timerfd, signalfd
* rebase to 2.6.32-rc8
* c/r thread/cpu state earlier
* misc fixes
Hoping to give it some testing to clean remaining rough edges, and in particular net-c/r patches whose folding wasn't trivial, as well as the other archs.
On the todo-list are:
* assorted cleanup/fixes
* more cleanups on ckpt_err/ckpt_msg
* support for x86-64
* unlinked files (if ready ?)
Now back to work...
Oren.
A message that is not in the archive (sent off-list or not yet received): Pine.LNX.4.64.0912012352420.5665-CXF6herHY6ykSYb+qCZC/1i27PF6R63G9nwVQlTi/[email protected]
PATCH 1/1 checkpoint: remove clone_with_pids remnants from arch/powerpc
We have eclone now so this stuff can go (and it doesn't build anyway).
--- arch/powerpc/include/asm/systbl.h | 1 - arch/powerpc/kernel/process.c | 18 ------------------ 2 files changed, 0 insertions(+), 19 deletions(-)
arch/powerpc/include/asm/systbl.h +0 -1
@@ -329,4 +329,3 @@ COMPAT_SYS(rt_tgsigqueueinfo) PPC_SYS(eclone) SYSCALL(checkpoint) SYSCALL(restart) -PPC_SYS(clone_with_pids)
arch/powerpc/kernel/process.c +0 -18
@@ -900,24 +900,6 @@ int sys_clone(unsigned long clone_flags, unsigned long usp,return do_fork (clone_flags, usp, regs,0 , parent_tidp, child_tidp); } -int sys_clone_with_pids (unsigned long clone_flags,unsigned long usp, -int __user *parent_tidp,void __user *child_threadptr, -int __user *child_tidp,void __user *upid_setp, -struct pt_regs *regs) -{ -CHECK_FULL_REGS (regs); -if (usp ==0 ) - usp = regs->gpr[1 ]; -#ifdef CONFIG_PPC64 -if (test_thread_flag (TIF_32BIT )) { - parent_tidp =TRUNC_PTR (parent_tidp); - child_tidp =TRUNC_PTR (child_tidp); - } -#endif -return do_fork_with_pids (clone_flags, usp, regs,0 , - parent_tidp, child_tidp, upid_setp); -} - int sys_eclone(unsigned long clone_flags_low, struct clone_args __user *uclone_args, size_t size,
Signature · 1 line
1.6.0.6
PATCH 1/1 checkpoint: fix net/checkpoint.c for 64-bit
In reply to a message not in the archiveWhen NET_SKBUFF_DATA_USES_OFFSET is defined:
net/checkpoint.c:100: error: invalid operands to binary - (have 'sk_buff_data_t' and 'unsigned char *')
net/checkpoint.c:106: error: 'struct sk_buff' has no member named 'transport_hdr'
--- net/checkpoint.c | 4 ++-- 1 files changed, 2 insertions(+), 2 deletions(-)
net/checkpoint.c +2 -2
@@ -97,13 +97,13 @@ static void sock_record_header_info(struct sk_buff *skb, h->mac_len = skb->mac_len; h->skb_len = skb->len; h->hdr_len = skb->data - skb->head; - h->lin_len = (skb->tail - skb->head); + h->lin_len = ((unsigned long )skb->tail - (unsigned long )skb->head); h->frg_len = skb->data_len; h->data_offset = (skb->data - skb->head);#ifdef NET_SKBUFF_DATA_USES_OFFSET - h->transport_header = skb->transport_hdr; + h->transport_header = skb->transport_header; h->network_header = skb->network_header; h->mac_header = skb->mac_header;#else
Signature · 1 line
1.6.0.6
Quoted 1 line · Oren Laadan
I put together ckpt-v19-rc2 (kernel and user)
I'm not sure yet whether this is a regression, but checkpoint seems to be unable to handle file mappings that extend past the end of the file. I noticed this with Fedora 11 userspace on powerpc (ld.so sometimes maps libraries this way). I did not see this failure with a v19-rc1-based kernel I tested earlier this week, but I haven't retested with that yet.
Here's an example - 8K mapping of a 4K file:
# stat -c%s /tmp/myfile 4096
# grep myfile /proc/5164/maps f7e2b000-f7e2d000 r--p 00000000 08:06 5103709 /tmp/myfile
# checkpoint 5164 > /tmp/mmap.ckpt checkpoint: Bad address
When we try to follow the mapping past the end of the file, we get VM_FAULT_SIGBUS from handle_mm_fault(); the stack trace from debugging code I added is:
.__get_dirty_page+0x4c/0x164 (unreliable) .checkpoint_memory_contents+0x134/0x5a4 .private_vma_checkpoint+0xf4/0x120 .filemap_checkpoint+0x198/0x1d0 .checkpoint_mm+0x3c4/0x4fc .checkpoint_obj+0x17c/0x1d0 .checkpoint_obj_mm+0x50/0x88 .checkpoint_task+0x710/0xaa0 .do_checkpoint+0x9c0/0xb24 .SyS_checkpoint+0xd0/0x11c
Attached is a testcase.
#include <errno.h>
#include <fcntl.h>
#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <sys/mman.h>
#include <sys/stat.h>
#include <sys/types.h>
static size_t page_sz; /* system page size */
static size_t file_sz; /* size of the file */
static size_t map_sz; /* size of the mapping to establish */
static const char *filepath = "myfile";
static void setup(void)
{
page_sz = sysconf(_SC_PAGESIZE);
file_sz = page_sz;
map_sz = file_sz + page_sz;
}static void bail_perror(const char *msg)
{
perror(msg);
exit(1);
}static void bail_msg(const char *msg)
{
fprintf(stderr, "%s\n", msg);
exit(1);
}int create_file(size_t sz)
{
const void *buf;
ssize_t written;
struct stat sb;
int fd;
int rc;
rc = unlink(filepath);
if (rc < 0 && errno != ENOENT)
bail_perror("unlink");
fd = open(filepath, (O_RDWR | O_CREAT | O_EXCL), S_IRUSR);
if (fd < 0)
bail_perror("creat");
buf = calloc(1, sz);
if (!buf)
bail_perror("calloc");
written = 0;
while (written < sz) {
ssize_t res;
/* unsigned char zero = '\0'; */
res = write(fd, buf + written, sz - written);
if (res < 0)
bail_perror("write");
written += res;
}
/* check size */
rc = fstat(fd, &sb);
if (rc < 0)
bail_perror("fstat");
if (sb.st_size != sz)
bail_msg("unexpected file size");
return fd;
}void map_file(int fd, size_t map_sz)
{
void *addr;
addr = mmap(NULL, map_sz, PROT_READ, MAP_PRIVATE, fd, 0);
if (addr == MAP_FAILED)
bail_perror("mmap");
}int main(int argc, char **argv)
{
int fd;
setup();
fd = create_file(file_sz);
map_file(fd, map_sz);
printf("ready\n");
pause();
return 0;
}_______________________________________________
Containers mailing list Containers-cunTk1MwBs9QetFLy7KEm3xJsTq8ys+cHZ5vskTnxNA@public.gmane.org
https://lists.linux-foundation.org/mailman/listinfo/containers
A message that is not in the archive (sent off-list or not yet received): 1259875869.8693.30.camel-bi+AKbBUZKY6gyzm1THtWbp2dZbC/[email protected]
Oh ... maybe that's what why checkpoint fails on my experimental x86-64 port: it worked well on 2.6.31 but fails with "bad address" on 2.6.32.
It may be related to a change between 2.6.31 and 2.6.32 in the arguments to follow_page(), see commit:
mm: FOLL_DUMP replace FOLL_ANON 8e4b9a60718970bbc02dfd3abd0b956ab65af231
?
Oren.
Quoted 38 lines · Nathan Lynch
Quoted 1 line · Oren Laadan
I put together ckpt-v19-rc2 (kernel and user)
I'm not sure yet whether this is a regression, but checkpoint seems to be unable to handle file mappings that extend past the end of the file. I noticed this with Fedora 11 userspace on powerpc (ld.so sometimes maps libraries this way). I did not see this failure with a v19-rc1-based kernel I tested earlier this week, but I haven't retested with that yet.
Here's an example - 8K mapping of a 4K file:
# stat -c%s /tmp/myfile 4096# grep myfile /proc/5164/maps f7e2b000-f7e2d000 r--p 00000000 08:06 5103709 /tmp/myfile# checkpoint 5164 > /tmp/mmap.ckpt checkpoint: Bad addressWhen we try to follow the mapping past the end of the file, we get
VM_FAULT_SIGBUSfromhandle_mm_fault(); the stack trace from debugging code I added is:.__get_dirty_page+0x4c/0x164 (unreliable) .checkpoint_memory_contents+0x134/0x5a4 .private_vma_checkpoint+0xf4/0x120 .filemap_checkpoint+0x198/0x1d0 .checkpoint_mm+0x3c4/0x4fc .checkpoint_obj+0x17c/0x1d0 .checkpoint_obj_mm+0x50/0x88 .checkpoint_task+0x710/0xaa0 .do_checkpoint+0x9c0/0xb24 .SyS_checkpoint+0xd0/0x11cAttached is a testcase.
A message that is not in the archive (sent off-list or not yet received): [email protected]
Thinking about it further, here are two possible scenarios:
A) Task maps a file beyond it's limit, never touches those extra page (if it did, it would get EFAULT)
B) Task maps a file and writes the last page, then the file gets truncated (by at least a page). The task may continue to access that extra page since it already became anonymous.
In kernel 2.3.31, FOLL_ANON flags would make follow_page() return the zero-page for case A. For case B, the actual page is returned.
In kernel 2.3.32, FOLL_ANON is gone. Instead FOLL_DUMP makes follow_page() return -EFAULT in case A, which we can use to conclude that the page is not interesting (unmodified). Case B is handled the same as before.
So checkpoint should work.
Now restart: case A is easy, because mmap() works as before, and those pages that were never touched will not be restored either, they will remain untouched.
For case B it may be problematic, since we will start with a map that goes beyond the current file size (originally it wasn't), and we will attempt to modify a page in the "forbidden" zone. I wonder if get_user_page() will respond favorably when restart
requests such a page ? let's see ...
Oren.
Quoted 57 lines · Oren Laadan
Oh ... maybe that's what why checkpoint fails on my experimental x86-64 port: it worked well on 2.6.31 but fails with "bad address" on 2.6.32.
It may be related to a change between 2.6.31 and 2.6.32 in the arguments to
follow_page(), see commit:mm: FOLL_DUMP replace FOLL_ANON 8e4b9a60718970bbc02dfd3abd0b956ab65af231?
Oren.
Quoted 34 lines · Nathan Lynch
Quoted 1 line · Oren Laadan
I put together ckpt-v19-rc2 (kernel and user)
I'm not sure yet whether this is a regression, but checkpoint seems to be unable to handle file mappings that extend past the end of the file. I noticed this with Fedora 11 userspace on powerpc (ld.so sometimes maps libraries this way). I did not see this failure with a v19-rc1-based kernel I tested earlier this week, but I haven't retested with that yet.
Here's an example - 8K mapping of a 4K file:
# stat -c%s /tmp/myfile 4096# grep myfile /proc/5164/maps f7e2b000-f7e2d000 r--p 00000000 08:06 5103709 /tmp/myfile# checkpoint 5164 > /tmp/mmap.ckpt checkpoint: Bad addressWhen we try to follow the mapping past the end of the file, we get
VM_FAULT_SIGBUSfromhandle_mm_fault(); the stack trace from debugging code I added is:.__get_dirty_page+0x4c/0x164 (unreliable) .checkpoint_memory_contents+0x134/0x5a4 .private_vma_checkpoint+0xf4/0x120 .filemap_checkpoint+0x198/0x1d0 .checkpoint_mm+0x3c4/0x4fc .checkpoint_obj+0x17c/0x1d0 .checkpoint_obj_mm+0x50/0x88 .checkpoint_task+0x710/0xaa0 .do_checkpoint+0x9c0/0xb24 .SyS_checkpoint+0xd0/0x11cQuoted 3 lines
Attached is a testcase.
_______________________________________________
Containers mailing list Containers-cunTk1MwBs9QetFLy7KEm3xJsTq8ys+cHZ5vskTnxNA@public.gmane.org
https://lists.linux-foundation.org/mailman/listinfo/containers
Quoted 14 lines · Oren Laadan
I put together ckpt-v19-rc2 (kernel and user) that includes:
* eclone
* INET sockets
* timerfd, signalfd
* rebase to 2.6.32-rc8
* c/r thread/cpu state earlier
* misc fixesHoping to give it some testing to clean remaining rough edges, and in particular net-c/r patches whose folding wasn't trivial, as well as the other archs.
Quoted 6 lines
On the todo-list are:
* assorted cleanup/fixes
* more cleanups on ckpt_err/ckpt_msg
* support for x86-64
* unlinked files (if ready ?)
Ok, I only just realized, through a segfault on restart, that current->fs->pwd isn't being restored. Were you thinking of supporting that in v19?
If we don't want to do 'proper' current->fs c/r, I htink it's worthwhile doing a cheap c/r of current->fs->pwd as a filename.
(I'd send the cheap patch, but am mostly out from now until wed)
thanks,
-serge