Skip to contentSkip to conversation

Unkillable process in last git -- 100% reproducible

DISCUSSION Paolo Ornati 3 participants 8 messages Jun 28, 2006 – Jun 29, 2006 1 message not in the archive

Subsystems (2)
lore
Share & tools
lore

A message that is not in the archive (sent off-list or not yet received): 20060628142918.1b2c25c3@localhost

#1 lore raw
Quoted 7 lines · On Wed, 28 Jun 2006 14:29:18 +0200 Paolo Ornati <[email protected]> wrote:
[  430.083347] localedef     R  running task       0  8577   8558  8578               (NOTLB)
[  430.083352] gzip          X ffff81001e612ee0     0  8578   8577                     (L-TLB)
[  430.083358] ffff81001395bef8 ffff81001fd1a310 0000000000000246 ffff81001e612ee0 
[  430.083362]        ffff81001e4c0080 ffff81001e612ee0 ffff81001e4c0258 0000000000000001 
[  430.083366]        0000000000000046 0000000000000046 ffff81001395bf18 0000000000000010 
[  430.083370] Call Trace: <ffffffff80227f6f>{do_exit+2378} <ffffffff802628e9>{vfs_write+288}
[  430.083379]        <ffffffff80228065>{sys_exit_group+0} <ffffffff80209806>{system_call+126}

do_exit() -- kernel/exit.c

0xffffffff80227f66 <do_exit+2369>:      mov    %rax,0x18(%rbp)
0xffffffff80227f6a <do_exit+2373>:      callq  0xffffffff8048b850 <schedule>
0xffffffff80227f6f <do_exit+2378>:      ud2a
0xffffffff80227f71 <do_exit+2380>:      pushq  $0xffffffff804b7821
0xffffffff80227f76 <do_exit+2385>:      retq   $0x3ba
0xffffffff80227f79 <do_exit+2388>:      jmp    0xffffffff80227f79 <do_exit+2388>
Signature · 2 lines
	Paolo Ornati
	Linux 2.6.17-ga39727f2-dirty on x86_64
#2 lore raw
Quoted 8 lines · On Wed, 28 Jun 2006 14:53:49 +0200 Paolo Ornati <[email protected]> wrote:
Quoted 7 lines
[  430.083347] localedef     R  running task       0  8577   8558  8578               (NOTLB)
[  430.083352] gzip          X ffff81001e612ee0     0  8578   8577                     (L-TLB)
[  430.083358] ffff81001395bef8 ffff81001fd1a310 0000000000000246 ffff81001e612ee0 
[  430.083362]        ffff81001e4c0080 ffff81001e612ee0 ffff81001e4c0258 0000000000000001 
[  430.083366]        0000000000000046 0000000000000046 ffff81001395bf18 0000000000000010 
[  430.083370] Call Trace: <ffffffff80227f6f>{do_exit+2378} <ffffffff802628e9>{vfs_write+288}
[  430.083379]        <ffffffff80228065>{sys_exit_group+0} <ffffffff80209806>{system_call+126}
Quoted 8 lines

do_exit() -- kernel/exit.c

0xffffffff80227f66 <do_exit+2369>:      mov    %rax,0x18(%rbp)
0xffffffff80227f6a <do_exit+2373>:      callq  0xffffffff8048b850 <schedule>
0xffffffff80227f6f <do_exit+2378>:      ud2a
0xffffffff80227f71 <do_exit+2380>:      pushq  $0xffffffff804b7821
0xffffffff80227f76 <do_exit+2385>:      retq   $0x3ba
0xffffffff80227f79 <do_exit+2388>:      jmp    0xffffffff80227f79 <do_exit+2388>

that is the end of the function:

...

        schedule();
        BUG();
        /* Avoid "noreturn function does return".  */
        for (;;) ;
}

So the questions are two:

1) why schedule() didn't work?

2) why the process is looping around "ud2a" (placed by BUG()) and presumibly throwing a lot of "invalid opcode" exceptions?

Signature · 2 lines
	Paolo Ornati
	Linux 2.6.17-ga39727f2-dirty on x86_64
#3 lore raw
Quoted 31 lines · On Wed, 28 Jun 2006 15:09:43 +0200 Paolo Ornati <[email protected]> wrote:
Quoted 8 lines · On Wed, 28 Jun 2006 14:53:49 +0200 Paolo Ornati <[email protected]> wrote:
Quoted 7 lines
[  430.083347] localedef     R  running task       0  8577   8558  8578               (NOTLB)
[  430.083352] gzip          X ffff81001e612ee0     0  8578   8577                     (L-TLB)
[  430.083358] ffff81001395bef8 ffff81001fd1a310 0000000000000246 ffff81001e612ee0 
[  430.083362]        ffff81001e4c0080 ffff81001e612ee0 ffff81001e4c0258 0000000000000001 
[  430.083366]        0000000000000046 0000000000000046 ffff81001395bf18 0000000000000010 
[  430.083370] Call Trace: <ffffffff80227f6f>{do_exit+2378} <ffffffff802628e9>{vfs_write+288}
[  430.083379]        <ffffffff80228065>{sys_exit_group+0} <ffffffff80209806>{system_call+126}
Quoted 8 lines

do_exit() -- kernel/exit.c

0xffffffff80227f66 <do_exit+2369>:      mov    %rax,0x18(%rbp)
0xffffffff80227f6a <do_exit+2373>:      callq  0xffffffff8048b850 <schedule>
0xffffffff80227f6f <do_exit+2378>:      ud2a
0xffffffff80227f71 <do_exit+2380>:      pushq  $0xffffffff804b7821
0xffffffff80227f76 <do_exit+2385>:      retq   $0x3ba
0xffffffff80227f79 <do_exit+2388>:      jmp    0xffffffff80227f79 <do_exit+2388>

that is the end of the function:

...

        schedule();
        BUG();
        /* Avoid "noreturn function does return".  */
        for (;;) ;
}
Quoted 6 lines

So the questions are two:

1) why schedule() didn't work?

2) why the process is looping around "ud2a" (placed by BUG()) and presumibly throwing a lot of "invalid opcode" exceptions?

It seems that the first mail didn't hit LKML (maybe it was too big), here it is (stripped):

------------------------------------------------------------------------
Running "localedef" triggers an infinite loop in kernel mode (or something) --> localdef becomes unkillable.

This is dmesg after pressing ALT+SysRQ+T three times:

[CUT]

KERNEL:

--

	Paolo Ornati
	Linux 2.6.17-ga39727f2-dirty on x86_64
	                      ^^^^^^^

The -dirty is because of a fix to PS2 Keyboard auto-repeat, nothing else.
------------------------------------------------------------------------

Full config attached.

#4 lore raw

Re: Unkillable process in last git -- Bisected

Quoted 48 lines · On Wed, 28 Jun 2006 15:19:55 +0200 Paolo Ornati <[email protected]> wrote:
Quoted 28 lines
Quoted 8 lines
Quoted 7 lines
[  430.083347] localedef     R  running task       0  8577   8558  8578               (NOTLB)
[  430.083352] gzip          X ffff81001e612ee0     0  8578   8577                     (L-TLB)
[  430.083358] ffff81001395bef8 ffff81001fd1a310 0000000000000246 ffff81001e612ee0 
[  430.083362]        ffff81001e4c0080 ffff81001e612ee0 ffff81001e4c0258 0000000000000001 
[  430.083366]        0000000000000046 0000000000000046 ffff81001395bf18 0000000000000010 
[  430.083370] Call Trace: <ffffffff80227f6f>{do_exit+2378} <ffffffff802628e9>{vfs_write+288}
[  430.083379]        <ffffffff80228065>{sys_exit_group+0} <ffffffff80209806>{system_call+126}
Quoted 8 lines

do_exit() -- kernel/exit.c

0xffffffff80227f66 <do_exit+2369>:      mov    %rax,0x18(%rbp)
0xffffffff80227f6a <do_exit+2373>:      callq  0xffffffff8048b850 <schedule>
0xffffffff80227f6f <do_exit+2378>:      ud2a
0xffffffff80227f71 <do_exit+2380>:      pushq  $0xffffffff804b7821
0xffffffff80227f76 <do_exit+2385>:      retq   $0x3ba
0xffffffff80227f79 <do_exit+2388>:      jmp    0xffffffff80227f79 <do_exit+2388>

that is the end of the function:

...

        schedule();
        BUG();
        /* Avoid "noreturn function does return".  */
        for (;;) ;
}
Quoted 6 lines

So the questions are two:

1) why schedule() didn't work?

2) why the process is looping around "ud2a" (placed by BUG()) and presumibly throwing a lot of "invalid opcode" exceptions?

It seems that the first mail didn't hit LKML (maybe it was too big), here it is (stripped):

------------------------------------------------------------------------
Running "localedef" triggers an infinite loop in kernel mode (or something) --> localdef becomes unkillable.

This is dmesg after pressing ALT+SysRQ+T three times:

[CUT]

KERNEL:

Quoted 7 lines

--

	Paolo Ornati
	Linux 2.6.17-ga39727f2-dirty on x86_64
	                      ^^^^^^^

The -dirty is because of a fix to PS2 Keyboard auto-repeat, nothing else.
------------------------------------------------------------------------

Hi everybody,

as you can see from the subject the problem is an unkillable process (localedef) that eats 100% CPU (the system is usable however).

After a git-bisect session I've found that reverting the following commit fixes the problem.

Any idea?

commit 6527c2bdf1f833cc18e8f42bd97973d583e4aa83

Author: Vladimir V. Saveliev <[email protected]>
Date:   Tue Jun 27 02:53:57 2006 -0700

[PATCH] generic_file_buffered_write(): deadlock on vectored write

generic_file_buffered_write() prefaults in user pages in order to avoid deadlock on copying from the same page as write goes to.

However, it looks like there is a problem when write is vectored:
fault_in_pages_readable brings in current segment or its part (maxlen). OTOH, filemap_copy_from_user_iovec is called to copy number of bytes (bytes) which may exceed current segment, so filemap_copy_from_user_iovec switches to the next segment which is not brought in yet. Pagefault is generated. That causes the deadlock if pagefault is for the same page write goes to: page being written is locked and not uptodate, pagefault will deadlock trying to lock locked page.

[[email protected]: somewhat rewritten]
Cc Neil Brown <[email protected]>
Cc Martin Schwidefsky <[email protected]>
mm/filemap.c +11 -7
index 9c7334b..d504d6e 100644
@@ -2095,14 +2095,21 @@ generic_file_buffered_write(struct kiocb
 	do {
 		unsigned long index;
 		unsigned long offset;
-		unsigned long maxlen;
 		size_t copied;
 
 		offset = (pos & (PAGE_CACHE_SIZE -1)); /* Within page */
 		index = pos >> PAGE_CACHE_SHIFT;
 		bytes = PAGE_CACHE_SIZE - offset;
-		if (bytes > count)
-			bytes = count;
+
+		/* Limit the size of the copy to the caller's write size */
+		bytes = min(bytes, count);
+
+		/*
+		 * Limit the size of the copy to that of the current segment,
+		 * because fault_in_pages_readable() doesn't know how to walk
+		 * segments.
+		 */
+		bytes = min(bytes, cur_iov->iov_len - iov_base);
 
 		/*
 		 * Bring in the user page that we will copy from _first_.
@@ -2110,10 +2117,7 @@ generic_file_buffered_write(struct kiocb
 		 * same page as we're writing to, without it being marked
 		 * up-to-date.
 		 */
-		maxlen = cur_iov->iov_len - iov_base;
-		if (maxlen > bytes)
-			maxlen = bytes;
-		fault_in_pages_readable(buf, maxlen);
+		fault_in_pages_readable(buf, bytes);
 
 		page = __grab_cache_page(mapping,index,&cached_page,&lru_pvec);
 		if (!page) {
Signature · 2 lines
	Paolo Ornati
	Linux 2.6.17.1 on x86_64
#5 lore raw

Re: Unkillable process in last git -- Bisected

Hello

Quoted 3 lines · Paolo Ornati
Quoted 3 lines

Running "localedef" triggers an infinite loop in kernel mode (or something) --> localdef becomes unkillable.

Can you, please, run localedef under strace -e set=open,write on a kernel having the below patch applied, so that we will see arguments of write which caused write to fall into endless loop.

Quoted 69 lines

commit 6527c2bdf1f833cc18e8f42bd97973d583e4aa83

Author: Vladimir V. Saveliev <[email protected]>
Date:   Tue Jun 27 02:53:57 2006 -0700

[PATCH] generic_file_buffered_write(): deadlock on vectored write

generic_file_buffered_write() prefaults in user pages in order to avoid deadlock on copying from the same page as write goes to.

However, it looks like there is a problem when write is vectored:
fault_in_pages_readable brings in current segment or its part (maxlen). OTOH, filemap_copy_from_user_iovec is called to copy number of bytes (bytes) which may exceed current segment, so filemap_copy_from_user_iovec switches to the next segment which is not brought in yet. Pagefault is generated. That causes the deadlock if pagefault is for the same page write goes to: page being written is locked and not uptodate, pagefault will deadlock trying to lock locked page.

[[email protected]: somewhat rewritten]
Cc Neil Brown <[email protected]>
Cc Martin Schwidefsky <[email protected]>
mm/filemap.c +11 -7
index 9c7334b..d504d6e 100644
@@ -2095,14 +2095,21 @@ generic_file_buffered_write(struct kiocb
 	do {
 		unsigned long index;
 		unsigned long offset;
-		unsigned long maxlen;
 		size_t copied;
 
 		offset = (pos & (PAGE_CACHE_SIZE -1)); /* Within page */
 		index = pos >> PAGE_CACHE_SHIFT;
 		bytes = PAGE_CACHE_SIZE - offset;
-		if (bytes > count)
-			bytes = count;
+
+		/* Limit the size of the copy to the caller's write size */
+		bytes = min(bytes, count);
+
+		/*
+		 * Limit the size of the copy to that of the current segment,
+		 * because fault_in_pages_readable() doesn't know how to walk
+		 * segments.
+		 */
+		bytes = min(bytes, cur_iov->iov_len - iov_base);
 
 		/*
 		 * Bring in the user page that we will copy from _first_.
@@ -2110,10 +2117,7 @@ generic_file_buffered_write(struct kiocb
 		 * same page as we're writing to, without it being marked
 		 * up-to-date.
 		 */
-		maxlen = cur_iov->iov_len - iov_base;
-		if (maxlen > bytes)
-			maxlen = bytes;
-		fault_in_pages_readable(buf, maxlen);
+		fault_in_pages_readable(buf, bytes);
 
 		page = __grab_cache_page(mapping,index,&cached_page,&lru_pvec);
 		if (!page) {
#6 lore raw

Re: Unkillable process in last git -- Bisected

Quoted 8 lines · On Wed, 28 Jun 2006 23:06:17 +0400 "Vladimir V. Saveliev" <[email protected]>…
Quoted 3 lines · Paolo Ornati
Quoted 3 lines

Running "localedef" triggers an infinite loop in kernel mode (or something) --> localdef becomes unkillable.

Can you, please, run localedef under strace -e set=open,write on a kernel having the below patch applied, so that we will see arguments of write which caused write to fall into endless loop.

The full strace is short enough, I think:

$ strace localedef --no-archive -c -i en_US -f ISO-8859-1 -A /usr/share/locale/locale.alias en_US

execve("/usr/bin/localedef", ["localedef", "--no-archive", "-c", "-i", "en_US", "-f", "ISO-8859-1", "-A", "/usr/share/locale/locale.alias", "en_US"], [/* 44 vars */]) = 0
uname({sys="Linux", node="tux", ...}) = 0
brk(0) = 0x549000
mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x2b0f18098000
access("/etc/ld.so.preload", R_OK) = -1 ENOENT (No such file or directory)
open("/etc/ld.so.cache", O_RDONLY) = 3
fstat(3, {st_mode=S_IFREG|0644, st_size=163137, ...}) = 0
mmap(NULL, 163137, PROT_READ, MAP_PRIVATE, 3, 0) = 0x2b0f18099000
close(3) = 0
open("/lib/tls/libc.so.6", O_RDONLY) = 3
read(3, "\177ELF\2\1\1\0\0\0\0\0\0\0\0\0\3\0>\0\1\0\0\0\360\306"..., 640) = 640
lseek(3, 64, SEEK_SET) = 64
read(3, "\6\0\0\0\5\0\0\0@\0\0\0\0\0\0\0@\0\0\0\0\0\0\0@\0\0\0\0"..., 616) = 616
lseek(3, 680, SEEK_SET) = 680
read(3, "\4\0\0\0\20\0\0\0\1\0\0\0GNU\0\0\0\0\0\2\0\0\0\6\0\0\0"..., 32) = 32
fstat(3, {st_mode=S_IFREG|0755, st_size=1268224, ...}) = 0
lseek(3, 64, SEEK_SET) = 64
read(3, "\6\0\0\0\5\0\0\0@\0\0\0\0\0\0\0@\0\0\0\0\0\0\0@\0\0\0\0"..., 616) = 616
mmap(NULL, 2261000, PROT_READ|PROT_EXEC, MAP_PRIVATE|MAP_DENYWRITE, 3, 0) = 0x2b0f1819a000
madvise(0x2b0f1819a000, 2261000, MADV_SEQUENTIAL|0x1) = 0
mprotect(0x2b0f182b9000, 1085448, PROT_NONE) = 0

mmap(0x2b0f183b8000, 24576, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_FIXED|MAP_DENYWRITE, 3, 0x11e000) = 0x2b0f183b8000
mmap(0x2b0f183be000, 16392, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_FIXED|MAP_ANONYMOUS, -1, 0) = 0x2b0f183be000
close(3)                                = 0
mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x2b0f183c3000
mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x2b0f183c4000

mprotect(0x2b0f183b8000, 12288, PROT_READ) = 0
mprotect(0x2b0f18198000, 4096, PROT_READ) = 0
arch_prctl(ARCH_SET_FS, 0x2b0f183c3ae0) = 0
munmap(0x2b0f18099000, 163137) = 0
open("/dev/urandom", O_RDONLY) = 3
read(3, "\363\300\254\335\34A\365\327", 8) = 8

close(3)                                = 0
brk(0)                                  = 0x549000
brk(0x56a000)                           = 0x56a000

open("/usr/lib64/locale/locale-archive", O_RDONLY) = -1 ENOENT (No such file or directory) open("/usr/share/locale/locale.alias", O_RDONLY) = 3
fstat(3, {st_mode=S_IFREG|0644, st_size=2586, ...}) = 0
mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x2b0f18099000 read(3, "# Locale name alias data base.\n#"..., 4096) = 2586

read(3, "", 4096)                       = 0
close(3)                                = 0
munmap(0x2b0f18099000, 4096)            = 0

open("/usr/lib64/locale/en_US.utf8/LC_MESSAGES", O_RDONLY) = 3
fstat(3, {st_mode=S_IFDIR|0755, st_size=4096, ...}) = 0
close(3) = 0
open("/usr/lib64/locale/en_US.utf8/LC_MESSAGES/SYS_LC_MESSAGES", O_RDONLY) = 3
fstat(3, {st_mode=S_IFREG|0644, st_size=52, ...}) = 0
mmap(NULL, 52, PROT_READ, MAP_PRIVATE, 3, 0) = 0x2b0f18099000
close(3) = 0
open("/usr/lib64/gconv/gconv-modules.cache", O_RDONLY) = 3
fstat(3, {st_mode=S_IFREG|0644, st_size=21546, ...}) = 0
mmap(NULL, 21546, PROT_READ, MAP_SHARED, 3, 0) = 0x2b0f1809a000
close(3) = 0
open("/usr/lib64/locale/en_US.utf8/LC_CTYPE", O_RDONLY) = 3
fstat(3, {st_mode=S_IFREG|0644, st_size=208464, ...}) = 0
mmap(NULL, 208464, PROT_READ, MAP_PRIVATE, 3, 0) = 0x2b0f180a0000
close(3) = 0
access("/usr/lib64/locale/en_US", W_OK) = 0
open("ISO-8859-1", O_RDONLY) = -1 ENOENT (No such file or directory)
open("/usr/share/i18n/charmaps/ISO-8859-1", O_RDONLY) = -1 ENOENT (No such file or directory) open("/usr/share/i18n/charmaps/ISO-8859-1.gz", O_RDONLY) = 3
fstat(3, {st_mode=S_IFREG|0644, st_size=3150, ...}) = 0
pipe([4, 5]) = 0
getrlimit(RLIMIT_NOFILE, {rlim_cur=1024, rlim_max=1024}) = 0
getrlimit(RLIMIT_NOFILE, {rlim_cur=1024, rlim_max=1024}) = 0
getrlimit(RLIMIT_NOFILE, {rlim_cur=1024, rlim_max=1024}) = 0
getrlimit(RLIMIT_NOFILE, {rlim_cur=1024, rlim_max=1024}) = 0
getrlimit(RLIMIT_NOFILE, {rlim_cur=1024, rlim_max=1024}) = 0
clone(child_stack=0, flags=CLONE_CHILD_CLEARTID|CLONE_CHILD_SETTID|SIGCHLD, child_tidptr=0x2b0f183c3b70) = 4258

close(5)                                = 0
close(3)                                = 0
fcntl(4, F_GETFL)                       = 0 (flags O_RDONLY)

fstat(4, {st_mode=S_IFIFO|0600, st_size=0, ...}) = 0
mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x2b0f180d3000
lseek(4, 0, SEEK_CUR) = -1 ESPIPE (Illegal seek)
read(4, "<code_set_name> ISO-8859-1\n<comm"..., 4096) = 4096
read(4, " LATIN CAPITAL LETTER Z\n<U00"..., 4096) = 4096
read(4, " SUPERSCRIPT THREE\n<U00B4"..., 4096) = 4096
read(4, "L LETTER U WITH CIRCUMFLEX\n<U00F"..., 4096) = 337

read(4, "", 4096)                       = 0
close(4)                                = 0
munmap(0x2b0f180d3000, 4096)            = 0
open("en_US", O_RDONLY)                 = -1 ENOENT (No such file or directory)

open("/usr/share/i18n/locales/en_US", O_RDONLY) = 3
--- SIGCHLD (Child exited) @ 0 (0) ---
fstat(3, {st_mode=S_IFREG|0644, st_size=5480, ...}) = 0
mmap(NULL, 5480, PROT_READ, MAP_SHARED, 3, 0) = 0x2b0f180d3000
lseek(3, 5480, SEEK_SET) = 5480
open("i18n", O_RDONLY) = -1 ENOENT (No such file or directory)
open("/usr/share/i18n/locales/i18n", O_RDONLY) = 4
fstat(4, {st_mode=S_IFREG|0644, st_size=95199, ...}) = 0
mmap(NULL, 95199, PROT_READ, MAP_SHARED, 4, 0) = 0x2b0f180d5000
lseek(4, 95199, SEEK_SET) = 95199
brk(0x58b000) = 0x58b000
mmap(NULL, 262144, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x2b0f180ed000 mremap(0x2b0f180ed000, 262144, 524288, MREMAP_MAYMOVE) = 0x2b0f180ed000
brk(0x5ac000) = 0x5ac000
brk(0x5d4000) = 0x5d4000

mmap(NULL, 135168, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x2b0f1816d000
mmap(NULL, 135168, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x2b0f183c5000
mmap(NULL, 135168, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x2b0f183e6000

brk(0x5c7000) = 0x5c7000
mremap(0x2b0f1816d000, 135168, 266240, MREMAP_MAYMOVE) = 0x2b0f18407000 mremap(0x2b0f183c5000, 135168, 266240, MREMAP_MAYMOVE) = 0x2b0f18448000 mremap(0x2b0f183e6000, 135168, 266240, MREMAP_MAYMOVE) = 0x2b0f18489000 mremap(0x2b0f180ed000, 524288, 1048576, MREMAP_MAYMOVE) = 0x2b0f184ca000 mremap(0x2b0f18407000, 266240, 528384, MREMAP_MAYMOVE) = 0x2b0f180ed000 mremap(0x2b0f18448000, 266240, 528384, MREMAP_MAYMOVE) = 0x2b0f183c5000 mmap(NULL, 528384, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x2b0f185ca000
brk(0x5a7000) = 0x5a7000
mremap(0x2b0f18489000, 266240, 528384, MREMAP_MAYMOVE) = 0x2b0f1864b000 mremap(0x2b0f184ca000, 1048576, 3674112, MREMAP_MAYMOVE) = 0x2b0f186cc000 mremap(0x2b0f186cc000, 3674112, 7344128, MREMAP_MAYMOVE) = 0x2b0f186cc000 mremap(0x2b0f180ed000, 528384, 1052672, MREMAP_MAYMOVE) = 0x2b0f18446000 mremap(0x2b0f183c5000, 528384, 1052672, MREMAP_MAYMOVE) = 0x2b0f18dcd000 mremap(0x2b0f185ca000, 528384, 1052672, MREMAP_MAYMOVE) = 0x2b0f18ece000 mremap(0x2b0f1864b000, 528384, 1052672, MREMAP_MAYMOVE) = 0x2b0f18547000

brk(0x5c8000)                           = 0x5c8000
brk(0x5e9000)                           = 0x5e9000
brk(0x60a000)                           = 0x60a000
brk(0x62b000)                           = 0x62b000
brk(0x64c000)                           = 0x64c000
brk(0x66d000)                           = 0x66d000
fstat(4, {st_mode=S_IFREG|0644, st_size=95199, ...}) = 0
munmap(0x2b0f180d5000, 95199)           = 0
close(4)                                = 0
open("iso14651_t1", O_RDONLY)           = -1 ENOENT (No such file or directory)
open("/usr/share/i18n/locales/iso14651_t1", O_RDONLY) = 4
fstat(4, {st_mode=S_IFREG|0644, st_size=47626, ...}) = 0
mmap(NULL, 47626, PROT_READ, MAP_SHARED, 4, 0) = 0x2b0f180d5000
lseek(4, 47626, SEEK_SET)               = 47626
brk(0x68e000)                           = 0x68e000
brk(0x68d000)                           = 0x68d000
brk(0x6ae000)                           = 0x6ae000
brk(0x6cf000)                           = 0x6cf000
brk(0x6f0000)                           = 0x6f0000
brk(0x711000)                           = 0x711000
brk(0x73d000)                           = 0x73d000

fstat(4, {st_mode=S_IFREG|0644, st_size=47626, ...}) = 0
munmap(0x2b0f180d5000, 47626) = 0
close(4) = 0
fstat(3, {st_mode=S_IFREG|0644, st_size=5480, ...}) = 0

munmap(0x2b0f180d3000, 5480)            = 0
close(3)                                = 0
open("translit_neutral", O_RDONLY)      = -1 ENOENT (No such file or directory)

open("/usr/share/i18n/locales/translit_neutral", O_RDONLY) = 3 fstat(3, {st_mode=S_IFREG|0644, st_size=7383, ...}) = 0 mmap(NULL, 7383, PROT_READ, MAP_SHARED, 3, 0) = 0x2b0f180d3000
lseek(3, 7383, SEEK_SET) = 7383
brk(0x75e000) = 0x75e000
fstat(3, {st_mode=S_IFREG|0644, st_size=7383, ...}) = 0

munmap(0x2b0f180d3000, 7383)            = 0
close(3)                                = 0
open("translit_circle", O_RDONLY)       = -1 ENOENT (No such file or directory)

open("/usr/share/i18n/locales/translit_circle", O_RDONLY) = 3 fstat(3, {st_mode=S_IFREG|0644, st_size=14116, ...}) = 0 mmap(NULL, 14116, PROT_READ, MAP_SHARED, 3, 0) = 0x2b0f180d3000
lseek(3, 14116, SEEK_SET) = 14116
brk(0x77f000) = 0x77f000
fstat(3, {st_mode=S_IFREG|0644, st_size=14116, ...}) = 0

munmap(0x2b0f180d3000, 14116)           = 0
close(3)                                = 0
open("translit_cjk_compat", O_RDONLY)   = -1 ENOENT (No such file or directory)
open("/usr/share/i18n/locales/translit_cjk_compat", O_RDONLY) = 3
fstat(3, {st_mode=S_IFREG|0644, st_size=59104, ...}) = 0
mmap(NULL, 59104, PROT_READ, MAP_SHARED, 3, 0) = 0x2b0f180d3000
lseek(3, 59104, SEEK_SET)               = 59104
brk(0x7a0000)                           = 0x7a0000
brk(0x7c1000)                           = 0x7c1000
brk(0x7e2000)                           = 0x7e2000
brk(0x803000)                           = 0x803000
fstat(3, {st_mode=S_IFREG|0644, st_size=59104, ...}) = 0
munmap(0x2b0f180d3000, 59104)           = 0
close(3)                                = 0
open("translit_compat", O_RDONLY)       = -1 ENOENT (No such file or directory)

open("/usr/share/i18n/locales/translit_compat", O_RDONLY) = 3 fstat(3, {st_mode=S_IFREG|0644, st_size=34361, ...}) = 0 mmap(NULL, 34361, PROT_READ, MAP_SHARED, 3, 0) = 0x2b0f180d3000
lseek(3, 34361, SEEK_SET) = 34361
brk(0x824000) = 0x824000
fstat(3, {st_mode=S_IFREG|0644, st_size=34361, ...}) = 0

munmap(0x2b0f180d3000, 34361)           = 0
close(3)                                = 0
open("translit_font", O_RDONLY)         = -1 ENOENT (No such file or directory)
open("/usr/share/i18n/locales/translit_font", O_RDONLY) = 3
fstat(3, {st_mode=S_IFREG|0644, st_size=58986, ...}) = 0
mmap(NULL, 58986, PROT_READ, MAP_SHARED, 3, 0) = 0x2b0f180d3000
lseek(3, 58986, SEEK_SET)               = 58986
brk(0x845000)                           = 0x845000
brk(0x866000)                           = 0x866000
brk(0x887000)                           = 0x887000
brk(0x8a8000)                           = 0x8a8000
fstat(3, {st_mode=S_IFREG|0644, st_size=58986, ...}) = 0
munmap(0x2b0f180d3000, 58986)           = 0
close(3)                                = 0
open("translit_fraction", O_RDONLY)     = -1 ENOENT (No such file or directory)

open("/usr/share/i18n/locales/translit_fraction", O_RDONLY) = 3 fstat(3, {st_mode=S_IFREG|0644, st_size=2380, ...}) = 0 mmap(NULL, 2380, PROT_READ, MAP_SHARED, 3, 0) = 0x2b0f180d3000
lseek(3, 2380, SEEK_SET) = 2380
fstat(3, {st_mode=S_IFREG|0644, st_size=2380, ...}) = 0

munmap(0x2b0f180d3000, 2380)            = 0
close(3)                                = 0
open("translit_narrow", O_RDONLY)       = -1 ENOENT (No such file or directory)

open("/usr/share/i18n/locales/translit_narrow", O_RDONLY) = 3 fstat(3, {st_mode=S_IFREG|0644, st_size=6210, ...}) = 0 mmap(NULL, 6210, PROT_READ, MAP_SHARED, 3, 0) = 0x2b0f180d3000
lseek(3, 6210, SEEK_SET) = 6210
brk(0x8c9000) = 0x8c9000
fstat(3, {st_mode=S_IFREG|0644, st_size=6210, ...}) = 0

munmap(0x2b0f180d3000, 6210)            = 0
close(3)                                = 0
open("translit_small", O_RDONLY)        = -1 ENOENT (No such file or directory)

open("/usr/share/i18n/locales/translit_small", O_RDONLY) = 3 fstat(3, {st_mode=S_IFREG|0644, st_size=3088, ...}) = 0 mmap(NULL, 3088, PROT_READ, MAP_SHARED, 3, 0) = 0x2b0f180d3000
lseek(3, 3088, SEEK_SET) = 3088
fstat(3, {st_mode=S_IFREG|0644, st_size=3088, ...}) = 0

munmap(0x2b0f180d3000, 3088)            = 0
close(3)                                = 0
open("translit_wide", O_RDONLY)         = -1 ENOENT (No such file or directory)

open("/usr/share/i18n/locales/translit_wide", O_RDONLY) = 3 fstat(3, {st_mode=S_IFREG|0644, st_size=5030, ...}) = 0 mmap(NULL, 5030, PROT_READ, MAP_SHARED, 3, 0) = 0x2b0f180d3000
lseek(3, 5030, SEEK_SET) = 5030
fstat(3, {st_mode=S_IFREG|0644, st_size=5030, ...}) = 0

munmap(0x2b0f180d3000, 5030)            = 0
close(3)                                = 0
brk(0x8ea000)                           = 0x8ea000
brk(0x910000)                           = 0x910000

unlink("/usr/lib64/locale/en_US/LC_CTYPE") = 0
creat("/usr/lib64/locale/en_US/LC_CTYPE", 0666) = 3
writev(3, [{"\25\21\3 X\0\0\0", 8}, {"h\1\0\0h\4\0\0h\n\0\0h\n\0\0h\20\0\0h\20\0\0h\24\0\0h\24"..., 352}, {"\2\0\2\0\2\0\2\0\2\0\2\0\2\0\2\0\2\0\2\0\2\0\2\0\2\0\2"..., 768}, {"\200\0\0\0\201\0\0\0\202\0\0\0\203\0\0\0\204\0\0\0\205"..., 1536}, {NULL, 0}, {"\200\0\0\0\201\0\0\0\202\0\0\0\203\0\0\0\204\0\0\0\205"..., 1536}, {NULL, 0}, {"\0\0\2\0\0\0\2\0\0\0\2\0\0\0\2\0\0\0\2\0\0\0\2\0\0\0\2"..., 1024}, {NULL, 0}, {NULL, 0}, {NULL, 0}, {NULL, 0}, {"upper\0", 6}, {"lower\0", 6}, {"alpha\0", 6}, {"digit\0", 6}, {"xdigit\0", 7}, {"space\0", 6}, {"print\0", 6}, {"graph\0", 6}, {"blank\0", 6}, {"cntrl\0", 6}, {"punct\0", 6}, {"alnum\0", 6}, {"combining\0", 10}, {"combining_level3\0", 17}, {"\0\0\0\0", 4}, {"toupper\0", 8}, {"tolower\0", 8}, {"totitle\0", 8}, {"\0\0\0\0", 4}, {"\20\0\0\0\1\0\0\0\7\0\0\0\377\1\0\0\177\0\0\0\30\0\0\0"..., 2328}, ...], 125

Signature · 2 lines
	Paolo Ornati
	Linux 2.6.17.1 on x86_64
#7 lore raw

Re: Unkillable process in last git -- Bisected

In reply to Paolo Ornati
Quoted 11 lines · On Wed, 28 Jun 2006 20:38:25 +0200 Paolo Ornati <[email protected]> wrote:

After a git-bisect session I've found that reverting the following commit fixes the problem.

Any idea?

commit 6527c2bdf1f833cc18e8f42bd97973d583e4aa83

Author: Vladimir V. Saveliev <[email protected]>
Date:   Tue Jun 27 02:53:57 2006 -0700

[PATCH] generic_file_buffered_write(): deadlock on vectored write

From: Andrew Morton <[email protected]>

The recent generic_file_write() deadlock fix caused generic_file_buffered_write() to loop inifinitely when presented with a zero-length iovec segment. Fix.

Note that this fix deliberately avoids calling ->prepare_write(), ->commit_write() etc with a zero-length write. This is because I don't trust all filesystems to get that right.

This is a cautious approach, for 2.6.17.x. For 2.6.18 we should just go ahead and call ->prepare_write() and ->commit_write() with the zero length and fix any broken filesystems. So I'll make that change once this code is stabilised and backported into 2.6.17.x.

The reason for preferring to call ->prepare_write() and ->commit_write() with the zero-length segment: a zero-length segment _should_ be sufficiently uncommon that this is the correct way of handling it. We don't want to optimise for poorly-written userspace at the expense of well-written userspace.

Cc "Vladimir V. Saveliev" <[email protected]>
Cc Neil Brown <[email protected]>
Cc Martin Schwidefsky <[email protected]>
Cc Chris Wright <[email protected]>
Cc Greg KH <[email protected]>

---

 mm/filemap.c |    9 ++++++++-
 mm/filemap.h |    4 ++--
 2 files changed, 10 insertions(+), 3 deletions(-)
mm/filemap.c renamed from mm/filemap.c~generic_file_buffered_write-handle-zero-length-iovec-segments-stable +8 -1
diff -puN mm/filemap.c~generic_file_buffered_write-handle-zero-length-iovec-segments-stable mm/filemap.c
@@ -2125,6 +2125,12 @@ generic_file_buffered_write(struct kiocb
 			break;
 		}
 
+		if (unlikely(bytes == 0)) {
+			status = 0;
+			copied = 0;
+			goto zero_length_segment;
+		}
+
 		status = a_ops->prepare_write(file, page, offset, offset+bytes);
 		if (unlikely(status)) {
 			loff_t isize = i_size_read(inode);
@@ -2154,7 +2160,8 @@ generic_file_buffered_write(struct kiocb
 			page_cache_release(page);
 			continue;
 		}
-		if (likely(copied > 0)) {
+zero_length_segment:
+		if (likely(copied >= 0)) {
 			if (!status)
 				status = copied;
 
mm/filemap.h renamed from mm/filemap.h~generic_file_buffered_write-handle-zero-length-iovec-segments-stable +2 -2
diff -puN mm/filemap.h~generic_file_buffered_write-handle-zero-length-iovec-segments-stable mm/filemap.h
@@ -88,7 +88,7 @@ filemap_set_next_iovec(const struct iove
 	const struct iovec *iov = *iovp;
 	size_t base = *basep;
 
-	while (bytes) {
+	do {
 		int copy = min(bytes, iov->iov_len - base);
 
 		bytes -= copy;
@@ -97,7 +97,7 @@ filemap_set_next_iovec(const struct iove
 			iov++;
 			base = 0;
 		}
-	}
+	} while (bytes);
 	*iovp = iov;
 	*basep = base;
 }

_

#8 lore raw

Re: Unkillable process in last git -- Bisected

Quoted 5 lines · On Wed, 28 Jun 2006 22:20:36 -0700 Andrew Morton <[email protected]> wrote:

From: Andrew Morton <[email protected]>

The recent generic_file_write() deadlock fix caused generic_file_buffered_write() to loop inifinitely when presented with a zero-length iovec segment. Fix.

It works for me :)

Signature · 2 lines
	Paolo Ornati
	Linux 2.6.17.1 on x86_64