OOPS on bringing down ppp

Hi all, consistently getting the following when pppd is terminated. Happens in 2.4.0-test11, fine in 2.4.0-test9, don't know about test10. Same happens for pppd 2.4.0b4 and 2.4.0, both recompiled for test11. Is this related to the modutils incompatability (modutils 2.3.19) ?

CONFIG_PPP and CONFIG_PPP_ASYNC are built in, CONFIG_PPP_DEFLATE and CONFIG_PPP_BSDCOMP as modules, but oopses whether they are loaded or not.

ksymoops 2.3.4 on i686 2.4.0-test11. Options used
-V (default)
-k /proc/ksyms (default)
-l /proc/modules (default)
-o /lib/modules/2.4.0-test11/ (default) -m /usr/src/linux/System.map (specified)

Warning (compare_maps): snd symbol pm_register not found in /lib/modules/2.4.0-test11/sound/snd.o. Ignoring /lib/modules/2.4.0-test11/sound/snd.o entry
Warning (compare_maps): snd symbol pm_send not found in /lib/modules/2.4.0-test11/sound/snd.o. Ignoring /lib/modules/2.4.0-test11/sound/snd.o entry
Warning (compare_maps): snd symbol pm_unregister not found in /lib/modules/2.4.0-test11/sound/snd.o. Ignoring /lib/modules/2.4.0-test11/sound/snd.o entry c0114c6c
*pde = 00000000

Oops:  0000
CPU:    0
EIP:    0010:[<c0114c6c>]

Using defaults from ksymoops -t elf32-i386 -a i386
EFLAGS: 00010246
eax: c4bb1544 ebx: c4e72000 ecx: c5fe83a0 edx: 00000000
esi: 00000006 edi: 00000000 ebp: c5fe83a0 esp: c4e73fb8
ds: 0018 es: 0018 ss: 0018
Process pppd (pid: 1099, stackpage=c4e73000)

Stack: 00000100 c4e6delc c4e6de4c c5b27e20 c5fe83a0 c11c35c0 c11c35c0
       c11c35c0 c11c35c0 c0114f60 c0225f60 c4e6dedc c4e6dec8 c0108914
       c4e6de4c 00000078 c4e6dedc
Call Trace: [<c0114f60>] [<c0108914>]
Code: 8b 4f 08 39 ca 7d 22 8b 47 14 83 3c 90 00 74 14 89 f0 89 d3

Trace; c0114f60 <exec_helper+14/18>
Trace; c0108914 <kernel_thread+28/38>
Code; c0114c6c <exec_usermodehelper+29c/368>
00000000 <_EIP>:
Code; c0114c6c <exec_usermodehelper+29c/368> <=====
0: 8b 4f 08 mov 0x8(%edi),%ecx <=====
Code; c0114c6f <exec_usermodehelper+29f/368>
3: 39 ca cmp %ecx,%edx
Code; c0114c71 <exec_usermodehelper+2a1/368>
5: 7d 22 jge 29 <_EIP+0x29> c0114c95
<exec_usermodehelper+2c5/368>
Code; c0114c73 <exec_usermodehelper+2a3/368>
7: 8b 47 14 mov 0x14(%edi),%eax
Code; c0114c76 <exec_usermodehelper+2a6/368>
a: 83 3c 90 00 cmpl $0x0,(%eax,%edx,4)
Code; c0114c7a <exec_usermodehelper+2aa/368>
e: 74 14 je 24 <_EIP+0x24> c0114c90
<exec_usermodehelper+2c0/368>
Code; c0114c7c <exec_usermodehelper+2ac/368>
10: 89 f0 mov %esi,%eax
Code; c0114c7e <exec_usermodehelper+2ae/368>
12: 89 d3 mov %edx,%ebx

3 warnings issued. Results may not be reliable.

Any ideas ?

Mark

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/

When pppd downs the ppp0 device, unregister_netdevice() is trying to run /sbin/hotplug in a new kernel thread. That thread's `files' structure is copied from pppd, but it is NULL. Presumably pppd's files pointer was also NULL.

Try this:

@@ -99,8 +99,10 @@
 	flush_signal_handlers(current);
 	spin_unlock_irq(&current->sigmask_lock);
 
-	for (i = 0; i < current->files->max_fds; i++ ) {
-		if (current->files->fd[i]) close(i);
+	if (current->files) {
+		for (i = 0; i < current->files->max_fds; i++ ) {
+			if (current->files->fd[i]) close(i);
+		}
 	}
 
 	/* Drop the "current user" thing */

Not my area, but I don't think exec_usermodehelper() should assume that current->files is always valid.

Al, is this correct? If so, does daemonize() also need this test? If not, then how did this thread get (current->files == NULL)? -
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/

exit_files() will leave it NULL yes. You may want to borrow (inherit ;)) the files from init_task

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/

I don't think so. I cannot reproduce this oops on 2.4.0-test11 with modutils 2.3.21, ppp 2.4.0. modutils 2.3.19 should be compatible with kernel 2.4.0-test11, the incompatibility is between modutils <= 2.3.15 and kernel 2.4.0-test11.

It was difficult to find the right area of code, my compile with egcs-2.91.66 and -march=i686 gives quite different Assembler, so take this analysis with a pinch of salt. Is there any chance that you are using the wrong compiler and/or compiler options?

The oops looks to be on line 102 of kmod.c

	for (i = 0; i < current->files->max_fds; i++ ) {

current->files is NULL. That has nothing to do with modutils, rather it points to an invalid or incomplete current context.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/

I get the same oops w/ 2.4.0-test11pre7, pppd-2.3.11 and pppd-2.4.0, not quite sure if it happens on _every_ pppd termination, but it does happen a lot. Modutils-2.3.2[01]. ksymoops below, notice that it is preceded by a waitpid failed, don't wether it is important or not... My kernel is compiled on a fairly stock RH-7.0 with kgcc.

system:

-- Versions installed: (if some fields are empty or look -- unusual then possibly you have very old versions) Linux fsck.ikr.org 2.4.0-test11 #2 Sun Nov 19 00:16:38 EST 2000 i586 unknown

Kernel modules         2.3.20
Gnu C                  egcs-2.91.66
Gnu Make               3.79.1
Binutils               2.10.0.18
Linux C Library        2.1.94
Dynamic linker         ldd (GNU libc) 2.1.94
Procps                 2.0.7
Mount                  2.10m
Net-tools              1.56
Console-tools          0.3.3
Sh-utils               2.0
Modules Loaded         ppp_deflate bsd_comp ppp_async ppp_generic slhc

ide-cd cdrom soundcore de4x5 raid1 BusLogic

Nov 23 16:05:13 fsck kernel: waitpid(1808) failed, -512
Nov 23 16:05:13 fsck kernel: Unable to handle kernel NULL pointer
dereference at
 virtual address 00000008
Nov 23 16:05:13 fsck kernel:  printing eip:
Nov 23 16:05:13 fsck kernel: c011556e
Nov 23 16:05:13 fsck kernel: *pde = 00000000
Nov 23 16:05:13 fsck kernel: Oops: 0000
Nov 23 16:05:13 fsck kernel: CPU:    0
Nov 23 16:05:13 fsck kernel: EIP:    0010:[exec_usermodehelper+734/944]
Nov 23 16:05:13 fsck kernel: EFLAGS: 00010246
Nov 23 16:05:13 fsck kernel: eax: 00000000   ebx: c1177b20   ecx: c1fb4fb0
edx: c1fb4fa4
Nov 23 16:05:13 fsck kernel: esi: c2694000   edi: c2694000   ebp: 00000000
esp: c2695fb0
Nov 23 16:05:13 fsck kernel: ds: 0018   es: 0018   ss: 0018
Nov 23 16:05:13 fsck kernel: Process pppd (pid: 1808, stackpage=c2695000)
Nov 23 16:05:13 fsck kernel: Stack: 00000100 c34e7d90 c34e6000 c1a640c0
c3eca9a0 c36fc160 c1177b20 c3eca9a0 
Nov 23 16:05:13 fsck kernel:        c3eca9a0 c1177b20 c3eca9a0 c3eca9a0
c0115890 c0222840 c34e7e4c c34e7e38 
Nov 23 16:05:13 fsck kernel:        c0108e87 c34e7dbc c34e7dbc c34e7e4c 
Nov 23 16:05:13 fsck kernel: Call Trace: [exec_helper+20/24]
[kernel_thread+35/48] 
Nov 23 16:05:13 fsck kernel: Code: 3b 68 08 7d 2a 8b 40 14 83 3c a8 00 74

15 b8 06 00 00 00 89

ksymoops:

ksymoops 2.3.5 on i586 2.2.16-22. Options used
-v /usr/src/linux/vmlinux (specified)
-K (specified)
-L (specified)
-o /lib/modules/2.4.0-test11/ (specified)
-m /usr/src/linux/System.map (specified)

No modules in ksyms, skipping objects

Nov 23 16:05:13 fsck kernel: Unable to handle kernel NULL pointer
dereference at virtual address 00000008
Nov 23 16:05:13 fsck kernel: c011556e
Nov 23 16:05:13 fsck kernel: *pde = 00000000
Nov 23 16:05:13 fsck kernel: Oops: 0000
Nov 23 16:05:13 fsck kernel: CPU:    0
Nov 23 16:05:13 fsck kernel: EIP:    0010:[exec_usermodehelper+734/944]
Nov 23 16:05:13 fsck kernel: EFLAGS: 00010246
Nov 23 16:05:13 fsck kernel: eax: 00000000   ebx: c1177b20   ecx: c1fb4fb0
edx: c1fb4fa4
Nov 23 16:05:13 fsck kernel: esi: c2694000   edi: c2694000   ebp: 00000000
esp: c2695fb0
Nov 23 16:05:13 fsck kernel: ds: 0018   es: 0018   ss: 0018
Nov 23 16:05:13 fsck kernel: Process pppd (pid: 1808, stackpage=c2695000)
Nov 23 16:05:13 fsck kernel: Stack: 00000100 c34e7d90 c34e6000 c1a640c0
c3eca9a0 c36fc160 c1177b20 c3eca9a0 
Nov 23 16:05:13 fsck kernel:        c3eca9a0 c1177b20 c3eca9a0 c3eca9a0
c0115890 c0222840 c34e7e4c c34e7e38 
Nov 23 16:05:13 fsck kernel:        c0108e87 c34e7dbc c34e7dbc c34e7e4c 
Nov 23 16:05:13 fsck kernel: Call Trace: [exec_helper+20/24]
[kernel_thread+35/48] 
Nov 23 16:05:13 fsck kernel: Code: 3b 68 08 7d 2a 8b 40 14 83 3c a8 00 74

15 b8 06 00 00 00 89
Using defaults from ksymoops -t elf32-i386 -a i386

Code; 00000000 Before first symbol
00000000 <_EIP>:
Code; 00000000 Before first symbol
0: 3b 68 08 cmp 0x8(%eax),%ebp
Code; 00000003 Before first symbol
3: 7d 2a jge 2f <_EIP+0x2f> 0000002f Before
first symbol
Code; 00000005 Before first symbol
5: 8b 40 14 mov 0x14(%eax),%eax
Code; 00000008 Before first symbol
8: 83 3c a8 00 cmpl $0x0,(%eax,%ebp,4)
Code; 0000000c Before first symbol
c: 74 15 je 23 <_EIP+0x23> 00000023 Before
first symbol
Code; 0000000e Before first symbol
e: b8 06 00 00 00 mov $0x6,%eax
Code; 00000013 Before first symbol
13: 89 00 mov %eax,(%eax)

-Marc

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/

Nope that didn't help. I'm using gcc 2.95.2, didn't think of it before since it has never caused me any problems. I'll get around to trying 2.91.66 at some point.

A quick aside, anyone know of a problem with the list, I seem to be being cut off at random intervals :)

Mark

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/