Forum for usage questions, reporting bugs, posting patches, and requesting features for the crash utility

Michael Holzheu | 15 Aug 2011 17:13

Picon

[PATCH] s390: Add restart stack support

Hi Dave,

With Linux 3.0.1 on s390 we have a new kernel stack for the PSW restart
interrupt. See kernel git commit 7dd6b3343fdc190712d1620ee8848d25c4c77c33
for details. With this patch the crash "bt" command will find this stack.

Signed-off-by: Michael Holzheu <holzheu <at> linux.vnet.ibm.com>
---
 s390x.c |   15 ++++++++++++---
 1 file changed, 12 insertions(+), 3 deletions(-)

--- a/s390x.c
+++ b/s390x.c
 <at>  <at>  -825,9 +825,16  <at>  <at>  static void get_int_stack(char *stack_na
 	unsigned long stack_addr;

 	*start = *end = 0;
-	if (!MEMBER_EXISTS("_lowcore", stack_name))
-		return;
-	stack_addr = ULONG(lc + MEMBER_OFFSET("_lowcore", stack_name));
+	if (strcmp(stack_name, "restart_stack") == 0) {
+		stack_addr = symbol_value("restart_stack");
+		if (!stack_addr)
+			return;
+		stack_addr = readmem_ul(stack_addr);
+	} else {
+		if (!MEMBER_EXISTS("_lowcore", stack_name))
+			return;
+		stack_addr = ULONG(lc + MEMBER_OFFSET("_lowcore", stack_name));
+	}

(Continue reading)

Michael Holzheu | 15 Aug 2011 17:16

Picon

[PATCH] s390: Print pt_regs for interrupt stack frames

Hi Dave,

With this patch instead of just printing the string " - Interrupt -",
we print the pt_regs for interrupts. The pt_regs contain all the current
registers and PSW of the interrupted CPU. This is very useful information
for debugging.

Old output:
-----------
 #4 [00017ea0] do_restart at 111c18
 #5 [00017eb8] psw_restart_int_handler at 5c1194
 - Interrupt -
 #6 [00017f58] vtime_stop_cpu at 5c043e
 #0 [00907eb0] cpu_idle at 106026
 #1 [00907ef8] start_kernel at 9799e8

New output:
-----------
 #4 [00017ea0] do_restart at 111c18
 #5 [00017eb8] psw_restart_int_handler at 5c1194
 PSW:  0706000180000000 00000000005c043e (vtime_stop_cpu+134)
 GPRS: 0000000000000000 00000000005c043e 00000000011e5228 00000000011e5248 
       0000000000907f08 00000000011e5b40 0000000000979324 0000000000000000 
       00000000009c0000 00000000009c0010 00000000009ab024 00000000011e5200 
       00000000011e5238 00000000005c6a00 0000000000106026 0000000000907e68 
 #0 [00907eb0] cpu_idle at 106026
 #1 [00907ef8] start_kernel at 9799e

Signed-off-by: Michael Holzheu <holzheu <at> linux.vnet.ibm.com>
---

(Continue reading)

Dave Anderson | 15 Aug 2011 18:14

Picon

Re: [PATCH] s390: Add restart stack support


----- Original Message -----
> Hi Dave,
> 
> With Linux 3.0.1 on s390 we have a new kernel stack for the PSW restart
> interrupt. See kernel git commit 7dd6b3343fdc190712d1620ee8848d25c4c77c33
> for details. With this patch the crash "bt" command will find this
> stack.
> 
> Signed-off-by: Michael Holzheu <holzheu <at> linux.vnet.ibm.com>
> ---

Queued for the next release (crash-5.1.8).

Thanks,
  Dave

> s390x.c | 15 ++++++++++++---
> 1 file changed, 12 insertions(+), 3 deletions(-)
> 
> --- a/s390x.c
> +++ b/s390x.c
>  <at>  <at>  -825,9 +825,16  <at>  <at>  static void get_int_stack(char *stack_na
> unsigned long stack_addr;
> 
> *start = *end = 0;
> - if (!MEMBER_EXISTS("_lowcore", stack_name))
> - return;
> - stack_addr = ULONG(lc + MEMBER_OFFSET("_lowcore", stack_name));
> + if (strcmp(stack_name, "restart_stack") == 0) {

(Continue reading)

Dave Anderson | 15 Aug 2011 18:15

Picon

Re: [PATCH] s390: Print pt_regs for interrupt stack frames


----- Original Message -----
> Hi Dave,
> 
> With this patch instead of just printing the string " - Interrupt -",
> we print the pt_regs for interrupts. The pt_regs contain all the current
> registers and PSW of the interrupted CPU. This is very useful information
> for debugging.
> 
> Old output:
> -----------
> #4 [00017ea0] do_restart at 111c18
> #5 [00017eb8] psw_restart_int_handler at 5c1194
> - Interrupt -
> #6 [00017f58] vtime_stop_cpu at 5c043e
> #0 [00907eb0] cpu_idle at 106026
> #1 [00907ef8] start_kernel at 9799e8
> 
> New output:
> -----------
> #4 [00017ea0] do_restart at 111c18
> #5 [00017eb8] psw_restart_int_handler at 5c1194
> PSW: 0706000180000000 00000000005c043e (vtime_stop_cpu+134)
> GPRS: 0000000000000000 00000000005c043e 00000000011e5228 00000000011e5248
> 0000000000907f08 00000000011e5b40 0000000000979324 0000000000000000
> 00000000009c0000 00000000009c0010 00000000009ab024 00000000011e5200
> 00000000011e5238 00000000005c6a00 0000000000106026 0000000000907e68
> #0 [00907eb0] cpu_idle at 106026
> #1 [00907ef8] start_kernel at 9799e
>

(Continue reading)

Michael Holzheu | 15 Aug 2011 18:28

Picon

[PATCH] s390: Add restart stack support (version 2)

Hi Dave,

Sorry, the restart stack patch is missing a check if the restart_stack
symbols exists. Without this check, we get error messages for older
kernels. Here the updated patch.

Sorry for the noise.

Signed-off-by: Michael Holzheu <holzheu <at> linux.vnet.ibm.com>
---
 s390x.c |   17 ++++++++++++++---
 1 file changed, 14 insertions(+), 3 deletions(-)

--- a/s390x.c
+++ b/s390x.c
 <at>  <at>  -825,9 +825,16  <at>  <at>  static void get_int_stack(char *stack_na
 	unsigned long stack_addr;

 	*start = *end = 0;
-	if (!MEMBER_EXISTS("_lowcore", stack_name))
-		return;
-	stack_addr = ULONG(lc + MEMBER_OFFSET("_lowcore", stack_name));
+	if (strcmp(stack_name, "restart_stack") == 0) {
+		stack_addr = symbol_value("restart_stack");
+		if (!stack_addr)
+			return;
+		stack_addr = readmem_ul(stack_addr);
+	} else {
+		if (!MEMBER_EXISTS("_lowcore", stack_name))
+			return;

(Continue reading)

Dave Anderson | 15 Aug 2011 19:28

Picon

Re: [PATCH] s390: Add restart stack support (version 2)


----- Original Message -----
> Hi Dave,
> 
> Sorry, the restart stack patch is missing a check if the restart_stack
> symbols exists. Without this check, we get error messages for older
> kernels. Here the updated patch.
> 
> Sorry for the noise.
> 
> Signed-off-by: Michael Holzheu <holzheu <at> linux.vnet.ibm.com>

OK -- I had checked it out only a live system, but that code
path wouldn't be exercised in that case, so I'll add in the 
symbol_exists() check to s390x_back_trace_cmd().

Also, the "if (!stack_addr)" check in get_int_stack() is an 
impossibility, right?

Thanks,
  Dave

> ---
> s390x.c | 17 ++++++++++++++---
> 1 file changed, 14 insertions(+), 3 deletions(-)
> 
> --- a/s390x.c
> +++ b/s390x.c
>  <at>  <at>  -825,9 +825,16  <at>  <at>  static void get_int_stack(char *stack_na
> unsigned long stack_addr;

(Continue reading)

Michael Holzheu | 16 Aug 2011 10:20

Picon

Re: [PATCH] s390: Add restart stack support (version 2)

Hello Dave,

On Mon, 2011-08-15 at 13:28 -0400, Dave Anderson wrote:
> OK -- I had checked it out only a live system, but that code
> path wouldn't be exercised in that case, so I'll add in the 
> symbol_exists() check to s390x_back_trace_cmd().

Thanks!

> Also, the "if (!stack_addr)" check in get_int_stack() is an 
> impossibility, right?

Correct, with the symbol_exists("restart_stack") check the "if (!
stack_addr)" should be always false.

Michael

Group

gmane.linux.kernel.crash-dump.crash-utility.

Project Web Page

Forum for usage questions, reporting bugs, posting patches, and requesting features for the crash utility

Search Archive

Page

1

Articles in period: 7

August 2011

Language

Change language

Options

Current view: Threads only / Showing only 20 lines / Not hiding cited text.
Change to All messages, whole messages, or hide cited text.

Post a message
NNTP Newsgroup
Classic Gmane web interface
XML

XML

RSS Feed
List Information

About Gmane

Recent entries

[Patch v4] Show module taint flags (19 May 2013)
EPPIC: fails to build in crash-7.0.0/gdb-7.6 environment (17 May 2013)
[ANNOUNCE] crash gcore command, version 1.2.1 is released (14 May 2013)
[ANNOUNCE] crash-7.0.0 is available (10 May 2013)
[PATCH v3] Show module taint flags (7 May 2013)
HEADS-UP: Linux kernel 3.9 debuginfo issues (30 Apr 2013)
How do I get IA64 register R32 and above? (23 Apr 2013)
fix bug of struct command (12 Apr 2013)
Threaded crash tool? Is it time? (12 Apr 2013)
'crash' compilation on Arch fails with -D_FORTIFY_SOURCE=2 (11 Apr 2013)
[ANNOUNCE] crash version 6.1.6 is available (4 Apr 2013)
Packaging eppic (2 Apr 2013)
[PATCH] kmem option to display pages per migrate type (29 Mar 2013)
[PATCH] fix bug of gdb involved by gdb-7.3.1.patch (29 Mar 2013)
[ANNOUNCE] crash-6.1.5 is available (28 Mar 2013)
A possible "dev -d" fix for newer kernels (27 Mar 2013)
[PATCH v2] Add -m option to kmem (27 Mar 2013)
Function parameters from stack frames (26 Mar 2013)
Throw read error on vmcore produced by ARM soc. (26 Mar 2013)
How to parse out kmem output? (24 Mar 2013)
Why module's global symbol cannot be displayed in crash? [ARM] (21 Mar 2013)

Archives

10	May 2013
32	April 2013
178	March 2013
42	February 2013
55	January 2013
35	December 2012
54	November 2012
72	October 2012
39	September 2012
106	August 2012
73	July 2012
68	June 2012
74	May 2012
86	April 2012
138	March 2012
201	February 2012
150	January 2012
61	December 2011
39	November 2011
74	October 2011
74	September 2011
7	August 2011
6	July 2011
45	June 2011
57	May 2011
34	April 2011
46	March 2011
59	February 2011
96	January 2011
16	December 2010
43	November 2010
57	October 2010
53	September 2010
61	August 2010
1	July 2010
39	June 2010
33	May 2010
30	April 2010
31	March 2010
69	February 2010
37	January 2010
23	December 2009
20	November 2009
25	October 2009
63	September 2009
32	August 2009
10	July 2009
35	June 2009
36	May 2009
61	April 2009
12	March 2009
38	February 2009
84	January 2009
28	December 2008
43	November 2008
99	October 2008
55	September 2008
59	August 2008
14	July 2008
3	June 2008
33	May 2008
40	April 2008
92	March 2008
40	February 2008
27	January 2008
39	December 2007
48	November 2007
30	October 2007
89	September 2007
31	August 2007
77	July 2007
11	June 2007
68	May 2007
28	April 2007
54	March 2007
30	February 2007
51	January 2007
65	December 2006
56	November 2006
99	October 2006
54	September 2006
62	August 2006
27	July 2006
9	June 2006

Design

Zawodny | Right Menu | Left Menu

Your Own Design

Paste the URL of your CSS below. Download CSS template